<?page no="0"?> Sprachvergleich Studien zur synchronen und diachronen Sprachwissenschaft Band 2 The book investigates multiple preverbs (PVs) in some ancient IE languages (Vedic, Homeric Greek, Old Church Slavic, and Old Irish). After an introduction, it opens with the theoretical framework and a typologically-oriented overview of PVs. It then gives quantitative data about multiple PV composites and carries out philological, formal, semantic, and syntactic analyses on them. The comparison among these languages suggests that a process of accumulation lies behind multiple PV composites. Also, PV ordering is explained by diƒerent factors: semantic solidarity between PVs and verbs PVs’ tendency to be speci†ed by event participants, PVs’ etymologies, in‡uence from other languages. The book also contributes to casting light on the reasons for PVs’ grammaticalization and lexicalization. These are two distinct reanalyses triggered by the same factor, i.e. the mentioned semantic solidarity, which makes PVs be felt as redundant. They are thus reassigned salient pieces of information as actional markers (grammaticalization) or reinterpreted as part of the verb (lexicalization). ISBN 978-3-8233-8274-4 Zanchi Multiple Preverbs Chiara Zanchi Multiple Preverbs in Ancient Indo-European Languages A comparative study on Vedic, Homeric Greek, Old Church Slavic, and Old Irish 18274_Umschlag.indd Alle Seiten 06.08.2019 11: 17: 27 <?page no="1"?> Multiple Preverbs in Ancient Indo-European Languages <?page no="2"?> herausgegeben von Katrin Schmitz (Wuppertal) Joachim Theisen (Athen) Carlotta Viti (Zürich) wissenschaftlicher Beirat Daniel Petit (Paris) Georges-Jean Pinault (Paris) Sabine Ziegler (Berlin) Sprachvergleich Studien zur synchronen und diachronen Sprachwissenschaft Band 2 <?page no="3"?> Chiara Zanchi Multiple Preverbs in Ancient Indo- European Languages A comparative study on Vedic, Homeric Greek, Old Church Slavic, and Old Irish herausgegeben von Katrin Schmitz (Wuppertal) Joachim Theisen (Athen) Carlotta Viti (Zürich) wissenschaftlicher Beirat Daniel Petit (Paris) Georges-Jean Pinault (Paris) Sabine Ziegler (Berlin) Sprachvergleich Studien zur synchronen und diachronen Sprachwissenschaft Band 2 <?page no="4"?> Bibliografische Information der Deutschen Nationalbibliothek Die Deutsche Nationalbibliothek verzeichnet diese Publikation in der Deutschen Nationalbibliografie; detaillierte bibliografische Daten sind im Internet über http: / / dnb. dnb.de abrufbar. Gedruckt mit Unterstützung des Förderungsfonds Wissenschaft der VG WORT . This book was published thanks to the support of Associazione Alunni - Fondazione Ghislieri and of Verwertungsgesellschaft WORT . © 2019 · Narr Francke Attempto Verlag GmbH + Co. KG Dischingerweg 5 · D-72070 Tübingen Das Werk einschließlich aller seiner Teile ist urheberrechtlich geschützt. Jede Verwertung außerhalb der engen Grenzen des Urheberrechtsgesetzes ist ohne Zustimmung des Verlages unzulässig und strafbar. Das gilt insbesondere für Vervielfältigungen, Übersetzungen, Mikroverfilmungen und die Einspeicherung und Verarbeitung in elektronischen Systemen. Internet: www.narr.de eMail: info@narr.de Satz: pagina GmbH, Tübingen CPI books GmbH, Leck ISSN 2569-2275 ISBN 978-3-8233-8274-4 (Print) ISBN 978-3-8233-9274-3 (ePDF) ISBN 978-3-8233-0125-7 (ePub) www.fsc.org MIX Papier aus verantwortungsvollen Quellen FSC ® C083411 ® <?page no="5"?> To the memory of my parents <?page no="7"?> Tale of contents 7 Tale of contents Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 List of abbreviations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17 Glosses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17 Languages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20 Authors, works, and manuscripts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20 Grammars and dictionaries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21 Symbols . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23 1.1. Motivations behind the present study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23 1.2. Aims and structure of the study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26 1.3. Corpora of the present study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29 1.3.1. The R̥g-Veda . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29 1.3.2. The Homeric poems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30 1.3.3. The Codices Zographensis , Marianus , and Suprasliensis . . . . . . . . 31 1.3.4. The Milan and the Priscian Glosses . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32 1.4. Brief methodological remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33 2. Theoretical background: Analytical tools for the study of preverbs . . . . . . . . . . . 37 2.1. Cognitive Grammar . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37 2.1.1. Space: the basic domain of human cognition . . . . . . . . . . . . . . . . . . 37 2.1.2. Going from spatial to abstract domains: metaphor and metonymy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38 2.1.3. The conceptualization of the spatial event . . . . . . . . . . . . . . . . . . . . 40 2.2. Grammaticalization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44 2.2.1. A brief history of grammaticalization studies . . . . . . . . . . . . . . . . . 44 2.2.2. Current approaches to grammaticalization . . . . . . . . . . . . . . . . . . . . 45 2.2.3. The continuum of grammaticalization . . . . . . . . . . . . . . . . . . . . . . . . 46 2.2.4. Grammaticalization: an abused theoretical concept? . . . . . . . . . . 51 2.2.5. Grammaticalization and lexicalization . . . . . . . . . . . . . . . . . . . . . . . . 54 2.3. Describing the meanings and the functions of preverbs . . . . . . . . . . . . . 55 2.3.1. Semantic Roles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55 2.3.2. Aspect, actionality, and transitivity . . . . . . . . . . . . . . . . . . . . . . . . . . . 59 <?page no="8"?> 8 Tale of contents 3. Preverbs: an overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65 3.1. Preverbs in Indo-European . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65 3.1.1. Preverbs: definition and functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65 3.1.2. The positional properties of preverbs . . . . . . . . . . . . . . . . . . . . . . . . . 74 3.1.3. The origin of preverbs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80 3.1.4. Preverbs as a terminological and a typological problem . . . . . . . 82 3.2. Multiple preverbs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87 3.3. Preverbs outside Indo-European . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89 3.3.1. Preverbs in Finno-Ugric and Caucasian languages . . . . . . . . . . . . 89 3.3.2. Preverbs in Amerindian languages . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91 3.3.3. Preverbs in Northern Australian languages . . . . . . . . . . . . . . . . . . . 92 4. Multiple preverbs in Vedic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95 4.1. Preverbs in Vedic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95 4.1.1. State of the art . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95 4.1.2. The categorial status of Vedic preverbs . . . . . . . . . . . . . . . . . . . . . . . 97 4.2. Multiple preverbs in numbers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117 4.2.1. Composites with multiple preverbs . . . . . . . . . . . . . . . . . . . . . . . . . . 117 4.2.2. Verbal roots modified by multiple preverbs . . . . . . . . . . . . . . . . . . 124 4.2.3. Attested combinations of preverbs . . . . . . . . . . . . . . . . . . . . . . . . . . . 129 4.3. The form of composites . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134 4.3.1. Sandhi phenomena . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134 4.3.2. The position of preverbs with respect to inflectional affixes . . 136 4.4. The semantics of multiple preverbs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 138 4.4.1. Preverbs with spatial, abstract and actional meanings . . . . . . . . 138 4.4.2. Same (combination of) preverbs, different meanings . . . . . . . . . 143 4.4.3. Different degrees of compositionality . . . . . . . . . . . . . . . . . . . . . . . . 145 4.4.4. Summarizing the meanings of preverbs in multiple preverb combinations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153 4.5. The syntax of multiple preverbs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 156 4.5.1. Movable preverbs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 156 4.5.2. Optional preverbs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 158 4.5.3. Vedic preverbs as transitivizing morphemes . . . . . . . . . . . . . . . . . 159 4.5.4. Composites taking no second argument . . . . . . . . . . . . . . . . . . . . . 162 4.6. Preverb ordering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 164 4.6.1. Previous accounts of preverb ordering . . . . . . . . . . . . . . . . . . . . . . . 166 4.6.2. An integrated account of preverb ordering . . . . . . . . . . . . . . . . . . . 168 5. Multiple preverbs in Homeric Greek . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171 5.1. Preverbs in Homeric Greek . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171 <?page no="9"?> Tale of contents 9 5.1.1. State of the art . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171 5.1.2. The unclear categorial status of preverbs in Homeric Greek . 172 5.2. Multiple preverbs in numbers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 182 5.2.1. Composites with multiple preverbs . . . . . . . . . . . . . . . . . . . . . . . . . . 182 5.2.2. Verbal roots modified by multiple preverbs . . . . . . . . . . . . . . . . . . 191 5.2.3. Attested combinations of preverbs . . . . . . . . . . . . . . . . . . . . . . . . . . . 196 5.3. The form of composites . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 199 5.3.1. Philological and metrical analyses of composites with multiple preverbs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 199 5.3.2. Sandhi phenomena . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 201 5.3.3. The position of preverbs with respect to inflectional affixes . . 201 5.4. The semantics of multiple preverbs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 203 5.4.1. Preverbs with spatial, abstract and actional meanings . . . . . . . . 203 5.4.2. Same (combinations of) preverbs, different meanings . . . . . . . . 206 5.4.3. Different degrees of compositionality . . . . . . . . . . . . . . . . . . . . . . . . 210 5.4.4. Summarizing the meanings of preverbs in multiple preverb combinations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 216 5.5. The syntactic status of multiple preverbs . . . . . . . . . . . . . . . . . . . . . . . . . . 217 5.5.1. Movable preverbs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 218 5.5.2. Optional preverbs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 220 5.5.3. Non ambiguous constructions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 220 5.5.4. Multiple preverbs as transitivizing morphemes . . . . . . . . . . . . . . 222 5.5.5. Composites taking no second argument . . . . . . . . . . . . . . . . . . . . . 224 5.6. Preverb ordering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 225 5.6.1. Previous accounts of preverb ordering . . . . . . . . . . . . . . . . . . . . . . . 226 5.6.2. Issues with previous approaches . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 227 5.6.3. An integrated account of preverb ordering . . . . . . . . . . . . . . . . . . . 229 6. Multiple preverbs in Old Church Slavic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 231 6.1. Preverbs in Old Church Slavic . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 231 6.1.1. State of the art: the Slavic prefixes . . . . . . . . . . . . . . . . . . . . . . . . . . . 231 6.1.2. The status of preverbs in Old Church Slavic . . . . . . . . . . . . . . . . . 234 6.1.3. The ongoing development of Slavic prepositions . . . . . . . . . . . . . 241 6.2. Multiple preverbs in numbers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 245 6.2.1. Composites with multiple preverbs . . . . . . . . . . . . . . . . . . . . . . . . . . 245 6.2.2. Verbal roots modified by multiple preverbs . . . . . . . . . . . . . . . . . . 250 6.2.3. Attested combinations of preverbs . . . . . . . . . . . . . . . . . . . . . . . . . . . 252 6.2.4. The Greek counterparts of Old Church Slavic composites . . . . 253 6.3. The form of composites . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 256 6.3.1. The actional suffixes of multiple preverb verbs . . . . . . . . . . . . . . . 256 <?page no="10"?> 10 Tale of contents 6.3.2. The alternations involving the suffix -(j)a- . . . . . . . . . . . . . . . . . . . 258 6.3.3. Triplets containing speech verbs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 259 6.3.4. The perfectivizing value of v ъ z- . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 261 6.4. The semantics of multiple preverbs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 261 6.4.1. Preverbs with spatial, abstract and actional meanings . . . . . . . . 261 6.4.2. Same preverbs, different meanings . . . . . . . . . . . . . . . . . . . . . . . . . . . 263 6.4.3. Different degrees of compositionality . . . . . . . . . . . . . . . . . . . . . . . . 271 6.4.4. Summarizing the meanings of preverbs in multiple preverb combinations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 276 6.5. The syntax of multiple preverb composites . . . . . . . . . . . . . . . . . . . . . . . . 277 6.5.1. The alternative constructions to multiple preverbs: scanty relics of a preceding stage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 278 6.5.2. Preverb repetition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 280 6.5.3. Preverbs as modifiers of verb argument structure . . . . . . . . . . . . 281 6.6. Preverb ordering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 282 6.6.1. Preverb ordering: the account of Modern Slavic . . . . . . . . . . . . . . 283 6.6.2. An integrated account of preverb ordering . . . . . . . . . . . . . . . . . . . 285 7. Multiple preverbs in Old Irish . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 287 7.1. Preverbs in Old Irish . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 287 7.1.1. State of the art . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 287 7.1.2. The status and functions of Old Irish preverbs . . . . . . . . . . . . . . . 290 7.2. Multiple preverbs in numbers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 301 7.2.1. Composites with multiple preverbs . . . . . . . . . . . . . . . . . . . . . . . . . . 301 7.2.2. Verbal roots modified by multiple preverbs . . . . . . . . . . . . . . . . . . 313 7.2.3. Attested combinations of preverbs . . . . . . . . . . . . . . . . . . . . . . . . . . . 323 7.2.4. The Latin counterparts of Old Irish composites . . . . . . . . . . . . . . 330 7.3. The form of composites . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 342 7.3.1. The allomorphy of Old Irish preverbs . . . . . . . . . . . . . . . . . . . . . . . . 342 7.3.2. Augment and preverbs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 345 7.4. The semantics of multiple preverbs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 347 7.4.1. Preverbs with spatial, abstract, and actional meanings . . . . . . . 347 7.4.2. Same preverbs, different meanings . . . . . . . . . . . . . . . . . . . . . . . . . . . 352 7.4.3. Different degrees of compositionality . . . . . . . . . . . . . . . . . . . . . . . . 354 7.4.4. Summarizing the meanings of preverbs in multiple preverb combinations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 363 7.5. The syntax of multiple preverb composites . . . . . . . . . . . . . . . . . . . . . . . . 366 7.5.1. No alternative constructions to multiple preverb composites . 366 7.5.2. Preverb repetition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 369 7.5.3. Preverbs as modifiers of verb argument structure . . . . . . . . . . . . 370 <?page no="11"?> Tale of contents 11 7.6. Preverb ordering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 372 7.6.1. McCone’s hierarchy of preverb ordering . . . . . . . . . . . . . . . . . . . . . 373 7.6.2. Preverb ordering in the Milan and Priscian Glosses: beyond McCone’s generalizations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 375 7.6.3. An integrated account of preverb ordering . . . . . . . . . . . . . . . . . . . 381 8. Multiple preverbs in ancient Indo-European languages: differences, similarities and concluding remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 385 8.1. Multiple preverbs: differences among Vedic, Homeric Greek, Old Church Slavic, and Old Irish . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 385 8.1.1. Multiple preverbs in numbers: a comparison . . . . . . . . . . . . . . . . . 385 8.1.2. Different degrees of univerbation, lexicalization, and grammaticalization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 387 8.2. Multiple preverbs: similarities among Vedic, Homeric Greek, Old Church Slavic, and Old Irish . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 390 8.2.1. Preverb ordering: the common reasons behind it . . . . . . . . . . . . . 390 8.2.2. Common process of formation of multiple preverb composites . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 394 8.2.3. Grammaticalization and lexicalization: the common reason behind two distinct developments . . . . . . . . . . . . . . . . . . . . . . . . . . . 395 8.2.4. Common semantic developments . . . . . . . . . . . . . . . . . . . . . . . . . . . . 397 8.3. Brief concluding remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 399 Web Resources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 401 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 403 <?page no="12"?> 12 List of Figures List of Figures Fig. 1: The profile-base asymmetry: wheel vs. spoke , rim , wheel (from Langacker 2008: 67) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40 Fig. 2: Kinship relations: profile-base and TR - LM asymmetries (Langacker 2008: 68) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42 Fig. 3: Vedic token data: ordering (from Papke 2010: 101) . . . . . . . . . . . . . . . . . . 166 Fig. 4: The link between Path and distributive meanings of po- . . . . . . . . . . . . 265 Fig. 5: The link between the spatial and ingressive meanings of v ъ z- . . . . . . 270 List of Tables Tab. 1: Language sample, texts, and dating . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34 Tab. 2: Vendler’s (1957) actional classes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62 Tab. 3: Hopper & Thompson’s parameters of transitivity (adapted from Hopper & Thompson 1980: 252) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64 Tab. 4: Terms employed for the elements of Northern Australian complex verbs (adapted from Schultze-Berndt 2003: 146) . . . . . . . . . . . . . . . . . . . . . . 85 Tab. 5: The positioning of Vedic multiple preverbs ((i) relative proximity to the verb) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100 Tab. 6: The positioning of Vedic multiple preverbs ((ii) relative positioning with respect to the verb) (adapted from Papke 2010: 84-89) . . . . . . . . . 100 Tab. 7: Possible orientations and positions of Vedic preverbs . . . . . . . . . . . . . . 108 Tab. 8: The positions and orientations of Vedic preverbs . . . . . . . . . . . . . . . . . . . 110 Tab. 9: Vedic composites with multiple preverbs and their frequency . . . . . . 121 Tab. 10: Vedic composites attested after the R̥g-Veda and their meanings . . 124 Tab. 11: Vedic verbal roots modified by multiple preverbs . . . . . . . . . . . . . . . . . 128 Tab. 12: Vedic combinations of preverbs and their frequencies . . . . . . . . . . . . . 131 Tab. 13: Examples of external sandhi in Vedic composites . . . . . . . . . . . . . . . . . . 134 Tab. 14: Examples of internal sandhi in Vedic composites . . . . . . . . . . . . . . . . . . 135 Tab. 15: Vedic partially compositional composites . . . . . . . . . . . . . . . . . . . . . . . . . 148 Tab. 16: Vedic fully compositional composites . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151 Tab. 17: Vedic non-compositional composites . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152 <?page no="13"?> List of Tables 13 Tab. 18: The meanings of Vedic preverbs in multiple preverb combinations. 156 Tab. 19: The positioning of Vedic preverbs and their frequencies . . . . . . . . . . 165 Tab. 20: Homeric composites with multiple preverbs and their frequency . . 186 Tab. 21: Composites attested in post-Homeric prose and their meanings . . . 187 Tab. 22: Homeric verbs and verb roots modified by multiple preverbs . . . . . 194 Tab. 23: Homeric combinations of preverbs and their frequencies . . . . . . . . . 197 Tab. 24: Types of split brought about by assumed metrical pauses . . . . . . . . . 200 Tab. 25: Homeric fully compositional composites . . . . . . . . . . . . . . . . . . . . . . . . . . 211 Tab. 26: Homeric non-compositional composites . . . . . . . . . . . . . . . . . . . . . . . . . . 211 Tab. 27: Homeric partially compositional composites . . . . . . . . . . . . . . . . . . . . . . 213 Tab. 28: The meanings of Homeric multiple preverbs . . . . . . . . . . . . . . . . . . . . . . 217 Tab. 29: The positioning of Homeric Greek preverbs and their frequencies 225 Tab. 30: Imbert’s synchronic semantic constraints on preverb ordering . . . . 226 Tab. 31: Grammaticalization of Slavic prefixes as markers of perfectivity . . 240 Tab. 32: Old Church Slavic composites with multiple preverbs . . . . . . . . . . . . . 247 Tab. 33: Composites attested in later (South) Slavic languages and their meanings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 249 Tab. 34: Old Church Slavic verbal roots modified by multiple preverbs . . . . 251 Tab. 35: Old Church Slavic preverb combinations and their frequencies . . . 252 Tab. 36: The Greek counterparts of Old Church Slavic composites . . . . . . . . . 254 Tab. 37: The actional suffixes of multiply preverbed verbs . . . . . . . . . . . . . . . . . 257 Tab. 38: The compositionality of Old Church Slavic composites . . . . . . . . . . . 273 Tab. 39: The meanings of Old Church Slavic multiple preverbs . . . . . . . . . . . . 277 Tab. 40: The positioning of Old Church Slavic preverbs and their frequencies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 283 Tab. 41: Lexicalized IP -V sequences and their Greek counterparts . . . . . . . . . 286 Tab. 42: Old Irish preverbs and their positions ( GOI ) . . . . . . . . . . . . . . . . . . . . . . . 297 Tab. 43: Preverbs featuring positions C and D only . . . . . . . . . . . . . . . . . . . . . . . . 298 Tab. 44: Old Irish composites with multiple preverbs ( Milan and Priscian Glosses ) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 312 Tab. 45: Old Irish verbal roots modified by multiple preverbs . . . . . . . . . . . . . . 320 Tab. 46: Old Irish combinations of preverbs and their frequency . . . . . . . . . . . 327 Tab. 47: The Latin counterparts of Old Irish composites . . . . . . . . . . . . . . . . . . . 339 Tab. 48: Allomorphy of Old Irish preverbs: adand aith- (from Anderson 2016: 220) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 342 Tab. 49: Allomorphy of Old Irish preverbs: inand ind- (from Anderson 2016: 218) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 343 Tab. 50: Allomorphy of Old Irish preverbs: es(s)and us(s)- (Anderson 2016: 221) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 343 <?page no="14"?> 14 List of Tables Tab. 51: Old Irish partially compositional composites . . . . . . . . . . . . . . . . . . . . . . 360 Tab. 52: Old Irish non-compositional composites . . . . . . . . . . . . . . . . . . . . . . . . . . 362 Tab. 53: The meanings of Old Irish multiple preverbs . . . . . . . . . . . . . . . . . . . . . . 366 Tab. 54: The positioning of Old Irish preverbs and their frequencies . . . . . . . 373 Tab. 55: Relative ordering of Old Irish preverbs in primary composition . . . 374 Tab. 56: Old Irish preverb combinations arranged as in McCone (1997) . . . . 379 Tab. 57: Multiple preverbs in numbers: a general overview . . . . . . . . . . . . . . . . 385 Tab. 58: Semantic parameters: a general overview . . . . . . . . . . . . . . . . . . . . . . . . . 387 Tab. 59: Syntactic parameters: a general overview . . . . . . . . . . . . . . . . . . . . . . . . . 388 <?page no="15"?> Acknowledgements 15 Acknowledgements This book is a revised version of my PhD thesis, which was written between 2014 and 2017, when I was a PhD student of the Doctoral School in Linguistic Sciencies at University of Pavia and University of Bergamo. Thus, first and foremost, my deepest thanks go to my PhD supervisor, Silvia Luraghi: what I owe her goes far beyond the scope of this work. I also sincerely thank my co-advisors. For Elisa Roma, it would be an understatement to define her contribution to the chapter on Old Irish as crucial. My gratitude also goes to Pierluigi Cuzzolin for his insightful comments, especially relating to the chapter on Old Church Slavic. Among my co-advisors, Brian D. Joseph deserves special mention: he warmly welcomed me to Ohio State University during the spring term 2017, where I spent four of the most scientifically inspiring and pleasant months of my academic life. This work benefited enormously from my one-year visit at Friedrich Schiller Universität Jena, for which my most sincere gratitude goes to Martin J. Kümmel. Among people in Jena, I also wish to thank Sergio Neri, Cassandra Freiberg, and Annemarie Verkerk. Cormac Anderson from the MPI in Jena merits special mention, as he was so kind as to provide me with the first group of Old Irish data, which greatly facilitated my continuing collection. Similarly, Hanne M. Eckhoff deserves special thanks for extracting from the TOROT Treebank the Old Church Slavic verbs of this work in 2015. During my stay in Jena, I had the honor of being invited to a special workshop on preverbs held at University of Mainz in May 2016. I thank Björn Wiemer for inviting me to this workshop, where I met Peter Arkadiev, Andrej Malchukov, and Kirill Kozhanov. This work has benefited from the discussions held with each of them. Among the several people with whom I had conversations relating to preverbs, some call for an explicit mention. Since my Erasmus in Salzburg dating back to 2013, Thomas Krisch has been showing his interest in my work on multiple preverbs, for which I am most grateful. He was one of the external evaluators of my PhD thesis as well: his insightful comments immensely enhanced the final version of this book. Besides being one of the editors of the present book series, Carlotta Viti was also one of the external evaluators of my PhD thesis: her enthusiasm for my work very much contributed in making this book possible. I take this opportunity to thank the co-editors of the book series, Katrin Schmitz and Joachim Theisen, as well as the members of its scientific committee, Daniel Petit, Georges-Jean Pinault, and Sabine Ziegler. In addition, I wish <?page no="16"?> 16 Acknowledgements to warmly thank Alexander Lubotsky: during my PhD defence, he provided me with constructive criticism, which especially contributed to improving the chapters relating to Vedic and to Old Church Slavic. Leonid Kulikov also gave me thoughtful suggestions on the chapter relating to Vedic. Craig Melchert, whom I met in Carrboro ( NC ) in March 2017, answered my many rapid questions on Indo-European preverbs. Pier Marco Bertinetto also took the time to explain to me how applicatives work in Ayoreo in a very long and extremely interesting Skype call. This is also the right forum in which to deliver my deepest thanks to all scholars, colleagues, and students who privately shared their own monographs, papers or theses with me: thanks to Luisa Ruvoletto, Cinzia Citraro, Marco Budassi, Costanza Conforti, Luisa Borchio, and Carlo Dalle Ceste. This book would never have been concluded without the support of my colleagues. I express my gratitude to Guglielmo Inglese, one of the most brilliant and amazingly sarcastic linguists I know and value. I thank Erica Pinelli for our neverending discussions on metaphors and metonymies and for the reciprocal support we wholeheartedly offer each other. Last but not least, I thank Chiara Naccarato: during our PhD school, I was lucky enough to gain her as an excellent colleague, as an ever-present next-door neighbor, and as a faithful companion in Romanian summer schools as well as in Russian international conferences. But, above all, my greatest fortune was to earn her precious friendship. <?page no="17"?> List of abbreviations 17 List of abbreviations Glosses abl ablative abs absolutive acc accusative adv adverb adp adposition all allative ana anaphoric aor aorist app applicative arg argument art article aug augment cl clitic conj conjunction d definite conjugation dat dative deict deictic dem demonstrative distr distributive du dual e enclitic em emphatic ep exterior preverb ex existential excl exclusive <?page no="18"?> 18 List of abbreviations f feminine fut future gen genitive ger gerund i indefinite conjugation imp imperative impf imperfect indf indefinite inj injunctive ins instrumental int interrogative intens intensive ip interior preverb ipfv imperfective loc locative mid middle mp medial preverb m / p medio-passive N noun n neuter neg negation nom nominative np noun phrase o object opt optative p preverb pass passive pfv perfective pl plural poss possessive <?page no="19"?> List of abbreviations 19 pot potential PP Prepositional Phrase p rep p Prepositional Preverb pret preterit p rev p Preverbal Preverb prot prototonic pst past ptc particle ptcp participle P2 second position rec reciprocal red reduplication refl reflexive rel relative s subject sbjv subjunctive sfx suffix sg singular sr semantic role sup supine V verb voc vocative VP verbal phrase 1 first person 2 second person 3 third person In glosses, the nominal number is specified only if it is plural or dual (singular is not indicated), gender is not indicated unless it is feminine or neuter. Among verbal categories, indicative mood and active voice are likewise not indicated. <?page no="20"?> 20 List of abbreviations Languages AG Ancient Greek BCS Bosnian-Croatian-Serbian Cypr. Cypriot Greek CS Common Slavic Germ. German Hitt. Hittite Hom.Gr. Homeric Greek IE Indo-European It. Italian Lat. Latin Lith. Lithuanian Lyd. Lydian MW Middle Welsh OCS Old Church Slavic OI r. Old Irish OL at. Old Latin OR Old Russian PIE Proto-Indo-European Ved. Vedic Vulg.Lat. Vulgar Latin Authors, works, and manuscripts AitBr . Aitareya-Brāhmana Arist. Aristotle Bes. Gregorii Magni papae Homiliae in Evangelia (Besědy na evangelije papy Grigorija Velikago) D. Demosthenes Euch. Euchlogium Sinaiticum <?page no="21"?> List of abbreviations 21 Fest. Sextus Pompeius Festus Har. Haravijaya Hdt. Herodotus Il . Iliad Jn John’s Gospel KUB Keilschrifturkunden aus Boghazköi Lk Luke’s Gospel Mar. Codex Marianus MB h. Mahābhārata Metaph. Metaphysics Mk Mark’s Gospel Ml. Milan Glosses Mt Matthew’s Gospel Od. Odyssey Plb. Polybius R̥V R̥g-Veda Sg. Priscian Glosses Supr. Codex Suprasliensis Wb. Würzburg Glosses Usp. Sbor. Uspenskij sbornik Xen. Xenophon Zogr. Codex Zographensis Grammars and dictionaries ACC Stokes 1899-1900 CGH O’Brien 1962 DELG Chantraine 1968 e DIL Electronic Dictionary of the Irish Language EWA ia Mayrhofer 1986-2001 <?page no="22"?> 22 List of abbreviations GOI Thurneysen 1946 IEW Pokorny 1959 KPV Schumacher 2004 LIPP I Dunkel 2014a LIPP II Dunkel 2014b LIV 2 Rix et al. 2001 LSJ The Online Liddell-Scott-Jones Greek-English Lexicon REW Vasmer 1953-1958 RIVELEX Krisch 2006 RIVELEX II Krisch 2012 TOROT Eckhoff & Berdicevskis 2015 VKG II Pedersen 1913 Symbols → become (synchronic rule) = clitic boundary + combination of elements […] constituent boundaries | context of a phonological rule / context of a sound change ~ correspond to > develop into (diachronic change) morpheme boundary _ position in a rule / sound change √ root # starting-/ endpoint of a sequence * unattested form <?page no="23"?> 1.1. Motivations behind the present study 23 1. Introduction πάντων γὰρ ὅσα πλείω μέρη ἔχει καὶ μή ἐστιν οἷον σωρὸς τὸ πᾶν ἀλλ᾿ ἔστι τι τὸ ὅλον παρὰ τὰ μόρια, ἔστι τι αἴτιον (Arist. Metaph. 8.1045a 9-10) ‘In all things which have a plurality of parts, and which are not a total aggregate but a whole of some sort distinct from the parts, there is some cause’ 1.1. Motivations behind the present study This work investigates multiple preverbs modifying verbs in some ancient Indo-European languages, specifically in Vedic, Homeric Greek, Old Church Slavic, and Old Irish. The construction under research is schematized in (1)a and exemplified in (1)b: (1) a. P n [exterior] (…P 2 [medial]) P 1 [interior] V b. Multiple preverb composites containing roots for ‘putting, laying, throwing’ PIE *d h eh 1 -, *kerH-, *leg h - ( LIV 2 136, 353, 398) Language Verb Actual meaning Literal meaning Ved. adhí ní √d h ā- 1 ‘deposit for’ ‘over-down-put’ Hom.Gr. ep-ana-títhēmi ‘shut’ ‘on-upward-put’ OCS prěd ъ -po-lagati ‘distribute to’ ‘in-front-along-lay’ OI r. do·aithchuiredar ‘return’ 2 ‘to-back-put’ (to-aith-√cori-) Each simplex base in (1)b is modified by more than one preverb, that is, a small uninflected morpheme with original spatial semantics and free-standing status. 1 Vedic multiple preverbs-verb combinations are written as separated items, based on the status of preverbs in this variety, which is thoroughly discussed in Chapter 4. 2 Though the Old Irish citation form morphologically represents a third person singular and is usually translated with an English third person singular in the literature on Old Irish, I here decided to consistently use the English citation form instead, namely the infinitive without to . <?page no="24"?> 24 1. Introduction The resulting formations can develop predictable or unpredictable semantics, given the concrete basic meanings of the elements that make them up. Preverbs and preverbation are two well-studied topics in Indo-European linguistics (cf. e. g. Rousseau 1995; Booij & Van Kemenade 2003; Chapter 3, and references therein), to such an extent that the notion of preverb itself emerged from these fields of study. However, far less attention has been paid to multiple preverb constructions of the type in (1)a-b, in which two or more such morphemes attach onto the same simplex verb. This gap in the literature may occur because the accumulation of preverbs, though possible, does not seem to be the favored procedure in ancient Indo-European languages (Kuryłowicz 1964: 174). In spite of this general observation, a number of scholars noticed the relatively exceptional presence of multiple preverbs in Old Irish. Thurneysen ( GOI 495) even wrote that “there is no restriction on the number of prepositions [i. e. preverbs] that may be employed in composition.” According to Kuryłowicz (1964: 174 ff.), in Old Irish, multiple preverbs are widespread as they do not constitute an ambiguous construction: the preverb farthest from the verbal stem is clearly separated from the rest of the verbal complex, as it retains a proclitic status. McCone (1997) offered an explanation of the ordering preferences of Old Irish preverbs in what he called “primary composition”, namely the inherited layer of composition, whereby more than one preverb simultaneously attached onto the same simplex verb. However, Trudy Rossiter, a student of McCone’s, in her doctoral thesis (Rossiter 2004), challenged this view: she showed that the vast majority of Old Irish verbs with multiple preverbs can be reduced by removing the outermost preverb. This fact points to a process of formation by incremental one-by-one accumulation of preverbs (the so-called “accretion” or “recomposition”), a scenario that McCone (2006) also later embraced (cf. Chapter 7). McCone’s (1997) monograph offered Papke (2010) a starting point to develop her comparison between Vedic and Old Irish preverb ordering, which she also extended to Homeric Greek. Papke concluded that, as there are strong correlations between preverb orderings, especially in Vedic and Old Irish, Vedic order must be historically motivated (cf. Chapters 4 and 7). In addition, in her view, Vedic verbs with multiple preverbs formed through a process that Rossiter and McCone would call “accretion” or “recomposition”: at first, only one preverb and a verbal base combine; afterwards, this established combination becomes available for the attaching of further preverb(s) (cf. Chapter 3). Caroline Imbert dedicated several studies to Homeric Greek multiple preverbs, to their historical sources, and to the synchronic constraints ruling preverb ordering (cf. e. g. Imbert 2008). Notably, Imbert’s works are typologically oriented: she applied to Homeric Greek the category of ‘relational preverbs’ (cf. Chapter 3), which Craig & Hale (1988) identified as being among the pre- <?page no="25"?> 1.1. Motivations behind the present study 25 verbs in Rama (a Chibchan language). Accordingly, Imbert argued that Homeric Greek multiple preverbs developed from previous postpositions, as Craig & Hale showed for Rama. Zanchi (2014) also investigated Homeric multiple preverbs and their origins, but came to different conclusions from Imbert’s: according to Zanchi, multiple preverbs are believed to have developed from original adverbs, rather than from postpositions (cf. Chapter 5). Until now, there have been no studies focusing specifically on multiple preverbs in Old Church Slavic, although both Fil’ (2011) and Zanchi & Naccarato (2016) take into account both Old Russian and Old Church Slavic data. Instead, multiple preverbs and their functions in modern Slavic languages have received much attention: for example, multiple preverbs are investigated in Czech by Filip (2003), in Bulgarian by Istratkova (2004), in Serbian by Milićević (2004), and in Russian, among others, by Babko-Malaya (1999), Filip (1999, 2003), Ramchand (2004), Romanova (2004), Svenonius (2004a, 2004b), and Tatevosov (2008, 2009). However, the system of multiple preverbs in modern Slavic turns out to be completely different from that of Old Church Slavic (cf. Chapter 6). Along with this relative dearth of studies on multiple preverbs, it is worth mentioning another crucial gap in the relevant literature. Specifically, virtually no current studies integrate scholars’ conclusions about the origin, functions, and developments of preverbs in different languages. In order to gain a precise, and at the same time more comprehensive, understanding of the common reasons behind the behavior and historical development of preverbs, it is imperative that such an investigation takes place. For example, the above-mentioned concept of “accretion” or “recomposition” was coined by Rossiter (2004) and McCone (2006) for Old Irish, and - to my knowledge - never extended beyond its original scope. A second case in point is the so-called “Vey-Schooneveld effect”, which basically accounts for the development of Slavic preverbs into aspectual markers as a reanalysis triggered by semantic redundancy. This hypothesis was developed within Slavic linguistics and virtually was never tested elsewhere (a limited exception is Latin linguistics; cf. Chapter 6, fn. 6). As a final example, Viti (2008a, 2008b) connected the development of Homeric preverbs to markers of actionality (and transitivity) with their ability to draw anaphoric reference to discourse-active (i. e. topical) participants. Although Boley (2004) and others also regarded preverbs as elements contributing to textual cohesion, and Friedrich (1987), Coleman (1991) and Cuzzolin (1995) spoke about “discourse-oriented grammaticalization” for Latin (and generally Indo-European) preverbs, similar analyses were never performed on a wider language sample. Thus, the choice to investigate relatively underrepresented phenomena such as multiple preverbs in a relatively wide sample of Indo-European languages aims to be a first contribution to fill the literature gaps outlined above. In par- <?page no="26"?> 26 1. Introduction ticular, Vedic and Homeric Greek were selected as they represent comparably early stages of development, in which preverbs retain most of their assumed original meanings, functions, and syntactic freedom (cf. Section 1.3 for the chronology of their attestation; the most ancient attested Indo-European language, Hittite, was not included in this investigation, as it represents a divergent and to some extent problematic development, on which see Chapter 3, fn. 31). By contrast, Old Church Slavic offers a glimpse into the initial steps toward one of the possible later developments of preverbs: specifically, their subsequent grammaticalization into fully-fledged aspectual markers. In parallel, Old Irish, with its flourishing usage of multiple preverbs, provides an excellent touchstone to assess another development that preverbs may undergo: specifically, their lexicalization into semantically idiosyncratic or unpredictable composite items. 1.2. Aims and structure of the study The aims of this work can be subcategorized as follows: (a) language-internal goals; (b) comparative goals; (c) wide-ranging goals. To begin with, for each language of the sample, the present investigation aims to (i) describe the full array of multiple preverb formations in terms of preverb combinations, verbal roots, and their frequencies; (ii) assess the extent to which multiple preverbs underwent lexicalization or grammaticalization; (iii) understand the morphosyntactic status of multiple preverbs; (iv) detect the meanings of preverbs in multiple preverb combinations; (v) provide insights into the formation process of verbs modified by multiple preverbs and preverb ordering. Regarding (b) goals, this work seeks to (i) compare multiple preverb formations, multiple preverb combinations, the verbal bases multiple preverb formations contain, and preverb ordering; (ii) compare the statuses of multiple preverbs in the above-mentioned languages; (iii) identify, describe, and motivate common semantic shifts. With the most general level (c) goals, the study aims to (i) provide, within a relatively limited data-sample, more detailed reasons preverbs underwent the well-known lexicalization and grammaticalization; (ii) identify the pattern of formation of multiple preverb verbs; (iii) integrate references that focus on different languages to acquire a more general view of the common processes of development and their motivations. In order to meet these goals, the present investigation takes into account a number of morphological, semantic, and syntactic parameters, which are briefly described below: <?page no="27"?> 1.2. Aims and structure of the study 27 (a) the position of preverbs with respect to that of the other pieces of preverbal morphology; the sandhi effects undergone by the elements that make up the formation; the position of the accent; when relevant, the metrical constraints that may influence the placement of preverbs and univerbation; (b) multiple preverb verbs’ degree of semantic compositionality; preverbs’ degree of polysemy in multiple preverb combinations; (c) preverbs’ potential displacement from the modified verbal base, the range, and the type of such displacement; preverbs’ obligatoriness; preverbs’ repetition outside the preverbal context; argument structure of multiple preverb verbs. The present work is organized as follows. This introduction continues with brief descriptions of the texts selected for the current investigation. I focus on philological metadata, including dating, geographical origin, author, content, and textual tradition, as far as these pertain to the linguistic amalgam that such written records transmit to us. The introduction ends with a few methodological caveats relating to the usage of the so-called “corpus languages” (in Cuzzolin & Haverling’s 2010 terms) for research on historical linguistics. Chapter 2 provides an overview of the theoretical background that the linguist needs to study preverbs and their developments in this language sample. Understandably, given the broad geographic and chronological distribution of these languages (cf. Sections 1.3 and 1.4 below), preverbs show very distinct behaviors and thus require varied analytical tools. On the one hand, the frameworks of Cognitive Grammar, of spatial relations, and of Semantic Roles are needed to analyze the basic meanings of preverbs and their paths of lexicalization. On the other hand, the categories of aspect and actionality are also crucial to frame the developments of preverbs that are due to grammaticalization. Chapter 3 provides the reader with an overview of preverbs both inside and outside Indo-European. The first section, which deals with Indo-European, opens with a working definition of preverbs and a description of their functions. It then discusses their positional properties in ancient and modern Indo-European languages and their origin. Lastly, it addresses preverbs as a typological and a terminological problem, introduces the terminology adopted in this work, and explains the reasons behind the outlined terminological choices (for the sake of clarity, it is worth establishing the fact that I consistently call multiple preverbs+verbs combinations ‘composites’ rather than ‘compounds’). Then, a section follows on the specific research topic: multiple preverbs. The last section of the chapter offers some typological insights on Finno-Ugric, Caucasian, Amerindian, and Northern Australian preverbs, insofar as their behavioral properties and paths of development may pertain to our understanding of Indo-European preverbs. <?page no="28"?> 28 1. Introduction The analysis unfolds in Chapters 4-7, starting with Vedic, going through Homeric Greek and Old Church Slavic, and concluding with Old Irish. All of these chapters are organized in a consistent way. The first section outlines the state of the art on preverbs and multiple preverbs in each language and provides information on their categorial status. The second section then goes on to display quantitative data on multiple preverbs: specifically, the catalogues of (i) multiple preverb composites, (ii) multiple preverb combinations, (iii) verbal roots modified by multiple preverbs together with their frequencies. For Old Church Slavic and Old Irish, the Greek and Latin counterparts of multiple preverb composites are also provided. The third section addresses various issues relating to the form of composites: (a) possible sandhi phenomena occurring between their elements; (b) the relative positioning of preverbs with respect to other pieces of preverbal morphology; (c) in the case of Old Church Slavic, the interaction between preverbs and Slavic secondary verbal suffixes expressing (im)perfectivity. The fourth section analyzes the semantics of preverbs occurring in multiple preverb composites and the semantic compositionality of multiple preverb composites. It also focuses on a number of particularly interesting semantic developments, in order to show how new abstract meanings relate to the basic one via cognitive metaphors. It concludes with a table summarizing the meanings of preverbs as they occur in multiple preverb composites. The fifth section follows dealing with the syntax of multiple preverbs and / or the argument structure of multiple preverb composites. It explores (a) the possible alternative constructions to multiple preverb combinations (i. e. the ‘mobility’ of preverbs); (b) their optionality; (c) their ability to function as transitivizing morphemes; (d) their inclination to be repeated outside the preverbal context; (e) their capacity for referring back to previously mentioned or discourse-active participants. The final section of Chapters 4-7 investigates the reasons behind preverb ordering, which is understood to originate from the interplay of different factors: (a) semantic and cognitive motivations, most notably redundancy; (b) textual motivations, such as the reference to discourse-active participants; (c) historical motivations, such as specific etymological origins of particular preverbs; (d) contact-relating motivation, including calquing from Greek or Latin. The final chapter of this work, Chapter 8, summarizes the analyses presented in the preceding four chapters, compares their results, and offers general concluding remarks. <?page no="29"?> 1.3. Corpora of the present study 29 1.3. Corpora of the present study 1.3.1. The R̥g-Veda The R̥gVeda (from √ r̥c- ‘praise, verse’ + veda ‘knowledge’) represents the most ancient Indian collection ( sam̥hitā ‘put together’) of hymns ( sūktas (literally) ‘well said’) addressed to the Vedic gods, mantras, magic spells, and sacred formulas. It belongs with the four canonical sacred texts of Hinduism, known as the Vedas, which also include the Sāmaveda ‘veda of chants’, the Yajurveda ‘veda of the sacrifices’, and the Atharvaveda ‘veda of the magic spells’. Together, they constitute the so-called “early Vedic” or “mantra language”, the most ancient variety of Old Indo-Aryan. Together with the Vedas, the Vedic corpus also comprises later prose texts: the Brāhmaṇas , the Āraṇyakas , the Upaniṣads , and the Sūtras . Early Vedic can be considered a north-western dialect and as such is close to Avestan, whereas later Vedic shows many features of the central Vedic dialects, which approximate this variety to Classical Sanskrit. In particular, the language of the Āraṇyakas and Upaniṣads seems to attest an intermediate stage to that of the Sūtras , which is very close to Classical Sanskrit (Macdonell 1916: 1 ff.). In this work, I only take into account the R̥gVeda , which is undoubtedly the most ancient among the sam̥hitās : while the R̥gVeda is mentioned in the Sāmaveda , the Yajurveda , and the Atharvaveda , the R̥gVeda itself contains no references to these three other Vedic collections. The R̥gVeda is also one among the oldest extant texts in any Indo-European language. Philological and linguistic evidence suggests that the R̥gVeda was composed in the north-western region of the Indian subcontinent, most likely between 1500-1200 BC, though a wider time frame of 1700-1100 BC has also been proposed (cf. further Witzel 1995; Mallory & Douglas 1997; Anthony 2007; Kulikov 2017). The R̥gVedic hymns certainly postdate the Indo-Iranian separation (about 2000 BC ) and probably the Indo-Aryan Mitanni documents (1400 BC ). The R̥gVeda is organized in ten books, known as maṇḍalas (literally) ‘circles’, of varying antiquity and length, for a total of 1028 hymns. The hymns in turn consist of individual stanzas ( r̥cas ‘praises’), which can be further subdivided into metrical units ( pāda ‘foot’) (cf. Chapter 4). Different sections of the R̥gVeda can be assigned to different chronological layers. The maṇḍalas II - VII (‘family books’) constitute the oldest and the shortest part of the collection (‘early R̥g-Vedic’); maṇḍalas I, X, and part of VIII (so-called Vālakhilya ) are the latest additions (‘late R̥g-Vedic’); maṇḍalas VIII-IX are chronologically heterogeneous. <?page no="30"?> 30 1. Introduction 1.3.2. The Homeric poems The Iliad and the Odyssey are two epic poems (for a total of about 28 000 lines) that chiefly recount the last weeks of the Trojan War and the tribulations that Odysseus experienced when returning to Ithaca after the fall of Troy. The Homeric epic is composed in hexameters, that is, lines made up of six ( héx ‘six’) feet, which in turn comprise regular alternations of light / short and heavy / long syllables, interrupted by regular patterns of metrical pauses (cf. Chapter 5). The variety transmitted by the epic tradition is basically an archaic eastern Ionic, enriched by an amalgam of Mycenean and Aeolic features, as well as by a number of other archaic traits that cannot easily be ascribed to any particular dialect or region (Horrocks 2010: 44). This arguably artificial admixture can be explained by taking into account that, although one usually refers to their author as Homer, the Iliad and the Odyssey are actually examples of oral poetic diction (Lord 1960; Parry 1971). Most of the early epic bards, likely going back to the Bronze Age (Horrocks 1997, 2010), repeated, or better artistically recomposed, the poems during public performances, by drawing on a conventional range of recurrent narrative themes and of ready-made dictions to fit such themes into the meter (so-called “formulas”, that is, “group[s] of words which [are] regularly employed under the same metrical conditions to express a given essential idea”; cf. Lord 1960: 30). As a consequence, though the Iliad and the Odyssey were probably recorded in writing during the 8 th century BC , they preserve more ancient layers of the Greek language, from at least two centuries earlier, in the shape of formulas, precisely by virtue of this peculiar process of composition. Therefore, the Homeric poems are of inestimable value for linguistic reconstruction (Watkins 1976). Through this passage from oral to written transmission, the texts are likely to have been updated by their editors, though without seriously damaging the poets’ traditional narrative and stylistic repertoire (Horrocks 2010: 46). The basis for the modern editions of the poems emerges from the versions produced by the Hellenistic philologists (4 th -1 st centuries BC ). They in turn had at their disposal different previous editions, which could have been either earlier or contemporary, either personal ( kat’ándra ‘according to a man’) or official ( katà póleis ‘according to towns’). All in all, the Homeric text was fluid: both the bards who put together the Homeric texts and the editors who established their official form used varieties of Greek different from the original language of the oral tradition. <?page no="31"?> 1.3. Corpora of the present study 31 1.3.3. The Codices Zographensis, Marianus, and Suprasliensis Old Church Slavic (or Slavonic) is the linguistic variety attested in some of the oldest Slavic written records, which date back to the 10 th -11 th centuries AD . These records were not contemporary with Constantine (i. e. Cyril) and Methodius’ mission of Christianization of ancient Moravia (a region located somewhere in the Danube Basin), which crucially triggered the translation of Christian sacred texts from the Greek of the Septuagint and Byzantine Greek into the language of the Slavs (Marcialis 2007). However, as first shown by August Leskien, a chronologically consistent and relatively old group of extensive manuscripts can be identified and employed as a canonical source to describe the system of Old Church Slavic. The Slavic variety attested in this canon does not represent any particular Slavic regional dialect, but rather a literary language used by Slavs of different regions as a shared linguistic conduit within the Christian community (cf. Drinka 2011). Nonetheless, it has the general flavor of an early eastern Balkan Slavic (or Bulgaro-Macedonian) variety, and as such has also been labeled as Old Bulgarian or Old Macedonian (Lunt 1965: 4). As noted above, Old Church Slavic texts are translations from original Greek sources, which boasted a prestigious literary tradition and outstanding authority. For these reasons, Old Church Slavic has been frequently considered to be deeply influenced by the Greek originals at different linguistic layers, ranging from syntax (Lunt 1977; MacRobert 1986) to the lexicon (cf. Chapter 6; see also Ziffer 2005; Drinka 2011). Among the manuscripts contained in the Old Church Slavic canon, this work examines the most ancient ones, i. e. Codex Zographensis and Codex Marianus , and the most extensive one, i. e. Codex Suprasliensis (Lunt 1965: 7, 9). The Codex Zographensis and Codex Marianus are two of the so-called tetraevangelia , that is, full versions of the Gospels, both primarily written in Glagolitic script (cf. Lunt 1965: 15 ff. for more information on this script). The Codex Zographensis is made up of 271 folia in standard Glagolitic, plus 17 folia in Macedonian Glagolitic, and later additions in Cyrillic. It covers the Gospels from Mt 3.11 to the end of John (though the section Mt 16.20-24.20 belongs with the Macedonian addenda). It can be regarded as being phonetically faithful to Cyril and Methodius’ language (i. e. probably south-eastern Macedonian), but it also displays a number of arguably younger morphological features. The Codex Marianus , made up of 147 folia, contains the Gospel text from Mt 5.23 to Jn 21.7. It shows a number of linguistic deviations from the Cyril and Methodius’ language, which can be possibly motivated either by northern Macedonian or by Serbian influence. The Codex Suprasliensis , written entirely in Cyrillic, includes as many as 285 folia and covers different narrative materials. <?page no="32"?> 32 1. Introduction It mainly comprises a menaeum for the month of March, that is, a collection of saints’ lives for daily reading, enriched by a number of sermons for Holy Week and Easter. The language variety that it employs comes from a region located somewhere in central or eastern Bulgaria and is undoubtedly later than the language of the two above-mentioned tetraevangelia . Its Greek sources have not come down to us, which makes it difficult to precisely identify the constructions clearly demonstrating Greek influence (cf. Chapter 6). 1.3.4. The Milan and the Priscian Glosses The fundamental sources for the linguistic study of Old Irish consist of glosses on Latin manuscripts, which have been assembled in the two volumes of the Thesaurus palaeohibernicus (Stokes & Strachan 1901-1903), of which the Würzburg Glosses on the Pauline Epistles, the Milan Glosses and the Priscian Glosses constitute the largest portions. These collections of glosses represent archaic prose texts, which came down to us in more or less contemporary manuscripts. Thus, they did not undergo the major morphosyntactic and orthographic updating that altered most texts surviving only via later transcriptions. Nevertheless, due to the nature of short texts, glosses may be fluid: (a) when copying brief notes, the scribe may both omit a gloss and also add further glosses; (b) additional glosses can also come from a manuscript different from the exemplar that was originally copied (cf. Hofman 1993). In this work, the largest collection, i. e. the Milan Glosses , and the Priscian Glosses , which are extremely important due to their lexicographic richness (cf. Chapter 7), are taken into account. The Milan Glosses contain Old Irish interlinear and marginal explanations on, and translation of, a Latin commentary on the Psalms (manuscript Ambrosianus C301, now preserved in Milan). The manuscript dates back to the end of the 8 th -9 th centuries AD and reached Milan via Bobbio, after being written down most likely in Ireland. The earlier Latin commentary and the slightly later glosses and translations into Old Irish seem to be carried out by two different hands, as the glossator occasionally expresses hesitations as to the reading of the Latin commentary. Later on, a third scribe, probably equipped with better Latin skills, added a few corrections and the incipits of two Old Irish poems, now hardly readable. The main scribe, who signs himself as Diarmait, has often been blamed for having worked with less precision than the scribes who compiled the Würzburg Glosses ; hence, unsupported spellings and slips of the pen are frequent (McCone 1985b; GOI 4-7). Based mainly on phonological evidence, the Irish variety of the Milan Glosses has been said to be later than that of the Würzburg Glosses , but earlier than that of the Priscian Glosses (however, this is at present still matter of debate; cf. McCone <?page no="33"?> 1.4. Brief methodological remarks 33 1985b; Roost 2013). As first shown by Strachan (1901), Latin massively influenced the Irish texts at different linguistic levels: most notably for the purposes of the present work, Latin arguably played a role in the coinage of new Irish words and composites (cf. further Chapter 7; Strachan 1901; Strokes & Strachan 1901-1903; McCone 1985b). The Priscian Glosses are made up of marginal and interlinear comments on a translation of Priscian’s Institutiones Grammaticae (5 th -6 th centuries AD ) into Old Irish. They survived to the present thanks to a number of manuscripts, among which Cod. Sang. 904 is the largest and contains all glosses that also occur in all other manuscripts. It comes from St. Gall and may have been written down during the 9 th century in Ireland. The St. Gall glosses were compiled by two hands, which transcribed from the same original, plus minor later addenda. The language of this collection is said to be heterogeneous; however, it is generally considered later than that of the Milan Glosses , though it also shows a number of archaisms, probably due to the fact that it was compiled from different sources of various ages (Strachan 1903: 470). 1.4. Brief methodological remarks This work is entirely based on inherently limited corpora: the texts that survived to the present, mainly owing to accidents of the textual tradition, are the only existing texts at linguists’ disposal. Cuzzolin & Haverling (2010: 25) addressed such varieties as “corpus languages”: “[they are] no longer anybody’s native language[s] and what we can know of [them] as […] living language[s] is to be traced in the written material still at our disposal.” Therefore, the picture of a particular language that such materials present is certain to be fragmentary. Joseph & Janda (2003: 19) effectively sum up the consequences of these issues as follows: “no matter how carefully we deal with documentary evidence from the past, we will always be left with lacunae in coverage, with a record that remains imperfect and so confronts us with major chasms in our understanding that must somehow be bridged.” To begin with, a lack of attestation does not necessarily imply actual absence in the grammar or in the lexicon of a certain language (cf. Joseph & Janda 2003: 15 ff. for some examples of “accidental gaps in the historical record”). Moreover, the textual tradition and the manuscript transmission of certain written sources can also be responsible for alterations or / and updating of the originals (cf. in particular Chapter 5). Thus, all above texts constitute instances of intrinsically diachronic corpora, in that they simultaneously attest to different chronological layers of a variety: on the one hand, texts at our disposal are the outcome of <?page no="34"?> 34 1. Introduction centuries of textual tradition; on the other hand, different sections of the same text can date back to different time periods (Sections 3.1-3.4). All in all, as is discussed in Chapter 2 (see especially Section 2.2.3.3), grammaticalization theory is the most appropriate theoretical tool to deal with such inherently diachronic data: the developments that can be subsumed under the rubric of grammaticalization can be understood as gradual diachronic processes that result in gradient synchronic linguistic categories. This point has proved to be of crucial importance especially for the analysis of Vedic and Homeric multiple preverbs (cf. Chapters 4-5). In addition, by employing the grammaticalization theory and its intrinsic diachronic character, one can also assess the overall development of Indo-European preverbs by analyzing their behavior in sub-varieties that belong to very widely dispersed chronological layers. Specifically, as shown in Table 1, Vedic and Homeric Greek, on the one hand, and Old Church Slavic and Old Irish, on the other, are divided by a time gap of more than one millennium. l anguage t exts and m anuscripts t ime period Vedic R̥g-Veda 18 th -12 th centuries BC Homeric Greek Iliad, Odyssey about 8 th century BC Old Irish Milan Glosses, Priscian Glosses 8 th -9 th centuries AD Old Church Slavic Codex Marianus, Codex Zographensis, Codex Suprasliensis 10 th -11 th centuries AD Tab. 1: Language sample, texts, and dating Furthermore, the R̥g-Veda , the Homeric poems, and the Old Church Slavic texts represent literary corpora, in terms of their content and aims. Thus, their variety most likely does not faithfully mirror the actual usages of everyday speech (cf. Joseph & Janda 2003: 17-19). In addition, the Vedic hymns and the Homeric poems constitute poetic corpora. As such, they have to meet relatively rigid metrical requirements, which possibly also contributed to distancing the language from daily usage. Occasionally, meter might both constrain syntax in general, and word order in particular, as well as motivate otherwise obscure lexical choices (cf. Chapters 4 and 5, for further discussion and relevant examples). For the Old Church Slavic texts and for the Old Irish glosses, one must take into account their undeniable interaction with the Greek sourceand the Latin <?page no="35"?> 1.4. Brief methodological remarks 35 main texts. Thus, further issues relating to the employment of parallel (or quasi-parallel) corpora come into play, which have been touched upon in Sections 1.3.3 and 1.3.4, and further discussed in Chapters 6-7. However, with regard to our understanding of the formation process of multiple preverb composites, Greek and Latin equivalents have proved to be crucial, in that they can provide access points to the various degrees of lexicalization and semantic change affecting multiple preverb composites. <?page no="37"?> 2. Theoretical background: Analytical tools for the study of preverbs 2.1. Cognitive Grammar In this work, I adopt the theoretical framework of Cognitive Grammar, in which grammatical forms are conceived as meaningful: the difference between grammatical and lexical meanings essentially lies in their degree of abstractness. Thus, syntax and semantics are understood as a continuum : “lexicon and grammar form a gradation consisting solely in assemblies of symbolic structures” (Langacker 2008: 5). The meanings of the elements of grammar emerge as the concepts associated with linguistic expressions. Such concepts are grounded on elementary semantic structures, which in turn are based on humans’ perception and spatio-physical experience (cf. e. g. Talmy 1983; Lakoff 1987; Langacker 1987), as shown in Section 1.1. 2.1.1. Space: the basic domain of human cognition Human beings’ perception, experience, and conceptualizations are mediated and constrained by human bodies: this is what is meant by “embodiment” and “embodied cognition”. Embodied experience gives shape to conceptual structure: the world, as sensed by organs of perception, constitutes the basis of conceptual structure, that is, of human thoughts and concepts (cf. among many others, Lakoff 1987; Lakoff & Johnson 1980; Langacker 1987; Svorou 1994). Furthermore, if embodied experience shapes our conceptual structure, it must also constitute the foundation of meaning, that is, of concepts expressed by means of human language. This implies that meaning must be mediated through human perception: all concepts, both concrete and abstract, are grounded in terms of spatio-physical experience. Therefore, in Cognitive Grammar, space is regarded as one of the basic domains of human cognition, as it does not seem to be understood through other cognitive domains, and provides the basis for understanding other more abstract domains (Lakoff & Johnson 1980). More generally, any set of concepts that cannot be described by means of another set of concepts can be regarded as a basic domain. By contrast, any domain that needs at least one other domain to be conceptualized is abstract (Croft 1993). <?page no="38"?> 38 2. Theoretical background: Analytical tools for the study of preverbs Linguistic forms, which are humans’ means for expressing thoughts and concepts, are initially associated with a concrete and spatial meaning, which constitutes the starting point for developing more abstract meanings and functions. The mapping from a concrete to an abstract conceptual domain is possible thanks to the cognitive mechanisms of metonymy and metaphor. Importantly, several metonymic and metaphorical meanings are regularly associated with specific linguistic sources, and later on conventionalized (cf. Section 2.1.2). In conventionalized lexical items, it can become difficult to trace back to the original spatial meaning, or to understand the links between the developed abstract and the basic spatial meanings from a synchronic point of view (cf. Section 2.2.5 on lexicalization; Chapters 6 and 7 on Old Church Slavic and Old Irish preverbs for cases in point). Thus, a given linguistic form is usually polysemous: each of its meanings can rely on the basic or on the abstract domains that pertain to that linguistic form. All meanings of a linguistic form are organized around its basic meaning in what can be called “structured polysemy” (e. g. Tyler & Evans 2003): meanings directly or indirectly relate to the center or to each other in a motivated radial structure (e. g. Lakoff 1987). 2.1.2. Going from spatial to abstract domains: metaphor and metonymy Metaphor is a way to conceptualize a cognitively difficult domain in terms of an easier domain; in other words, through metaphor it is possible to understand “conceptually complex phenomena in terms of less complex ones” (Claudi & Heine 1986: 299). Thus metaphor establishes a mapping between two different conceptual domains that, despite being equated, remain distinct (Croft 1993). Examples (1)-(2) show two different uses of the English preposition in : (1) Luke is in the kitchen. (2) Luke is in love . Sentences (1) and (2) contain similar linguistic items: the proper name Luke , the third person singular of the verb to be , the preposition in , and a common noun, kitchen in (1), determined by the article the , and love in (2). However, while the noun kitchen denotes a real Location where Luke is, the noun love denotes a state that Luke experiences. Thus, the same verb to be and the same preposition in express a spatial relation in (1), but a metaphorical relation in (2). This shift toward the abstract plane is accounted for by conceptual metaphor: the room kitchen physically contains Luke ; in a comparable way, the state of being in love is understood as a container in which Luke is metaphorically located. As Lakoff & Johnson (1980: 32 ff.) point out, the same metaphors can be responsible <?page no="39"?> 2.1. Cognitive Grammar 39 for multiple semantic changes: for example, states are often conceptualized as containers (the so-called “Container metaphor”). Within this work, this tenet is borne out both within a single language and also across different languages: Sections 4.4, 5.4, 6.4, and 7.4 are devoted to the semantics of multiple preverbs, Chapters 4-7 show that morphemes with similar basic spatial meanings also tend to develop similar abstract meanings (cf. further Chapter 8). Metonymy occurs when an entity of a certain conceptual domain is referred to by means of an entity belonging to a contiguous or identical conceptual domain (Lakoff & Johnson 1980: 29; Croft 1993). Within the same domain, such entities are connected by means of humans’ encyclopedic experience (Lakoff 1987). For example, in (3) below, the expression the ham sandwich does not refer to an actual sandwich, rather to the person who ordered it. The entity ham sandwich belongs to the conceptual domain that can be labelled as [ customer ], because a customer in a restaurant is presumed to order something to eat or drink. 1 (3) The ham sandwich is waiting for his check . Notably, example (3) cannot be regarded as a case of personification metaphor, given that human qualities are not ascribed to the said sandwich. Rather, the sandwich is a part of the conceptual domain of the person ordering it. By contrast, example (4) contains an istance of personification: (4) Inflation has attacked the foundation of our economy. Inflation has pinned us to the wall. In example (4), a non-human entity, inflation , is conceived or conceptualized as human owing to the metaphor inflation is a human being (Lakoff & Johnson 1980: 28 ff.). 2 This metaphoric extension only selects one feature of the source-entity, specifically ‘a human being can be an adversary’, according to the personification inflation is an adversary . Categories of entities, including human beings, show a number of properties that can be either viewed as a whole or observed one by one. Categories of entities seem to be organized as so-called Gestalten , that is, structures in terms of which our perception of the world is given a shape, and that exhibit a number of properties, including that of being “at once holistic and analyzable” (Lakoff 1977: 246). Thus, metaphors can also be based on a single property possessed by a category of entities, as in (4). 1 The notation with small caps and square brackets means that [ customer ] should be interpreted as a conceptual domain and not as an entity. 2 Small caps are Lakoff & Johnson’s (1980) and indicate that the above statement is a cognitive metaphor. <?page no="40"?> 40 2. Theoretical background: Analytical tools for the study of preverbs 2.1.3. The conceptualization of the spatial event Given that spatial concepts are cognitively basic for human beings (cf. Section 2.1.1), it is worth discussing how they are conceptualized. Spatial phenomena can be viewed from different standpoints, and consequently conceptualized in different ways. One of the most important varying parameters, in terms of conceptualization, is “prominence” (in Langacker’s 1987, 2008 terms). Prominence is a kind of asymmetry related to the focus of attention, that is, to what a linguistic expression describes as foreground or background (Langacker 2008: 68). A discussion of different types of prominence follows. 2.1.3.1. The first type of prominence: the profile-base asymmetry The concept of profile is introduced by Langacker (1987) by means of the word spoke. In order to understand the meaning of spoke, Langacker writes, one must also be aware of what a wheel is. The segment spoke is defined in relation with the structure of the whole wheel . Langacker describes the relation holding between the spoke and the wheel as the relation of a “profile” of a concept with respect to its “base”. The profile is the precise and narrow concept expressed by a word, whereas the base can be defined as the encyclopedic knowledge or conceptual structure presupposed by the said word. 3 As Croft (1993) points out, the profile and the base comprise an inseparable pair: a profile needs a base against which it is individuated. Symmetrically, a base cannot be individuated without the profiles that are defined with respect to it. The verb “to profile” corresponds to the noun “profile”. Thus, both these formulations are possible: spoke functions as a profile of the base wheel , or spoke profiles a certain part of the base wheel . In a similar way, the meaning of wheel is also the base for hub and rim , as shown in Figure 1 below: Fig. 1: The profile-base asymmetry: wheel vs. spoke , rim , wheel (from Langacker 2008: 67) 3 Langacker (1987) and Lakoff (1987) also use the term “domain” to indicate the base, whereas Fillmore (1982) calls it “frame”. <?page no="41"?> 2.1. Cognitive Grammar 41 An expression can profile either a thing, as in Figure 1, or a relationship. Therefore, the concept of profile can also be employed to describe spatial and non-spatial relations and thus the meaning of preverbs. For example, the Homeric motion verb eis-ana-baínō ‘go up to’ profiles the movement of an entity going along a trajectory in a certain direction. This verb contains two preverbs: the former, eis- ‘to’, profiles the direction of motion (Goal); the latter, ana- ‘up’, instead profiles its orientation and Path, specifying that the verb indicates an upward motion. The whole spatial relation expressed by the compound eis-ana-baínō implies that there are a path, an entity that moves along that path, and an entity to be reached, which constitute the basis of the spatial relation. 2.1.3.2. The second type of prominence: the Trajector-Landmark asymmetry As anticipated in discussing the meaning of the verb eis-ana-baínō ‘go up to’, entities are usually located with respect to other entities functioning as reference points (Talmy 1983; Langacker 1987). This way of locating entities implies a further asymmetrical relation holding between the located entity and the reference-entity. Talmy (1983) introduces the terms “Figure” and “Ground”, borrowed from Gestalt psychology (Köhler 1929; Koffka 1935), to describe this asymmetrical relation: the Figure is the object to be located, while the Ground is the object with respect to which the Figure is located. In reference works, other terms are also used to identify the participants in a spatial relation, including the pairs “locans”-“locatum” and “referent”-“relatum” (e. g. Rappaport & Levin 1985; Levinson 1996). In this work, I opt for Langacker’s (1987, 2008) terminology, which describes Talmy’s Figure and Ground in terms of focus of attention. Langacker argues that, while profiling a spatial relation holding between two entities, one of such entities is always more focused than the other one. Langacker calls the more prominent and located entity the Trajector (henceforth TR ), and the less prominent reference-entity Landmark (henceforth LM ). Langacker employs the concepts of TR and LM beyond the cognitive domain of space. Following his lead, let us take as an example the kinship relations of having a child and having a parent . Both relations share the base, that is, the domain of kinship relations . In addition, they profile the same kinship relation, as they involve two participants, of whom one is the child or the parent of the other. What changes is the directionality of the relation, and thus their TR - LM alignment: having a child is primarily concerned with parents, who thus function as a TR . By contrast, having a parent is a predication concerning the child, who in turn functions as a TR . Figure 2 shows both the profile-base and TR - LM asymmetries: <?page no="42"?> 42 2. Theoretical background: Analytical tools for the study of preverbs Fig. 2: Kinship relations: profile-base and TR - LM asymmetries (Langacker 2008: 68) In Figure 2 (a)-(d), bold highlights the profile. Both in (a) and in (b), the profile is a human entity, either the parent or the child . They are both characterized by the relative role that they play in the kinship relation, which is conceptualized as the base. However, in both (c) and (d), the shared kinship relation is profiled. The semantic contrast between have a parent and have a child resides in their opposite directionalities, and thus their TR-LM alignments. 2.1.3.3. The parameters of the spatial event So far, I have discussed static spatial events, in which a TR is located with respect to a LM . However, a spatial event can also imply motion: in such events, the TR moves with respect either to a stable or to another moving entity ( LM ). In each case, one recognizes an asymmetrical relation between a TR and a LM . Several parameters can contribute to such asymmetrical relation, including the number of the moving entities, the direction of movement, the path of movement, containment, contact, orientation, or a combination of these (Svorou 1994: 24). Motion events can be conceptualized as having directionality, or a deictic orientation. TR s can be directed toward or away from LM s: for example, the English verb to go implies a motion away from the speaker, whereas to come implies a direction toward the speaker. Furthermore, the directionality of certain entities can be specified on a vertical axis, such as in the following Italian verbs: salire ‘to go up’ entails an upward motion, while scendere ‘to go down’ a downward motion. In addition, the conceptualization of a motion event subsumes a trajectory, that is, the path covered by a TR with respect to a LM . However, the trajectory can either be profiled or remain in the base. For example, the Ancient Greek verbs pḗgnumi ‘fix’ and bállō ‘throw’ can take either the dative or the accusative, both expressing the direction of the motion. The dative and the accusative cases differ in their profile: the dative only profiles the endpoint of the trajectory, while the accusative profiles it as a whole. <?page no="43"?> 2.1. Cognitive Grammar 43 Both location and motion events can involve the containment of the TR inside the LM , conceptualized as a container. As for location events, the TR can either be placed inside, near, or in contact with the LM (“ in essive”, “ ad essive”, and “ super essive” location, respectively). As regards motion events, the TR can move either toward the inside of a LM , toward its vicinity, or toward a contact position with the LM (“ il lative”, “ al lative”, and “ super allative” motion). Conversely, the TR can move from the inside of the LM , from its vicinity, or from a contact position with the LM (“ e lative”, “ ab lative”, “ super lative” motion, respectively). As discussed for examples (1) and (2), both spatial and metaphorical LM s can be conceived as containers: the Container metaphor is one of the most widespread means for shifting from the concrete to the abstract plane in conceptualizing events (Lakoff & Johnson 1980: 32 ff.). The Container metaphor is based on the fact that the human body itself has its dimensionality and can be conceptualized as a container. In contexts in which containment is not involved, contact instead can be relevant to identify the position of a TR with respect to a LM . The opposition based on contact is not expressed through morphological cases in Indo-European languages. However, in English as well as in ancient Indo-European languages, a number of preverbs-adpositions (cf. Chapter 3 on this terminology) in fact express this contrast: in English both on and over indicate that the TR is vertically located with respect to the LM , and differ only in the presence / lack of contact (Brugmann 1988); in parallel, for example, the basic meanings of Ancient Greek epí ‘on’ and hupér ‘over’ express superiority, the former involving contact, the latter lacking this implication (Luraghi 2003: 24). Another relevant parameter is the “plexity” (in Talmy’s 2000: 177-254 terms) of the TR , of the LM and of the trajectory. TR and LM can consist of separate items (multiplex), or of a non-analyzable single entity (uniplex); in parallel, trajectory can be unidirectional (uniplex) or multidirectional (multiplex). Remarkably, the plexity is not an inherent feature of the participants in a spatial relation, but rather depends on how these participants are conceptualized. A further distinction only concerns multiplex entities: they can be either continuous or discontinuous. Discontinuous entities are typically plural count nouns, which profile the existence of a number of individuated entities; conversely, continuous entities are usually mass nouns and collectives, which profile an undifferentiated mass. Thus, discontinuous and continuous entities, as well as plural and mass nous, are distinguished only by their degree of individuation (Langacker 1987: 294). <?page no="44"?> 44 2. Theoretical background: Analytical tools for the study of preverbs 2.2. Grammaticalization 2.2.1. A brief history of grammaticalization studies At least since the early 19 th century, it has been a common observation that independent lexical items constitute the ultimate source for bound grammatical forms (e. g. Bopp 1816; Humboldt 1822; Wüllner 1831). 4 However, Antoine Meillet was the one who presumably coined the term “grammaticalization”, which increasingly acquired more generally accepted usage in later research. Meillet (1912) described grammaticalization as the development of an autonomous word into a grammatical element, that is, as a linguistic change whereby lexical items enter the grammatical system. 5 Meillet (1912: 140-141) later on compared the process of grammaticalization to a spiral: first, languages add non-obligatory elements to a given expression for the sake of intensification; then, these elements increasingly undergo semantic bleaching (i. e. they lose or reduce their meaning), lose their autonomy, and eventually enter the grammatical system. Further new expressive elements thereafter come into play and in turn undergo bleaching in a neverending cycle. 6 Crucially for later research (cf. Section 2.2.3), Meillet assigned gradualness to grammaticalization: focusing on the French verb ‘to be’, he individuated different stages within its path from a lexical locative-existential verb ( je suis chez moi ‘I am home’), through a copula ( je suis malade ‘I am sick’), into a tense auxiliary ( je suis allé ‘I went’). Building on Meillet’s ideas, Kuryłowicz (1975[1965]: 52) was among the first scholars to undertake new studies framed within the theory of grammaticalization. Kuryłowicz’s broadened definition of grammaticalization later became traditional: “Grammaticalization consists in the increase of the range of a morpheme advancing from a lexical to a grammatical or from a less grammatical to a more grammatical status, e. g. from a derivative formant to an inflectional one.” Thus, when grammaticalization takes place, not only must a previously lexical item enter the grammatical system, but also it becomes possible for a less grammatical item to acquire new and more grammatical functions. Interest in Kurylowicz’s work on grammaticalization revived in the 1970’s, when Givón (1971, 1979) gave new life to grammaticalization studies, by stress- 4 On the history of grammaticalization studies, cf. Lehmann (1995[1982]), Heine, Claudi & Hünnemeyer (1991), and Hopper & Traugott (2003). 5 “[…] le passage d’un mot autonome au rôle d’élément grammatical” (Meillet 1912: 131). 6 Here are Meillet’s (1912: 140-141) words: “Les langues suivent ainsi une sorte de développement en spirale: elles ajoutent des mots accessoires pour obtenir une expression intense; ces mots s’affaiblissent, se dégradent et tombent au niveau de simples outils grammaticaux; on ajoute de nouveaux mots ou des mots différents en vue de l’expression; l’affaiblissement recommence, et ainsi sans fin.” <?page no="45"?> 2.2. Grammaticalization 45 ing the fact that a language structure can only be understood in the light of its past (Givón, indeed, is the source of the famous slogan “today’s morphology is yesterday’s syntax”). However, works consistently framed within grammaticalization theory only started flourishing during the 1980’s (cf. Lehmann 1995[1982]; Heine & Reh 1982), and persist to the present (cf. among many others Heine, Claudi & Hünnemeyer 1991; Traugott & Heine 1991a, 1991b; Hopper & Traugott 1993; Heine 1997; Rissanen et al. 1997; and Fischer et al. 2000). In the last decades, however, grammaticalization theory also has undergone serious criticism (cf. Section 2.2.4). 2.2.2. Current approaches to grammaticalization While undergoing serious criticism, the grammaticalization theory has also extended its traditional scope of application in different directions. While in the 20 th century the grammaticalization theory was mostly confined to grammatical, semantic, and pragmatic analyses carried out by functionally-oriented scholars from Europe and North America, it is now employed as a theoretical framework in phonology, language acquisition, and sociolinguistics, as well as in studies performed by formal linguists in regions including East Asia and South America (Narrog & Heine 2011: 2-3). This diversity had the consequence of multiplying the number of approaches to and definitions of grammaticalization (cf. Traugott 2010). One possible way to sort out these variations is by clustering the approaches into two main groups based on the inclusiveness / exclusiveness of grammaticalization. The older and more restrictive approach to grammaticalization puts strong emphasis on the parameter of obligatorification (cf. e. g. Lehmann 1995[1982]; Haspelmath 2004), which implies reduction in transparadigmatic variability, increased dependency, and tightening of boundaries. Haspelmath’s (2004: 26) definition is often cited to exemplify this approach: “A grammaticalization is a diachronic change by which the parts of a constructional schema come to have stronger internal dependencies” (Haspelmath 2004: 26). Lehmann (2004: 155) also describes grammaticalization as a reduction of autonomy: “Grammaticalization of a linguistic sign is a process in which it loses in autonomy by becoming more subject to constraints of the linguistic system” (Lehmann 2004: 155). In both descriptive and theoretical literature on grammaticalization, this term is commonly understood in the narrow sense to denote a loss: loss of meaning, loss of phonology, and loss of independence. By contrast, more inclusive approaches regard grammaticalization as an expansion of grammar, and generally as an enrichment of languages (cf. e. g. von Fintel 1995; DeLancey 2001; Himmelmann 2004; Brinton & Traugott 2005; <?page no="46"?> 46 2. Theoretical background: Analytical tools for the study of preverbs Croft 2006). Among others, an advocate of this view is DeLancey (2001): “The word grammaticalization […] implies a process of becoming ‘grammatical.’ The reference can be taken as being to lexical morphemes becoming grammatical ones, or, more broadly, to any linguistic construct (a morpheme, a syntactic construction, or a discourse pattern) becoming part of the grammatical system of a language.” Importantly, in DeLancey’s view, grammaticalization not only affects morphemes, but also linguistic units embedded within greater constructions or patterns; in addition, it not only causes the development of morphology, but also of all structures of languages. Brinton & Traugott (2005: 99) adopt a similarly broad definition: “Grammaticalization is the change whereby in certain linguistic contexts speakers use parts of a construction with a grammatical function. Over time the resulting grammatical item may become more grammatical by acquiring more grammatical functions and expanding its host-classes.” Thus, to the foundational element of expansion, Brinton & Traugott add that of context expansion, or paradigmaticization. Croft’s (2006: 366) definition is even wider: grammaticalization is “the process by which grammar is created.” As Fischer (2011b) notes, a number of linguists even employ “grammaticalization” as a synonym for “change”, though not every linguistic change, for example sound change, can qualify as a grammaticalization. The analysis of the developments of Indo-European preverbs can take advantage both of the inclusive and of the exclusive approach to grammaticalization outlined so far. Specifically, on the one hand, the criterion of obligatoriness is crucial for understanding the developments of Vedic and Homeric Greek preverbs (cf. Chapters 4 and 5). We will see that the occurrence of a preverb gains increasing obligatoriness to express certain types of spatial and non-spatial meanings. On the other hand, the wider approach, by taking into consideration semantic bleaching and the concomitant context expansion undergone by grammatical items, is of particular importance for Chapters 6 and 7, in which Old Church Slavic and Old Irish preverbs are analyzed. 2.2.3. The continuum of grammaticalization As DeLancey (2001) points out, a grammaticalization process entails changes at every level of language. Notably, although all changes outlined below typically occur in grammaticalization processes, independently of one another, they are not essential for grammaticalization to take place (Bybee, Perkins & Pagliuca 1994). At the semantic and pragmatic level, a grammaticalization process involves changes such as semantic bleaching, semantic specialization through metaphor, reanalysis through pragmatic inference, lexicalization - the process by which <?page no="47"?> 2.2. Grammaticalization 47 originally independent lexemes become parts of new lexical items (cf. Section 2.2.5), and “referent conflation”. A referent conflation is a semantic change in which two conceptually distinct referents are reanalyzed as one. For example, in the development of an adposition from a relational noun, e. g. atop + noun from *on the top of + noun , one can see the conflation of two referents into one ( top + noun > noun ) (DeLancey 2001). In the layers of syntax and morphology, changes such as reanalysis and alterations in constituent structure also can occur. In addition, as noted earlier (Section 2.2.2), grammaticalization often involves obligatorification (Lehmann 1995[1982]). However, some scholars, such as Heine & Kuteva (2007: 34), point out that, though obligatorification seems to be important, it is not necessary for grammaticalization to take place, and that it also occurs in language changes different from grammaticalization. Heine et al. (1991) and Heine & Kuteva (2007: 40-41) also speak of decategorialization in grammaticalization processes. Through decategorialization, a linguistic expression is likely to lose morphological and syntactic properties that were characteristic of its initial category but which are not central for the new grammatical function. For example, in English a number of gerund forms, such as barring, concerning, and considering, came to acquire prepositional functions. This development led such forms to lose a number of properties normally associated with the morphological category of verbs: e. g. the possibility of taking auxiliaries, and of being inflected for tense and aspect. At the phonological level, morphemes can undergo phonetic erosion or cliticization (Heine & Kuteva 2007: 42 ff.). In Himmelmann’s (2004) view, such changes result in a set of expansions (cf. Section 2.2.2): host-class expansion, in which expressions are generalized to more collocational contexts, syntactic expansion, in which expressions come to acquire new syntactic functions, and semantic-pragmatic expansion. 2.2.3.1. Gradualness and the stages of grammaticalization Gradualness is a concept that affects linguistic change in general and grammaticalization in particular, and as such it is spread throughout diachronic stages. Gradualness can be understood as a two-fold concept. First, it describes how a linguistic change propagates among new groups of speakers, different situational contexts, and various stages of language. Second, it concerns the structural propagation of change: “gradualness refers to the fact that most change involves (a series of) micro-changes” (Traugott & Trousdale 2010: 23). Each step in this process represents an intermediate construction type in structural terms (Croft 2001: 313). In fact, by virtue of gradualness, more than one intermediate step <?page no="48"?> 48 2. Theoretical background: Analytical tools for the study of preverbs in change may coexist in the same individual or community of speakers (cf. Section 2.2.3.3). In contrast with Traugott & Trousdale (2010), who regard grammaticalization as a sequence of discrete changes, no matter how small these are, DeLancey (2001) conceives of the whole grammaticalization as a continuum : the distinguishing of a number of discrete processes only results from linguists’ idealization. Such processes need not be laid out in a strict serial order. However, some of them are likely to trigger other successive processes by providing them with the necessary conditions for development. In this work the stages of grammaticalization are described mainly using DeLancey’s model (2001). First, the essential precondition for grammaticalization is a productive syntactic construction: a lexeme or some lexemes must frequently occur in certain constructions owing to some semantic or pragmatic motivations. By virtue of their frequency, such constructions undergo conventionalization, or in DeLancey’s terms “undergo a functional specialization” (e. g. the face of NP , finish VP , and so on). In what one may call usage-based models of language, frequency is invoked as one of the main forces, or even as the main force, driving grammaticalization, or even linguistic change in general (cf. e. g. Bybee & Hopper 2001; Bybee 2011). At a later stage, such a construction undergoes semantic bleaching (as defined above), and consequently acquires the possibility of being used in a wider range of contexts. Notably, such a construction also occurs in contexts conflicting with its original and more specific meaning. For example, when the noun front becomes a part of the adposition in front of , it loses the portion of its meaning that directly refers to the human body (Heine & Kuteva 2007: 40). The next stage involves decategorialization (or re categorialization; cf. Heine et. al. 1991; Heine & Kuteva 2007; Section 2.2.3): the construction loses (some of) the morphosyntactic behaviors characteristic of its original category. For example, as discussed in Chapter 3, as soon as Indo-European local adverbs start developing into preverbs and / or into adpositions, they lose the typical syntactic freedom of adverbs. Such categorial reassessment can have two results. The developed form can enter one of the existing morphosyntactic categories of the language. Alternatively, the form comes to show behaviors that differentiate it from all the other items of the language. In this latter case, the said form gives rise to a new morphosyntactic category. This is the case of Proto-Indo-European local adverbs undergoing functional bifurcation into preverbs and adpositions. Finally, two other (non-obligatory) stages of grammaticalization are cliticization and morphologization. In these, the grammaticalized form loses its independency, both at the phonological and syntactic levels. For example, the grammaticalization path of Indo-European local adverbs includes the following <?page no="49"?> 2.2. Grammaticalization 49 loss of independency: local adverbs first develop into clitic preverbs, and subsequently into prefixes that cannot be displaced from their hosting verbs (cf. Chapter 3). 2.2.3.2. Gradience The concept of grammaticalization is grounded on the dichotomy of “lexical” vs. “grammatical” forms. However, grammaticalization studies contributed to blurring this clear-cut distinction (e. g. Lehmann 1985, 1995[1982]; Hopper & Traugott 1993; DeLancey 2001). In assuming a continuum between lexicon and grammar, most versions of grammaticalization theory go hand in hand with functionalist and usage-based models of language, including Cognitive Grammar (Narrog & Heine 2011: 9; Section 2.1). As with the term gradualness, the term gradience - here regarded as a synchronic phenomenon relating to the continuum of categoriality and grammaticalness (Traugott & Trusdale 2010: 22) - allows for a two-fold interpretation. To begin with, the members of a certain category do not fit equally into that category; rather, there can be better or worse representatives of the said category. This interpretation of gradience involves only a single category. In contrast, the second interpretation involves at least two categories, which are not understood as discrete and separated by clear-cut boundaries, but instead as constituting a continuum . Going back to the traditional dichotomy between lexicon and grammar, as certain as it is that there do exist linguistic elements that are either clearly grammatical (e. g. case inflections) or clearly lexical (e. g. nouns and verbs), it can be difficult to draw a sharp line dividing the categories of grammatical and lexical forms. Rather, linguistic forms are likely to be placed on a continuum of grammaticality and, accordingly, to be given a degree of grammaticality (cf. further the discussion on prototypical categories and prototypes at Section 2.3.1). Thus, one and the same linguistic element can be polysemous, in that it simultaneously expresses different lexical meanings, and also serves grammatical functions. In such cases, a number of scholars speak of “polygrammaticalization” (e. g. Craig 1991), whereby a single lexical item gives rise to multiple distinct paths of development. It is important to note that lexical morphemes develop into grammatical morphemes, but also that specific grammatical morphemes tend to emerge regularly from specific lexical sources cross-linguistically. Thus, grammaticalization is based on non-random linguistic sources (e. g. Traugott 1988; Givón 1979; Bybee 1988; Bybee, Perkins & Pagliuca 1994; Heine et. al. 1991; various papers in Traugott & Heine 1991a, 1991b). For example, future tense constructions frequently develop from verbs that originally mean ‘want’, ‘go’, or ‘have’. One such case is found in the Romance languages, in which the emergence of <?page no="50"?> 50 2. Theoretical background: Analytical tools for the study of preverbs synthetic future conjugations results from the univerbation of the auxiliary habere ‘have’ with an infinitive (e. g. It. canterò ‘I will sing’ < Vulg.Lat. cantare : inf habeo : prs .1 sg , which originally means ‘to sing’ + ‘I have’; cf. Benveniste 1968; Hopper & Traugott 1993). Causative morphemes regularly develop from serialized or complement-taking verbs with meanings like ‘make’, ‘give’ or ‘send’ (DeLancey 2001). Such systematic developments also suggest that grammatical functions must include a portion of semantic content, rather than being purely structural, which is one of the main claims of Cognitive Grammar (cf. Section 2.1). 7 In addition, as all these developments are gradual (cf. Section 2.2.3.1), there must be a diachronic stage in which a given form shows multiple functions, and thus displays an intermediate and uncertain categorial status (cf. the discussions on Vedic and Homeric Greek preverbs in Chapters 4 and 5). However, the categorial status of such items is uncertain or ambiguous only from the linguists’ standpoint, not from that of speakers, who naturally employ a certain form or a certain construction with its different meanings and functions. Therefore, not only can categories be said to be non-discrete, but they are also not even a priori given to linguists; rather, linguistic categories emerge from speakers’ language use and can be abstracted by the linguists through data observation (cf. Haspelmath 2007b). 2.2.3.3. The intersection between gradience and gradualness As discussed in Section 2.3.2, at a synchronic level, it is often impossible unambiguously to assign a morpheme or a construction to a discrete grammatical category (e. g. Hopper 1987; Givón 1979; Haspelmath 2011). Such synchronic gradience can be seen as a side-effect of diachronic gradualness (Hopper 1987; Bybee & Hopper 2001). In other words, “since grammaticalization is generally regarded as a gradual diachronic process, it is expected that the resulting function words form a gradient from full content words to clear function words” (Haspelmath 2001: 16 539). This relies on the continuum that characterizes grammaticalization, which provides a more useful basis for understanding even the synchronic gradience structure. In this light, grammaticalization approaches also call into question the traditional dichotomy between synchronic vs. dia- 7 These common developments prompt the question of whether grammaticalization paths can be borrowed (Heine & Kuteva 2005, 2006 speak of “grammaticalization contact zones”). A putative case in point might be the emergence of a future marker based on a volition verb in the Balkans. Upon closer inspection, however, the steps and processes needed to explain the emergence of such future markers in different Balkan languages vary to such an extent that it is difficult to maintain the grammaticalization contact zone as a meaningful construct ( Joseph 2011). <?page no="51"?> 2.2. Grammaticalization 51 chronic analyses (cf. Mithun 2011 for a discussion and a case-study on Navajo, a Na-Dené language spoken in Arizona, New Mexico, Utah, and Colorado). Such an approach to grammaticalization, which is based on the assumption that its diachronic gradualness results in a synchronic gradience, is particularly helpful to analyze the development of preverbs, especially in a sample including languages that are diachronically distributed (roughly, Vedic: 18 th -12 th centuries BC ; Homeric Greek: 8 th century BC ; Old Irish: 7 th -9 th AD ; Old Church Slavic: 9 th -11 th centuries AD ; cf. Section 1.3), and whose corpora are intrinsically diachronic, such as those of Vedic, Homeric Greek, Old Church Slavic and Old Irish (cf. Chapter 1). As discussed in Chapters 4 and 5, the Vedic and Homeric languages especially mirror the stratification of different stages of Old Indo-Aryan and Ancient Greek. Thus, a purely synchronic analysis would have led to a multiplicity of categories, as a number of linguistic forms, including preverbs, show multiple functions and, from the linguists’ viewpoint, seem to belong to more than one category. Rather, the uncertain behaviors shown by a number of forms arguably reflect the gradual steps in their grammaticalization path, which their diachronically stratified corpora show us. 2.2.4. Grammaticalization: an abused theoretical concept? As touched upon in Section 2.2.2, the whole enterprise of grammaticalization has undergone serious critique in recent years. Joseph (2011: 193-194) summarizes the major themes that drew criticism as follows: (a) grammaticalization as a process or result; (b) privileging one cluster of developments over others; (c) alternative outcomes / results; (d) unidirectionality. 8 The first criticism questions the nature of grammaticalization itself, by claiming that it is neither a distinct process nor a theoretically primitive concept, but rather merely a way of representing a whole series of independent linguistic processes or mechanisms of change, including for example sound change, semantic change, and reanalysis. Crucially, each of the mentioned changes also manifests itself independently of grammaticalization (cf. Campbell 2001; Joseph 2001, 2003, 2004, 2006, 2011, 2014; Newmeyer 2001). As Joseph discusses in several papers, by keeping the grammaticalization process separated from the other mechanisms of change, linguists needlessly increase the dimensions of language change, which are ultimately three: (i) the physiological dimension, which is relevant for sound change; (ii) the cognitive dimension, which is the basis for analogy (of which metaphor is understood as a subtype, that is, as a conceptual 8 Joseph (2011: 202 ff.) also adds the issue that he calls “grammaticalization and language contact”, which I do not thoroughly discuss here (cf. the hints in fn. 7). <?page no="52"?> 52 2. Theoretical background: Analytical tools for the study of preverbs analogy) and reanalysis; (iii) the social dimension, which is critical for the diffusion of language changes. Should we add to these dimensions / mechanisms of change the grammaticalization process as a distinct type of development? Or, rather, should grammaticalization be merely a label for naming the result of independently occurring processes? (cf. Joseph 2011). Moreover, if one assumes that, indeed, grammaticalization constitutes a distinct process, it remains difficult to assess how many grammaticalizations produce a certain gram, i. e. how many processes of change result in a new grammatical item or category (cf. also Janda 2001; Fischer 2011b). A second major critical theme expresses skepticism as to whether the series of sub-processes that are grouped under grammaticalization (on which cf. Section 2.2.3) really deserve to be treated separately as special, i. e. being assigned their own label, meriting dedicated conferences, textbooks, and so forth, with respect to other (clusters of) changes with which historical linguistics is concerned. This also relates to the issue of when exactly grammaticalization happens. For example, when does grammaticalization occur in the development of the Romance future of the type of Italian canterò ‘I will sing’ from Vulgar Latin cantare habeo ‘I have to sing’ (cf. Section 2.2.3.2)? Is it when the locution cantare habeo comes to show a conventionalized future meaning even without concomitant phonetic reduction? Or rather, are phonetic reduction and univerbation necessary ingredients for grammaticalization? Or, should these two processes (semantic change and univerbation) even be regarded as two distinct grammaticalizations, though they result in a single one grammatical item and category? Joseph’s (2011) elaboration on theme (b) directly leads us to his point (c), in which the author further emphasizes that “much more goes on in language change than just the often cited movement of lexical / somewhat-grammatical to (more) grammatical that characterizes grammaticalization” ( Joseph 2011: 198). Joseph goes to discuss a number of morphological developments that cannot be subsumed under the grammaticalization rubric. These are schematized below: (a) derivational morphology > derivational morphology (e. g. resegmentation of morpheme boundaries); (b) inflectional morphology > inflectional morphology (e. g. remaking of verbal endings based on other endings); (c) inflectional morphology > derivational morphology (e. g. reanalysis of an inflectional ending as part of the stem onto which verbal endings are further added, as in the instances of the so-called Watkins’ Law; cf. Arlotto 1972; Collinge 1985). The last line of criticism that I discuss here is one of the cornerstones of grammaticalization, specifically the so-called “unidirectionality hypothesis”: changes falling under the rubric of grammaticalization always go from less grammatical <?page no="53"?> 2.2. Grammaticalization 53 to more grammatical (Rosenbach 2004: 73; Börjars & Vincent 2011). From Givón (1971) onward, different works fairly strictly embraced the unidirectionality hypothesis, including Lehmann (1995[1982]: 16), Hopper & Traugott (1993: Chapter 5), Traugott (2001), Heine (2003), and Brinton & Traugott (2005: Chapter 4.3). In usage-based models, unidirectionality is motivated by frequency: “Changes related to increases in frequency all move in one direction and even decreases in frequency do not condition reversals: there is no process of de-automatization or de-habituation, subtraction of pragmatic inferences, etc. Once phonetic form and semantic properties are lost, there is no way to retrieve them. Thus grammaticization [i. e. grammaticalization] is unidirectional” (Bybee 2008: 348). Other functionalist approaches also suggest that social factors drive grammaticalization, such as “the invisible hand” (Keller 1990), the communicative goal of expressiveness, and speakers’ will to speak in such a way that they are noticed (so-called “extravagance” in Haspelmath 1999: 1043) and later on imitated by other speakers (so-called “conformity”). By contrast, formalists explain unidirectionality based on some universal principles relating to the principle of Economy (cf. van Gelderen 2004, 2011), including that of “grammar optimization”, that is, the elimination of unmotivated grammatical complexity or idiosyncrasy (Kiparsky 2011). Evidence against the unidirectionality hypothesis gathered in recent decades (e. g. among many others Ramat 1992; Haspelmath 2004; Willis 2007; Norde 2009; Kiparsky 2011) was seized upon to argue against the existence of grammaticalization as a distinct and pervasive process of linguistic change, precisely because of the said occurrence of many counterexamples (e. g. Janda 2001; Joseph 2001; Newmeyer 2001). As in the case of grammaticalization (cf. Section 2.2.2), there exist wider and narrower definitions for the opposite process, often called “degrammaticalization”, “antigrammaticalization” or “countergrammaticalization”. For example, Newmeyer (2001: 205) considers an instance of degrammaticalization any increase in lexical content or morphological independence. In a more precise way, Norde (2010: 126) defines degrammaticalization as a change that affects several linguistic layers: “Degrammaticalization is a composite change whereby a gram in a specific context gains in autonomy or substance on more than one linguistic level (semantics, morphology, syntax, or phonology).” This view is shared by other scholars, including Willis (2007: 273), who offers a catalogue of different linguistic changes operating at different linguistic levels that can be ascribed to degrammaticalization: (i) phonological strengthening, (ii) change rightward along the cline: affix > clitic > independent word; (iii) categorial reanalysis from grammatical to lexical; (iv) metaphorical shift from abstract to concrete; (v) pragmatic inferencing from abstract to concrete. Importantly, changes (i)-(v) <?page no="54"?> 54 2. Theoretical background: Analytical tools for the study of preverbs mirror their opposite counterparts in grammaticalization. The idea that degrammaticalization must be described in parallel to grammaticalization is common to other scholars. Haspelmath (2004: 27) for example writes as follows: “By this [i. e. antigrammaticalization] I mean a change that leads from the endpoint to the starting point of a potential grammaticalization and also shows the same intermediate stages.” 2.2.5. Grammaticalization and lexicalization Within the literature on grammaticalization (and degrammatilicalization), a central issue is assessing whether a linguistic change constitutes a real instance of grammaticalization, or whether it can be better captured within some alternative fields of analysis. For example, degrammaticalization of the type (iii) (cf. Section 2.2.4), that is, the categorial reanalysis from grammatical to lexical, implies new entries in the lexicon. Thus, what is the point in differentiating between type (iii) of degrammaticalization and “lexicalization”, which, according to some definitions, includes any type of lexical enrichment (cf. e. g. Hopper & Traugott 1993: 127; van der Auwera 2002; Lightfoot 2011: 438 ff.)? 9 Thus, the first reason for ambiguity as regards the lexicalization is that, since the beginning of grammaticalization studies (Kuryłowicz 1975[1965]: 52), linguists have regarded lexicalization as the reverse of a grammaticalization. One way of distinguishing degrammaticalization of the (iii) type from lexicalization is emphasizing that degrammaticalization must mirror the same steps as those of grammaticalization in reverse order, while this is not true of lexicalization. Moreover, as pointed out by Anttila (1989[1972]: 151), any new item entering the lexicon undergoes lexicalization, including those that instantiate developments whereby they increase their grammaticality (e. g. open-class noun > closed-class adverb). This observation somewhat anticipates the second reason for confusion between grammaticalization and lexicalization: the fact that most functionalist scholars recognize that lexicon and grammar are not two discrete categories, but rather they operate on a continuum (cf. Sections 2.1 and 2.2.3.2) . 9 In this Section, I only discuss the diachronic definitions for “lexicalization”. However, lexicalization is differently understood in works embracing a synchronic perspective. For example, for many synchronic researchers, including lexical typologists, “lexicalization” and “lexicalize” mean “has a segmental expression”, or “is used to pack a certain portion of meaning” (Brinton & Traugott 2005; Traugott & Trausdale 2013: 33 ff.). Talmy’s (e. g. 1985, 2000) well-known lexical typology of motion verbs, for example, distinguishes verbs that express (“lexicalize”) the manner of motion, from verbs that encode the path of motion. <?page no="55"?> 2.3. Describing the meanings and the functions of preverbs 55 Along this continuum, lexical words also contain grammatical information and, conversely, grammatical words also include some lexical meaning. In this view, both features of grammaticality and lexicality are gradual, and a certain form can be either more / less grammatical or more / less lexical (cf. e. g. Lehmann 2002; Hopper & Traugott 2003; Fischer 2008; Trousdale 2008a, 2008b). In fact, more recent narrower definitions of lexicalization strongly emphasize the fact that lexicalization must also be understood as a sequential process, thus excluding items such as acronyms, conversions, and metatalk from the array of lexicalized items, due to their abrupt entry into the lexicon. This definition that regards gradualness as a crucial feature decisively complicated the whole picture (cf. Lehmann 1989; Wischer 2000; Brinton & Traugott 2005). Certainly, though, gradualness of development is not the only feature shared between lexicalization and grammaticalization. According to Lehmann (e. g. 1989) and Giacalone Ramat (1998: 121), for example, both processes entail a reductive component, understood as loss of autonomy and univerbation. This theoretical chaos may be sorted out to some extent by highlighting the fact that the semantic developments undergone by grammaticalized and lexicalized items are in fact opposite. On the one hand, grammaticalization implies generalization, and consequent host-class expansion (i. e. grammaticalized items are promiscuous, and usually show high pattern and token productivity). On the other hand, lexicalization produces opacity (or fossilization, non-compositionality, idiosyncrasy), which results in host-class reduction (i. e. decrease in pattern productivity and possibly in token productivity) (cf. Brinton & Traugott 2005: 96-110; Haas 2007: 34). Thus, for example, the Old Irish perfective marker roand empty preverb noare grammaticalized, in that they fairly systematically occur in certain morphosyntactic contexts (cf. Chapter 7). By contrast, there is no general consensus as to whether the development of Slavic preverbs into “bounder perfectives” (in Bybee, Perkins & Pagliuca’s 1994 terms) can be regarded as an actual grammaticalization: first, there is not a single marker of perfectivity; second, specific preverbs allegedly perfectivize verbs belonging to specific semantic classes (cf. Chapter 6). 2.3. Describing the meanings and the functions of preverbs 2.3.1. Semantic Roles In this work, the spatial and non-spatial meanings of Vedic, Homeric Greek, Old Church Slavic, and Old Irish preverbs are described in terms of Semantic Roles (henceforth SR s), that is, roles played by the participants in an event (Kittilä, <?page no="56"?> 56 2. Theoretical background: Analytical tools for the study of preverbs Västi & Ylikoski 2011: 7). Only when SR s labels are regarded as being too generic, more specific definitions will also be provided. According to Haspelmath (1997) and Luraghi (2003), SR s are distinguished by their encoding strategies: when a SR is expressed in a specific way that it is not shared with related SR s, it is considered to be independent. Such choice is motivated by the fact that “it would be very difficult to base such a list [of SR s] on semantic criteria alone, because then there would be no way of constraining the possible proliferation of functions” (Haspelmath 1997: 10-11). SR s are also assumed to be organized in prototypical categories (cf. Luraghi 2003). Prototypicality assumes groups of entities to be representative for the categories in question; the “prototype” is the most representative entity for a certain category (Lakoff 1987). 10 For example, the features of the prototypical Agent are humanity, causation, intentionality, control, and saliency (Givón 1984: 107; Fillmore 1968: 24-25; Jackendoff 1972: 32; Andrews 1985: 68; Luraghi 2003: 30). However, it is far from clear that an Agent consistently displays all such features, as the following examples illustrate: (5) Luke ate an apple. (6) Luke accidentally broke a vase. Both in (5) and in (6), the Agent is canonically encoded by the subject ( Luke ), as usually happens in languages in which the nominative case (in the broad sense, cf. Haspelmath 2006) is highly grammaticalized, including Indo-European languages. However, the events described (5) and (6) are different: in (5), Luke intentionally causes a change of state to the apple . In (6), Luke similarly brings about a change of state to the vase ; however, the action of breaking is not intentional, as the occurrence of the adverb accidentally shows. Therefore, in (6), Luke is a non-prototypical Agent, in that it lacks the features of intentionality and control. in ancient Indo-European languages, both spatial and non-spatial relations are expressed by means of morphological cases, preverbs, and adpositions (cf. Luraghi & Narrog 2014: 1-22). Importantly, the relative semantic contributions brought about by cases, preverbs and adpositions in carrying out this function can undergo variation along the diachrony of a certain language (cf. the relevant sections in Chapters 4-7). Notably, such historical developments are consistent with the organization of meanings in prototypical categories proposed by Cognitive Grammar. In addition, Cognitive Grammar offers the insight 10 Prototype theory was at first developed by the cognitive psychologist Eleanor Rosch in the 1970s (cf. e. g. Rosch 1973, 1975, 1978). For the application of the concept of prototype in linguistics, see Lakoff (1977) and Taylor (1989). <?page no="57"?> 2.3. Describing the meanings and the functions of preverbs 57 that metaphor and metonymy are cognitive tools by which human beings can conceptualize non-basic domains in terms of more basic ones such as space (cf. Section 2.1). In this work, I limit my scope to providing a non-exhaustive catalogue of SR ’s and to associating each of them with a prototypical definition. Owing to constraints of time and space, it is not possible to discuss all Vedic, Homeric Greek, Old Chruch Slavic, and Old Irish SR s and their coding strategies here (see Chapters 4-7 and references therein). - Agent : a prototypical Agent is an entity that performs and causes an action. Furthermore, Agent prototypically exercises a force over another entity (i. e. Patient) and is characterized by intentionality and control (cf. above in this Section; Section 2.3.2.3). - Patient: prototypically, the SR of the entity that undergoes a change of state or location, performed by an Agent. It is usually coded by the accusative case in nominative-accusative languages. Its typical feature is a high degree of affectedness (cf. Section 2.3.2.3). - Recipient: this SR is played by an animate entity that receives another entity. It is usually taken by the third argument of some trivalent predicates, typically of verbs of ‘giving’ (transaction verbs). - Addressee: the SR played by an entity, most frequently a human being, which is the goal of a communication event. - Experiencer : the SR of the (animate) entity that experiences a physical or a psychological process triggered by another entity or event ( Stimulus ). The Experiencer is often coded as a subject (cf. Luke in (7)), but occasionally it can also be associated with direct objects, as me in (8). (7) Luke always gets my jokes. (8) Luke annoys me to no end. - Possessor : the SR of the (animate) entity that possesses another entity. In Indo-European languages, possession is expressed by different encoding strategies, including the genitive case, the external possessor construction (Luraghi forthc. a), whether it is in the double accusative case (Luraghi & Zanchi forthc. on Ancient Greek) or in the dative case. 11 - Beneficiary : this SR is taken by the (human) entity in favor of whom an action is carried out. A sub-type of Beneficiary is Malefactive, which is the human entity against which an action is performed. Another sub-type of Beneficiary 11 In external possession constructions, two NPs that are not in a relation of syntactic dependency indicate a possessor and a possessum (Luraghi forthc. a). <?page no="58"?> 58 2. Theoretical background: Analytical tools for the study of preverbs is Substitute, which relies on the idea of replacement: if x acts in y ’s behalf, x is conceived as acting in y ’s place. - Cause: the SR of the referent that originates an event. Such referent can be a natural force, an emotion, an abstract notion, another type of inanimate entity, a human being, a state of affairs or an event. Cause expressions are often grounded on Source expressions according to the metaphor causes are origins (Lakoff & Johnson 1980), on Time expressions (an event that precedes another event can be understood as the cause of the subsequent event), and on Area expressions (cf. Area below). - Purpose : this SR is taken by an entity, often a state of affairs, aimed by the intentional activity of an Agent. Such SR is often expressed through allative markers, or markers of Beneficiary or of Cause. - Instrument: this SR is prototypically taken by an inanimate and manipulated entity that is used by an Agent to carry out an action. Being an inanimate entity, Instrument differs from Agent in the properties of intentionality and control. - Comitative : this SR prototypically involves an animate Agent performing an action together with another animate individuated entity, which carries out the same action. - Location: the SR of an entity ( LM ) that gives information about the position or the spatial orientation of another entity ( TR ). - Goal : the SR of an entity ( LM ) that individuates a portion of space toward which another entity ( TR ) moves along a trajectory. - Source : the SR of an entity ( LM ) that individuates a portion of space away from which another entity ( TR ) moves along a trajectory. - Path : this SR describes the portion of space that a TR covers while performing a motion. - Time: this SR describes either the temporal placement or the Duration ( LM ) of a certain event (TR). Notably, the event conceived as a TR appears either to be located within / before / after or to last a given amount of time, conceived as a LM . Time cross-linguistically tends to be conceptualized in terms of space (Haspelmath 1997). These two conceptual domains are mapped by means of the metaphor time is space (Lakoff & Johnson 1980; Claudi & Heine 1986). - Manner : this SR refers to the way in which an action, a state of affairs, an experience or a process is brought about. - Area: the SR of the thematic context or field within which an event is seen; it specifies the extent to which the state of affairs denoted by the verb applies. - Location : the SR of an inanimate LM that gives information about the position or the spatial orientation of another entity or event ( TR ). <?page no="59"?> 2.3. Describing the meanings and the functions of preverbs 59 2.3.2. Aspect, actionality, and transitivity 2.3.2.1. Aspect and actionality: terminological and conceptual issues Since their introduction into linguistic terminology and discussion, the notions of “aspect” and “actionality” have long overlapped and been confused (cf. among others Viti 2008a; 2008b; Danesi 2009: 9 ff.). The term aspect is the translational equivalent for Russian vid, which was first employed in Greč’s (1827) grammar of Russian to describe the ability of certain affixes to derive perfective and imperfective verbs from the same root. Later on, Curtius (1846) extended the notion of aspect to Indo-European languages other than Slavic, specifically to the Ancient Greek verbal system. Curtius’ terminology comprises the terms Zeitart (literally) ‘temporal modality’, including past, present and future tenses, and Zeitstufe (literally) ‘temporal layer’, including notions such as durativity, ingressivity, and completion. Brugmann (1885) and Delbrück (1897) introduced the terms Aktion and Aktionsart (literally) ‘actional modality, actionality’ , distinguishing among punctual, iterative, durative, and terminative events. The term Aktionsart was also used by Streitberg (1900), who ascribed to this category five different values, including the perfective. In the early 20 th century, the term Aspekt also officially entered the linguistic discussion thanks to the slavist Sigurd Agrell. In his grammar of Polish (Agrell 1908), he distinguished the categories of Aspekt ‘aspect’, which refers to the completeness-incompleteness of the action, and of Aktionsart , which describes the way in which the action is carried out (e. g., according to Agrell, definitive, effective, augmentative, perdurative, and others). These two notions were later on kept distinct, for example by Jacobsohn (1926), who assigned a subjective character to Aspekt , but an objective value to Aktionsart (cf. Section 2.3.2.2) . Krause (1953) also separated the two notions in a similar manner. After the introduction of the concept of Aspekt early in the 20 th century, the two categories of Aspekt ‘aspect’ and Aktionsart ‘actionality’ continued to be confused, and scholars exhibited - and to some extents still do - little agreement on the definition, scope, and values of these terms (cf. Bertinetto 1986; Bybee, Perkins & Pagliuca 1994; Sasse 2002; Tatevosov 2002; Croft 2012; and Filip 2012 for a thorough discussion of this issue). The confusion was especially apparent in 20 th century discussions of the Ancient Greek verbal system, in particular when it came to the description of the dichotomy between aorist and present tenses, and to the role of preverbs (cf. Brugmann &Thumb 1913; Meillet 1922; Meillet &Vendryes 1924; Ruipérez 1954). Precisely because of this terminological fuzziness, perfectivizing value was ascribed to Ancient Greek (but also to Vedic) preverbs (cf. Chapters 4 and 5), based on the much more grammaticalized <?page no="60"?> 60 2. Theoretical background: Analytical tools for the study of preverbs perfective-imperfective opposition expressed through Slavic preverbs (i. e. “prefixes”, cf. Chapter 3 on terminology; Chapter 6 on Old Church Slavic). In recent decades, the so-called bidimensional approach to aspect, which clearly distinguishes Aspekt and Aktionsart , has been the most widely employed within the field of Indo-European studies (cf. Strunk 1994; García Ramón 2002; Sasse 2002; Napoli 2006; Dahl 2010; Cotticelli-Kurras 2015; by contrast, cf. Inglese 2016, who, in his analysis of Hittite lexical aspect, adopted Croft’s cognitive approach, on which see below). Within the bidimensional approach, on the one hand, aspect, or more precisely “grammatical aspect”, encodes the grammatical expression of speakers’ viewpoint on events. On the other hand, Aktionsart, or “lexical aspect”, or “actionality”, is usually regarded as an objective, or inherent, feature of verbal roots, which can be classified into different categories accordingly (cf. Vendler 1957; Section 2.3.2.2). Recently, the bidimensional approach has been brought into question within various theoretical frameworks, including Cognitive Grammar, which does not allow for any clear-cut distinction between the layers of lexicon and syntax (cf. Section 2.1). In Croft’s (2012) recent monograph, for example, grammatical and lexical aspects are not distinguished, as they both emerge from the interaction between the idiosyncratic facets of verbal semantics, and the specific semantics of the constructions in which verbs occur. In the following sections, I provide very brief definitions of both aspect and actionality as they are used in this work, and discuss their interplay (Section 2.3.2.2), as well as the interaction between actionality and transitivity (Section 2.3.2.3). 2.3.2.2. Aspect and actionality: definitions, values, and their interplay The category of grammatical aspect expresses the way in which speakers view an event, that is, speakers’ subjective perspective on the event, encoded through grammatical means (cf. Vendler 1957; Comrie 1976; Bertinetto 1986; Bybee, Perkins & Pagliuca 1994; Rothstein 2004). The notion of grammatical aspect is related to that of time, which expresses the relation between the moment of speaking and the time of the situation described. Present tense is seen as simultaneous to the moment of speaking, past tense as previous to the moment of speaking, and future tense as subsequent to the moment of speaking (Comrie 1976: 2). Thus, time is a deictic category, in that it is anchored to the spatiotemporal context created by the participants in a conversation (Lyons 1977: 637). In contrast, grammatical aspect has nothing to do with deixis, but rather interacts with speakers’ perspective in a different way: “aspects are different ways of viewing the internal temporal constituency of a situation” (Comrie 1976: 3). Comrie thus points out two major factors defining the category, specifically (a) speakers’ viewpoint, (b) the internal setting of a certain event itself (and not its temporal reference) (cf. also Bertinetto 1986: 76). <?page no="61"?> 2.3. Describing the meanings and the functions of preverbs 61 The two main values of aspect are what we call perfective ( John sang a song ) and imperfective ( John was singing a song ) aspects. On the one hand, perfective aspect implies a global, complete, and external viewpoint upon the event, including its starting point, its carrying out, and most importantly its endpoint. On the other hand, imperfective aspect presupposes an internal viewpoint upon the event, in which an open window is profiled only, without any additional information on its temporal boundaries (e. g. Comrie 1976; Bertinetto 1986). Among tenses, the present is usually regarded as inherently imperfective (i. e. ongoing), see Comrie (1976). 12 The main instantiations of the imperfective are the progressive ( John is singing ) and the habitual ( John sings every Wednesday night ) aspects. The two values are distinct, but can co-occur in the same sentence, as in John used to be writing poems (Comrie 1976: 33): any single occurrence of a certain situation contributes to building the progressive aspect, whereas the sum of all these occurrences is conceptualized as habitual. Other scholars identified further language-specific values for the imperfective, including the so-called “continue” (cf. Bertinetto 1986: 172 ff. on Italian), the “continuative” (cf. Bybee, Perkins & Pagliuca 1994: 127 on English), and the “non-progressive continuative” aspects (cf. Comrie 1976: 35 on English). In traditional descriptions, actionality or lexical aspect has a semantic nature, and is defined as the internal temporal structure of a state of affairs denoted by verbs, regardless of speakers’ view on the said state of affairs. As actionality is inherently associated with verbal roots, verbs denoting different states of affairs can be grouped together based on actionality. Building on Vendler’s (1957) traditional proposal displayed in Table 2, based on the actional traits of durativity, dynamicity, and homogeneity (or telicity), different scholars later refined his classification, taking into account several semantic criteria and syntactic tests (e. g. Bertinetto 1986; Botne 2003; Croft 2012; Bertinetto & Civardi 2015). 13 12 Comrie (1976) argues for a primary time distinction between present and past, to which the perfective-imperfective aspectual distinction is added. This view is not supported by Dahl (1985), who first distinguishes perfective and imperfective, and secondarily superimposes the layer of time distinctions. 13 The mentioned actional traits of homogeneity and telicity (called also “boundedness” in Jackendoff 1990 and “delimitedness” in Tenny 1994) overlap at the practical classificatory level. However, these labels profile two different actional properties of events: in homogeneity, the lack of a change of state, and in telicity, the lack of temporal boundaries. <?page no="62"?> 62 2. Theoretical background: Analytical tools for the study of preverbs d urative d ynamic h omogenous ( atelic ) e xample State + - + John stands still Activity + + + John walks Achievement - + - John heard the news Accomplishment + + - John ate the apple Tab. 2: Vendler’s (1957) actional classes Thus, the notions of aspect and actionality are well distinguished from one another at a theoretical level. However, when it comes to the empirical analysis, bidimensional approaches often encounter difficulties, which highlight the deep interplay of the two domains (cf. Tatevosov 2002). To begin with, depending on the construction in which they occur, verbs can instantiate different actional classes, a behavior that Bertinetto (1986) calls ‘aspectual hybridism’: (9) John sings ( activity ) vs. John sang a song ( accomplishment ). As shown in (9), the addition of an event participant can turn an atelic activity into a telic accomplishment (cf. also Dowty 1979: 61). For this reason, a number of authors suggested ascribing the trait of telicity to verbal phrases, and not simply to verbs, thus treating telicity as a compositional phenomenon (cf. Dowty 1979; Hinrichs 1985; Verkuyl 1972, 1993, 2005). Furthermore, specific features of event participants can also contribute to building actionality, though in principle it should be stored in the lexicon: (10) John sang a song ( accomplishment ) vs. John sang songs ( activity ). The examples in (10) show that the actional value is determined by the verbal objects: when they are numerable and specific, the actionality is telic; however, indefinite plurals and mass nouns are not able to change an activity into a telic predicate (in other languages, such as Hungarian, this distinction is expressed morphologically, via accusative-genitive alternation; cf. Heinämäki 1984). Finally, non-homogeneity and telicity are almost exclusively realized in perfective contexts (cf. the discussion on the “Slavic-style aspect” in Chapter 6). This is what Bertinetto (2001) calls the “telicity paradox”: (11) John drew a circle ( telic ) vs. John was drawing a circle ( atelic ). Overall, examples from (9) to (11) illustrate that the clear-cut distinction drawn by the proponents of bidimensional approaches between the categories of aspect and actionality is not easy to maintain, as it largely relies on non-trivial theo- <?page no="63"?> 2.3. Describing the meanings and the functions of preverbs 63 retical assumptions. First, the categorial distinction between aspect and actionality mirrors a more general separation between the grammar and the lexicon, which has been called into question by Cognitive Grammar and grammaticalization theory (cf. Sections 2.1 and 2.2). Second, though Vendler’s classification in Table 2 is often regarded as cross-linguistically valid, typological studies have shown that this is not always the case (cf. Tatevosov 2002; Botne 2003). 2.3.2.3. Telicity and transitivity The aspectual notion of telicity is connected with prototypical transitivity. Traditionally speaking, transitivity is understood as a global transfer of an activity from an Agent to a Patient, carrying out a number of effects on the Patient. Dating from the work by Hopper & Thompson (1980), this traditional notion of transitivity has been analyzed into the various parameters displayed in Table 3. The higher the degree reached by each parameter in Table 3, the higher or the more prototypical is the transitivity of a certain sentence. <?page no="64"?> 64 2. Theoretical background: Analytical tools for the study of preverbs t ransitivity high low ( a ) p articipants 2 or more participants, A and O 1 participant ( b ) k inesis action non-action ( c ) a spect ( actionality ) telic atelic ( d ) p unctuality punctual non-punctual ( e ) v olitionality volitional non-volitional ( f ) a ffirmation affirmative negative ( g ) m ode realis irrealis ( h ) a gency A high in potency A low in potency ( i ) a ffectedness of o O totally affected O not affected ( j ) i ndividuation of o O highly individuated O non-individuated Tab. 3: Hopper & Thompson’s parameters of transitivity (adapted from Hopper & Thompson 1980: 252) 14 Among such parameters, telicity (parameter (c), highlighted in grey in Table 3) also plays a role, which is explained in the following terms: “An action viewed from its endpoint, i. e. a telic action, is more effectively transferred to a Patient than one not provided with such an endpoint. In the telic sentence I ate up , the activity is viewed as completed, and the transferral is carried out in its entirety; but in the atelic I am eating it , the transferral is only partially carried out” (Hopper & Thompson 1980: 252). In addition, with telic predicates, given the entirety of transferral, the object is also strongly affected (parameter (i)), and individuated (parameter (j)). Both completeness and individuation in turn contribute to linking telicity and transitivity with topicality, as argued in Viti (2008a, 2008b). This linkage is crucial to account for the pragmatic properties of and the grammaticalization undergone by ancient Indo-European preverbs, which find a thorough general discussion in Chapter 3, language-specific treatments in Chapters from 4 to 7, and a summary in Chapter 8. 14 Hopper & Thompson (1980) employ Dixon’s (1979) terminology: O(bject) and A(gent) refer to the participant in the transitive event. (b) k inesis is the ability of certain events (e. g. actions) of being transferred from one participant to another. (d) p unctuality refers to the feature of some predicates that have no intermediate phase between inception and completion. Alongside with parameter (c), p unctuality can also be regarded as relating to actionality. <?page no="65"?> 3.1. Preverbs in Indo-European 65 3. Preverbs: an overview 3.1. Preverbs in Indo-European 3.1.1. Preverbs: definition and functions Preverbs are synchronically uninflected morphemes that occur in front of a verbal stem and modify its meaning, as shown in (1). 1 (1) Simplex verb Composite verb a. Ved. √ i- ‘go, walk’ pra √i- ‘come forth, go on, begin’ b. AG eîmi ‘come, go’ pró-eimi ‘go forward, advance’ c. OCS iti ‘go, come’ pro-iti ‘go through’ d. OI r. √icc- ‘reach’ ro-√icc- ‘come, attain, reach, succeed’ Examples (1)a-d contain the reflexes of the Proto-Indo-European local adverb *pr-ó ‘forward, forth’ ( LIPP II 636-637) occurring in front of a motion verb and modifying its meaning accordingly. 2 Vedic, Ancient Greek, and Old Church Slavic motion verbs of (1)a-d go back to the same PIE root, i. e. *h 1 ei̯- ‘go, walk’ ( IEW 293-296; LIV 2 232), whereas the Old Irish one is related to PIE *h 2 nek̂- ‘reach’ ( IEW 316-318; LIV 2 282). Though the notion of preverb was coined and is traditionally employed in the field of Indo-European studies (Booij & Van Kemenade 2003: 1; the term “preverb” itself is a calque from the post-Classical Latin lemma prae-verbium , attested in Varro, De lingua latina 6.38, and in Gellius 6.7.5), many non-Indo-European languages also exhibit an array of uninflected morphemes that have the same functions as Indo-European preverbs (cf. Section 3.3). The functions of Indo-European preverbs are described in what follows (cf. Bader 1997; Booij & Van Kemenade 2003). In the first place, preverbs operate at the semantic level: they modify the meaning of the verb onto which they attach in different ways. Primarily, they 1 Instead, at the level of reconstruction, IE preverbs most likely go back to petrified case forms (cf. in particular Section 8.2.1 with references). 2 Example (1)d shows one of the few Old Irish composites in which the preverb roretains its etymological spatial semantics. Usually, rofunctions as an aspectual marker in Old Irish (cf. Chapter 7). <?page no="66"?> 66 3. Preverbs: an overview provide a verbal root with spatial information: preverbs can give indication to location events, or specify the direction of motion. For example, the Ancient Greek and Old Irish simplex caused motion verbs phérō ‘bring, carry’ and beirid ‘carry, bring’ (< PIE * b h er- ‘carry, bring’; cf. IEW 128 ff.; LIV 2 76 ff.) can be combined with a number of preverbs, resulting in the different composites displayed in (2) and (3) (this list of composites is illustrative but not exhaustive for either language; the added preverb and its semantic contributions are highlighted in bold): (2) AG composites with phérō ‘bring, carry’ a. anaphérō ‘bring up, bring back, report’ b. apophérō ‘bring away, bring back, hand in’ c. diaphérō ‘carry in different ways, differ’ d. ekphérō ‘carry out of’ e. emphérō ‘carry in’ f. eisphérō ‘carry in(to), contribute, introduce’ g. epiphérō ‘bring upon’ h. kataphérō ‘bring down’ i. metaphérō ‘carry across, translate’ j. prophérō ‘bring forward, utter’ k. prosphérō ‘bring to, offer, pay’ l. huperphérō ‘carry over, be preeminent’ (3) OI r. composites with beirid ‘bring, carry’ a. ar· beir (literally) ‘before bring’ → ‘live, eat, use’ b. as· beir (literally) ‘out_of bring’ → ‘say, speak’ c. con· beir ‘bring together, conceive’ d. do· beir ‘bring to, give’ e. fo· beir ‘bring under, subdue’ f. for· beir (literally) ‘bring over’ → ‘grow, surpass’ g. fris· beir (literally) ‘bring against’ → ‘oppose, resist, obstruct’ h. imm· beir ‘carry around, put, employ’ Preverb-verb combinations can result in non-compositional (idiomatic or unpredictable) meanings: put another way, the meaning of the resulting composite verbs cannot always be inferred from the sum of the meanings of their elements. Among composites in (2) and (3), the following show non-compositional semantics: AG ana-phérō ‘report’, apo-phérō ‘hand in’, dia-phérō ‘differ’, meta-phérō ‘translate’ , pro-phérō ‘utter’, and huper-phérō ‘be preeminent’; OI r. ar·beir ‘live, eat, use’, as·beir ‘say, speak’, con·beir ‘conceive’, for·beir ‘grow, surpass’, imm·beir ‘employ’. Several similar non-compositional developments occur across different languages: for example, in both Ancient Greek and Old Irish, the root for <?page no="67"?> 3.1. Preverbs in Indo-European 67 ‘bring’, combined with different preverbs ( AG pro- ‘forth, forward’, OI r. ess- ‘out of ’), produces a communication verb (pro-phérō ‘utter’(2)j; as·beir ‘say, speak’ (3)b; cf. also Rus. pro-iz-nosit’ ‘forth-out_of-bring’ → ‘say, pronounce, utter’ that contains the Slavic equivalents for both preverbs, and a verbal base for bringing). Arguably, these similarities are byproducts of the lexicalization of the Container metaphor, according to which human body is conceptualized as a container (cf. Chapter 2). In a number of formations, preverbs develop lexical meanings beyond the etymological spatial ones: for example, the preverb apo- ‘away from’ basically expresses ablativity, but it comes to mean ‘back’ in apo-phérō ‘bring back’. The same preverb also gains more abstract meanings. For example, in combination with the communication verb eîpon ‘say’, apoexpresses refusal, such as in ap-eîpon ‘deny’ (< apo- ‘ away’ + eîpon ‘say’). Importantly, preverbs can also carry out more grammatical functions than those outlined above. To begin with, preverbs frequently modify the actionality or lexical aspect of the verb onto which they attach, from durative to punctual or from atelic to telic. This function has been ascribed to preverbs of Indo-European languages, including Vedic, Homeric Greek, Old Church Slavic, and Old Irish (cf. the relative discussions in Chapters 4-7). Relevant examples also come from other Indo-European languages, both ancient and modern, as is shown in (4). (4) The telic value of preverbs a. ‘drink’ ‘drink up’ Lat. bibo con-bibo Germ. trinken aus-trinken b. ‘eat’ ‘eat up’ Lat. edo com-edo 3 Germ. essen auf-essen In addition, preverbs can bring about other types of actional meanings such as iterative (e. g. AG ana-metrḗsthai ‘start again’), distributive ( OR po-jati ‘take for multiple times’), delimitative ( OR po-sěděti ‘sit for a while’), and ingressive (e. g. Ved. prá √an ‘start breathing’, ní √svap- ‘fall asleep’; AG hupoperkázō ‘begin inch by inch to assume a dark color’; OCS vъs-po-męnǫti ‘start remembering’; OR raz-bolětisja ‘fall ill’). In Slavic languages, it is noteworthy that preverbs not only modify the lexical aspect of verbs, but have developed into fully-fledged markers of grammatical 3 On the actional value of Latin preverbs, cf. Haverling 2003, 2008, 2010; Pompei 2010. <?page no="68"?> 68 3. Preverbs: an overview aspect: preverbs underwent grammaticalization into “bounder perfectives” (Bybee 1985; Dahl 1985; Bybee & Dahl 1989; Bybee, Perkins & Pagliuca 1994; cf. the thorough collection of references contained in Ruvoletto 2016: 8-33). Such a typologically unusual development and its motivations are discussed in Chapter 6: even in the most anciently attested variety of Slavic, Old Church Slavic, one can find evidence for the subsequent expansion of the so-called “Slavic-style aspect” (cf. also Eckhoff & Haug 2015; Wiemer & Seržant 2017). As is shown in Chapter 7, Old Irish preverbs also carry out grammatical functions: specifically, the preverb ro (< PIE *pr-ó ‘forward, forth’) is paradigmaticized to express perfectivity; the contentless preverb no (< PIE * nú ‘now’; LIPP II 577) behaves as a verbal auxiliary under certain morphosyntactic conditions. Since the publishing of the work of Hopper & Thompson (1980), it has been acknowledged that a linkage exists between telicity-perfectivity and transitivity: in particular, a high degree of telicity-perfectivity is implied by prototypical transitivity (cf. Section 2.3.2.2). Therefore, as preverbs can mark telicity and perfectivity, they are also candidates to function as so-called “applicative” markers. Applicatives are overt verbal morphemes that “allow the coding of a thematically peripheral argument or adjunct as a core-object argument” (Peterson 2007: 1). In (5), an example from Ainu (isolated, Hokkaido, Japan) is shown: (5) Applicative construction in Ainu (from Shibatani 1996: 159) a. poro cise ta horari big house in live b. poro cise e-horari big house app -live ‘He lives in a big house.’ The same state of affairs is denoted in (5)a-b: in the former example, Location is expressed through a postpositional phrase introduced by ta ‘in’; in the latter sentence" Location is promoted to direct object of the composite verb e-horari , which contains the crucial applicative prefix e-. A typological parallel has been drawn between Indo-European preverbs and applicatives, as Indo-European preverbs also seem able to promote an adjunct to argumental status (on Vedic, cf. Chapter 4; Danesi 2009; and reference therein; on Ancient Greek, cf. Chapter 5; Horrocks 1981; Viti 2008a; on Lithuanian, cf. Kozhanov 2016). In (6), it is shown that Lithuanian preverbs can function in the same way as the applicatives in (5): the simplex verb eiti ‘go’ is intransitive, and takes an optional PP per+ acc expressing Path in (6)a. By contrast, the <?page no="69"?> 3.1. Preverbs in Indo-European 69 same morpheme per occurs in front of eiti as a preverb in (6)b: accordingly, the Path-participant is promoted to direct object. (6) Applicative construction in Lithuanian (from Kozhanov 2016: 370) a. ei-ti per gatv-ę goinf across streetacc ‘to go across the street’ b. per-ei-ti gatv-e app -goinf streetacc ‘to cross the street’ The preverb / applicative perof Lithuanian is etymologically connected with the corresponding preposition per : they both go back to the same Proto-Indo-European adverb * pér ( LIPP II 607). The linkage just outlined between Lithuanian preverbs / applicatives and prepositions is common within the Indo-European languages, in which preverbs can usually also function as adpositions (that is, preand postpositions) and adverbs (cf. Sections 3.1.2 and 3.1.3). This linkage also displays typological parallels: the same morphemes that function as applicatives can also occur out of the preverbal context, as adpositions or adverbs, in non-Indo-European languages, including for example Ayoreo (Zamucoan, Paraguay, Bolivia; Ciucci 2014 with references). Switching to the diachronic plane, it is worth mentioning that adpositions constitute one of the possible sources for applicatives (cf. Seiter 1979 on Oceanic languages; Weir 1986 on Nadëb, Nadahup, Amazonas, Brazil; Craig & Hale 1988 on Algoquian languages; Peterson 2007: 125 ff.). This development is accounted for as a discourse-oriented grammaticalization: according to Craig & Hale (1988), what has triggered the reanalysis of an adposition as a preverb (i. e. applicative) is the null anaphora of the noun phrase taken by the preposition. Though a grammaticalization path triggered by null anaphora is difficult to hypothesize for Indo-European preverbs, due to the presence of a developed case system complicating the whole picture, a number of scholars also believe that pragmatic features, such as topicality, are essential to account for certain developments of Indo-European preverbs (cf. Sections 3.1.2 below and 3.1.3; Viti 2008a, 2008b). The assessment of the actual role played by Indo-European preverbs in promoting an adjunct to argumental status is complicated by the fact that Proto-Indo-European and many daughter languages possess the previously mentioned case systems. On the one hand, it is true that preverbs can centralize an adjunct, thus aligning themselves with applicatives. On the other hand, the centralized <?page no="70"?> 70 3. Preverbs: an overview adjunct frequently does not exhibit all the coding and behavioral properties of direct objects proper: often, it does not receive the usual coding and does not play the usual role for direct objects, i. e. an accusative case expressing the Patient (e. g. Viti 2008a); in addition, these putative direct objects cannot always be passivized (e. g. Kulikov 2012). Moreover, in Proto-Indo-European and in a number of ancient Indo-European languages, cases not only serve the grammar (i. e. by distinguishing subjects from objects), but also retain parts of their concrete meanings (cf. especially Kuryłowicz 1964: 179 ff. on the distinction between grammatical and concrete cases). Accordingly, in many contexts, preverbs are not strictly necessary to allow for an adpositionless second argument, though they certainly contribute to clarifying the meaning of verbs and adpositionless cases. For example, in Ancient Greek, the usage of the adpositionless genitive poliês halòs ‘gray sea’ is allowed both with a composite containing ex- ‘out of ’ (ex-ana-dúomai ‘emerge from’) and with a composite lacking it ( ana-dúomai ‘emerge’): (7) hathróai heudousin, poliês halòs close_together. nom . pl sleep. prs .3 pl grey. gen sea. gen ex-anadûsai from_emerge. ptcp . aor . nom . pl ‘(The seals) sleep close together, emerging from the grey sea.’ ( Od .4.405) (8) karpalímōs d’ anédu poliês halòs ēǘt’ omíkhlē quickly ptc rise. aor .3 sg grey. gen sea. gen like fog. nom ‘And (Thetis) rose from the grey sea like the fog.’ ( Il .1.359) In addition, the extension of the transitive construction seems not always to be triggered by the occurrence of preverbs, but rather by the own frequency of this construction. This is true, for example, for Ancient Greek (cf. Luraghi 2010). Horrocks (1981: 41), alleging the applicative usage of Ancient Greek preverbs, mentions the composite pros-eîpon, which can take an accusative argument ( tòn ‘him’ expressing the Addressee) in the sense of ‘address someone’, as in (9): (9) tòn d’ aûte pros-éeipe theà him. acc ptc then toward-say. aor .3 sg goddess. nom ‘Then the goddess addressed him.’ ( Il .1.206) <?page no="71"?> 3.1. Preverbs in Indo-European 71 Horrocks remarks that the simplex eîpon cannot take a different direct object from the cognate épos ‘word’ (or its non-cognate synonym, mûthon ) or the indefinite pronoun ti ‘something’. However, the example in (10) contests this claim: the simplex verb eîpon takes the accusative of the Addressee, that is, Héktora . (10) dḕ tóte Pouludámas thrasùn Héktora eîpe ptc then P . nom daring. acc H . acc say. aor .3 sg parastás stand_by. ptcp . aor . nom ‘Then Polydamas, standing by, addressed daring Hector.’ ( Il .12.210) In fact, it is undeniable that preverbs are connected with valency-related formations, as they are involved in making up reciprocal constructions in different ancient Indo-European languages, as shown in (11)a-c: (11) a. Hittite (from Inglese 2017) GAL lÚ.meŠ UŠ . BAR SÍG BABBAR SÍG SA 5 anda chief weaver( pl ) wool white wool red rec immiyazi mix. prs .3 sg ‘The chief of the weaver mixed together white (and) red wool.’ ( KUB 21.20 i 9-10) b. Vedic (from Kulikov 2007: 718) añjáte ví añjate sám anoint. prs .3 pl . mid rec anoint. prs .3 pl . mid together añjate krátuṃ rihanti anoint. prs .3 pl . mid mental_power. acc lick. prs .3 pl . mid mádhunā abhí añjate sweetness( n ). inst on anoint. prs -3 pl . mid ‘They (= waters) anoint themselves (with Soma), anoint each other (? ), mix together with each other (? ), lick (Soma’s) mental power, anoint themselves with (his) sweetness.’ ( R̥V 9. 86. 43 ab) 4 4 For the translation of difficult Vedic passages, besides relying on my own interpretations, I also consulted Jamison & Brereton (2014). <?page no="72"?> 72 3. Preverbs: an overview c. Ancient Greek (Inglese & Zanchi forthc.) amphì d’ ár’ autòn Trôes héponth’ around ptc ptc 3 sg . acc T . nom . pl follow. prs .3 pl . m / p ‘And around him the Trojans crowded.’ ( Il .11.473-474) (literally, ‘The Trojans followed him from all sides.’) d. Old Irish (from Dedio & Widmer 2017) ni-mu-n-ˈaccamar neg rec -1 pl -see. pst .1 pl ‘we have not seen one another’ (Wb.18d3) In (11)a from Hittite, the verb imiya- ‘mix’ and the preverb anda ‘in’ indicate an object-oriented spatial reciprocal situation. In (11)b from Vedic, a canonical intransitive reciprocal is expressed by means of the preverb ví ‘in two spaces, in two times, in different directions’. In the Homeric passage in (11)c, the reciprocal meaning emerges from the preverb-verb construction with amphí ‘around’, detached from the plural verb héponth’(i) ‘follow’. In the Old Irish passage in (11)d, a personal reciprocal construction is formed with the verb taking the plural verbal endings, and preceded by the preverb imm- , which in this context surfaces as -mu- . 5 Together with the semantic and syntactic functions outlined above, preverbs can also serve discourse-related purposes. Specifically, Boley (2004: 56-58) describes Hittite, Vedic, and Homeric preverbs (i. e. place words in her terms; cf. Section 3.1.4.1) as elements able to draw anaphoric reference, and thus to contribute to textual cohesion. The following example from the Odyssey is instructive in this respect: (12) en dé hoi askòn éthēke theà in ptc dem . dat skin. acc put. aor .3 sg goddess. nom mélanos oínoio tòn héteron, héteron d’ húdatos dark. gen wine. gen dem . acc one. acc other. acc ptc water. gen 5 According to Thurneysen (GOI 516-518) and O’Brien (1938: 242-244), the reciprocal marker imm-(a- N ) is identical with the lexical preverb imm- ‘around, about’. Though the two are undoubtedly etymologically related (GOI 517; LIPP II 36), Dedio & Widmer (2017) assume them to constitute synchronically two different lexemes. <?page no="73"?> 3.1. Preverbs in Indo-European 73 mégan, en dè kaì êïa big. acc in ptc and provisions_for_a_journey( pl ). acc kōrúkōi: en dé hoi ópsa títhei sack. dat in ptc dem . dat cooked. acc . pl put. impf .3 sg menoeikéa pollá satisfying. acc . pl many. acc . pl ‘On (the raft) the goddess put for him one wineskin of black wine, and a big one of water; then, on (the raft she put) grain in a wallet; on (the raft) she put many cooked delicacies.’ ( Od .5.265-267) In (12), no overt referent occurs accompanying the preverb en in clause-initial position (cf. Section 3.1.2), though one understands from the preceding context that the preverb en refers to a previously mentioned raft. Boley’s remark is supported by the quantitative analysis carried out by Viti (2008a) on Homeric poems: Viti showed that the majority of referents linked to preverbs are topical, and thus either previously mentioned in the discourse context, or known within speakers’ encyclopedic knowledge. The cohesive function of preverb repetition in Vedic and Homeric Greek has been investigated by Dunkel (1976, 1979), and Klein (e. g. 1987, 2007, 2008). A relevant Vedic example follows: in (13), the preverb ní ‘down’ is repeated three times in the stanza, specifically at the beginning of each verse. (13) ní grā́māso a-vikṣata down inhabitant( pl ). nom to-enter. aor .3 pl ní padvánto ní pakṣíṇaḥ down having_feet. nom . pl down winged. nom . pl ní śyenā́saś cid arthínaḥ down hawk. nom . pl even greedy. nom . pl ‘The villagers have gone into (their homes), into (their homes) the creatures with feet, into (their homes) those with wings, into (their homes) even the greedy hawks.’ ( R̥V 10.127.5) <?page no="74"?> 74 3. Preverbs: an overview 3.1.2. The positional properties of preverbs In Indo-European, the same morpheme that functions as a preverb can also occur outside the preverbal context as an adverb or as an adposition. Garde (2004: 104-105) lists a number of such morphemes from five Indo-European languages: Ancient Greek, Latin, German, Russian, and Lithuanian. To Garde’s language sample, I add Vedic, Old Church Slavic, and Old Irish, which are relevant for this work (see Whitney 1955[1879]: 396 ff.; and Renou 1952: 316 ff. for Vedic; Aitzetmüller: 1991: 154 ff.; and Lunt 1965: 82, 151 on Old Church Slavic; VKG II 242 ff.; and GOI 495 ff. on Old Irish). Following Garde, I mark the morphemes that can only function as prepositions differently from those that only occur in preverbal position: the former are assigned the label Prep, the latter Prev. Morphemes carrying both labels, Prep-Prev, also feature both functions. 6 (14) Preverb catalogue in a number of Indo-European languages a. Vedic (cf. Chapter 4) Prep-Prev - 17: áchā, áti, ádhi, ánu, ántar, ápa, ápi, abhí, áva, úpa, tirás, nís, pári, purás, prá, práti, sám Prep - 3: upári, parás, púrā Prev - 4: úd, ní, párā, ví b. Ancient Greek (cf. Chapter 5) Prep-Prev - 19: amphí, aná, antí, apó, diá, eis, ek, en, epí, katá, metá, pará, perí, pró, prós, sún, hupó, hupér c. Old Church Slavic (cf. Chapter 6) Prep-Prev - 16: bed ъ , do, iz( ъ ), na, nad ъ , o(b), ot( ъ ), po, pod ъ , prěd ъ , pri, s ъ , v ъ (n), v ъ z( ъ ), u, z а Prep - 1: k ъ Prev - 4: prě-, pr о -, raz-, vyd. Old Irish (cf. Chapter 7) Prep-Prev - 15: air, com, dí, ess, eter, fo, for, frith, íar, imb, in, re, sech, tar, tri Prep - 1: ó Prev - 6: ad-, aith-, ne-, uss-, ro-, toe. Latin Prep-Prev - 14: ab, ad, ante, cum / con-, de, ex, in, ob, per, prae, pro, sub, super, trans 6 For a full catalogue of Proto-Indo-European preverbs with cognates, see Beekes (1995: 247 ff.). In this respect, the indices collecting all particles of Indo-European languages contained in LIPP I are very useful as well. Since the boundaries of the category of preverbs are not clearcut, there is sometimes disagreement as to what items should be included in the preverb catalogue: for example, Vedic áchā ‘to, toward’ is included by Renou (1952), but not by Whitney (1955[1879]). <?page no="75"?> 3.1. Preverbs in Indo-European 75 Prep - 1: post Prev - 2: dis-, redf. German Prep-Prev - 15: ab, an, auf, bei, durch, in / ein-, mit, nach, über, um, unter, vor, wider, zu Prep - 5: für, gegen, ohne, seit, von Prev - 5: be-, ge-, ver-, zerg. Russian Prep-Prev - 15: v, do, za, iz, na, nad, o / ob, ot, pered / pred-, po, pod, pri, pro, s, u Prep - 3: bez, k, čerez Prev - 4: vz-, vy-, pere-, razh. Lithuanian Prep-Prev - 10: ant-, apie / ap-, į, iš, nuo, po / pa-, per, prie, su, už Prep - 4: arti, be, iki, tils PrevP - 1: at- In a number of ancient Indo-European languages - notably in Hittite, Vedic, Old Avestan, Homeric Greek, Early Latin, Archaic Old Irish, and Gothic - preverbs can be separated from the verb that they modify, despite retaining their semantic cohesion with it. This apparent split is usually called “tmesis” (< AG témnō ‘cut’), and is exemplified in (15): (15) Tmesis in Vedic (from Danesi 2013: 61) a. prá vām brahmāṇi kāravo bharante forth 2 du . dat prayer( n ). acc . pl poet. nom . pl bring. prs .3 pl . mid ‘To you the poets offer their prayers.’ ( R̥V 7.72.4b) b. devo devebhir ā́ gamat god. nom god. ins . pl to go. sbjv . aor .3 sg ‘The god has come with the gods.’ ( R̥V 1.1.5c) In (15)a, the preverb prá modifies the meaning of the root √bhr̥- ‘carry, bring’, resulting in ‘offer’. The two elements of the composite are separated from one another: prá is sentence-initial, whereas the main verb bharante is sentence-final. In (15)b, instead, the preverb ā́, which reverses the deictic orientation of √gam- ‘go to, approach’, shows up immediately in front of the verb gamat, but remains a constituent independent of it, as is proved by the accentuation. As demonstrated by Watkins (1963, 1964), in some ancient Indo-European languages, tmesis is not a literary artifact, but rather a testimony of the fuzzy categorial <?page no="76"?> 76 3. Preverbs: an overview status of a class of morphemes which can function as free adverbs, adpositions or preverbs (on tmesis, cf. further De Angelis 2004 and references therein). 7 Hence, “tmesis” (literally) ‘cutting apart’ is a misleading label assigned by ancient grammarians to the apparent split exemplified in (15)a. The papers by Watkins (1963, 1964) describe the following types of verbal tmesis: 8 1. # PN (E) …V(…)# (cf. (15)a). 9 This type of tmesis has already been described by Wackernagel (1924: 171 f.; translation mine): “From ancient times, tmesis mostly occurs if the preverb shows up at the beginning of the sentence.” 10 In addition to Vedic, Hittite, Homeric Greek, Archaic Latin, and Archaic Old Irish allow this pattern (Watkins 1964). 2. #N(E) … PV (…)# (cf. (15)b). This type of tmesis is difficult to detect, as the preverb immediately precedes the verb, though the two are only juxtaposed, and do not constitute a single word. This type of tmesis is also allowed in Hittite, Homeric Greek, and Archaic Old Irish. 3. #N(E) … VP (…)#, as in (16). In this passage, the preverb sám ‘with’ occurs in post-verbal position, and adds a telic nuance of meaning to the simplex verb. Watkins suspects this pattern of possibly being a purely poetic innovation of Vedic and Homeric Greek. This pattern is also called “reverse tmesis” or “verbal anastrophe” (cf. Petit 2017). (16) jáyema sáṃ yudhí spŕ̥dhaḥ defeat. opt .1 pl with battle. loc enemy. acc . pl ‘May we conquer our enemies completely in battle.’ ( R̥V 1.8.3c) Later on, preverbs underwent what Watkins descriptively labelled as univerbation (Watkins 1963, 1964). In other words, preverbs developed into bound morphemes, prefixes inseparable from the verb, occasionally with concomitant 7 Among the languages investigated by Watkins (1963), Old Irish emerges as being problematic: there is no general consensus as to whether tmesis attests to an inherited but residual clausal configuration, or if it is a mere literary artifact (cf. Chapter 7). 8 The concept of tmesis was also applied to PPs (“prepositional tmesis”, whereby a preposition is split from the noun that it takes by one lexical word), and to words other than verbs, including nouns or adverbs (“defusional tmesis”; this terminology was coined by Petit (2017)). 9 Watkins’ (1964) abbreviations follow: # = sentence border, N = sentence connective, E = (enclitic) pronominal element, V = verb form, P = preverb. 10 “Von alters her tritt Tmesis am ehesten ein, wenn das Präverbium zugleich an der Spitze des Satzes steht.” <?page no="77"?> 3.1. Preverbs in Indo-European 77 stress shifts, phonetic erosion, or even ultimate disappearance. Pinault (1995) represents the grammaticalization path covered by preverbs as in (17). (17) independent preverb > left member of a verbal compound > prefix > (zero) The path in (17) can be illustrated through Latin preverbs. Commenting on some passages from early Latin prayers, Cuzzolin (1995) and Vincent (1999) discuss the fact that, while in Early Latin preverbs must be analyzed as independent constituents, these morphemes later on became left members of verbal compounds, and eventually developed into (bound) prefixes. These stages can be discerned based on two remarks on the early prayers by the grammarian Festus (Cuzzolin 1995: 130; Vincent 1999: 1118): (18) Sub vos placo’, in precibus fere under 2 pl . acc reconcile. prs .1 sg in prayer. dat . pl mostly cum dicitur, significat id, quod when say. prs .3 sg . pass mean. prs .3 sg dem . acc rel . acc ‘supplico’ implore. prs .1 sg ‘When people say, mostly in prayers, sub vos placo , it means the same as supplico [‘implore’].’ (Fest. 402; ed. Lindsay 1913) (19) Ob vos sacro’, in quibusdam precationibus against 2 pl . acc devote. prs .1 sg in certain. dat . pl prayer. dat . pl est, pro ‘vos obsecro’, ut ‘sub be. prs .3 sg instead 2 pl . acc beseech. prs .1 sg as under vos placo’, pro ‘supplico’ 2 pl . acc reconcile. prs .1 sg instead implore. prs .1 sg ‘ Ob vos sacro in certain prayers stands for vos obsecro , just as sub vos placo stands for supplico .’ (Fest. 206; ed. Lindsay 1913) Festus points out that, though in the early varieties of Latin sub ‘under’ and ob ‘to, toward’ could occur outside the immediate preverbal context, these displaced preverbs already constituted single lexical units with the verbs placo ‘reconcile’ and sacro ‘consacrate’ . Festus’ remark finds an effective reformulation in Romagno (2004: 68): “univerbation is only the end point of a grammaticalization, whose semantic effects are far earlier visible.” Accordingly, Cuzzolin (1995: 133) <?page no="78"?> 78 3. Preverbs: an overview sets out to replace the misleading term “tmesis” with costituenza discontinua ‘discontinuous constituency’: in fact, there are no univerbated compounds proper to ‘cut apart’; rather, preverb-verb discontinuous and continuous combinations are on their way to actual univerbation. Even if displaced, preverbs can modify the semantics of simple verbs, adding spatial or aspectual information: the meaning of the resulting composites can no longer be considered strictly compositional. Going back to (18)-(19), how can the meanings of supplico ‘kneel down / humble one’s self, pray / beg humbly, beseech, implore, supplicate’ and obsecro ‘beseech, entreat, implore, supplicate, conjure’ be straightforwardly derived from sub + placo and ob+sacro ? Along the path of development in (17), preverbs can partially or entirely lose their local and concrete meanings (De Angelis & Gasbarra 2010: 153): for example, sub no longer means ‘under’ in supplico , just as ob no longer means ‘to, toward’ in obsecro. To be sure, univerbation does not necessarily imply semantic bleaching: both in Sanskrit and in Classical Greek, verbal composites can be fully univerbated but still retain spatial and compositional meanings. Thus, the so-called tmesis attests to a step within the grammaticalization process in (17), in which an independent morpheme has started gravitating toward a verbal stem and modifying it semantically, but does not yet constitute a bound morpheme with it. Examples from Vedic (15) and Latin (18)-(19) represent two different stages in the historical development of preverbs: on the one hand, we a have a tmesis in which full lexical word(s) intervene(s) in between the preverb-verb combination. On the other hand, the Latin composites are split only by an enclitic pronoun. Precisely considering the type of material (lexical vs. grammatical) that interrupts the preverb-verb combinations, Bertrand (2014) distinguishes “lexical” and “non-lexical tmeses” (“non-lexical tmesis” also being called “clitic interposition”). Arguably, lexical tmesis mirrors a less advanced stage along the continuum toward the univerbation and grammaticalization than that attested by non-lexical tmesis, as I explain below. One among the few generally accepted syntactic statements about Indo-European is Wackernagel’s law, claiming that enclitics originally occupied the second position in the sentence (Watkins 1964: 1036). Wackernagel’s formulation is based on the evidence from Greek, Indo-Iranian, and Latin; its existence is also evinced in Celtic, and, abundantly in Anatolian and Balto-Slavic. Arguably, in cases of non-lexical tmesis, composites are “split”, exactly because the tendency toward univerbation gives way to the stronger Wackernagel’s law. In this light, non-lexical tmesis is a weaker indicator of independent constituency than lexical tmesis and is allowed by a wider range of languages than those outlined above, including Hittite, Gothic, Old Irish, Lithuanian, and Ossetic for example: <?page no="79"?> 3.1. Preverbs in Indo-European 79 (20) a. Hittite (adapted from Petit 2017) Para=[m]a=aš pa-iz-zi p =and=3 sg . nom go. prs .3 sg ‘He goes forward.’ b. Gothic (adapted from Conforti 2014: 17) us=nu=gibit þo kaisaris kaisara back= ptc =pay. imp .2 pl dem . acc . pl . n C. gen . sg C. dat . sg jah þo gudis guda. conj dem . acc . pl . n G. gen . sg G. dat . sg ‘Pay back to Caesar what (is) Caesar’s, to God what (is) God’s.’(Lc. 20.25) c. Old Irish at-[t]at=chigestar a=dæ to-2sg=see. fut .3 sg . pass o = G. voc ‘You will be seen, o God.’ (Ml.59c12) d. Old Lithuanian / Non-standard Lithuanian (from Rosinas 1995: 10) ap-mi-šviesk akis up / on / to-1sg-illuminate eyes ‘illuminate my eyes (literally, the eyes on / to me)’ e. Ossetic (Digor dialect) (adapted from Petit 2017) Æra=sæ=farsta. p =3 pl . acc =ask. pst .3 sg ‘He asked them.’ In (20)a from Hittite, the interposition of the clitics =ma= ‘and’ and =aš ‘he’ occurs in between para- …paizzi ‘he goes forward’. In (20)b from Gothic, the enclitic connective =nu intervenes between the preverb usand the verb =gibit (for a thorough catalogue of the elements that can occur in tmesis position in Gothic, cf. Conforti 2014). In (20)c from Old Irish, a second person enclitic pronoun separates at- ‘to, toward’ from =chigestar ‘will be seen’. In (20)d from Old / Non-standard Lithuanian, a first person pronoun encoding the animate Goal-participant ‘cuts’ the composite (in Modern Standard Lithuanian, reflexive pronoun only can occur in tmesis position; cf. Chapter 6). In (20)e from Ossetic, the composite æra= …=farsta ‘he asked’ is split by the interposition of the clitic = sæ= ‘them’. <?page no="80"?> 80 3. Preverbs: an overview Composite verbs and movable preverbs of some modern Germanic languages, in particular West-Germanic, which have received attention in the recent literature (cf. e. g. Ackerman & Webelhuth 1998; McIntyre 2000, 2001a, 2001b, 2003; Dehé and Wanner 2001; Lüdeling 2001; Zeller 2001; Booij 2002a, 2002b; Dehé et al. 2002; Van Kemenade & Los 2003), seem to represent a different development. Van Kemenade & Los (2003: 79 ff.) call the productive preverb-verb combination in such languages “separable complex verb”. “Separable complex verbs in the present-day West-Germanic languages typically consist of a verbal base, and a non-verbal part, often but not always a ‘particle’” (Van Kemenade & Los 2003: 80). Such a prefix / particle is part of a separable complex verb that, though constituted by morphemes separable by syntactic processes, operates as a single lexical unit. Dutch and German, for example, feature two different word orders, specifically SVOP in main clauses (21)a, and SOV in subordinate clauses (21)b. By way of example, consider the Dutch verb op-bellen ‘call up’ in (21)a-b: (21) Word orders in Dutch: main vs. subordinate clauses a . Jan belt zijn moeder morgen op J. rings his mother tomorrow up ‘John will phone his mother tomorrow.’ b . Jan zegt dat hij morgen zijn moeder op-belt J. says that he tomorrow his mother up-rings ‘John says that he will phone his mother tomorrow.’ (adapted from Booij 1990) As Booij & Van Kemenade (2003: 6) remark, the separability of separable complex verbs also manifests itself in the position of the infinitival particle te, which occurs between the two elements of separable complex verbs (e. g. op te bellen ‘to call up’), as well as in the form of the perfect passive participle, with the prefix geplaced in between the preverb and the verbal stem ( op-ge-beld ‘called up’). Separable complex verbs also behave similarly in derivational morphology: for example, the genominalization of opbellen is op-ge-bel ‘phone call’, with the nominalizing prefix occurring between the preverb and the verbal base. 3.1.3. The origin of preverbs As discussed in Sections 3.1.1 and 3.1.2, in a number of Indo-European languages, the same set of morphemes can occur in adverbial, preverbal and adpositional (preand post-positional) position. From Kuryłowicz (1964: 171) onward, the <?page no="81"?> 3.1. Preverbs in Indo-European 81 relation among adverbs, preverbs and adpositions has been interpreted at the diachronic level as follows: “The fact that in the Indo-European languages many an indeclinable may function both as preverb and as preposition has been a sufficient reason for attributing to them an adverbial origin. Such an assumption fully accounts for their subsequent functional bifurcation. On the one hand, a group consisting of adverb + verb may develop into a compound […]. On the other hand, within a construction [(verb + adverb) + oblique case] a syntactic shift may entail a new articulation [verb + (adverb + oblique case)], i. e. (preposition + oblique case)” (italics are Kuryłowicz’s). Before Kuryłowicz, the adverbial origin of Indo-European preverbs was also posited by Brugmann (1890: 758 ff.), Meillet (1903: 158 ff., 193), Saussure (1922: 246 ff.), and Wackernagel (1924: 153 ff.). After Kuryłowicz, many scholars embraced his theory, including Chantraine (1953: 82 ff.), Baldi (1979), Krisch (1984: Ch. 3) Beekes (1995: 218), Rousseau (1995: 12), Hoenigswald (1998: 257), and Fortson (2004: 139). Thus, in the early stages of Indo-European, preverbs were independent constituents of an adverbial character. Later on, these adverbs started gravitating semantically toward a verb, or toward an inflected NP : adverbs either came to modify the meaning of a verb, or to specify the value of a case. After such adverb+verb or adverb+ NP combinations became conventionalized, adverbs began to lose their autonomy, and to be part of proper syntactic constituents with verbs or noun phrases: they underwent grammaticalization. The functional bifurcation of local adverbs into preverbs or adpositions is mirrored by their increasing obligatorification for the structure of the constituents in which they occur. From the adpositional side, such an obligatorification undergone by previously free-standing adverbs has been interpreted as a piece of evidence demonstrating the rise of configurational syntax among Indo-European languages (cf. Hewson & Bubenik 2006; Luraghi 2009, 2010, forthc.b, forthc.c; Reinöhl 2016, and references therein). The grammaticalization of adpositions can thus be framed within a larger picture, whereby non-configurational languages, exhibiting originally (i) free constituent order, (ii) discontinuous nominal expressions, and (iii) null anaphora (Hale 1983), such as ancient Indo-European languages, undergo a series of changes that bring about features of configurationality. In fact, a number of scholars (Meillet & Vendryes 1924: 520; Hewson & Bubenik 2006) regard the creation of adpositional phrases as crucial in creating configurationality in Indo-European. As even in the oldest Indo-European languages it is relatively unusual for these morphemes to display the purely adverbial function (cf. e. g. Dunkel 1976, 1980; Luraghi 2001), preverb+verb and adposition+ NP combinations have been reconstructed Proto-Indo-European itself. Specifically, in his pioneering works on Old Indo-Aryan and Proto-Indo-European word order, Delbrück (1878: <?page no="82"?> 82 3. Preverbs: an overview 13, 1888: 15-16) formulated the following rule: preverb / adposition usually precedes its verb, but follows its case (cf. further McCone 2006: 41 ff.). However, Delbrück’s observation on word order, whether correct or not (cf. fn. 15 in this Chapter, and Chapter 4 on Vedic), does not imply that either bound preverbs (i. e. prefixes) proper or adpositions proper (i. e. obligatory adpositions selecting the case that accompanies them) can be assumed for Proto-Indo-European (on the development of configurational syntax and the internal structure of Indo-European PP s, see Hewson & Bubenik 2006; Luraghi 2009, 2010, forthc.b, forthc.c; Reinöhl 2016, and references therein). Rather, in the oldest branches of Indo-European, the relationship that holds between preverbs+verbs and adpositions+ NP s can be described primarily as modification, rather than as specification or government (cf. further Boley 2004: 34; Chapters 4 and 5). In her work on tmesis in Hittite, Vedic and Homeric Greek, Boley (2004: 52) summarizes this view as follows: “the preverb / preposition [“place word” in her terminology] is clearly an addition to what we regard as a basic sentence.” By contrast, in most Indo-European languages less ancient than Vedic and Homeric Greek, including Old Church Slavic and Old Irish (Chapters 6 and 7), adpositions seem to already function as heads of the phrases in which they occur: their omission brings about agrammaticality and / or alterations in meaning. 3.1.4. Preverbs as a terminological and a typological problem 3.1.4.1. The terminological side of the issue As discussed in Sections 3.1.2 and 3.1.3, the same Proto-Indo-European morphemes that developed into preverbs also developed into adpositions in later languages. Additionally, these morphemes originally functioned as free constituents of an adverbial character, thus being able to modify whole sentences, nouns and verbs. Furthermore, preverbs encompass a wide range of meanings and functions, different from their basic contribution of adding spatial specifications to verbal stems (cf. Section 3.1.1). This multiplicity of functions is mirrored in a high variability in the terminology referring to these morphemes, as I discuss in what follows. Precisely because of such widely varying placements, meanings, and functions, Cuzzolin et al. (2006) consider these morphemes to constitute a problematic morphological category. Accordingly, the authors call the morphemes that belong to this category “adverbs-adpositions-preverbs” ( adv sadp sprev s). Equally descriptive labels are employed by Bolinger (1971) and Friedrich (1987): the former, in his work on English, uses the abbreviation “adprep” to combine their adverbial and prepositional function; the latter employs the same label to refer to these morphemes across the whole Indo-European language family. <?page no="83"?> 3.1. Preverbs in Indo-European 83 Friedrich (1987), in the heading of the same paper, explains what he means by “adprep”: in the subheading, within brackets, he adds the term “spatio-temporal auxiliaries”. The “spatio-temporal” part specifies the original semantics of these elements, whereas the “auxiliaries” part indicates that that they are usually added as adverbial modifiers to a predicate or a noun. The primary spatial value of Hittite, Vedic, and Homeric adv sadp sprev s also supports Boley’s (2004) choice, who calls these elements “place words”, as mentioned above . The fact that adv sadp sprev s originally functioned as modifiers, and not as syntactic heads, is also emphasized in Ivanov (1973) and in Friedrich (1976), who opt for “locative auxiliaries”. The label “satellites” also points out the fact that these elements functionally gravitate toward a verbal center as its modifiers. The term “satellite” is used in the literature on the typology of motion events as well (from Talmy 1983 onward), and occasionally in works on the Old Irish verbal complex (e. g. McCone 2006: vii titles a section of his monography “The Verb and its Satellites in Proto-Indo-European”). Patri (2007) also focuses on the modifier character of preverbs, by calling them determinant adverbial(s) ‘adverbial determiner(s)’. As Papke (2010: 4) shows, in works on Old Indo-Aryan, the morphemes under discussion are named in very different ways, arguably owing to the functional ambiguity that they exhibit in the oldest varieties of this branch: Präpositionen ‘prepositions’ (Delbrück 1888); “adverbial prepositions” (Macdonell 1910, 1916); “words of direction”, “elements of an adverbial character”, “the so-called prepositions”, “verbal prefixes” (Whitney 1955[1879]; Kulikov 2012); “preverbs”, or even “semi-autonomous verbal morphemes” (Kulikov 2012); and Lokalpartikeln ‘local particles’ in the various publications by Hettrich, Casaretto, and Scheider (cf. Chapter 4, Table 8 for exact references). Each among these terms points out a specific aspect of these morphemes: their ability to modify nouns ( Präpositionen , “adverbial prepositions”, “the so-called prepositions”), their etymological spatial meaning (“words of direction”), their original syntactic autonomy (“elements of an adverbial character”, “semi-autonomous verbal morphemes”), or their preverbal placement (“verbal prefixes”). Interestingly, no author describes these morphemes as “postpositions”, though the Indo-Aryan branch later on develops secondary postpositions (cf. fn. 15 in this Chapter; Reinhöl 2016; Chapter 4). In the Indian grammatical tradition (e.g. in Pāṇini’s Aṣṭādhyāyī ), the category of upa-sargas 'placed before' also includes the prefixes dusand su- , which express 'negativity' and 'positivity' respectively, and do not belong with preverbs proper. As Pompeo (2002) remarks, Chantraine (1953), in his Homeric grammar, tends to use the term prépositions ‘prepositions’, which is the one generally used in the literature on Ancient Greek. As Chantraine (1953: 82) already pointed out, the label pré-position , a calque from the Greek term pró-thesis ( pro- ‘in front of, <?page no="84"?> 84 3. Preverbs: an overview forth’+ a derivative from the root *d h eh 1 - ‘put’), is inadequate, as it is a cover term that comprises petits mots invariables ‘small uninflected words’, which specify the value of cases and are variably used as adverbs, preverbs or adpositions. Nevertheless, Chantraine only switches to the label particules ‘particles’ in the passages in which he discusses the adverbial origins of these morphemes (Chantraine 1953: 82 ff.). The term “particle” is often regarded as especially confusing because of its lack of precision. This issue is clearly discussed by Schourup (1999: 229): the term “is sometimes used to refer to elements of those traditional word classes that are uninflecting (‘invariable’), such as conjunctions, prepositions, interjections, and adverbs; at other times it is applied to all invariables except adverbs, conjunctions, and prepositions (see Hartmann 1994: 2953); more often, though, […it] is applied to items that do not fit easily into any well-established word class.” A number of authors, such as Hettrich and colleagues, try to remedy such inspecificity by adding the modification Lokal- ‘locative’ . Horrocks (1981) and Luraghi (2003) also opt for “local particle”, whenever they do not want to specify the part of speech of these small uninflected words. However, the term “local particles” can also be ambiguous: in Hittite, for example, this label is employed to describe a specific class of P2 clitics, which result from a grammaticalization process undergone by Proto-Indo-European adv sadp sprev s (cf. Luraghi 2001; fn. 15 in this Chapter). In works on Slavic, the term “verbal prefixes” is the most widespread, based on the fact that preverbs have shown the status of bound morphemes since the earliest attestations of this branch (i. e. tmesis is not attested; cf. Chapter 6). The term “prefixes” usually comprises more items than the adv sadp sprev s category, so as to include other types of preverbal morphemes such as the negation ne- . The same applies to the term “preverbs” in the literature on Old Irish, in which “preverb” is a catch-all term for all morphemes occurring in preverbal position, including the negation, and the paradigmaticized perfective marker ro- ( VGK II ; Vendryes 1923; Lewis & Pedersen 1961[1937]; grammaticalized rois instead called “augment” in McCone 1997, 2006). As their Ancient Greek counterparts, Old Irish adv sadp sprev s are frequently called “prepositions” as well (e. g. Pokorny 1914; GOI ). Preverbs also represent a terminological challenge because of the variety of functions that preverbal morphemes of different origin exhibit outside Indo-European. As Schultze-Berndt (2003: 145-146) points out, this issue surfaces in the literature on Northern Australian languages: along with the terms “preverb” and “verb” , various others are also employed by some authors, but none is generally accepted to date. Without elaboratiang upon the functions of preverbal morphemes in Northern Australian languages (cf. Section 3.3), it is sufficient to <?page no="85"?> 3.1. Preverbs in Indo-European 85 illustrate the range of terminological variation: Table 4 charts the selection of terms provided by Schultze-Berndt referring to the elements that constitute the Northern Australian verbal complex. u ninflecting element i nflencting element r eferences Preverb Verb Nash 1982, 1986; Simpson 1991 Verbal particle Verb / Auxiliary Hoddinott & Kofod 1976; Merlan 1994 Coverb Verb Kofod 1996; Wilson 1999; Carr 2000; Schultze-Berndt 2000, 2001 Uniflecting verb Inflecting verb McGregor 2002 Participle Finite verb Cook 1988 Base Auxiliary Capell 1979 (Main) verb Auxiliary Reid 1990; Walsh 1996 Tab. 4: Terms employed for the elements of Northern Australian complex verbs (adapted from Schultze-Berndt 2003: 146) In addition, while the label “preverbs” is broadly accepted as adequate to define the preverbal morphemes of Algonquian languages (cf. Section 3.3), such term is far from being precise. Accordingly, a number of scholars call all elements that appear before a verb preverbs, while others restrict the definition based on semantic, functional, or etymological criteria. 3.1.4.2. Terminology adopted in this work As this work deals with different ancient Indo-European languages, including Vedic, Homeric Greek, Old Church Slavic, and Old Irish, in which these morphemes exhibit quite different statuses and behaviors, terminological choices turned out to be particularly problematic: specifically, non-neutral terms such as “place words” or “prefixes” would be particularly inadequate. For example, “place words” as well as “local particles” are inadequate to name Slavic and Old Irish preverbs, which are quite advanced in their grammaticalization and lexicalization processes, and thus only infrequently show mere spatial semantics. Conversely, the label “prefixes” is adequate for Slavic, in which preverbs exhibit the morphological status of bound morphemes. However, this choice would be totally confusing for Vedic and Homeric Greek, in which preverbs partly retain their syntactic independence, and partially misleading for Old Irish, in which the outermost preverb holds instead a clitic status. <?page no="86"?> 86 3. Preverbs: an overview Therefore, I opted for what I regard as the most neutral terminological choices. The morphemes belonging to the category of adv sadp sprev s are called “preverbs” (also in tmesis position), unless clearly noun-oriented. If noun-oriented, these morphemes are variably called “adpositions”, “prepositions” or “postpositions”. Preverb-verb combinations of all languages are referred to as “composites”, rather than “compounds”, as the latter term usually implies a certain degree of univerbation. 11 In Vedic and Homeric Greek, however, it is not the case that these preverb-verb combinations actually represent single words in all contexts: thus, the term “composites” seems to me more appropriate due to its neutrality as to univerbation. 3.1.4.3. The typological side of the issue The terminological peculiarity relating to the category of adv sadp sprev s is discussed by Garde with an eye to its typological implications (Garde 2004: 103 ff.). Both adpositions and preverbs are named after their positional properties. 12 However, in the languages that feature both preverbs and prepositions, this terminological consistency conceals a typological oddity. Specifically, the category of “prepositions-preverbs” represents a two-fold paradox for word order typology. In the first place, the predominant phrase structure is reconstructed as centripetal (i. e. right-headed) for Indo-European (Garde 2004: 109). 13 However, prepositions defy this generalization: prepositions function as heads of the phrases to which they belong, but are nevertheless placed on their left; by contrast, exclusively centripetal languages such as Turkish usually allow for postpositions only (cf. Garde 2004: 111). In the second place, in the majority of centrifugal (i. e. left-headed) languages, prepositions and preverbs are both allowed. However, in centrifugal languages, preverbal morphology usually plays a far greater role than in Indo-European; for example, prefixation is usually employed for derivational purposes, e. g. in Indonesian, in which the prefix pederives deverbal nouns (cf. Teselkin & Aleva 1960: 18, 57-58; Garde 2004: 111). 11 The term “compound” seems to carry the idea of univerbation within the field of nominal composition as well. In fact, multi-word expressions such as Italian ferro da stiro ‘iron’ and the like tend not simply to be called “compounds”, but rather “phrasal or prepositional compounds” (or “phrasal lexemes” as in Masini 2009). 12 “La préposition se signale d’abord, parmi toutes les «parties du discours» des grammaires classiques, par une particularité terminologique: dès les origines, elle a été nommée uniquement par la place qu’elle occupe” (Garde 2004: 103 ff.). 13 The terms “centripetal” and “centrifugal” were coined by Tesnière (1959: 22). Though they sound a bit old-fashioned, I retain them in this Section, in order to conform to Garde’s (2004) terminology. <?page no="87"?> 3.2. Multiple preverbs 87 Furthermore, Garde (2004: 111) points out a paradox within the paradox: specifically, “the preposition and the preverb, though both preposed, exhibit divergent roles in the dependency relation, and accordingly in the order (i. e. centripetal or centrifugal) featured by the constituents in which they occur” (Garde 2004: 111, translation mine). 14 The preposition functions as a head of the PP (i. e. centrifugal order), whereas the preverb works as a verbal modifier in the composite verb (i. e. centripetal order). I suggest that such an apparent typological paradox can be resolved by keeping in mind the adverbial origin that prepositions and preverbs share (cf. Friedrich 1976; Section 3.1.3). In combination with both nouns and verbs, preverbs-adpositions started out as adverbial modifiers, occurring in front of the modified noun or verb. In this light, the word order featured by prepositions and preverbs is consistently centripetal (i. e. right-headed): the modifier precedes the modified. 15 Later on, these modifier-modified combinations underwent conventionalization, and accordingly the mutual relationships holding between these elements were arguably tightened until the eventual creation of a new continuous constituent or a bound composite. 3.2. Multiple preverbs Thus far, several key characteristics of preverbs have been discussed, including their functions, their positional properties, their origin, and their murky categorial and typological statuses, reflected in the correspondingly unclear terminology (cf. Section 3.1). Another critical issue requiring attention is the number of preverbs that can occur in front of a verbal stem. According to my definition, “multiple preverbs” occur every time two or more of such morphemes are placed in front of a verbal base. Among Indo-European 14 “[…] la PP et le PV, tous deux préposés, s’opposent par leur place dans le rapport de dépendance, et donc par l’ordre (centripète ou centrifuge) appliqué dans les syntagmes qui les contiennent. Dans le syntagme prépositionnel la PP est le régissant du nom (ordre centrifuge), et dans le verbe préverbé le PV le subordonné du verbe (ordre centripète) […].” 15 The fact that two of the oldest Indo-European branches, that is, Anatolian and Old Indic, are postpositional is not really an issue because in both subgroups, postpositions are arguably secondary formations partly going back to nominal forms (cf. Luraghi 2001 on Anatolian; Reinhöl 2016 and Chapter 4 on Indo-Aryan). In Anatolian and Indo-Aryan, the counterparts of Indo-European adv sadp sprev s only underwent grammaticalization as P2 clitics or as preverbs proper, respectively (not all scholars agree on the outlined grammaticalization of Anatolian adv sadp sprev s; for updated references, see Luraghi 2001; Melchert 2009). <?page no="88"?> 88 3. Preverbs: an overview languages, single preverbs are widespread, whereas multiple preverbs are not. However, Old Indic, Ancient Greek, Slavic and Old Irish, are noteworthy for allowing multiple preverbs. An example from each of the mentioned Indo-European branches is shown in (22); notably, all composites contain a Proto-Indo-European root for seeing or looking. 16 (22) Multiple preverb composites containing roots for ‘seeing, looking’ *ku̯ei̯s-,*spek̂-,*u̯ei̯d- ( LIV 2 381, 575, 665) Ved. ábhi ví √spaś- ‘look at, view’ ( R̥V 3, 62, 9; 10, 187, 4) AG eis-an-eîdon ‘look upward to’ ( Il. 16.232, 24.307) OCS pro-po-věděti ‘proclaim, predict’ (Mar. Mk. 1.38, 16.15 etc.) OI r. imm·accai ( imm-ad-√kwis-) ‘look after, examine, consider’ (Ml.114a15, 17b6) The composites of (22) can be schematized as follows: (23) P n [exterior] (…P 2 [medial]) P 1 [interior] V Multiple preverbs can be either different, as in (22), or identical, as in AG pro-pro-kulíndomai ‘keep rolling in front of ’ and OI r. as·éirig (ess-ess-√rig-) ‘rise again’. The preverb farthest from the verbal stem (P n ) is called “exterior preverb” ( EP ), whereas the closest to it “interior preverb” ( IP ). 17 All preverbs occurring between the EP and IP are “medial preverbs” ( MP s). For example, the Ancient Greek composite ex-apo-baínō ‘step out ofʼ contains ex- ‘out of ’, which is the EP , and apó- ‘away from’, which is the IP . In case of more than two preverbs, such as in the composite ex-up-an-ístēmi ‘start up from under’ , exis the EP , hupo- ‘under’ the MP , and an(a)is IP . The cumulation of preverbs exemplified in (22) and represented in (23), though possible, does not seem to reflect the usual phenomenon in old Indo-European languages (Kuryłowicz 1964: 174 ff.). In this respect, Old Irish constitutes an exception: three, four, and even five preverbs are occasionally put in front of a verbal base (cf. Chapter 7; Rossiter 2004; McCone 2006). Kuryłowicz accounts for the productivity of verbal composition in Old Irish as follows: in Old Irish, the EP is clearly separated from the rest of the composite by a clear-cut 16 Lexicalized composites containing multiple preverbs are also found in Latin: e. g. exponō ‘put out, set out’< *(ex-)po-znō < *(ex-)po-snō < *(ex-)po-sinō (Dunkel 1981b: 230 fn. 29; De Vaan 2008: 479). 17 I avoid the terms “external” and “internal” preverbs, as they are specifically used in the literature on Modern Slavic to distinguish two set of preverbs, featuring divergent semantic and behavioral properties. Multiple preverbs are very widespread in Modern Slavic languages, though they seem to constitute a quite different phenomenon from that investigated in this work (cf. Chapter 6). Cumulation of preverbs is also evidenced in the closest branch to Slavic, that is, Baltic (on Lithuanian, e. g. cf. Nevis & Joseph 1992: 96). <?page no="89"?> 3.3. Preverbs outside Indo-European 89 juncture, # EP n = MP 2 - IP 1 -V#, and has the status of a proclitic, instead of a prefix. Furthermore, the addition of a new preverb ( MP 3 ) prompts such a juncture: # EP = MP 3 - MP 2 - IP 1 -V#. Therefore, while in principle the other Indo-European languages including Vedic, Homeric Greek, and Old Church Slavic allow two interpretations for sequences such as # EP - IP -V#, namely # EP -[ IP -V]# and #[ EP - IP ]-V#, only the former interpretation is possible for Old Irish (on this issue, cf. also McCone 2006: 177-189). Thus, multiple preverbs in Old Irish do not belong to an ambiguous structure: the EP is always perceived as a modifier of the remaining verbal form taken as a whole (cf. also the concept of “recomposition” or “accretion”, introduced by McCone 1997, 2006, and in this work, mentioned in Chapter 1 and discussed in Chapter 7). This lack of ambiguity arguably favored the productivity of multiple preverbs in Old Irish. In addition, the earliest attestations of this language (4 th -5 th centuries onward) go back to a chronological phase, when composites were advanced in their lexicalization process: this means that a good number of lexicalized composites must already have been available for recomposition or accretion (cf. Chapter 7). 3.3. Preverbs outside Indo-European 3.3.1. Preverbs in Finno-Ugric and Caucasian languages In Eurasia, preverbs do not only occur in the Indo-European family: Finno-Ugric languages and the languages of the Caucasus also feature preverbs (concerning Finno-Ugric, cf. among others Metslang 2001 on Estonian; Wälchli 2001 on Livonian and Southern Finnic in general; Sgarioto 1999 with references, and Knittel 2015 on Hungarian; concerning Caucasian, cf. Harris 2003 on Georgian; Daniel & Rind-Pawlowski 2014 on Khinalug; Maisak 2014 on Agul). Hungarian preverbs are detachable preverbal morphemes with a primary spatial meaning (in (24)-(26) below, cf. át- ‘across’, el- ‘away’, and be- ‘into’), but, at a synchronic level, their main function is to modify the lexical or the grammatical aspect of the verbs onto which they attach (cf. also Kiefer 1994). Examples of Hungarian composites are provided in (24)-(27): (24) a. Péter át-men-t-ø2 a híd-on. P. across-gopst -3 sg .i art bridgesup ‘Peter went across the bridge.’ <?page no="90"?> 90 3. Preverbs: an overview b. Be-járta az erdő-t into-walk. pst .3 sg . d the forestacc ‘He roamed the forest.’ (25) Péter el-alud-t. P. away-sleeppst .3 sg .i ‘Peter fell asleep.’ (26) Péter be-csomagol-t-a az ajándék-ot. P. into-wrappst -3 sg . d art giftacc ‘Peter wrapped the gift.’ (27) Meg-főztem P-cook. pst .1 sg i ‘I cooked (the meal).’ (adapted from Knittel 2015: 1-2) Notably, in (24)a-b, the preverbs átand bebring about changes in the argument structure of simplex verbs (i. e. function as applicatives, cf. Section 3.1.1): they are responsible for the obligatorification of the locative arguments a hídon ‘the bridge’ and az erdő-t ‘the forest’ in the superessive and accusative case, respectively (Nyéki 1988: 144). In (25), the preverb elchanges the actionality of the verb: whereas alszik means ‘to sleep’, elalszik has the inchoative meaning of falling asleep. In contrast, the preverbs in (26) and (27) express telicity and perfectivity. From among the Caucasian languages, I take Georgian (Kartvelian, South Caucasian) preverbs as an example (cf. Harris 2003: 61-78 for a thorough description of their meaning and functions). Georgian preverbs synchronically exhibit a number of functions strikingly similar to those of their Indo-European counterparts: preverbs modify the meaning of a verbal stem by adding spatial or aspectual information. In addition, more than one preverb can attach onto a verbal base: for example, directional preverbs can occur in association with mo- ‘hither’, which is also a preverb, making up together a “complex preverb”. Such functional similarity is mirrored in the similarity of preverbs’ sources in these two families. In Georgian and sibling languages, a number of preverbs have relatively recently developed from earlier adverbs. Two-fold evidence confirms this claim: (i) the etymological sources of Kartvelian preverbs are related to adverbs; (ii) Old Georgian allows for tmesis, which suggests a residual independent constituency status for preverbs. In Modern Georgian, tmesis is no longer possible. <?page no="91"?> 3.3. Preverbs outside Indo-European 91 3.3.2. Preverbs in Amerindian languages Beyond the boundaries of Eurasia, preverbs show a wider range of uses than their Indo-European, Finno-Ugric, and Caucasian counterparts. This is the case with some Amerindian languages, particularly those belonging to the Algonquian family. Pentland (2005) provides an exhaustive description of the differences and similarities between Indo-European and Algonquian preverbs. In fact, for the most part, they have similar functions. Notably, as in Indo-European, Algonquian preverbs are also used as adpositions. However, as Pentland (2005) remarks, the Indo-European counterparts of Algonquian preverbs are generally limited to forms with spatio-temporal meanings. Instead, Algonquian preverbs display many other meanings, as shown through some composites from Severn Ojibwe (Ojibwe, Northern Ontario and Manitoba) in (28): (28) pimi-taacipo ʻcrawl alongʼ vs. taacipo ʻcrawlʼ nihta-nikamo ʻskilled at singingʼ vs. nikamo ʻsingʼ ishi-naakosi ʻlook like thisʼ vs. naakosi ʻbe visibleʼ (from Slavin 2006b) The preverb pimiis an adverbial element whose meaning is ‘along’. The preverb nihtaderives from a verbal root that means ‘do often or habitually’ and ‘be good at’. Nihtais one of the so-called “modal preverbs”, which refer to an agent’s inclination or ability to cause an event to happen (Slavin 2006a: 4). The preverb ishi- ‘in a certain time / place / manner’ belongs to the so-called class of “relative preverbs”. These preverbs that require an antecedent in the sentence, and “function to relate the verbal event to various associate circumstances, such as way in which it takes place, where it takes place, the reason it takes place, how long since it has taken place, how many times and how often” (Valentine 2001: 160). Research on another Amerindian language, specifically Rama (Chibchan, Nicaragua), leads Craig & Hale (1988) to introduce a new type of preverbs: the so-called “relational preverbs”. These diverge from the opposite group of “satellite preverbs” by virtue of their syntactic and ordering properties rather than because of their semantic properties (cf. Imbert & Grinevald 2004; Imbert 2008, 2009). “Satellite preverbs” are not syntactically linked to any particular argument in the sentence: instead, they are bound to the verb, and specify its lexical or semantic content. In a “satellite preverb”+verb combination, the verb functions as the syntactic head of the argument. Conversely, “relational preverbs” are elements that behave morphologically as preverbs, but syntactically as adpositions. In other words, relational preverbs select the argument taken by the verb (and determine its case). Precisely based on the evidence from “relational <?page no="92"?> 92 3. Preverbs: an overview preverbs”, Craig & Hale (1988) argue for the postpositional origin of preverbs in Rama: “relational preverbs” are the endpoint of a grammaticalization process starting from previous postpositions. 3.3.3. Preverbs in Northern Australian languages Preverbs occur in a number of languages of different genetic affiliation throughout Northern Australia (Schultze-Berndt 2003). In a large part of the linguistic area defined by the presence of preverbs, they form complex predicates that look strikingly similar to the separable complex verbs of Germanic languages. As Schultze-Berndt (2003: 145) points out, examples (29)a-b from Jaminjung (Australian, region around the Victoria River) have “straightforward translation equivalents in English”: (29) a. Jaminjung: a spatial preverb jag yirr-ijga-ny binka-bina down 1 pl . excl -gopst riverall ‘We went down to the river.’ b. Jaminjung: an actional preverb mangarra burrb nganthi-w-iya! plant_food finish 2 sg / 3 sg pot -eat. impf ‘You should have eaten up your food! ’ (adapted from Schultze-Berndt 2003: 145) However, while Germanic preverbs form a closed word class, Northern Australian preverbs constitute an open class comprising hundreds of members, including recent loans. The sources for preverbs are diverse: spatial adverbs (as in Indo-European), nouns, and verbal roots or stems. Owing to their open class status, Northern Australian preverbs cover a wider range of meanings than their Indo-European counterparts: in addition to expressing spatial path or completion, they are able to encode manner of motion, change of state, impact, very specific actions, and communicative events (for relevant examples and further discussion, see Schultze-Berndt 2003: 149-150). Interestingly, preverbs in Jaminjung and in all the neighboring languages, do not assign a deictic orientation to verbs, as their Indo-European counterparts do. In Northern-Australian languages, the abundance of preverb meanings just outlined also correlates with an even wider range of morphological and syntactic properties. With regard to morphology, Australian preverbs can consti- <?page no="93"?> 3.3. Preverbs outside Indo-European 93 tute the input for several types of derivations, including reduplication, lexical aspect-changing derivations, and nominalizations. With regard to syntax, preverbs usually occur immediately to the left of the verb, but they may also be separated from the verb by other constituents (tmesis position), and appear postverbally. As in Indo-European, multiple preverbs are occasionally found within a single clause. In addition, Northern Australian preverbs not only function as constituents of complex verbs but also show a wider range of syntactic functions, including that of main predicates in dependent clauses (cf. the socalled absolute usage of Vedic and Homeric preverbs, discussed in Chapters 4 and 5). Northern Australian preverbs can also affect the argument structure of composites. <?page no="95"?> 4.1. Preverbs in Vedic 95 4. Multiple preverbs in Vedic 4.1. Preverbs in Vedic 4.1.1. State of the art Traditional grammars of Vedic and Sanskrit usually include a section devoted to preverbs, which are either called “preverbs” or “prepositions”: this inconsistent terminological treatment mirrors their ambiguous categorial status (e. g. Delbrück 1888: 432 ff. uses the label Präpositionen ; Renou 1935: 109 ff. préverbes ; Whitney 1955[1879]: 414 “prepositions”; Macdonell 1916: 208, 210, 254, 266 “prepositions” and “verbal prefixes”; cf. Sections 3.1.4, 4.1.2.1-4.1.2.4). These grammars generally provide a list of preverbs with their basic usages as well as brief discussions on their phonetic, accentual, and positional properties. In addition, grammars demonstrate the usages of each preverb in combination with a number of verbal stems, as well as with morphological cases. Later investigations draw on the suggestions provided by traditional grammars to elaborate on specific features of Vedic preverbs. For example, these grammars generally observe that occasionally preverbs completely change the meaning of base verbs; i. e. preverb-verb combinations can result in non-compositional composites (e. g. Whitney 1955[1879]: 396; Thumb 1905: 444; Renou 1935: 145). Building upon this observation, Sturm (2014) uses a construction-based approach to study preverb-verb combinations whose meaning is no longer transparently the sum of their parts. Sturm’s goal is to identify recurrent patterns of idiomaticization for Vedic preverb-verb constructions. Renou (1935: 141) assigns actional values to Vedic preverbs: e. g. Renou describes abhí ‘to, unto, against’ and nís ‘out, forth’ as terminative, ā́ ‘to, unto, at’as resultative, and úpa ‘to, unto, toward’ as inchoative. 1 Delbrück (1897: 146-147) also notes that Vedic and Sanskrit preverbs show these functions, but refers to them as “perfectivizing functions”: he thus regards preverbs as able to modify the grammatical, rather than the lexical, aspect of verbs. Gonda (1962) reacts to Delbrück’s approach, remarking that the function of preverbs pertains to the lexicon, and not to the grammar. Danesi’s (2009) dissertation claims also proceed along these lines: she describes the meanings of Vedic preverbs in association with a selected list of roots that can be ascribed to different actional 1 The basic translations of preverbs are Whitney’s (1955[1879]: 396 f.). <?page no="96"?> 96 4. Multiple preverbs in Vedic classes (e. g. stative, activity, and resultative verbs). Danesi also addresses the issue of the valency-changing ability of Vedic preverbs (on which, cf. further Kulikov 2012). Delbrück (1888: 434-437) dedicates two sections of his grammar to combinations of two and three preverbs (the latter only attested in Classical Sanskrit). 2 Multiple preverbs are also the focus of Papke’s (2010) dissertation, which is a diachronic investigation of these constructions from Vedic to Classical Sanskrit. Papke’s aim is to understand the motivations for Sanskrit preverb ordering. To meet her goal, Papke compares the preverb ordering of Classical Sanskrit with that of Vedic and of other ancient Indo-European languages, including Homeric Greek and Old Irish (note, however, that her Homeric and Old Irish data are taken secondarily from Imbert 2008 and McCone 1997, respectively). A systematic investigation of almost all Vedic preverbs is offered in a series of papers by Hettrich, Casaretto, and Schneider, which have been published as outcomes of a project focused on local particles in the R̥g-Veda (Casaretto 2010a, 2010b, 2011a, 2011b, 2011c, 2011d, 2012, 2013; Hettrich 1991, 1993, 2002; Hettrich et al. 2010; Schneider 2009, 2010a, 2010b, 2011, 2012, 2013a, 2013b, 2013c). 3 Apart from the pioneering papers by Hettrich (1991, 1993, 2002), all cited publications are consistently organized according to one single compositional scheme. After providing quantitative data on the frequency of each preverb, a section follows on the semantics of local particles in their adnominal and adverbial functions. Then, the authors describe the syntactic positioning of preverbs as well as their obligatoriness for the syntax of the Vedic sentence. Subsequently, the semantic modifications brought about by the local particle in its preverbal usage are examined. Next, issues concerning the interaction of the local particle with similar morphemes and of the iteration of particles are addressed. A brief discussion on the etymology of the preverb closes each paper. In addition to the systematic works mentioned above, many other papers have focused on different aspects of Vedic preverbs, related to their syntax, diachrony, semantics, and etymology. For example, Kulikov (2012) investigates the (in)transitivizing potential of Vedic preverbs. Another series of papers, including Renou (1956), Watkins (1964), Sathe (1993), and Pinault (1995), are concerned with the positional properties of preverbs, and occasionally employ the general framework of word order typology (e. g. Andersen 1979). Dunkel (1981a) and Klein (2003) concentrate on preverb iteration, whereas Klein (e. g. 1987, 2008, 2 Thus, in Vedic, what I call “multiple prevers” are actually sequences of two preverbs only. 3 One can find the complete list of publications pertaining to the project at the following link: https: / / www.phil.uni-wuerzburg.de/ fileadmin/ 04080400/ Projekt_Publikationen.pdf. <?page no="97"?> 4.1. Preverbs in Vedic 97 2012) deals with the pragmatic function of verse-initial preverb repetition within the Vedic discourse. Danesi (2013) investigates the grammaticalization of Vedic preverbs through the case study of ápa ‘away, forth, off’: in particular, she shows that preverb-verb constructions constitute a semantic unit, though both the preverb and the verb retain much of their syntactic independence from one another. Casaretto & Schneider (2015) is a more general study on the grammaticalization of Vedic local particles into adpositions and preverbs. An even broader view on the grammaticalization of postpositions and the rise of configurational syntax in Indo-Aryan is offered by Reinöhl (2016), whose results are summarized in Section 4.1.2.7. A number of etymological studies on Vedic preverbs also appeared in the literature: for example, Dunkel (1982a; LIPP II , and references therein) who deals with the origin of Vedic ā́ ‘to, unto, at’, and Gonda (1968) who investigates the relatedness of the different usages of the preverb / particle ápi ‘unto, close upon, on’/ ‘and, too, moreover, also’. 4.1.2. The categorial status of Vedic preverbs In Vedic, preverbs display a high degree of syntactic and functional ambiguity, variously appearing as adverbs, preverbs, and even adpositions. Such ambiguity leads Renou to express skepticism as to the possibility, and even usefulness, of disambiguating the function of preverbs in the Vedic sentence: “it [i. e. assigning a clear-cut categorial function to preverbs] is a superimposition of our minds facing a linguistic structure that is virtually ambiguous” (Renou 1956: 119, translation mine) (cf. Section 2.2.3.2 on the non-existence of theoretically given linguistic categories). 4 This ambiguity is further explored in the next subsections, both from the standpoint of syntactic placement (Sections 4.1.2.1, 4.1.2.3, 4.1.2.4), and from that of accentuation (Section 4.1.2.5). Further, I briefly touch upon the issues of the interaction between Vedic meter and preverb placement (Section 4.1.2.6), and of the grammaticalization of Indo-Aryan postpositions (Section 4.1.2.7). 4.1.2.1. The free positioning of Vedic preverbs Differing from Classical Sanskrit preverbs, which are univerbated to the modified verb, R̥g-Vedic preverbs still exhibit wide syntactic freedom and are not always univerbated to verbal stems. In particular, Vedic preverbs allow for the 4 “[…] qu’elle est une surérogation de notre esprit devant une structure linguistique qui était fondamentalement ambiguë.” <?page no="98"?> 98 4. Multiple preverbs in Vedic following placements (extensively described, among many others, by Delbrück 1888: 44-46; Macdonell 1910; Renou 1935; Watkins 1963: 1037; Pinault 1995). (a) Tmesis position: # P N(E) …V(…)# . 5 preverb(s) is(are) displaced from the verb, which occurs in sentence final position, and fronted to the beginning of the clause (1). 6 In case of multiple preverbs, one is sentence initial, whereas the other may occur in various positions: rarely, it occurs at the beginning of the clause together with the former preverb (# PP …V#); more frequently, it appears immediately in front of the verb (#P…PV#); occasionally, it occurs in front of the verb, but not immediately so (#P…P…V#; cf. (2)). (1) prá nū́ mahitváṃ vr̥ṣabhásya vocaṃ forward now greatness( f ).acc bull. gen say. inj . aor .1 sg ‘Now I proclaim the greatness of the bull [=Indra].’ ( R̥V 1.59.6a) (2) prá vām máhi dyávī abhí úpastutim forward 2 du . dat great. voc . n sky. voc . du to praise. acc bharāmahe bring. prs .1 pl . mid ʻGreat (Earth,) Sky, we bring our praise forward to you two.ʼ ( R̥V 4.56.5ab) 7 For multiple preverbs, Papke (2010: 84-85) provides the number of tokens and the percentages for each position described above (cf. Table 5). 8 (b) Postverbal position: #N(E) …VP(…)#. Preverb(s) occur(s) immediately or non-immediately after the verb that they modify (3). In case of multiple preverbs, only one usually occurs after the verb, whereas the other occurs imme- 5 Cf. fn. 9, Chapter 3. These abbreviations also apply to the discussion of examples (21) and (22) in this Chapter. 6 The text of the R̥ g Veda is the online version of the metrically restored text published by van Nooten & Holland (1994). 7 Example (2) displays problematic nominal syntax and irregularities: in particular, dyávī must be the only example of a neuter form made from the stem dyu- . 8 The total number of Papke’s occurrences was gathered from 387 Vedic passages. Papke’s (2010) data are taken from a manual search through Grassmannʼs Wörterbuch zum Rig-Veda (1936[1873]). This is possible, as Grassmann’s dictionary provides, for each Vedic verbal root, the catalogue of preverbs or multiple preverbs that can modify it. Thus, Papke’s data rely on Grassmann’s judgements as to whether a specific preverb has nominal or verbal orientation. However, this matter is very controversial, and was not systematically debated until the series of papers by Hettrich and colleagues, as explained in Section 4.1.2.4. Therefore, Grassmann’s evaluations, though valuable, must be taken with caution. <?page no="99"?> 4.1. Preverbs in Vedic 99 diately before it (# …PV(…)P…#; cf. (4)) or is separated from the verb by one or more words (# …P…V(…)P…#). (3) jáyema sáṃ yudhí spŕ̥dhaḥ conquer. opt .1 pl together battle( f ). loc enemy( f ). acc . pl ‘May we conquer our enemies completely in battle.’ ( R̥V 1.8.3c) (4) utá śúṣṇasya dhr̥ṣṇuyā́ prá mr̥kṣo abhí védanam and S. gen bodly forward anoint. inj . aor .2 sg to property. acc ʻAnd you boldly grab the property of Śuṣṇa.ʼ ( R̥V 4.30.13ab) Papke’s (2010: 85-89) dissertation gives the number of tokens and the percentages for each position involving at least one preverb in postverbal placement. As the following quantitative data show, the orders whereby both preverbs occur after the verb are extremely rare (each is < 1 %) (cf. Table 5). (c) Immediate preverbal position without actual univerbation: #N(E) … PV(…)#. The preverb (5) or preverbs (6) directly occur(s) in front of the verb that it(they) modifies(modify). The occurrence or absence of univerbation depends on various factors, including the type of clause (main vs. subordinate) and the type of verbal form (finite vs. non-finite) (cf. Section 4.1.2.5). (5) índrasya nú vīriā̀ṇi prá vocaṃ I . gen now manly_deed( n ). acc . pl forward tell. inj . aor .1 sg ‘Now I proclaim the manly deeds of Indra.’ ( R̥V 1.32.1a) (6) tuvā́m abhí prá ṇonumo jétāram 2 sg . acc to forward roar. intens . prs .1 pl conqueror. acc áparājitam unconquered. acc ‘We keep crying out to you, the unconquered conqueror.’ ( R̥V 1.11.2cd) As Papke (2010: 84) points out, the order in (6), with two immediately preverbal preverbs, is the most common one (153 occurrences out of 387 (40 %)). Overall, the quantitative data provided by Papke (2010) show that multiple preverbs tend to occur in immediate or non-immediate preverbal position. By contrast, the postverbal position is usually selected by only one preverb, and only rarely by both of them. <?page no="100"?> 100 4. Multiple preverbs in Vedic Papke’s data are summarized in Table 5 and in Table 6. Each table provides the reader with a different viewpoint on the same Vedic data: Table 5 focuses on preverbs’ proximity to the verb, whereas Table 6 on preverbs’ relative positioning with respect to the verb. p ositioning of preverbs f requency (a) Tmesis position # PP …V# #P … PV # #P …P …V# 172 (45 %) 7 (2 %) 127 (33 %) 38 (10 %) (b) Postverbal position # … PVP …# # … PV …P …# # …P … VP …# # …P …V …P …# # …V … PP …# # … VP …P …# # … VPP …# 62 (15 %) 21 (5 %) 15 (4 %) 12 (3 %) 11 (3 %) 1 (< 1 %) 1 (< 1 %) 1 (< 1 %) (c) Immediate preverbal position # … PPV # 153 (40 %) t otal 387 (100 %) Tab. 5: The positioning of Vedic multiple preverbs ((i) relative proximity to the verb) p ositioning of preverbs f requency Both preverbs before the verb One preverb after the verb Both preverbs after the verb 325 (84 %) 59 (15 %) 3 (1 %) t otal 387 Tab. 6: The positioning of Vedic multiple preverbs ((ii) relative positioning with respect to the verb) (adapted from Papke 2010: 84-89) 4.1.2.2. Vedic composites: syntactic or lexical units? In Vedic, preverb-verb combinations still seem to operate at the syntactic rather than the lexical level: verbal composition is a quite productive process within Vedic verbal system. As Whitney (1955[1879]: 395) highlights, in Vedic, virtually <?page no="101"?> 4.1. Preverbs in Vedic 101 every verbal root combines with preverbs. Delbrück (1888: 433) provides a short catalogue of verbal roots that are never attested in combinations with preverbs. 9 Moreover, as Danesi (2013: 62) points out, the free positioning of preverbs described in Section 4.1.2.1 has no effect on the whole meaning of the composite. The comparison between (1) and (5) is instructive in this respect: both examples contain the root √ vac- ‘speak’ in combination with the preverb prá ‘forward, onward, forth, fore-’, resulting in the meaning ‘announce, proclaim’ (‘fore-tell, tell openly’ → ‘proclaim’). In both passages, the poet is telling, in a solemn way, about the heroic actions of god Indra. As Pinault (1995: 47, translation mine) puts it, “[…] the semantic modification undergone by the verb as the result of the combination with a preverb does not depend on the immediate proximity of the preverb and the verb” (cf. also Romagno’s 2004: 68 remark, at Section 3.1.2). 10 Thus, semantic shifts to the metaphorical plane do not require univerbation. This also applies to more lexicalized composites than prá √vac- ‘proclaim’, such as adhi √i- ‘go above’ → ‘approach mentally’ → ‘be aware’: in example (8) below, the preverb immediately precedes the verb, but the same composite with the same meaning can also occur with adhi ‘above’ in tmesis position (e. g. R̥ V 10.33.7ab). Correspondingly, univerbation does not imply either semantic shift from the basic spatial meaning or lexicalization, as shown in (7): (7) ápāhan [ápa-ahan] vr̥trám paridhíṃ nadī́nām away-strike. impf .3 sg V. acc encloser. acc river( f ). gen . pl ‘He has smashed away Vr̥tra, the encloser of the rivers.’ ( R̥V 3.33.6b) In (7), the preverb ápa ‘away, forth, off’ and the verb ahan ( impf .3 sg ) ‘has stricken’ occur together in initial position. Nevertheless, they do not make up a non-compositional composite: the semantic contributions brought about by both elements are still recognizable. Accordingly, Casaretto & Schneider (2015: 9 Delbrück’s (1888: 433) list of non-preverbed roots consists of √ īś - ‘be master’, √ kari - ‘praise, commemorate’, √ krudh - ‘be angry’, √ gras - ‘devour’, √trā - ‘rescue’, √ tviṣ - ‘be stirred’, √ dakṣ - ‘be able’, √ dhraj - ‘sweep’, √ dhvan - ‘sound’, √ dhvar - ‘injure’, √ nāth - ‘seek aid’, √ niṃs - ‘kiss’, √ bhand - ‘be greeted with praise’, √ bharv - ‘chew’, √ bhām - ‘be angry’ (probably a non-existent root extracted from a denominative, see EWAia II 261), √ bhikṣ - ‘beg’, √ bhrī - ‘hurt’, √ bhreṣ - ‘totter’, √ raṃh - ‘hasten’, √ ran - ‘rejoice’, √ rup - ‘feel pain (especially of stomach pain)’, √ vrādh - ‘be great’, √ śad - ‘fall’, √ śami - ‘be quiet’, √ sas - ‘sleep’, √ sparh - ‘be eager’, √ sridh - ‘blunder’, √ sriv - ‘fail (of miscarriage)’, and √ hrī - ‘be ashamed’. 10 “[…] la modification sémantique du verbe par le préverbe ne dépend pas de la proximité immédiate du préverbe et du verbe.” <?page no="102"?> 102 4. Multiple preverbs in Vedic 232) call combinations of this type ‘syntactic compounds’, as they are still analyzable and do not result in new lexical entries. 11 The evidence provided so far suggests that Vedic composites are better analyzed as syntactic units, rather than as lexical units. By contrast, however, there are also composites whose meanings are non-compositional (i. e. idiomatic; cf. Section 3.1.4.2 on terminology). Non-compositionality points toward lexicalization and irregular semantic shifts that blur connection with the base verb. One such composite is shown in (8): (8) yádi stutásya maruto adhīthá [adhi-ithá] if praise. gen M. voc . pl above-go. prs .2 pl ‘If you, O Maruts, are aware of the praise …’ ( R̥V 7.56.15a) In (8), the combination of adhi- ‘above, over, on, onto (literally, ‘from under’)’ and ithá (go: prs .2 pl ) ‘go’ results in the unpredictable meaning of being aware. 12 Lexicalized composites of this kind are by no means infrequent in Vedic (cf. Sturm 2014). Alongside with the lexicalization, the incipient grammaticalization undergone by a number of Vedic preverbs also indicates closeness to the verb. A number of Vedic preverbs, such as prá ‘forward, onward, forth, fore-’ and sám ‘along, with, together’, can modify lexical aspect, bringing about notions such as those of ingressivity or completion (Renou 1935; Gonda 1962; Danesi 2009; Casaretto & Schneider 2015: 246 ff.). In particular, prá occasionally develops ingressive meanings (Delbrück 1888: 460; Renou 1935: 144; Gonda 1962: 232-241): in (9), the composite prá √imeans ‘begin, start’, whereas the simplex verb √imeans ‘go’ ( prá ‘forward, onward’ → ‘at the beginning of ’). 13 (9) pra-yatí yajñé asmín forward-go. ptcp . prs . loc sacrifice. loc dem . loc ‘… during this sacrifice which begins.’ (R̥V 3.29.16a) 11 On the difference between syntactic composition, tmesis, univerbation, on the one hand, and lexicalization, on the other hand, cf. Hettrich (2007: B.b.IV.2), Pinault (1995: 42-43), and Watkins (1963). See also Section 2.2.5. 12 The fact that the composite exhibits an unpredictable meaning does not imply that the semantic shift undergone by adhi-√i- ‘above-go’ → ‘approach mentally’ → ‘be aware’ cannot be retrospectively explained. The composite might describe the metaphorical movement of a metaphorical TR, i. e. Maruts’ mind, toward a metaphorical LM, i. e. the praise. It is not infrequent that preverbs having the basic spatial meaning of ‘over, above’ come to introduce the semantic role of Area (cf. AG hupér ; Engl. over ; Germ. über ). 13 On this passage, see also Delbrück (1888: 461), and Danesi (2013: 66). <?page no="103"?> 4.1. Preverbs in Vedic 103 Instead, sám ‘with, together’ often expresses completion, even displaced from the verb with which it combines (Gonda 1962: 225 ff.). This actional meaning is based on the common semantic shift according to which completion can be thought of as togetherness (Gonda 1962: 225 defines these as “two realizations of the same basic meaning”). In (10), the composite sám √dahmeans ‘consume by fire’, whereas dahper se means ‘burn’; the completive meaning of sám is further emphasized by the indefinite víśvaṃ : acc ‘every’. 14 (10) víśvaṃ sám atríṇaṃ daha every. acc with demon. acc burn. imp .2 sg ‘Burn completely every demon.’ ( R̥V 1.36.14b) Moreover, Vedic preverbs can sometimes behave as applicatives: the addition of preverbs has sometimes the side-effect of centralizing the peripheral argument(s) taken by a verb. 15 A case in point is the transitive composite abhi √vr̥t- ‘roll against > overcome’ (11), which contains the preverb ábhi ‘to, unto, against’, and the intransitive manner of motion verb √vr̥t- ‘roll’. 16 (11) abhi-vŕ̥tya sapátnān abhí yā́ no árātayaḥ to-roll. abs rival. acc . pl to rel ( f ). nom / acc . pl 1 pl . gen evil_spirit. nom . pl ‘Having overcome the rivals and (having overcome those) who (are) our enemies …’ ( R̥V 10.174.2ab) As pointed out by Danesi (2013: 67) and Casaretto & Schneider (2015: 244 ff.), transitivity is the consequence of the semantic contribution brought about by the preverb abhí ‘to, toward, over’, which centralized the Goal-participant, thus 14 As mentioned in Section 4.1.1, Delbrück (1897) assigns more of a perfectivizing function (i. e. pertaining to grammatical aspect) to Vedic preverbs. Gonda (1962: 229) firmly opposes to Delbrück’s view, and remarks that “‘Perfektivierung’ is a question of semantics and lexicology.” Such a confusion between lexical and grammatical aspect has long characterized the discussions on preverbs and perfectivization / telicization (e. g. in Ernout & Thomas 1964; Brunel 1939; for similar remarks, see 2.3.2). 15 Applicatives are “a means some languages have for structuring clauses which allow the coding of a thematically peripheral argument or adjunct as a core-object argument. Such constructions are signaled by overt verbal morphology” (Peterson 2007: 1; cf. Section 3.1.1). 16 Example (11) also contains abhí ‘to’ in absolute position (on this usage of Vedic preverbs, see Section 4.1.2.3). Moreover, in this context, this free-standing preverb seems to have the function of anaphorically recalling the just mentioned spatial relation (on the pragmatic functions of preverbs, cf. Section 3.1.1). <?page no="104"?> 104 4. Multiple preverbs in Vedic building a transitive composite out of an intransitive manner of motion verb (for further remarks on this issue, cf. Section 4.5.3). 4.1.2.3. Vedic preverbs with adnominal, adverbial, and absolute positions Besides modifying a verb, a number of Vedic preverbs can also function as noun modifiers, i. e. as quasiadpositions. 17 As pointed out e. g. by Whitney (1955[1879]: 414) and Casaretto & Schneider (2015), no Vedic preverb seems to function as an adposition proper, in that no preverb combines the following features: (i.) ability of determining the case of the noun to which it is juxtaposed; (ii.) syntactic obligatoriness; (iii.) compulsory adjacency to the noun that it modifies (on the criteria for detecting prepositional phrases, cf. also e. g. Hagège 2010; Luraghi 2010). Nevertheless, many Vedic preverbs occur in constructions that show one of the features just outlined (Casaretto & Schneider 2015: 233 ff.). As nominal modifiers, preverbs can occur both before (12) and after (13) the modified noun (Casaretto & Schneider 2015: 241), though the pre nominal position is usual for Vedic preverbs with nominal orientation (Reinöhl 2016: 75, who summarizes results obtained by Hettrich and his research group). (12) dvimātā́ hótā vidátheṣu born_of_two_mothers. nom hotar. nom worship( n ). loc . pl samrā́ṭ ánu ágraṃ cárati sovereign_king. nom along top( n ). acc walk. prs .3 sg kṣéti budhnáḥ remains. prs .3 sg bottom. nom ‘The hotar (priest, or the sacrificial flame), born of two mothers, is the sovereign king during the sacrifices; he goes to the top, the foundation remains (still).’ ( R̥V 3.55.7ab) (13) svastí pánthām ánu carema happily path. acc along walk. opt . prs .1 pl 17 As pointed out in Sections 3.1.4.1 and 4.1.1, for this reason, Vedic preverbs are often called “prepositions” in reference works: “prepositional prefixes” (Whitney 1955[1879]: 396); “prepositions compounded with roots” (Macdonell 1916: 265); die Präpositionen in Verbindung mit Verben ‘prepositions in connection with verbs’ (Delbrück 1888: 440). <?page no="105"?> 4.1. Preverbs in Vedic 105 sūryācandramásāv iva sun_and_moon. nom . du like ‘Happily we may walk along (our) path like the sun and the moon.’ ( R̥V 5.51.15ab) In (12), the preverb ánu ‘after, along, toward’ precedes the noun in the accusative case ágraṃ ‘top’. The combination of ánu+ acc expresses Goal. In (13), instead, the postposed ánu ‘after, along, toward’ contributes to the expression of Path together with the accusative of extension pánthām ‘path’. Both the accusative of Goal and the accusative of Path can also occur adpositionless. However, without the further specification of a preposition, the usual reading for an adpositionless accusative is that of Goal. Thus, in order to express the Path, the adpositionless accusative can be regarded as an exception to the far more frequent construction involving an adposition, such as ánu ‘after, along, toward’ (Hettrich 2007; Casaretto 2011a: 39-40; Casaretto & Schneider 2015: 235). In other passages, the presence of an adposition is not syntactically compulsory, but adds the specification of the spatial region to the generic directional meaning of an adpositionless case. For example, this is the case in (14) below, in which ádhi ‘above, over, on, onto’ clarifies that the direction of movement is ‘onto’, rather than simply ‘to’ (Casaretto & Schneider 2015: 226; for similar considerations, cf. also Boley 2004). (14) tíṣṭhā rátham ádhi táṃ stand. imp .2 sg chariot. acc above dem . acc ‘Mount that chariot! ’ ( R̥V 5.33.3c) Without an accompanying noun phrase, Vedic preverbs can also behave as adverbs. The function of úpa ‘to, unto, toward’ in (15) is adverbial: (15) tā́ni narā jujuṣāṇā dem . n . acc . pl hero. voc . du enjoy. ptcp . prf . nom . du . mid úpa yātam toward drive. imp .2 du ‘Having enjoyed those [praises], o heroes, drive hither! ’ ( R̥V 2.39.8c) The syntactic freedom of Vedic preverbs is also shown by their usage in what is generally called absolute position, i. e. as “substitutes” of verbal forms. “Omit- <?page no="106"?> 106 4. Multiple preverbs in Vedic ted” verbal forms can be recovered either from the previous linguistic material (16) or from the extralinguistic context (17). (16) saptá svásāro abhí mātáraḥ śíśuṃ seven. nom . pl . f sister( f ). nom . pl to mother( f ). nom . pl baby. acc ‘The seven sisters (rush as) mothers to the baby.’ ( R̥V 9.86.36a) (17) prá te nā́vaṃ ná sámane vacasyúvam forward 2 sg . dat ship( f ). acc like assembly( f ). loc eloquent. acc . f ‘In the assembly, I (set in motion my chant) forward to you as an eloquent ship.’ (R̥V 2.16.7a) For the passage in (16), Renou (1935: 60) assumes the verbal form arṣanti : prs .3 pl ‘flow’ (from √r̥ṣflow’), which can be recovered from the preceding stanza. Instead, for (17), it is not so easy to retrieve the omitted verbal form: a possibility is iyarmi : prs .1 sg (from √r̥- ‘go, move, set in motion’), as suggested by Renou (1935: 61) based on similar formulaic expressions occurring elsewhere in the hymns. According to Danesi (2013: 65), the absolute usages just outlined tell, on the one hand, that there is a high degree of syntactic autonomy holding between preverbs and verbs; on the other hand, that certain preverb-verb combinations must be stored as single units in speakers’ mental lexicon: otherwise, the recovery of the missing verbal form would have been impossible. In my opinion, Danesi’s latter suggestion is not completely convincing. All in all, it takes for granted the compulsory presence of a verbal form in every context. Moreover, for passages such as (16) above, mental storage is an unnecessary assumption, as the missing verbal form shows up in the immediate previous material, and anaphoric reference is thus easy to draw. Then, for cases such as (17), the assumed motion verb could be easily replaced by another generic motion verb, thus making it difficult to argue for lexical storage. 18 18 In order to clarify my position on the matter, it can be useful to provide an example from a contemporary language, such as Russian (example (i) comes from the Spoken Corpus of the Russian National Corpus (Aleksej Popogrebskij. Prostye vešci , k / f [2006]), see http: / / www.ruscorpora.ru/ en/ . (i) a. “Ty kuda? ” (Vasin, Ivan Švedov, Muž, 37, 1969) 2 sg . nom to_where b .“V metro” (Sergej, Sergej Puskepalis, Muž, 40, 1966) into metro. acc ‘“Where (are) you (going)? ” “To the subway.”’ In spoken Russian, in many contexts such as that in (i), the overt expression of motion via a motion verb is unnecessary. In (i)a, the interrogative adverb kuda unambiguously <?page no="107"?> 4.1. Preverbs in Vedic 107 4.1.2.4. The ambiguous status of Vedic preverbs In Sections 4.1.2.1-4.1.2.3, I have shown that Vedic preverbs still exhibit (a) a high degree of syntactic freedom as to their positioning; (b) no binding syntactic relations with any other element within the sentence. For these reasons, Vedic preverbs frequently represent functionally ambiguous elements, whose nominal or verbal orientation is difficult to determine. An instructive passage in this respect is provided in (18): (18) síndhor ūrmā́v ádhi śritáḥ river. gen wave. loc above lean. ptcp . pst . nom . pass ‘(The wise one) leaning on the wave of the river.’ ( R̥V 9.14.1b) As Hettrich et al. (2010: 20) explain, one cannot decide whether, in (18), ádhi ‘above, over, on, on to’ is constructed with the preceding noun ( ūrmā́v : loc ‘wave’) or with the following verb ( śritáḥ : ptcp . pst . nom . pass ‘leaning’). Though examples such as (18) are frequent in Vedic, and thus it can be very difficult, or even pointless, to assign Vedic preverbs a clear-cut functional category (Renou 1956), attempts have been made in this direction, in particular by Hettrich, Casaretto, and Schneider. In a series of papers, Hettrich and colleagues try to perform such a difficult classificatory task, based on the following assumptions (Hettrich et al. 2010: 20): (i) clause-initial position is the marked option for verbal orientation (cf. (1), (2)); (ii) immediate preverbal position is the un marked option for verbal orientation (cf. (5), (6)); according to Reinöhl & Casaretto (2018), these two positions strongly correlate with lexicalization of the composite; (iii) adjacent placement to a noun, either preor post-nominal, indicates nominal orientation (cf. (12)-(14)); The orientations and possible positions of Vedic preverbs are summarized in Table 7. Positions from 2a to 4 indicate verbal orientation, with different degrees of markedness. Hettrich and colleagues regard as ambiguous positions 5a-c even in contexts where, from a semantic standpoint, preverbs clearly display verbal or nominal orientation. expresses Goal, as does the prepositional phrase v + acc in (i)b. Further specifications of motion are not required. <?page no="108"?> 108 4. Multiple preverbs in Vedic Abbreviation Orientation Position 1a nominal immediate prenominal 1b nominal immediate postnominal 2a verbal immediate postverbal, univerbated 2b verbal immediate postverbal 3 verbal clause-initial (tmesis #P(E) …V#) 4 verbal other position within the sentence 5a ambiguous NP _preverb_V 5b ambiguous preverb_ NP …V 5c ambiguous V_preverb_ NP Tab. 7: Possible orientations and positions of Vedic preverbs The results of the analysis performed by Hettrich and colleagues are summarized in Table 8. Overall, the verbal orientation prevails over the nominal orientation in Vedic. The preverbs showing a strong preference for the verbal orientation are highlighted in gray (i. e. ápa, áva, ā́, úd, ní, nís, párā, purás, prá, sám, ví ). <?page no="109"?> 4.1. Preverbs in Vedic 109 Preverb p osition o rientation Frequency 1a 1b 2a 2b 3 4 5a 5b 5c Nominal (1a-b) Verbal (2a-b, 3, 4) Unclear (5a-c) áchā 25 57 17 16 26 13 11 9 0 82 72 20 174 áti 50 17 49 2 7 15 32 8 19 67 73 59 199 ádhi 93 83 37 0 42 22 13 0 0 176 101 13 290 ánu 165 83 99 9 34 29 66 22 0 248 171 88 507 antár 28 44 10 1 4 28 5 0 0 72 43 5 120 ápa 2 1 107 2 80 35 3 4 1 3 224 8 235 ápi 14 10 39 3 6 8 17 9 1 24 56 27 107 ábhi 165 70 260 11 70 52 99 74 0 235 393 173 801 áva 8 10 133 4 47 12 22 7 0 18 196 29 243 ā́ 34 27 323 5 126 37 97 56 0 61 491 153 705 úd 0 1 162 2 125 3 15 4 0 1 292 19 312 úpa 98 34 126 11 17 13 81 45 0 132 167 126 425 tirás 33 5 3 2 2 4 0 4 6 38 11 10 59 ní 23 3 468 11 100 9 31 20 0 26 588 51 665 nís 6 3 62 3 31 6 13 4 0 9 102 17 128 párā 0 1 62 0 23 1 7 0 0 1 86 7 94 parás 17 8 5 0 6 6 0 0 0 25 17 0 42 <?page no="110"?> 110 4. Multiple preverbs in Vedic Preverb p osition o rientation Frequency 1a 1b 2a 2b 3 4 5a 5b 5c Nominal (1a-b) Verbal (2a-b, 3, 4) Unclear (5a-c) pári 54 58 150 1 44 30 84 ? 0 112 225 84 421 purás 6 3 12 8 1 21 1 0 0 9 42 1 52 púrā 12 1 0 0 0 46 0 0 0 13 46 0 59 prá 54 14 579 466 156 34 69 0 68 1201 103 1372 práti 52 9 91 2 32 32 25 21 0 61 157 46 264 sám 8 3 390 19 161 19 41 42 6 12 589 89 690 ví 15 5 624 41 179 94 73 13 5 20 938 91 1049 Tab. 8: The positions and orientations of Vedic preverbs 19 19 The data of Table 8 are taken from Casaretto (2010a, 2010b, 2011a, 2011b, 2011c, 2011d, 2012, 2013), Hettrich et al. (2010), and Schneider (2009, 2010a, 2010b, 2011, 2012, 2013a, 2013b, 2013c) and Reinöhl & Casaretto (2018: 246). The data on ā́ cover only books II-V of the R̥ gVeda (all R̥ g- Vedic occurrences of ā́ however are reported in the relevant entry of the RIVELEX II). The data related to ádhi , ántar and pári are published (Hettrich 1991, 1993, 2002), but originally not analyzed according to the categories described in Table 7 and employed in later publications. Reinöhl & Casaretto (2018: 246) however more or less normalized such data as well (but cf. their fn. 13 and 14). <?page no="111"?> 4.1. Preverbs in Vedic 111 4.1.2.5. The accentual properties of Vedic preverbs As Vedic grammars point out (e. g. Macdonell 1910: 76 ff.; Whitney 1955[1879]: 28 ff.), Vedic has a pitch accent of musical nature. Moreover, as a basic rule, every word bears its own accent. This rule however admits a number of exceptions: there are words that never take an accent (i. e. enclitic pronouns and particles including =ca ‘and’, =u ‘on the other hand’, iva ‘like’, etc.), as well as words that lose their accent under certain syntactic conditions. Both Vedic verbs and Vedic preverbs belong to the latter group; their accentual properties can vary and are interrelated, as explained below. Vedic verbs bear no accent in main clauses, except when they occur in sentenceor pāda -initial position. 20 By contrast, verbs are accented in subordinate clauses (Macdonell 1910: 107ff.). Accordingly, in main clauses, whatever their positioning, preverbs usually behave as independent words and bear an independent accent, while the verbal form is unaccented. By contrast, in subordinate clauses, preverbs tend to lean onto the verbs that they modify, which are accented, and to make up actual compounds with them. As far as multiple preverbs are concerned, they usually also instantiate the pattern outlined above and schematized below: (a) main clauses: verb → unaccented; preverb(s) → accented (19); 21 (b) subordinate clauses: verb → accented; preverb(s) → unaccented (20). (19) yunájmi te bráhmaṇā keśínā yoke. prs .1 sg 2 sg . gen prayer( n ). ins hairy. acc . du hárī úpa prá yāhi fawn-coloured. acc . du toward forward proceed. imp .2 sg ‘With holy prayer, I yoke your long-maned pair of Bays: drive toward (them).’ ( R̥V 1.82.6ab) 20 A pāda , or foot, is the minimal unit of the Vedic meter (see Section 4.1.2.6). 21 Ancient Greek verbs show similar enclitic behavior, as in the well-known retraction of the accent in finite verbal forms. <?page no="112"?> 112 4. Multiple preverbs in Vedic (20) yūyáṃ hí devīr r̥tayúgbhir 2 pl . nom . f for goddess( f ). voc . pl properly_harnessed. ins . pl áśvaiḥ pari-pra-yāthá bhúvanāni horse. ins . pl around-forward-proceed. prs .2 pl being( n ). acc . pl sadyáḥ on_the_same_day ‘For you, o goddesses, with your steeds yoked in due time, proceed around the living beings in one day’. ( R̥V 4.51.5ab) In (19), the preverbs úpa ‘to, unto, toward’ and prá ‘forward, onward, forth, fore-’ precede the verb that they modify, i. e. the unaccented yāhi ‘drive, go, proceed’. Despite their position, which would call for univerbation, they retain their accent, and are not compounded with the verb. In (20), instead, the subordinate clause introduced by hí ‘for, because’ contains the accented verbal form yāthá ‘(you) drive, go, proceed’, onto which the preverbs pari- ‘round about, around’ and pra- ‘forward, onward, forth, fore-’ attach. There are exceptions to the tendency outlined in (a) for main clauses. When the preverb ā́ ‘to’ modifies a verb together with a further preverb occurring in exterior position (# EP ā́ V#), the EP (here upa- ) loses its accent, and undergoes univerbation with ā́ : (21) imáṃ yajñám idáṃ váco dem . acc sacrifice. acc dem . acc . n speech( n ). acc jujuṣāṇá upā́gahi enjoy. ptcp . prf . nom . mid come_near. imp . aor .2 sg ‘Enjoying this sacrifice and this praise, (O Soma,) come near, (and stay close to make us prosper).’ ( R̥V 1.91.10ab) However, when the EP ends with -i (# EP [-i] ā́ V#) , the univerbation shown in (21) does not occur, as exemplified in (22): <?page no="113"?> 4.1. Preverbs in Vedic 113 (22) úd agne tiṣṭha práti ā́ tanuṣva ní amítrām̐ up A. voc stand. imp .2 sg against to extend. imp .2 sg down enemy. acc . pl oṣatāt tigma-hete burn. imp .2 sg of_sharp-weapons. voc ‘Rise up, O Agni! Stretch out against (the enemy)! Burn down the foes, O (god) with the sharp weapons! ’(R̥V 4.4.4ab) In a single passage, the combination of úpa ‘to, unto, toward’ and áva ‘down, off’ behaves as a combination containing ā́ ‘to, unto, at’, in that úpa loses its accent and is univerbated to the following áva ( upā́vasr̥ja : imp .2 sg , R̥V 10.110.10a). As for verbs occurring in subordinate clauses, besides featuring the usual order shown in (20), they also allow for the displacement of the EP ( sám ‘along, with, together’ in (23)), which is separated from the remaining composite IP-V , and retains its accent: 22 (23) só agnír yó vásur gr̥ṇé sáṃ yám dem . nom A. nom rel . nom V. nom call. prs .3 sg with rel . acc āyánti dhenávaḥ come. prs .3 pl milk_cow( f ). nom . pl ‘He is Agni, who is praised as the Vasu, to whom the milk-cows come together.’ ( R̥V 5.6.2ab) However, separations such as that in (23) are by no means frequent, and in any case never go further than one pāda (Renou 1935: 51; Danesi 2013). 23 In addition, when separation occurs, the linguistic element splitting the composite frequently happens to be the subordinator (e. g. R̥V 5.56.4; 8.6.8; etc.). Furthermore, in cases of multiple preverbs, the word order in which both preverbs are accented and separated from the verb is extremely rare in subordinate clauses. One such rare example is provided in (24): (24) prá yát stotā́ jaritā́ forward when praising. nom invoker. nom 22 For the composite sám ā́ √i - ‘come together’, the order [EP IP V] is also attested in main clauses (R̥ V 7. 40. 70; 10. 85. 33). 23 Cf. fn. 20. <?page no="114"?> 114 4. Multiple preverbs in Vedic tū́rṇyartho vr̥ṣāiámāṇa úpa pursuing_an _object. nom eager_as_a_bull. ptcp . prs . nom . mid to gīrbhír ī́ṭṭe song( f ). ins . pl implore. prs .3 sg . mid ‘… what time the praising invoker, keen of purpose and eager as a bull, with songs implores you.’ (R̥V 3.52.5cd) Preverb(s) is(are) usually univerbated in negative clauses, as well as with non-finite verbal forms (e.g. on the position of preverbs with participles, cf. Lowe 2015: 130). In case of multiple preverbs, however, the EP can be displaced, thus keeping its own accent (#P …P-V [non-finite] #, cf. R̥V 5.1.1cd; 7.104.21ab). Rarely, both preverbs are either separated from the verb (#P …P …V [non-finite] #, cf. R̥V 10.70.9cd), or separated and univerbated to one another (#P-P …V [non-finite] #, cf. R̥V 10.70.9cd). To sum up, in main clauses, the accentual properties of preverbs strongly suggest that they retain much of their original adverbial status. This also holds true for other types of clauses, though to a lesser extent. In subordinate and negative clauses, as well as with non-finite verbal forms, preverbs are usually unaccented and univerbated, and only occasionally occur displaced from the verb that they modify. 4.1.2.6. Stylistic and metrical reasons for preverb placement In Sections 4.1.2.1-4.1.2.5, I explained that the constraints driving the placement of preverbs are mainly syntactic in nature. We have also seen that the development of non-compositional meanings is not necessarily linked to the univerbation of the composite. In this Section, I briefly tackle the issue as to what extent the positioning of preverbs can be influenced by the stylistic and metrical features of the Vedic hymns. The R̥g-Veda consists of poetic texts composed in metrical structure and organized in books, hymns, stanzas, and verses (cf. Section 1.3.1). Their basic unit is the pāda ‘foot’, that is, each verse, or line, that constitutes a stanza. 24 Such verses are formed by five (rarely), eight, eleven, or twelve syllables. The more or less regular alternations of long and short syllables constitute the Vedic meter. Even syllables (second, fourth, etc.) are generally long, and the general look of the Vedic meter is iambic. The last part of the Vedic verses, called “cadence”, is clearly separated from the rest. Verses of eleven and twelve syllables are also split by a further metrical pause after the fourth or fifth syllable (Macdonell 24 The Vedic ‘foot’ is not the same metrical unit as the Ancient Greek ‘foot’. <?page no="115"?> 4.1. Preverbs in Vedic 115 1916: 436 ff.). The most common Vedic stanzas are formed by three or four dimeter (eight syllables) or trimeter (eleven / twelve syllables verses). Does this metrical structure influence the placement of preverbs? A general answer is that the position of preverbs does not seem to depend on the meter (Papke 2010: 98 ff.). Monosyllabic preverbs can virtually occur in any position that allows for their quantity. In addition, the sandhi rules of Vedic can lengthen a short final syllable of a disyllabic preverb, if required by the metrical structure (cf. Section 4.3.1). The displacement of preverbs seems to be more a matter of style and poetic diction. As Renou (1935: 52-53) points out, for example, preverbs are very likely to be fronted in invocations or prayers, thus gaining the pragmatic function of emphasizing the initial impetus. This is the case of examples (1) and (2) above in this Chapter, both starting with the preverb prá ‘forward, onward, forth, fore-’. In addition, the pragmatic value of preverb repetition in initial position has been investigated by Dunkel (1979) and Klein (e. g. 1987, 2007, 2008, 2012), who describe this anaphora as a kind of cohesive process of human language. In any case, the displacement of preverbs usually leads them in clause-initial (or pāda -initial) position, which is the position in which preverbs in general also tend to occur. Thus, the possibilities of displacement remain within the boundaries of Vedic grammar. 4.1.2.7. The ongoing grammaticalization of Vedic adpositions As remarked in Section 4.1.2.3, in Vedic, there are no binding syntactic relations between adpositions and the noun phrases that they modify (e. g. Hewson & Bubenik 2006: 102: ff.; Casaretto & Schneider 2015; Reinöhl 2016: 65-84, and references therein). Further indications in this respect are provided in what follows. To begin with, adpositions are non-obligatory to express many spatial and non-spatial concepts. In parallel, Vedic morphological cases preserve their concrete values better than most other ancient Indo-European languages (Macdonell 1916: 298 ff.; Hewson & Bubenik 2006: 102 ff.; Hettrich 2007; Casaretto & Schneider 2015, and references therein). As mentioned in Section 4.1.2.3, the prepositionless accusative can express Goal (frequently, e. g. R̥V 1.162.21), Path (rarely, e. g. R̥V 2.16.3), and Duration (rarely, e. g. R̥V 10.161.4). The instrumental case is also employed to express Path (e. g. R̥V 2.33.1) and Duration (e. g. R̥V 1.86.6), besides Comitative and Instrument. The ablative can mean separation or distance based on the semantics of the verb that takes it (dynamic or stative, cf. Hettrich 2007: C.a. IV , 2; e. g. R̥V 2.33.1). The locative case indicates Location (e. g. R̥V 1.32.2), Goal (cf. below), and Time (e. g. R̥V 10.53.3). <?page no="116"?> 116 4. Multiple preverbs in Vedic Frequently, the function of the adposition is only to specify the geometric shape of the spatial region in which a certain event occurs (cf. example (14) above, in which the preverb ádhi ‘on’ clarifies that the motion is directed ‘onto’ a certain LM , and not simply ‘to’ it). In such examples, the preverb is syntactically unnecessary, though it provides a clear semantic contribution. In contrast, this is not the case for the following passage, in which the allative semantic contribution of úpa ‘toward’ overlaps with the allative semantics of the accusative case ( índram ) (possibly, úpa indicates close contact in this context). (25) gíro ma índram úpa yanti praise( f ). nom . pl 1 sg . gen I. acc toward go. prs .3 pl ‘My praises go toward Indra.’ ( R̥V 3.51.2b) There are only a few contexts in which the lack of a preverb totally changes the meaning of the sentence. A case in point follows: (26) imé jīvā́ ví mr̥taír dem . nom . pl living. nom . pl asunder dead. ins . pl ā́vavr̥tran hither_turn. aor .3 pl . pass ‘These living ones have separated themselves from the dead ones.’ ( R̥V 10.18.3a) In combination with ví ‘ asunder’, the instrumental case indicates separation, whereas the adpositionless instrumental would express a Comitative meaning. Instead, the adpositionless case employed to convey a similar meaning of separation would be the ablative (Casaretto & Schneider 2015: 240), as shown in example (27): (27) mā́ naḥ sū́ryasya saṃdŕ̥śo yuyothāḥ neg 1 pl . acc sun. gen sight. abl keep_away. sbjv . prs .2 sg . mid ‘Don’t keep us away from the sight of the sun! ’ ( R̥V 2.33.1b) Overall, the few Vedic preverbs that preferably select nominal orientation (e. g. ánu ‘along, after’, tirás ‘over, through’, parás ‘off’; cf. Table 8) tend to occur in pre nominal position, though postnominal placement is by no means infrequent (cf. Sections 4.1.2.3 and 4.1.2.4). Despite this tendency to prenominal placement in Vedic, Indo-Aryan languages later on developed fully grammaticalized <?page no="117"?> 4.2. Multiple preverbs in numbers 117 post positions, rather than prepositions. These post positions, however, do not continue the ancient Vedic preverbs, but rather go back to different etymological sources: cf. e. g. Hindi mẽ ‘in’ < Ved. mádhye : loc ‘in the middle’ (Casaretto & Schneider 2015: 254; Reinöhl 2016: 65-84, and references therein). Thus, Indo-Aryan does not fit the general Indo-European pattern of development, whereby original free-standing locative adverbs undergo a functional bifurcation into preverbs or adpositions (counter to Hewson & Bubenik 2006: 102 ff.). The reasons that in Vedic local adverbs do not develop into adpositions is explained by Reinöhl (2016: 80 ff.) in terms of a mismatch between the prosody and the semantics of local adverbs. In Vedic, even noun-oriented local adverbs could encliticize onto verbal forms because of Vedic prosodic rules (cf. Section 4.1.2.5). Such a mismatch between functional and prosodic affiliation arguably blocked the grammaticalization of Vedic local adverbs into proper adpositions. 4.2. Multiple preverbs in numbers 4.2.1. Composites with multiple preverbs Table 9 contains Vedic multiple preverb composites. In order to identify such composites, I used Grassmann’s dictionary (1936[1873]) as a starting point (cf. fn. 8 in this Chapter). Then, I exclusively selected those combinations of multiple preverbs-verbs that at least once attest to both preverbs in preverbal position (#P_P_V#). This methodology is based on Hettrich et al.’s (2010: 20) remark, according to which direct preverbal position is the unmarked option for local adverbs with verbal orientation (cf. Section 4.1.2.4). This selection process resulted in 116 composites occurring in 186 R̥g-Vedic passages. Thus, my criteria are stricter than Papke’s (2010), who individuated as many as 387 composites (cf. fn. 8 in this Chapter). c omposite m eaning f requency abhí prá √arcsing loudly of 1 áchā párā √igo away toward 1 ánu áva √igo down after, follow 1 ánu párā √igo away after 1 ánu prá √igo after, follow 1 ápa pára √igo off 1 abhí prá √igo near to, approach 3 <?page no="118"?> 118 4. Multiple preverbs in Vedic c omposite m eaning f requency abhí sám √iapproach together, come together at 2 abhí ví √icome toward from different parts 1 abhí ā́ √icome to, approach 1 ā́ áva √irush down upon 1 ā́ nís √igo off, depart 1 úpa prá √imarch on, go toward 9 nís ā́ √igo off, depart 3 pári ā́ √icirculate 1 pári prá √irun through on all sides 1 práti úd √irise and go toward 1 sám ā́ √icome together, approach together, meet at / in / with 2 ví párā √igo back (home) again 1 ví prá √igo forth in different directions, disperse, spread out 1 ní ā́ √īrset someone down 5 sám ā́ √īrput together, create 1 sám prá √īrcome forth together 1 ní ā́ √kr̥hold back 1 sám ā́ √kr̥bring together, gather, prepare 3 ví ā́ √kr̥undo, sever, divide, separate from 1 adhí ví √kṣarpour out, flow out 1 abhí prá √gāhdig into, penetrate 1 áva ā́ √gamundertake, begin 1 ádhi sám √gamgo up to, approach together 1 úpa ā́ √gamcome near, come to 2 abhí ā́ √gāapproach, come to 1 úpa prá √gāstep near to, proceed to 3 abhí prá √gā 2 - (√gai-) encourage to start singing about, begin to praise 5 <?page no="119"?> 4.2. Multiple preverbs in numbers 119 c omposite m eaning f requency abhi prá √cakṣsee 1 ánu sám √carwalk alongside, visit, seek after 2 abhí ā́ √carcome up, approach 1 abhí úd √carrise over 1 abhí sám √cargo together to, seek for 5 úd ā́ √carrise out of 1 úpa ā́ √carcome near to, attend upon 2 úpa prá √jinvplease or gratify in approaching 1 ánu prá √jñātrace, discover 1 áti nís √tanpenetrate with rays 1 práti ā́ √tanextend in the direction of, shine upon / against 1 abhí ā́ √taptorment, pain 1 abhí ā́ √tr̥ ̄pass through to, come up to 1 ádhi ā́ √dātake away from above 1 abhí ā́ √diśaim at (in hostile manner) 1 ā́ nís √duhcreate out of 1 abhí prá √dr̥ ̄put forth by bursting or opening 1 ā́ prá √drurun forth here 1 pári prá √dhanvflow forth around 1 adhí sám √dhāstore up 1 adhí ní √dhādeposit for 1 abhí sám √dhācompose the mind at 1 antár ā́ √dhāreceive into, contain 1 ánu ā́ √nūsound here through 1 abhí prá √nūpraise highly to 6 abhí sám √nūrejoice together at 5 úpa ní √padlie down beside 1 ánu ā́ √phaṇjump 1 antár ví √bhāshine in different directions between 1 <?page no="120"?> 120 4. Multiple preverbs in Vedic c omposite m eaning f requency ánu prá √bhūspread over 2 abhí prá √bhūassist, help 1 abhí sám √bhūenter, reach, come to 1 ánu prá √bhūṣ- Serve 1 abhí prá √bhr̥bring forth to, offer to 1 pári ā́ √bhr̥carry near, fetch from 1 abhí prá √mandfeverishly await, confuse, infatuate 4 ánu prá √muclet loose successively 1 abhí prá √mr̥śseize, grasp 1 prá abhí √mr̥śseize, grasp 1 ánu prá √yajwin for oneself 1 abhí ā́ √yamaim at 1 sám ā́ √yamdraw, pull, stretch 1 sám prá √yamoffer together / mutually, give to 1 áti ā́ √yādrive by 2 ábhi sám √yāvisit, approach to 1 úpa ā́ √yācome near, approach 2 úpa prá √yāproceed toward 2 pári prá √yāgo forth around 1 prá ā́ √yācome near, approach 1 práti prá √yāgo back, return 2 ánu sám √rabhtake hold of 1 abhí sám √rabhtake hold of 2 ánu ā́ √labhlay hold of, grasp, handle, take in the hand 1 ápa ní √līhide oneself, disappear completely 1 ánu prá √vahgo, get forward 1 abhí ā́ √vanstrive, seek to win 1 ánu prá √vidunderstand backward and forward 1 ā́ ví √vidknow by distinguishing 1 <?page no="121"?> 4.2. Multiple preverbs in numbers 121 c omposite m eaning f requency úpa prá √vidunderstand 1 ā́ pári √vr̥surround with 1 ápa ā́ √vr̥jwipe out, bring away 1 ánu ā́ √vr̥troll near along 1 ánu prá √vr̥tproceed along / after 1 abhí ā́ √vr̥troll toward, hurry toward 4 pári ā́ √vr̥tturn round, turn away from, return to 2 práti ā́ √vr̥tturn against 1 sám ā́ √vr̥tturn back, come back, return 1 áti prá √vr̥dhoutgrow 1 antár pári √vyā - hide in 1 áti prá √śr̥dhbring in front of in excess 1 abhí prá √sadsit down, settle along 1 ā́ ní √sadsit down on, cause to sit down, establish 12 áti prá √sr̥outstrip, surpass 1 ví ā́ √sr̥run through 1 ví prá √sr̥spread 1 úpa áva √sr̥jreach over, give, bestow 1 ánu ví √sthāextend over 1 ábhi prá √sthāadvance toward, reach, surpass 3 ábhi ví √spaślook at, view, look hither 2 pári prá √syandgush around, flow forth or round 2 abhí prá √hanoverpower 1 pári sám √hā 2 rise up from 1 t otal 186 Tab. 9: Vedic composites with multiple preverbs and their frequency 25 25 In all the tables of this Chapter, composites are sorted by root. This choice is motivated by the fact that, in Vedic, univerbation is only at its onset. Within a group of composites containing the same root, the EP determines the order. The last criterion for ordering is the IP. The alphabetical order is that of the Devanāgarī script. <?page no="122"?> 122 4. Multiple preverbs in Vedic The high number of composites (116), the high number of roots modified by multiple preverbs (56, cf. Table 10), and the high number of preverb combinations that can accompany many different verbal roots (52, cf. Table 11) all suggest that multiple preverbs did not constitute an infrequent pattern in Vedic. As shown in Table 9, 88 out of 116 composites occur only once in the R̥g-Veda , and only one composite, i. e. ā́ ní √sad- ‘sit down on, cause to sit down, establish’, shows up in more than 10 occurrences (on the productivity of verbal composition in Vedic, cf. Danesi 2013: 62 and Section 4.1.2.2). Table 10 contains those composites that are also attested elsewhere than in the R̥g-Veda , along with their R̥g-Vedic and post-R̥g-Vedic meanings. As Table 10 shows, out of 116 composites, only 45 are attested in later texts. Despite dealing with a corpus language (cf. Section 1.4), post-R̥g-Vedic attestation is significant, as it suggests a certain degree of conventionalization and stability in the lexicon. In addition, Table 10 allows for interesting comparisons. Going from the Vedic to the post-Vedic period, composites often come to acquire less concrete and more non-compositional meanings. Some cases in point are the following: abhí prá √i- ‘go near to, approach’ > ‘think of, aim, intend’; ví ā́ √kar- ‘undo, sever, divide, separate from’ > ‘explain, predict, declare’; abhí ā́ √gā- ‘approach, come to’ > ‘visit, begin to’; abhí ā́ √car- ‘come up, approach’ > ‘undertake, practice’; ánu ā́ √vart- ‘roll near / along, move after / along’ > ‘revolve, move after, follow, change’; abhí ā́ √vart- ‘roll toward, hurry toward’ > ‘repeat’; pári ā́ √vart- ‘turn round, turn away from, return to’ > ‘be changed into, get possessed of ’. In other cases, from the Vedic to the post-Vedic period, composites gain a more specialized meaning: sám ā́ √i- ‘come together, approach together, meet at / in / with’ > ‘unite in marriage, enter, emulate, form an alliance with’; abhí sám √bhū- ‘enter, reach, come to’ > ‘obtain the shape of ’. Semantic shifts of this type are expected: all of them involve semantic bleaching of the elements making up the composite, which lose part of their original spatial value. c omposite M eaning in the R̥V P ost -R̥V edic Meaning ánu párā √igo away after follow in walking off ánu prá √igo after, follow follow in death, seek after abhí ā́ √icome to, approach go near, come to, approach abhí prá √igo near to, approach think of, aim, intend abhí sám √iapproach together, come together at invade úpa prá √imarch on, go toward march on, go toward, undertake an activity <?page no="123"?> 4.2. Multiple preverbs in numbers 123 c omposite M eaning in the R̥V P ost -R̥V edic Meaning nís ā́ √igo off, depart go off, depart pári ā́ √icirculate roam about, go around, return práti úd √irise and go toward ascend to sám ā́ √icome together, approach together, meet at / in / with unite in marriage, enter, emulate, form an alliance with ví prá √igo forth in different directions disperse, spread out go away, depart sám prá √īrcome forth together ( caus ) drive, push forward sám ā́ √kr̥bring together, gather, prepare bring together, gather, prepare ví ā́ √kr̥undo, sever, divide, separate from explain, predict, declare úpa ā́ √gamcome near, come to come back, approach, enter in a condition, be subject to, occur abhí ā́ √gāapproach, come to visit, begin to úpa prá √gāstep near to, proceed to step near to, proceed to abhí prá √gāhdig into, penetrate immerse abhí ā́ √carcome up, approach undertake, practice úpa ā́ √carcome near to, attend upon come near to, attend upon práti ā́ √tanextend in the direction of, shine upon / against extend in the direction of, shine upon / against abhí prá √dr̥ ̄put forth by bursting or opening ( pass ) be scattered / divided asunder abhí sám √dhācompose the mind at take aim at, overcome, win, associate with úpa ní √padlie down beside lie down beside abhí sám √bhūenter, reach, come to obtain the shape of abhí ā́ √yamaim at lengthen, draw, pull, assume sám ā́ √yamdraw, pull, stretch draw together, contract sám prá √yamoffer together / mutually, give to give in marriage, give back ábhi sám √yāvisit, approach to approach in hostile manner <?page no="124"?> 124 4. Multiple preverbs in Vedic c omposite M eaning in the R̥V P ost -R̥V edic Meaning úpa ā́ √yācome near, approach come near, approach, undergo úpa prá √yāgo toward, proceed toward go toward, proceed toward ánu sám √rabhtake hold of take hold of mutually ánu ā́ √labhlay hold of grasp, handle, take in the hand lay hold of grasp, handle, take in the hand ápa ní √līhide oneself, disappear completely hide oneself, disappear ánu ā́ √vr̥troll near along revolve, move after, follow, change abhí ā́ √vr̥troll toward, hurry toward repeat pári ā́ √vr̥tturn round, turn away from, return to be changed into, get possessed of práti ā́ √vr̥tturn against return, come back sám ā́ √vr̥tturn back, come back, return return home, approach, succeed, perish, dismiss, repeat abhí prá √sadsit down, settle along ( caus ) cause to be gracious ví prá √sr̥spread spread úpa áva √sr̥jreach over, give, bestow dismiss toward, let loose, let go toward ábhi prá √sthāadvance toward, reach, surpass start / advance toward, reach, surpass ábhi ví √spaślook at, view, look hither look at, view, look hither abhí prá √hanoverpower overpower Tab. 10: Vedic composites attested after the R̥g-Veda and their meanings 4.2.2. Verbal roots modified by multiple preverbs Table 11 displays the 56 Vedic verbal roots modified by multiple preverbs occurring in immediate preverbal position and their meanings, their PIE roots with their meanings, and their frequencies, that is, the number of composites containing each root. The rightmost column specifies the verb type. For my purposes, a rough classification of verbs suffices. 26 26 Other scholars, including Levin (1993) and Sausa (2015), proposed more fine-grained semantic classifications for English and Ancient Greek verbs, respectively. <?page no="125"?> 4.2. Multiple preverbs in numbers 125 v erbal r oot m eaning pie r oot ( LIV 2 page) m eaning f requency v erb t ype arc- (ṛc-) shine, sing, praise (Gōto: 97-99) *h 1 erk u̯ (240-241) shine, sing 1 emission ay- (i-) walk, go *h 1 ei̯- (232) go 19 motion īrgo, move, rise, arise from *h 3 er- (299) put in motion 3 caused motion kr̥do, make *k u̯ er- (391) cut off, carve 3 creation kṣarflow, stream *g u̯ ĝ h er- (213) drift in water, flow 1 manner of motion gamgo to, approach *g u̯ em- (209) go, come (to) 3 motion gāgo to, approach *g u̯ eh 2 - (205) step, put your foot on 2 motion gā 2 - (gai-) sing, recite *g (u̯) eH(i̯)- (183) sing 1 communication gāhdive into, bathe in, plunge into *g (u̯) eĝ h - (183) force into water 1 caused motion cakṣshine, emit light *k u̯ ek̂- (383) see, catch sight of 1 emission car i go, walk *k u̯ elh 1 - (386) turn 6 motion jinvmove one’s self , annoy *g u̯ i̯eh 3 - (215) live 1 caused motion jñāknow *ĝneh 3 - (168) discern 1 mental activity tanextend, be diffused (as light) *ten- (626) stretch 2 motion tapgive out heat, be hot, shine (as the sun) *tep- (630) be warm / hot 1 emission tar i pass across / over, cross over *terh 2 - (633) come through, cross 1 motion <?page no="126"?> 126 4. Multiple preverbs in Vedic v erbal r oot m eaning pie r oot ( LIV 2 page) m eaning f requency v erb t ype dāgive *deh 3 - (105) give 1 transfer diśshow a direction *dei̯k̂- (108) show, point 1 communication duhmilk, squeeze *d h eu̯g h - (148) meet 1 emission drārun, make haste *dreh 2 - (127) run (away) 1 manner of motion drurun, hasten *dreu̯- (129) run 2 manner of motion dhancause move quickly *d h enh 2 - (144) put in rapid motion 1 caused motion dhāput *d h eh 1 - (136) put 4 putting nū- (nu-) sound, shout *neu̯H- (456) yell 3 emission padfall *ped- (458) step, fall, sink 1 motion phaṇgo, leap ? *(s)p(h)en-(d)- *spar (i) - ( EWA ia II 199-200) flutter move fast, lively 1 manner of motion bhāshine, glow *b h eh 2 - (68) shine 1 emission bhūbecome, happen, be *b h u̯eh 2 - (98) grow, become 3 existence bhūṣstrive after, use efforts for *b h u̯eh 2 - (98) grow, become 1 mental activity bhṛbear, bring *b h er- (76) bring, carry 2 caused motion mandrejoice *med- (423) become full 1 mental activity mucloose, set free from *meu̯k- (443) loose, take off 1 caused motion <?page no="127"?> 4.2. Multiple preverbs in numbers 127 v erbal r oot m eaning pie r oot ( LIV 2 page) m eaning f requency v erb t ype mṛśtouch, consider *Hmelk̂- (226) spread, touch 1 contact yajvenerate a god, sacrifice *Hi̯aĝ- (224) admire 1 mental activity yamsustain, hold *i̯em- (312) extend, stretch 3 holding / keeping yāgo, move *i̯eh 2 - (309) pull to, lead 7 motion rabhtake hold of, grasp ? *rebh- (496) move 2 removing labhtake, seize, catch ? *rebh- (496) move 1 removing līset free *lei̯H- (405) cuddle 1 caused motion vahcarry, convey *u̯eHĝ h - (663) recognize, respect 1 caused motion vidknow, understand *u̯ei̯d- (665) catch sight of 3 mental activity vivāsattract (desiderative of van- ‘like’) *u̯en- (680) overpower, win 1 mental activity vr̥cover, screen *Hu̯er- (227) shut, put in(to), 1 putting vr̥jturn, twist off *h 2 u̯erg- (290) turn 1 manner of motion vr̥troll *u̯ert- (691) turn 6 manner of motion vr̥dhincrease, strengthen *Hu̯eRd h - (227) bind 1 change of state vyā-, vīcover, wrap *u̯i̯eh 1 - (695) wrap, envelop 1 putting śr̥dhmock at ? uncertain ( EW- A ia 619-620) ? uncertain 1 communication <?page no="128"?> 128 4. Multiple preverbs in Vedic v erbal r oot m eaning pie r oot ( LIV 2 page) m eaning f requency v erb t ype sadsit down *sed- (513) sit 2 posture sarrun, flow *sel- (496) shoot out, jump 3 manner of motion sr̥jlet go, let fly, discharge *selĝ- (528) loose, send to 1 send to sthāstand *steh 2 - (590) step to, stand up 2 posture spaś- (paś-) see, observe *spek̂- (575) look at 1 perception verb syad- (syand-) move or flow on rapidly, drive ? uncertain ? spand ( EWA ia II 781-782) ? uncertain 1 caused motion hanstrike, beat *g u̯h en- (218) hit 1 contact / impact hāstart or spring forward *ĝ h eH- (172) change one’s posture, move 1 motion Tab. 11: Vedic verbal roots modified by multiple preverbs 27 27 Roots are cited as in Grassmann’s (1936[1873]) dictionary; alternative forms in brackets are those of Monier-Williams (1899), in case they differ from Grassmann’s. <?page no="129"?> 4.2. Multiple preverbs in numbers 129 In addition to designating motion or location verbs proper (e. g. √i- ‘walk, go’), motion verbs here are considered to include manner of motion verbs (e. g. √muc- ‘loose, set free from’), verbs of caused motion (e. g. √dhan- ‘cause to run’), transfer verbs (which can be assimilated to verbs of caused motion; e. g. √dā- ‘give’), and verbs of putting and removing (which can be assimilated to verbs of caused motion; e. g. √dhā- ‘put’, √labh- ‘take’). Location verbs include posture verbs (e. g. √sad- ‘sit’), verbs of existence (e. g. √bhū- ‘be, become, happen’), and verbs of holding / keeping (e. g. √yam- ). These verb classes comprise more than half of Vedic verbal roots modified by multiple preverbs (35 out of 56). Adding to location and motion verbs proper, one change of state verb is also included, i. e. √vr̥dh- ‘increase, grow’: this verb can be easily assimilated with motion verbs, as increasing and growing can be metaphorically regarded as an upward motion. The remaining verbal roots indicate events in which a certain concrete or abstract TR is directed away from or toward (or both) a LM . The following verb classes belong to this group: (a) perception verbs, as eyes can follow a certain direction ( √spaś- ‘watch, see, observe’; cf. Danesi 2009: 107-116); (b) emission verbs, by which the TR is the warmth, the light, a substance, or a sound ( √arc- ‘shine’ , √tap- ‘give out heat’ , √duh- ‘milk’ , √nū- ‘sound’, √bhā- ‘shine’, √cakṣ- ‘shine, emit light’) (cf. also RIVELEX II 32, fn. 3; Danesi 2009: 64-75); (c) communication verbs, as words and utterances go from the speaker toward his / her addressee ( √gā 2 - ‘sing’, √diś- ‘show a direction’, √śr̥dh- ‘mock at’; Danesi 2009: 119-121); (d) verbs of impact, in which hits or blows behave as moving entities ( √han- ‘strike, beat’; cf. Danesi 2009: 158-175); (e) creation verbs, in which the event of creating is directed toward a certain Beneficiary ( √kr̥- ‘do, make’); (f) verbs of mental state, whereby emotions or other mental states, such as effort, joy, veneration, and attraction, can be directed toward a certain LM ( √bhūṣ- ‘strive after, use efforts for’, √mand- ‘rejoice’, √yaj- ‘venerate’; √vivās- ‘attract’; Danesi 2009: 60-64; 76-83). Two other verbs of mental state remain: one of them, √vid- ‘know’, goes back to a PIE root with the meaning of seeing (perception verb > verb of mental state), whereas the other one, √jñā- ‘know’, goes back to a PIE root meaning ‘discern, distinguish’ (and is thus similar to verbs indicating removing/ separation). 4.2.3. Attested combinations of preverbs In Vedic, there are as many as 52 combinations of multiple preverbs. These are displayed in Table 12, together with their frequencies, that is, the number of composites that contain a certain combination. <?page no="130"?> 130 4. Multiple preverbs in Vedic e xterior preverb i nterior preverb f requency áchā párā 1 áti ā́ 1 áti níṣ 1 áti prá 3 adhí ā́ 1 adhí ní 1 ádhi sám 2 adhí ví 1 ánu ā́ 4 ánu áva 1 ánu párā 1 ánu prá 9 ánu sám 2 ánu ví 1 abhí ā́ 9 antár ā́ 1 antár ví 1 antár pári 1 ápa ā́ 1 ápa ní 1 ápa párā 1 abhí prá 14 abhí sám 7 abhí úd 1 abhí ví 2 áva ā́ 1 ā́ áva 1 ā́ ní 1 ā́ nís 2 <?page no="131"?> 4.2. Multiple preverbs in numbers 131 e xterior preverb i nterior preverb f requency ā́ pári 1 ā́ prá 1 ā́ ví 1 ní ā́ 2 nís ā́ 1 pári ā́ 3 pári prá 4 pári sám 1 prá abhí 1 prá ā́ 1 práti ā́ 2 práti úd 1 práti prá 1 sám ā́ 5 sám prá 2 úd ā́ 1 úpa ā́ 3 úpa áva 1 úpa ní 1 úpa prá 5 ví ā́ 2 ví párā 1 ví prá 2 Tab. 12: Vedic combinations of preverbs and their frequencies 28 28 The first combination in Table 12 includes the preverb áchā ‘to, unto’, with which e. g. Maddonell (1916: 352) deals separately from the other preverbs based on its more restricted use (this preverb is “tolerably frequent in R̥ V […], but already very rare in AV”). <?page no="132"?> 132 4. Multiple preverbs in Vedic As shown in Table 12, in the R̥g-Veda , combinations of more than two preverbs are not attested. 29 Most combinations (32 out of 52) only occur with one verb; 9 out of 52 combinations are attested in two composites; the remaining 11 combinations are instantiated by more than two verbs (the most frequent combination has a frequency of 14 composites). These data also suggest a low degree of conventionalization and a high degree of productivity for multiple preverbs. The most frequent combinations, that is, abhí+prá ‘to, unto, against+forward, onward, forth, fore-’ (14 times), abhí+ā́ ‘to, unto, against+to, unto, at’ (9 times), ánu+prá ‘after, along, toward+forward, onward, forth, fore-’ (9 times), úpa+prá ‘to, unto, toward+ forward, onward, forth, fore-’ (5 times), and sám+ā́ ‘along, with, together+to, unto, against’ (5 times), contain either prá or ā́ as IP s. The preverbs prá and ā́ are the most frequent ones in the R̥g-Veda : prá occurs 1372 times, whereas ā́ shows up 3347 times (Casaretto 2012, 2013). These preverbs seem to be particularly prone to stack: 11 out of 56 combinations contain prá; 22 out of 56 combinations contain ā́; together they occur in more than a half of the combinations. Their high frequency as multiple preverbs might be related to their absolute high frequency. In addition, their generic semantics might also have played a role in their tendency to combine with other spatial specifications. The preverb prá indicates a generic forward motion (Path) without any inherent Goal. In contrast, the preverb ā́ expresses a generic movement directed toward the speaker, or toward the general location of the event (RIVELEX II 1 ff.; Casaretto 2013: 15; Grassmann 1936[1873]). No Vedic preverbs constitute stable double prepositions or adverbs. However, two of the most frequent preverb combinations also occur together outside the immediate preverbal context, as exemplified in (28) and (29): (28) Non-immediately preverbal abhí+prá nr̥ ̄ṇā́m u tvā nŕ̥tamaṃ gīrbhír ukthaír abhí man. gen . pl and 2 sg . acc most_manly. acc praise. ins . pl verse. ins . pl to prá vīrám arcata sabā́dhaḥ forward man. acc sing. imp .2 pl priest. nom . pl ‘Priests, glorify you, the hero, the most heroic of the heroes, with songs and praises.’ ( R̥V 3.51.4ab) 30 29 Combinations of three preverbs are known in later Vedic, as well as in Classical Sanskrit (Delbrück 1888: 435-437). 30 Cf. also R̥ V 8.49.1 and 8.69.4. <?page no="133"?> 4.2. Multiple preverbs in numbers 133 (29) Non-immediately preverbal ánu+prá pū́ṣann ánu prá gā́ ihi P. voc after forward cow. acc . pl go. imp .2 sg yájamānasya sunvatáḥ sacrifice. ptcp . prs . gen . mid press_out. ptcp . prs . gen ‘O Pūṣan, go forth after the cows of him who sacrifices and presses Soma.’ (R̥V 6.54.6ab) In the R̥g-Veda, no preverb is iterated in preverbal position. In other positions, however, preverb iteration is relatively common, as shown by Dunkel (1981a). If iterated, preverbs are univerbated, and treated as compound words proper, called āmreḍita in the Indian grammatical tradition (e. g. ápāpa R̥V 5.34.3; abhyàbhi R̥V 9.110.5; údud R̥V 4.21.9; úpopa R̥V 1.126.7; 8.51.7; 8.74.9; párāparā R̥V 1.38.6; prápra R̥V 1.40.7; 1.129.8; 1.138.1; 1.150.3; 3.9.3; 5.5.5; 5.58.5; 6.48.1; 7.6.3; 7.8.4; 8.69.1; 9.9.2; sáṃsam R̥V 10.191.1.). In spite of the lack of preverb iteration in preverbal position, there are composites in which two quasi-equivalent preverbs modify a single verbal root. A case in point is the combination ápa+párā ‘away+away’ in ápa pára √i- ‘go off’, shown in example (30), in which two preverbs indicate Source. Both the composite containing only the EP ápa √i- (31)a and the composite containing only the IP pára √i- (31)b show the similar meaning of ‘run away’ (Grassmann 1936[1873]: 192 ff.). (30) abhāgáḥ sán ápa páreto asmi having_no_share. nom be. ptcp . prs . nom away go_away. ptcp . prf . nom be. prs .1 sg ‘I have departed without a portion.’ (R̥V 10.83.5a) (31) a. Composite with the EP only: ápa √iápa-eti asyāḥ prati-cákṣiyeva [praticákṣiya iva] away_go. prs .3 sg 2 sg . gen . f against-shine. abs like ‘She goes away like a girl to be gazed upon.’ (R̥V 1.124.8b) b. Composite with the IP only: párā √ipárā ca yánti púnar ā́ ca yanti away and go. prs .3 pl back to and go. prs .3 pl ‘(The Dawns) go away and come again.’ (R̥V 1.123.12c) In addition, the combinations abhí+ā́ ‘to+to’ and úpa+ā́ ‘to+to’ contain two Goal preverbs with partially overlapping meanings (Casaretto 2010b: 98, fn. 3, and <?page no="134"?> 134 4. Multiple preverbs in Vedic references therein). These combinations are instantiated in eight composites (i. e. abhí ā́ √gā- ‘approach, come to’, abhí ā́ √car- ‘come up, approach’, abhí ā́ √tr̥ ̄- ‘pass through to, come up to’, abhí ā́ √yam- ‘aim at, attract’, abhí ā́ √vr̥t- ‘roll toward’; úpa ā́ √gam- ‘come near, come to’, úpa ā́ √yā- ‘approach, drive near’, úpa ā́ √car- ‘come near to, attend upon’). 4.3. The form of composites In Section 4.1.2.5, I discussed the accentual properties of Vedic preverbs. In what follows, I describe the rules of vowel and consonant combinations affecting Vedic composites (Section 4.3.1), and the principles governing the interactions of preverbs with other preverbal morphology (Section 4.3.2). 4.3.1. Sandhi phenomena The juxtaposition of preverbs and verbal stems results in various consonant and vowel clusters undergoing assorted assimilatory effects, i. e. the so-called sandhi effects (< sám ‘along, with, together’ + √dhā ‘put’). Old Indo-Aryan is notable for its marking of some of these sandhi effects in the script (for an overview of Old Indo-Aryan sandhi rules, see e. g. Whitney 1955[1879]: 34 ff.; Macdonell 1916: 20 ff.; Renou 1935: 32 ff.). Between the elements that constitute multiple preverb composites, the behavior of consonant and vowel clusters follows the rules of external (i. e. occurring at word-boundaries) sandhi, whenever they mismatch those of internal (i. e. occurring at morpheme-boundaries) sandhi. 31 Some combinations are exemplified in Table 13: s andhi r ule c omposite v erbal form l ocus a + i → -ení ā́ √īrnyeriré : prf .3 pl . mid R̥V 4.1.1b a + a → -āúpa ā́ √yāupā́yātaṃ : imp .2 du R̥V 7.71.2a m → ṃ | _C ádhi sám √gamádhi sáṃgata : ptcp . prf . nom R̥V 7.76.5a d → c | _c abhí úd √carabhy úc cara : imp .2 sg R̥V 8.25.21c r → ḥ | _C (-voice) antár pári √vyā - antáḥ párivīta : ptcp . prf . nom R̥V 4.1.7c Tab. 13: Examples of external sandhi in Vedic composites 31 This is not often the case, as most preverb-preverb combinations involve vowel clusters. The rules governing the coalescence of vowels are nearly the same both in internal and in external sandhi (Whitney 1955[1879]: 42). <?page no="135"?> 4.3. The form of composites 135 Occasionally, however, sandhi effects that Macdonell (1916: 42, 45) classifies as internal also occur between the IP and the verbal stem (cf. also Whitney 1955[1879]: 63). These are shown in Table 14: s andhi r ule c omposite v erbal form l ocus C (dental) → C (cerebral) | C (cerebral) áti níṣ √tanáti níṣ ṭatanyuḥ : prf . opt .3 pl R̥V 1.141.13d s → ṣ | V (ā̆) , k, r, s ā́ ní √sadā́ ní ṣīda : imp .2 sg R̥V1.104.1b 32 Tab. 14: Examples of internal sandhi in Vedic composites It is of particular significance that the boundary between the IP ní ‘down, in, into’ and the root √sad- ‘sit down’ (Table 14) is somehow perceived as internal. The semantics of ní and √sadis characterized by high solidarity, to such an extent that ní is able to push the preverb ā́ ‘to, unto, at’ farther from the verbal root (whereas ā́ usually selects the interior position, cf. Section 4.6; Papke 2010: 101; for similar effects in Homeric Greek and Old Irish composites, see McCone 2006: 181; cf. further Chapters 5 and 7). 33 The sound coalescence of sandhi occasionally obscures the morphological segmentation of composites, as in (32). Taken in isolation, the form of (32) allows for two morphological analyses: a former (a) including ā́ ‘to, unto, at’ as IP ; a latter (b) lacking it. 34 (32) Morphological ambiguity due to sandhi upā́carat ‘came here’ (R̥V 1.46.14b) - upa-ā-acarat p p -walk. impf .3 sg - upa-acarat p -walk. impf .3 sg 32 Cf. also R̥ V 1.22.8, 3.35.6, 6. 9.4, 6.40.1, 9.63.2, 9.99.8, 9.104.1, 10.104.5, 10.15.2, 10.73.9, 10.80.6, all attesting to the same composite noted in Table 14. 33 For a similar behavior, see also the composite pári ‘around’ + √ svaj- ‘embrace’: it gives forms such as pariṣvajat : imp .2 sg (R̥ V 6. 60. 10, cf. also 8.41.3 and 10.133.2), which also shows internal sandhi effects. It is again noteworthy that the meaning of the preverb pari ‘around’ is subsumed by that of the verbal stem that it accompanies. 34 Perhaps the form in (32) allows for a third reading, if one considers the injunctive with two preverbs. <?page no="136"?> 136 4. Multiple preverbs in Vedic However, the form upā́carat is by no means ambiguous in main clauses, such as in R̥V 1.46.14b: without the IP , the accent would have occurred on the preverb úpa ‘to, unto, toward’ in a main clause (cf. Section 4.1.2.5). 4.3.2. The position of preverbs with respect to inflectional affixes As happens in Homeric Greek (cf. Chapter 5), the position of Vedic preverbs interacts with other elements of the preverbal morphology. In Vedic, preverbal morphology includes reduplication and augment. Preverbs usually occur more externally than either, as exemplified in (33): (33) a. Preverbs and reduplication abhí pra-ta-sth-úḥ ep ip red -stand-3 pl ‘(they) advanced toward’ (R̥V 10.65.15b) b. Preverbs and augment úpa prā́gāt [pra-a-gā-t] ep ip aug -go-3 sg ‘(he) proceeded to’ (R̥V 1.162.7a) Reduplication appears in the following formations: the present stems of a certain class (e. g. píparti : prs .3 sg ‘brings over / to’, class III stem from √pr̥- ‘bring over / to’), nearly all perfect stems (e. g. (33)a), a large number of aorist stems (e. g. ájījanat: aor .3 sg ‘(he) has generated’, a reduplicated aorist from √jan- ‘generate’), and the intensive and desiderative secondary conjugations. For roots beginning with consonants, reduplication consists of the initial consonant of the verbal root and a vowel . 35 With roots beginning with vowels, it consists of that vowel, either alone or with a following consonant (Whitney 1955[1879]: 222; Macdonell 1916: 147). Thus, the phonetic shape of reduplication depends on that of the reduplicated verbal root (cf. further Janda & Joseph 1991). Because of this phonetic consistency with the root, reduplication not surprisingly constitutes the innermost piece of verbal morphology. As shown by the form in (34), in which two preverbs, the augment, and reduplication simultaneously occur, reduplication is even more interior than the augment: 35 If the initial consonant of the root is aspirated, reduplication contains the corresponding non-aspirated consonant (Grassmann’s Law). <?page no="137"?> 4.3. The form of composites 137 (34) Preverbs, reduplication, and augment pári prā́siṣyadat [pra-a-si-ṣyada-t] ep ip aug red -flow_rapidly-3 sg ‘(he) flowed around’ (R̥V 9.14.1) The augment marks past time reference, and characterizes the Vedic imperfect (33)b, pluperfect, and aorist, along with the conditional mood (Whitney 1955[1879]: 220 ff.). It possibly goes back to an independent particle ( *h 1 e- ‘then, at that time’; e. g. Beekes 2011: 252), which assumes the shape of a short ain Old Indic. The augment usually occurs between the preverb(s) and the verbal stem. In the R̥g-Veda , there are no exceptions to this rule. However, in later Vedic and in Sanskrit , the augment infrequently occurs between the EP and the IP (but never before the EP ) (Whitney 1955[1879]: 400). A number of these anomalous formations are given in (35): (35) a. ud-a-pra-patat : impf .3 sg ‘flew forth onto’ (R̥V in AitBr. ) b. anv-a-saṃ-carat : impf .3 sg ‘walked alongside’ ( MB h. ) c. abhy-a-ni-mantrayat : impf .3 sg ‘invited’ ( Har. ) In the later forms in (35)a-c, the placement of the augment points to a strong association between the IP and the verb. What, then, does the positioning of the augment suggest regarding the morphological status of R̥g-Vedic multiple preverbs? First, preverbs were probably not considered part of the verb, as they occur more externally than the usual outermost piece of verbal morphology, i. e. the augment. 36 Second, as prosodic (Section 4.1.2.5), and syntactic (4.5) evidence confirms, multiple preverbs mostly have the status of clitics in Vedic (for similar considerations on Classical Sanskrit and Homeric Greek, see Whitney 1955[1879]: 354 ff.; Papke 2010: 9, 94; Chapter 5). 36 The robust cross-linguistic tendency according to which inflectional affixes are the farthest from the root is usually referred to as “relevance” or “scope principle”, on which see e. g. Bybee 1985; Rice 2000. <?page no="138"?> 138 4. Multiple preverbs in Vedic 4.4. The semantics of multiple preverbs 4.4.1. Preverbs with spatial, abstract and actional meanings From a semantic standpoint, Vedic preverbs behave in the following ways: (a) they retain their spatial basic usages; (b) they develop further spatial usages or other types of lexical meanings, i. e. they make up non-compositional composites together with the verbs that they modify; (c) they develop actional - and, in particular, telic / atelic - meanings, thus undergoing grammaticalization into actional markers. These different functions are not mutually exclusive: on the contrary, by adding an inherent point to a spatial event (or to an event which is comparable to a spatial event; cf. Section 4.2.2), preverbs at once add telic / atelic nuances to (non-)telic predicates (cf., in particular, Viti 2008a, 2008b on Homeric Greek; Danesi 2009 on Vedic; Ruvoletto 2016 on Old Russian). In combination with a motion verb (e. g. √i- ‘walk, go’) or with a verb comparable to a motion verb (e. g. √bhā- ‘shine’; see Section 4.2.2), multiple preverbs can either profile two different portions of the trajectory (36), or add semantic specifications to the same portion of it (37). 37 In the latter case, their meanings can partially overlap. For example, both composites in (37)a and (38) contain two Goal-preverbs, that is, antár+ví ‘between, among, within+apart, asunder, away, out’ (37)a and úpa+ā́ ‘to, unto, toward+to, unto, at’ (38). Whereas the preverbs of the combination in (37)a profile two different spatial regions of the Goal (i. e. ‘inside’, ‘toward different directions’) and thus do not semantically overlap, the preverbs of the combination in (38) show high semantic solidarity with respect to each other. (36) a. Goal+Source: abhí ví √i- ‘go toward from different parts’ ékaṃ krátum abhí ví yanti sādhú single. acc purpose. acc to asunder go. prs .3 pl straight ‘(They), from different parts, go straightly toward a single purpose.’ (R̥V 6.9.5d) b. Goal+Path: úpa prá √i- ‘march on, go toward’ úpa prá yantu marútaḥ sudā́navaḥ to forth go. imp .3 pl M. nom . pl munificent. nom . pl ‘May the munificent Maruts come forth to (us).’ (R̥V 1.40.1c) 37 In Homeric Greek, multiple preverbs usually, though by no means exclusively, profile the same portion of the trajectory (cf. Section 5.4.1; Iacobini et al. 2017). <?page no="139"?> 4.4. The semantics of multiple preverbs 139 c. Path+Source: ánu párā √i- ‘come away along’ páram mr̥tyo ánu párehi [parā-ihi] pánthāṃ far. acc death. voc along away-go. imp .2 sg way. gen . pl ‘O Death, go away farther along the way …’ (R̥V 10.18.1a) (37) a. Goal+Goal: antár ví √bhā- ‘shine in different directions between’ dyā́vā-kṣā́mā rukmó antár ví bhāti heaven-earth( f ). acc . du golden. nom between asunder shine. prs .3 sg ‘Golden, (he) shines in between the heaven and the earth.’ (R̥V 1.96.5c) 38 b. Source+Source: ápa párā √i- ‘pass over, go off’ (R̥V 10.83.5 = (30)) (38) úpa ā́ √gam- ‘come near to’ (R̥V 1. 91. 10 = (21)) In other composites, the IP - typically ā́ ‘to, unto, at’ - possibly reverses the deictic orientation of the base motion verb, whereas the EP adds a further spatial specification (39). Danesi (2009: 212-240), however, argues that ā́ only implies the achievement of a Goal, and is not able to change the deictic orientation of motion verbs from itive to venitive. Danesi conjectures that the venitive reading is a byproduct of ā́, frequently indicating a Goal located in the speaker’s vicinity. In fact, ā́ clearly changes the deictic orientations of the transfer verb √dā- ‘give’ into ā́ √dā- ‘take, receive, such as in ádhi ā́ √dā- ‘take away from above’ (39)c, and of the verb of putting √dhā- ‘put’ into ā́ √dhā- ‘receive’ (39)d. 39 (39) a. nís ā́ √i- ‘come out, come forward’ vs. nís √i- ‘go out’ b. úpa ā́ √gam- ‘come near, come to’ vs. úpa √gam- ‘go near to’ c. ádhi ā́ √ dā- ‘take away from above’ vs. √dā- ‘give’ d. antár ā́ √dhā- ‘receive into, contain’ vs. √dhā- ‘put’ It is also possible for a preverb retaining a spatial meaning to combine with another preverb developing a lexical meaning of a different type. Neither meaning exclusively selects the interior or the exterior position. So, for example, in 38 As pointed out by Leonid Kulikov (p.c.), in this context, antár and ví might also profile Path, rather than Goal. In fact, the distinction between these two spatial roles is not always clear-cut. 39 The composite ā́ √dhāalso has other meanings such as ‘put down, place on’ in the R̥ g- Veda that do not imply a change in its orientation. <?page no="140"?> 140 4. Multiple preverbs in Vedic (40), the EP s show abstract meanings, whereas the IP s show spatial meanings; for 41, the opposite is true. However, the EP usually retains a spatial meaning when the spatial notion is further specified by a noun phrase, such as ananté : loc ‘boundless (region)’ in (41)d. (40) EP = non-spatial; IP = spatial a. Comitative+Goal: sám ā́ √i- ‘come together, approach together’ b. Again+Path: práti prá √yā 2 - ‘go back, return’ c. Successively+Path: ánu prá √muc- ‘let loose successively’ yát sīm ánu prá mucó badbadhānā́ when 3 sg / pl . acc after forward loose. inj . aor .2 sg bind. ptcp . prs . acc . pl . f . mid ‘When you let go forth the prisoned ones successively.’ (R̥V 4.22.7c) (41) EP = spatial; IP = non-spatial antár pári √vyā- ‘hide in’ ananté antáḥ pári-vīta ā́gāc boundless( n ). loc between covering-hide. ptcp . pst . nom . pass come_toward. impf .3 sg ‘He came hidden in the boundless (region).’ (R̥V 4.1.7cd) In (40)a, the EP sám displays its usual Comitative meaning; in (40)b, práti ‘in reverse direction’ develops the Time meaning of ‘again, returning activity’; in (40)c, ánu, whose basic meaning is ‘after’ in a spatial sense, comes to mean ‘successively’: the ‘after’ notion is shifted from the spatial to the temporal conceptual domain. In (41), the semantics of the IP pári ‘round about, around’ is partially bleached: its meaning of ‘around’ is subsumed by the verbal root √vyā- ‘envelop’ (cf. Section 4.4.3). Thus, the IP arguably comes to imply a lack of visibility, according to the following inference: being all around ( pári ) can imply obscuring the visual access to a certain entity ( LM ). In addition, two non-spatial but still lexical preverbs can combine, as shown in examples (42): (42) Two non-spatial lexical preverbs a. Addressee+‘loudly’: abhí prá √arc- ‘sing loudly to’ tám u abhí prá arcata 3 sg . acc ptc to forth sing. imp .2 pl ‘And Sing loud to him [= Indra].’ (R̥V 8.92.5a) <?page no="141"?> 4.4. The semantics of multiple preverbs 141 b. Excess+metaphorical Path: áti prá √vr̥dh- ‘outgrow’ c. Again+back: ví párā √i- ‘go back (home) again’ In (42)a, the IP prá means ‘loudly, openly’: the semantic shift from ‘forward’, forth’ to ‘loudly, openly’ can also be observed in Old Church Slavic (cf. Chapter 6), as well as in other Indo-European languages (e. g. Lat. pro-clamō ‘cry out, appeal noisly, pro-claim’). In (42)b, the same IP instead describes the metaphorical Path of growing ( √vr̥dh- ‘increase, augment’). The EP s of (42)a-b develop the following meanings: abhí ‘to, unto, against’ centralizes the Addressee-participant; áti ‘across, beyond, past’ expresses excess , that is, going beyond ( áti ) certain metaphorical limits ( LM ). In (42)c, the Source-preverb párā comes to indicate the way ‘back’ (cf. the Greek Source-preverb apo- ‘away from’, which can also undergo the same semantic shift). The EP ví ‘apart, asunder, away, out’ is etymologically connected to a PIE adverb *du̯is ‘in two’ (cf. Kulikov 2007: 723). Accordingly, it frequently means ‘in two different places’ or - shifting to the temporal domain - ‘in two different times’, as it seems to do in (42)c. 40 As noted in Section 4.1.2.2, Vedic preverbs occasionally also behave as actional markers. This is particularly evident for Goal-preverbs, which are able to specify the endpoint of a motion event. The following comparison is instructive between the composites ā́ prá √dru- (43)a and prá √dru- (43)b: (43) a. prá tú drava mátsvā sutásya forth ptc run. imp .2 sg rejoice. imp . aor .2 sg . mid soma_juice. gen gómataḥ mixed_with_milk. gen ‘Run forth, rejoice of the juice mixed with milk! ’ (R̥V 8.13.14ab) 41 b. ā́ prá drava harivo to forth run. imp .2 sg possessing_bay_horses. voc ‘Run forth to (us), Lord of Bays! (Be not ungracious: visit us, lover of goldhued oblation.)’(R̥V 5.31.2a) 40 As observed by Leonid Kulikov (p.c.), the composite ví párā √ionly occurs in R̥ V 10. 85. 33, the so-called wedding-hymn. Thus, as a hápax , its meaning is not completely straightforward: e. g. Geldner (1951-1957) interprets it as ‘go away and asunder’. In the latter case, the two preverbs would profile two different portions of the trajectory, i. e. Source+Goal (cf. (36)a). 41 The composite prá √druis interrupted by the particle tú , in this passage. This type of interruption, or tmesis, is less significant than a tmesis involving lexical words. On the <?page no="142"?> 142 4. Multiple preverbs in Vedic The root √druencodes an intransitive and atelic manner of motion verb (Danesi 2009: 135). In combination with the Path-preverb prá ‘forward, onward, forth, fore-’, the predicate remains atelic, and means ‘go / come forth’ (43)a. With the further addition of the Goal-preverb ā́ ‘to, unto, at’, however, a Goal of motion close to the speaker’s position becomes inherent: in (43)b, the Goal is the first person plural pronoun naḥ : 1 pl . acc (which occurs in the following pāda ). Not only Goal-preverbs express resultativity or telicity in Vedic. Source-preverbs can also have this function, as happens with ápa ‘away, forth, off’ in ápa ní √lī ‘disappear completely’: (44) bhíyaṃ dádhānā hŕ̥dayeṣu śátravaḥ fear. acc put. ptcp . prs . nom . pl . mid heart. loc . pl enemy. nom . pl párājitāso ápa ní layantām conquer. ptcp . pst . nom . pl . pass away down dissolve. imp .3 pl . mid ‘And let (our) enemies, who got terror in their minds, disappear away defeated.’(R̥V 10.84.7cd) The reason Source-preverbs can also express completion involves events being conceptualized as locations : departing from an event (i. e. a location ) implies that such an event is completed (Zanchi 2017b). 42 Preverbs expressing actional meanings are not linked to the exterior position. In the composite pári sám √hā 2 - ‘rise up from’, the IP has the actional meaning of intensification / completeness (cf. Section 4.1.2.2), whereas the EP still exhibits the spatial value of Source. In (45), the Source component of movement is further specified by the noun phrase vidyúto : abl ‘flash of lightning’. (45) vidyúto jyótiḥ pári saṃjíhānam flash_of_lightning( f ). gen light( n ). nom around spring_out. ptcp . prs . nom . n ‘light springing out from a flash of lightning’ (R̥V 7.33.10a) Other types of actional meanings can also be expressed by both EP s and IP s. In (46) for example, the EP prá ‘forward, onward, forth, fore-’ has ingressive value. difference between the so-called “lexical” and “non-lexical” tmesis in Greek, see Chapter 5 and Bertrand (2014). 42 In this case, in addition, the semantic solidarity between ápa and √ lī arguably favored the reanalysis of the EP as an actional marker (so-called Vey-Schooneveld effect: see especially Chapter 6). <?page no="143"?> 4.4. The semantics of multiple preverbs 143 (46) índram abhí prá gāyata I. acc to forth sing. imp .2 pl ‘(Here, come here, sit down.) Start singing to Indra! ’ (R̥V 1.5.1b) 4.4.2. Same (combination of) preverbs, different meanings Preverbs are polysemous elements, whose meaning can change based on the verbal root to which they attach: for example, we have seen that prá can indicate Path (cf. (36)b and (43)a-b), and metaphorical Path (cf. (42)b) with motion verbs or verbs that can be assimilated to motion verbs. The same preverb can also mean ‘loudly’ with communication verbs (cf. (42)a), or undergo grammaticalization into a marker of ingressivity (cf. (46)b). The range of functions that prá performs in the R̥g-Veda extends far beyond those outlined thus far (cf. the summary in Table 18). From its basic usage of ‘forward, onward’, two abstract meanings derive: (a) ‘before’ in the sense of being preposed to something, as in abhí prá √bhū- ‘to_before_be’ → ‘assist, help’; (b) ‘before’ in the sense of being metaphorically ‘in front of ’, as in áti prá √śr̥dh- ‘bring in front of in excess’ (this composite is used to describe an extremely zealous priest, who boldly sings his praises to Indra; cf. R̥V 8.13.6b). The latter meaning ‘in front of ’ provides the link from the basic meaning of prá and its Beneficiary reading, attested in ánu prá √yaj- ‘win for oneself ’. In this composite, ánu has a resultative meaning, whereas prá centralizes the Beneficiary of the event of winning or offering (cf. prá √yaj- ‘offer’(+ acc )). Finally, prá is bleached so as to function as a mere intensifier in abhí prá √mand- ‘feverishly await, confuse, infatuate’ (cf. √mand- ‘rejoice’). Owing to space limitations, it is impossibile to discuss all semantic shifts undergone by each preverb occurring in multiple preverb composites. These developments are summarized in Table 18. Here, I only focus on particularly interesting cases, i. e. those of ní ‘down, in, into’, pári ‘round about, around’, and ví ‘apart, asunder, away, out’. One intriguing preverb is ní. It retains its basic spatial usage ‘downward’ both in exterior and in interior positions, as in ní ā́ √īr- ‘set someone down’ and in ā́ ní √sad- ‘sit down on, cause to sit down, establish’. Shifting to the metaphorical plane, ní can indicate lack of control, according to the following cluster of metaphors: having control or force is up , being subject to control or force is down (Lakoff & Johnson 1980: 15). Such a metaphorical meaning is instantiated in ní ā́ √kr̥- ‘hold back’ (cf. ā́ √kr̥- ‘drive near). In another metaphorical extension, an entity (TR) placed in a lower position can be difficult to see: hence, the meaning that ní exhibits in ápa ní √lī- ‘hide oneself, disappear completely’ (cf. ní √lī- ‘hide oneself, conceal oneself from (+ abl )). <?page no="144"?> 144 4. Multiple preverbs in Vedic The preverb pári also shows very interesting semantic developments. Its basic usage is instantiated in the composite pári prá √dhanv- ‘flow forth around’, in which the Goal-preverb pári ‘around’ and the Path-preverb prá ‘forth’ are added to the manner of motion verb √dhanv- ‘flow, move rapidly’. It also conveys a telic nuance of meaning in ā́ pári √vr̥- (literally) ‘to_around_cover’ → ‘surround with’: surrounding meaning totally encompassing a certain entity (the telic reading is probably triggered by the overlap of the semantics of pári ‘around’ with the meaning of the verbal root √vr̥- ‘cover’; cf. Section 4.6.2 and Chapter 6 on the so-called Vey-Schooneveld effect). Another composite, that is, antár pári √vyā- ‘hide in’ demonstrates that ní ‘down, in, into’ is not the only preverb associated with the notion of lack of visibility. The preverb pári also carries this semantic contribution, via the metaphorical shift that can be schematized as follows: around > all around > covering > lack of visibility (cf. Section 4.4.1). Lastly, pári can indicate the Source of movement, as in pári sám √hā 2 - ‘rise up from’. However, this meaning is tied to the simultaneous occurrence of a noun phrase in the ablative case (cf. (45); see also Sections 4.1.2.7 and 4.5.2 on the still widespread concrete usages of Vedic adpositionless cases). As already remarked in Section 4.4.1, the preverb ví can probably be traced back to PIE *du̯is- ‘in two’ (cf. Lubotsky 1994: 202 ff.; Kulikov 2007: 723; contra EWA ia II 550). Accordingly, it can indicate an activity oriented toward (two) different Goals, as in ví prá √i- ‘go forth in different directions, disperse, spread out’ (in exterior position), or a Source from different directions, as in abhí ví √i- ‘come toward from different parts’ (in interior position). From this basic meaning, the notion of covering can be easily derived via the following shift: ‘in two directions’ > ‘in all directions’, as happens in ví prá √sr̥- ‘spread’ (cf. Gr. amphí ‘ on both sides’ > ‘on all sides’ > ‘around’; Luraghi 2003: 256; Chapter 5). The meaning of ‘through’, instantiated in ví ā́ √sr̥- ‘run through’, is also connected to the basic meaning: the way ‘through’ is the shortest way connecting Goal 1 and Goal 2 (cf. Zanchi 2017b). Moreover, the preverb ví indicates division or separation, as in ví ā́ √kr̥- ‘divide, separate from’: ‘in two directions’ > ‘in two different directions’. Its usage as a Source-preverb (cf. adhí ví √kṣar- ‘pour out, flow out’) is linked to the generic meaning of separation just outlined. Separation and division can also result in distinction, which is another possible semantic development of ví, exemplified in ā́ ví √vid- ‘know by distinguishing’. Moving from the spatial to the temporal plane, i. e. ‘in two directions’ > ‘in two times’ (cf. fn. 38 in this Chapter), one easily gets to the meaning of ‘again’, as in the composite ví párā √i- ‘go back (home) again’. Both this temporal meaning and its Source usage contribute to explaining the very widespread reversative employment of ví. This use is instantiated in the polysemous composite ví ā́ √ <?page no="145"?> 4.4. The semantics of multiple preverbs 145 kr̥-, which can among others mean ‘undo’ (some of the meanings just outlined are also discussed in Kulikov 2007). Clearly, as preverbs are polysemous morphemes, their combinations are also polysemous. For example, abhí+prá shows different meanings if combined with motion verbs, such as √i- ‘walk, go’, or with communication verbs, such as √arc- ‘sing’ or √nū- ‘roar, yell’. With √i- , the preverbs indicate Goal+Path, whereas with √arcand √nūthe EP introduces the Addressee of singing, and the IP means ‘loudly’ (cf. (42)a). With another communication verb, i. e. √gā 2 - (gai-) ‘sing’, the same combination has a different meaning, as the IP expresses ingressivity (cf. (46)). With √cakṣ- ‘shine, see, appear’, abhí points to the entity that is seen (Stimulus), whereas prá indicates the fictive Path of the eye directed toward that particular Stimulus. Together, these elements make up the composite abhí prá √cakṣ- ‘look at, see’. All in all, what these examples clearly suggest is that each preverb independently develops a range of meanings, and that these meanings contribute to the formation of multiple preverb composites as independent units. There do not seem to be specific meanings associated to specific combinations of preverbs. 4.4.3. Different degrees of compositionality It is not always easy to assess the degree of compositionality of Vedic composites. To begin with, multiple preverb composites can show different meanings in different contexts, which can also exhibit various degree of compositionality. Among such polysemous composites are the following: ví prá √i- ‘go forth in different directions (compositional), disperse, spread out (non-compositional)’; sám ā́ √kr̥- ‘bring together, gather (compositional), prepare (non-compositional)’; abhí sám √car- ‘go together to (compositional), seek for (non-compositional) (cf. examples (47)a-b)’; upá ā́ √car- ‘come near to (compositional), attend upon (non-compositional)’; ā́ ní √sad- ‘sit down on, cause to sit down, establish (non-compositional)’. (47) a. Compositional abhí sám √car- ‘go together to’ samānáṃ vatsám abhí saṃ-cárantī same. acc calf. acc to with-go. ptcp . prs . nom . du . f ‘They go together to (their) common calf.’ (R̥V 1.146.3a) <?page no="146"?> 146 4. Multiple preverbs in Vedic b. Non-compositional abhí sám √car ‘seek for’ anyásya cittám abhí saṃ-caréṇiyam another. gen thinking. acc to with-go. ptcp . fut . nom / acc ‘(We must) seek for another’s thought.’ (R̥V 1.170.1c) In addition, though a certain composite results in a non-compositional fomation, it might be the case that the semantic contribution brought about by each of its parts is still traceable (certainly by the linguist, and possibly by the speaker as well). For example, this is the case with anú ví √sthā- (literally) ‘along_in different directions _stay’ → ‘extend over’ and áti prá √sr̥- (literally) ‘beyond_forth_run’ → ‘outstrip, surpass’. Lastly, the semantic contribution of certain preverbs, though detectable, is redundant. One good example for that is antár pári √vyā- (literally) ‘between_around_envelop’ > ‘hide in’, in which the meaning of covering ( √vyā- ) partly subsumes the semantic contribution brought about by the IP pári ‘round about, around’ (cf. Section 4.4.2). I also consider as partially compositional all composites in which at least one element retains a detectable meaning - be it the EP , the IP , or the verbal root. The partially compositional composites are displayed in Table 15. All composites that defy clear-cut categorization are classified as ‘partially compositional’. Not surprisingly, about half of Vedic composites (51 out of 116) belong to this group. c omposite m eaning ápa pára √igo off abhí ā́ √icome to, approach ā́ nís √igo off, depart nís ā́ √igo off, depart pári ā́ √icirculate ví prá √igo forth in different directions, disperse, spread out ní ā́ √īrset someone down sám ā́ √īrput together, create sám ā́ √kr̥bring together, gather, prepare ví ā́ √kr̥undo, sever, divide, separate from adhí ví √kṣarpour out, flow out ádhi sám √gamgo up to, approach together <?page no="147"?> 4.4. The semantics of multiple preverbs 147 c omposite m eaning úpa ā́ √gamcome near, come to abhí ā́ √gāapproach, come to ánu sám √carwalk alongside, visit, seek after abhí ā́ √carcome up, approach abhí sám √cargo together to, seek for úd ā́ √carrise out of úpa ā́ √carcome near to, attend upon práti ā́ √tanextend in the direction of, shine upon / against abhí ā́ √tr̥ ̄pass through to, come up to abhí ā́ √diśaim at (in hostile manner) ā́ nís √duhcreate out of abhí prá √dr̥ ̄put forth by bursting or opening adhí sám √dhāstore up antár ví √bhāshine in different directions between ánu prá √muclet loose successively abhí prá √mr̥śseize, grasp prá abhí √mr̥śseize, grasp áti ā́ √yādrive by úpa ā́ √yācome near, approach prá ā́ √yācome near, approach ánu sám √rabhtake hold of abhí sám √rabhtake hold of ánu ā́ √labhlay hold of grasp, handle, take in the hand ánu prá √vahgo, get forward abhí ā́ √vanstrive, seek to win ā́ pári √vr̥surround with ápa ā́ √vr̥jwipe out, bring away abhí ā́ √vr̥troll toward, hurry toward pári ā́ √vr̥tturn round, turn away from, return to práti ā́ √vr̥tturn against <?page no="148"?> 148 4. Multiple preverbs in Vedic c omposite m eaning sám ā́ √vr̥tturn back, come back, return áti prá √vr̥dhoutgrow antár pári √vyāhide in abhí prá √sadsit down, settle along ā́ ní √sadsit down on, cause to sit down, establish ví ā́ √sr̥run through ánu ví √sthāextend over ábhi ví √spaślook at, view, look hither pári sám √hā 2 rise up from Tab. 15: Vedic partially compositional composites As noted earlier in this Section, a number of partially compositional composites show redundancy of some kind. Either the meanings of the preverbs overlap with each other (cf. (37) above, (48)); or the meaning of one of the preverbs, usually the IP , shows semantic solidarity with that of the verbal stem to which it attaches (49). (48) Composites containing preverbs with overlapping meanings abhí ā́ √i- ‘come to, approach’ (Goal+Goal) ánu prá √vah- ‘go, get forward’ (Path+Path) (49) Composites containing a redundant preverb a. ví prá √i- ‘go forth in different directions, disperse, spread out’ b. ví prá √sr̥- ‘spread’ c. abhí ā́ √diś- ‘aim at (in hostile manner)’ d. ā́ nís √duh- ‘create out of ’ e. antár ví √bhā- ‘shine in different directions between’ f. ánu sám √rabhtake hold of ’ g. ā́ pári √vr̥- ‘surround with’ h. ā́ ní √sad- ‘sit down on, cause to sit down, establish’ i. úpa ní √pad- ‘lie down beside’ In (49)a-b, the acts of going ( √i- ) and running (√ sr̥- ) imply a Path ( prá ‘forward, onward, forth, fore-‘). The root of the composite in (49)c, √ diś- ‘point out’, is per se Goal-oriented, and the composite further contains two Goal-preverbs. By contrast, the root in (49)d, √duh- ‘milk, extract’, is Source-oriented, and is combined with a Source-preverb, i. e. nís ‘out, forth’. In (49)e, √ bhā- ‘shine, be bright’ <?page no="149"?> 4.4. The semantics of multiple preverbs 149 already implies the idea of emission and diffusion, reinforced by the addition of ví ‘in two directions > in all directions’ (cf. Section 4.4.2). In (49)f, the concept of togetherness, expressed by sám ‘with, together’, is subsumed by √rabh- ‘grasp’. The notion of covering expressed by √vr̥- ‘cover’ contains the meaning of pári ‘round about, all around’(49)g. In (49)h-i, the downward direction ( ní ) is likewise implicit in the acts of sitting ( √sad- ) and of lying ( √pad- ). Non-compositionality does not always originate from the redundancy: in many composites, the meaning of the IP is bleached, which also results in partial compositionality. Cases in point are sám ā́ √īr- ‘put together, create’ and sám ā́ √kr̥- ‘bring together, gather, prepare’, in which only the meaning of togetherness provided by the EP is still detectable, whereas the IP ā́ is bleached (on the tendency of ā́ to undergo semantic bleaching, cf. Section 4.2.3). Other examples containing ā́ are the composites pári ā́ √i- ‘circulate’ and práti ā́ √tan- ‘extend in the direction of, shine upon / against’, in which only the EP pári ‘round about, around’ and práti ‘in reversed direction, back to, against, in return’ retain fully detectable meanings. The same can be also said for the many multiple preverb composites built on ā́ √vr̥t- ‘turn near, turn toward’, such as those exemplified in (50): 43 Musterdatei NFA_Basis_A.dot 145 Non-compositionality does not always originate from the redundancy: in many composites, the meaning of the IP is bleached, which also results in partial compositionality. Cases in point are sám ā́ √īr- ‘put together, create’ and sám ā́ √kr̥- ‘bring together, gather, prepare’, in which only the meaning of togetherness provided by the EP is still detectable, whereas the IP ā́ is bleached (on the tendency of ā́ to undergo semantic bleaching, cf. Section 4.2.3). Other examples containing ā́ are the composites pári ā́ √i- ‘circulate’ and práti ā́ √tan- ‘extend in the direction of, shine upon/ against’, in which only the EP pári ‘round about, around’ and práti ‘in reversed direction, back to, against, in return’ retain fully detectable meanings. The same can be also said for the many multiple preverb composites built on ā́ √vr̥t- ‘turn near, turn toward’, such as those exemplified in (50): 76 abhí ā́ √vr̥t- ‘roll toward’ (50) √vr̥t- → ā́ √vr̥t- → pári ā́ √vr̥t- ‘turn round’ ‘turn’ ‘turn near, toward’ práti ā́ √vr̥t- ‘turn against’ A similar process of accretion to that in (50) also lies behind the composite ápa ā́ √vr̥j- ‘wipe out, bring away’: the base verb √vr̥jmeans ‘bend, turn, divert’. If combined only with ā́, which expresses speaker’s vicinity, it comes to acquire the meaning of ‘bring in the possession of’: the IP defines the orientation of the motion expressed by √vr̥j- . The further addition of ápa then reverses such an orientation, resulting in ‘wipe out, bring away’. The composites that I regard as fully compositional and as non-compositional are displayed in Table 16 and in Table 17, respectively: these two groups include 37 and 26 verbs, respectively. C OMPOSITE M EANING abhí prá √arcsing loudly of áchā párā √igo away toward ánu áva √igo down after, follow ánu párā √igo away after ánu prá √igo after, follow 76 Unless underwise specified, verbs in (50) should be understood as intransitive: ā́ √vr̥ tand pári ā́ √vr̥ tcan be also used transitively, whereas ā́ √vr̥ t-, pári ā́ √vr̥ tand práti ā́ √vr̥ talso occur in the causative stem. A similar process of accretion to that in (50) also lies behind the composite ápa ā́ √vr̥j- ‘wipe out, bring away’: the base verb √vr̥jmeans ‘bend, turn, divert’. If combined only with ā́, which expresses speaker’s vicinity, it comes to acquire the meaning of ‘bring in the possession of ’: the IP defines the orientation of the motion expressed by √vr̥j- . The further addition of ápa then reverses such an orientation, resulting in ‘wipe out, bring away’. The composites that I regard as fully compositional and as non-compositional are displayed in Table 16 and in Table 17, respectively: these two groups include 37 and 26 verbs, respectively. 43 Unless underwise specified, verbs in (50) should be understood as intransitive: ā́ √vr̥ tand pári ā́ √vr̥ tcan be also used transitively, whereas ā́ √vr̥ t- , pári ā́ √vr̥ tand práti ā́ √vr̥ talso occur in the causative stem. <?page no="150"?> 150 4. Multiple preverbs in Vedic c omposite m eaning abhí prá √arcsing loudly of áchā párā √igo away toward ánu áva √igo down after, follow ánu párā √igo away after ánu prá √igo after, follow abhí prá √igo near to, approach abhí sám √iapproach together, come together at abhí ví √icome toward from different parts ā́ áva √irush down upon úpa prá √imarch on, go toward pári prá √irun through on all sides práti úd √irise and go toward sám ā́ √icome together, approach together, meet at / in / with ví párā √igo back (home) again sám prá √īrcome forth together úpa prá √gāstep near to, proceed to abhí prá √gā 2 - (√gai-) encourage to start singing about, begin to praise abhí úd √carrise over ádhi ā́ √dātake away from above ā́ prá √drurun forth here pári prá √dhanvflow forth around antár ā́ √dhāreceive into, contain ánu ā́ √nūsound here through abhí prá √nūpraise highly to abhí sám √nūrejoice together at úpa ní √padlie down beside abhí prá √bhr̥bring forth to, offer to pári ā́ √bhr̥carry near, fetch from sám prá √yamoffer together / mutually, give to <?page no="151"?> 4.4. The semantics of multiple preverbs 151 c omposite m eaning ábhi sám √yāvisit, approach to úpa prá √yāproceed toward pári prá √yāgo forth around práti prá √yāgo back, return ánu ā́ √vr̥troll near along ánu prá √vr̥tproceed along / after pári prá √syandgush around, flow forth or round áti prá √śr̥dhbring in front of in excess Tab. 16: Vedic fully compositional composites As one might expect, fully compositional composites (Table 16) contain motion, manner of motion, caused motion and location verbs (√ i- ‘walk, go’, √ īr- ‘go, move’, √ gā- ‘go to’, √ car- ‘go walk’, √ dru- ‘run’, √ dhanv- ‘cause to run’, √ pad- ‘fall’, √ bhr̥- ‘bear’, √ yā- ‘go, move’, √ vr̥t- ‘turn’, √ syand- ‘move, flow on rapidly’), as well as verbs that can be assimilated with them, such as communication verbs (√ gā 2 - ‘sing’, √ śr̥dh- ‘mock at’), verbs of emission (√ arc- ‘shine, sing’, √ nū- ‘sound’), transfer verbs (√ dā- ‘give’), verbs of putting / taking (√ dhā- ‘put’), and verbs of holding / keeping (√ yam- ‘hold, sustain’) (cf. Section 4.2.2). c omposite m eaning ní ā́ √kr̥hold back áva ā́ √gamundertake, begin abhi prá √cakṣsee úpa prá √jinvplease or gratify in approaching ánu prá √jñātrace, discover áti nís √tanpenetrate with rays abhí ā́ √taptorment, pain abhí sám √dhācompose the mind at ánu ā́ √phaṇjump ánu prá √bhūspread over abhí prá √bhūassist, help abhí sám √bhūenter, reach, come to <?page no="152"?> 152 4. Multiple preverbs in Vedic c omposite m eaning ánu prá √bhūṣserve abhí prá √mandfeverishly await, confuse, infatuate ánu prá √yajwin for oneself abhí ā́ √yamaim at sám ā́ √yamdraw, pull, stretch ápa ní √līhide oneself, disappear completely ánu prá √vidunderstand backward and forward ā́ ví √vidknow by distinguishing úpa prá √vidunderstand áti prá √sr̥outstrip, surpass ví prá √sr̥spread úpa áva √sr̥jreach over, give, bestow ábhi prá √sthāadvance toward, reach, surpass abhí prá √hanoverpower Tab. 17: Vedic non-compositional composites The group of non-compositional composites (Table 17) includes the following sub-groups: (a) motion or location verbs that develop non-spatial meanings in combination with preverbs; (b) verbal roots that cannot be assimilated with motion or location verbs ( √vid- ‘know’ and √jñā- ‘know’) . A good example for the group (a) is áva ā́ √gam- , which literally means ‘down_to_go’, but actually comes to mean ‘under-take, begin’. The outcome of the root √dhā- ‘put’ in combination with abhí ‘to, unto, against’ and sám ‘along, with, together’ is also interesting: the composite abhí sám√dhāmeans (literally) ‘put (the mind) together to’ → ‘acknowledge unanimously’. Another instructive example is abhí ā́ √yam- : in this composite, the Goal-preverbs cause the stative verb √yam- ‘hold’ to acquire directional nuances. The composite abhí ā́ √yammeans ‘aim at’, i. e. indicates a mental metaphorical motion toward a LM . Something similar can also be observed in ábhi prá √sthā- ‘advance toward, reach, surpass’, in which the stative root √sthā- ‘stand’ comes to indicate motion in combination with a Goal- ( ábhi ) and a Path-preverb ( prá ) (on the ability of Vedic preverbs to provide stative roots with dynamic nuances of meaning, cf. also Danesi 2009). Group (b) includes mental verbs, to which preverbs add the traits of intensification or telicity. For example, the root √mandmeans ‘rejoice’; if compounded <?page no="153"?> 4.4. The semantics of multiple preverbs 153 with ábhi ‘to, unto, against’ and prá ‘forward, onward, forth, fore-’, it indicates the activity of feverishly awaiting (intransitive) or infatuating (transitive). In both cases, the exact semantic contributions of the preverbs are difficult to trace. With stative verbs of knowing, such as √jñāand √vid-, preverbs also have the effect of adding dynamic nuances: ánu and prá, two Path-preverbs, result in the meaning of tracing or discovering in combination with √jñā- . The events of tracing and discovering can be interpreted as acts of knowing resulting from following a metaphorical Path toward knowledge. The root √vidinstead is found in various composites, among which ánu prá √vid-, which means ‘understand backward and forward’, i. e. ‘understand completely’. A last interesting example is ápa ní √lī- ‘hide oneself, disappear completely’. The root per se means ‘dissolve, melt, disappear’. It combines with ní ‘down, in, into’, possibly introducing the idea of dissolving down to nothingness (cf. 4.4.2); then, the EP ápa modifies the whole composite, by adding the actional meaning of completion (cf. example (44)). 4.4.4. Summarizing the meanings of preverbs in multiple preverb combinations Table 18 displays the meanings of Vedic multiple preverbs. Each meaning is exemplified by at least one composite. Many semantic developments have been discussed in Sections 4.4.1-4.4.3; those that have been omitted either because of space limitations or because they are assumed to be obvious without explanation, are also included in Table 18. p reverb m eaning e xample áti beyond over (diffusion) over (excess) áti ā́ √yā- ‘drive by’ áti nís √tan- ‘penetrate with rays’ áti prá √vr̥dh- ‘outgrow’ ádhi up for (Beneficiary) from above (Source) ádhi sám √gam- ‘go up to, approach together’ ádhi ní √dhā- ‘deposit for’ ádhi ā́ √dā- ‘take away from above’ ánu after along (Path) along / over (covering) over (Beneficiary) agreement (concede a point) resultative ánu párā √i- ‘go away after’ ánu ā́ √vr̥t- ‘roll near along’ ánu prá √bhū- ‘spread over’ ánu prá √bhūṣ- ‘serve’ ánu prá √jñā- ‘trace, discover’ ánu prá √yaj- ‘win for oneself ’ <?page no="154"?> 154 4. Multiple preverbs in Vedic p reverb m eaning e xample antár between (Goal) inside (Location) antár ví √bhā- ‘shine in different directions between’ antár pári √vyā- ‘hide in’ ápa away off away+telic ápa ā́ √vr̥j- ‘wipe out, bring away’ ápa párā √i- ‘go off’ ápa ní √lī- ‘hide oneself, disappear completely’ abhí to (Goal) to (Addressee) to (Stimulus) to (Maleficiary) to (Beneficiary) to (Recipient) to (Purpose) against over (Goal) intensive abhí ā́ √i- ‘come to, approach’ abhí prá √arc- ‘sing loudly of ’ abhi prá √cakṣ- ‘see’ abhí ā́ √tap- ‘torment, pain’ abhí prá √bhū- ‘assist, help’ abhí prá √ bhr̥- ‘bring forth to, offer to’ abhí sám √dhā- ‘compose the mind at’ prá abhí √mr̥ś- ‘attack, tackle’ abhí úd √car- ‘rise over’ abhí prá √han- ‘overpower’ áchā toward (Goal) áchā párā √i- ‘go away toward’ áva downward (Path) off (Source) ingressive ā́ áva √i- ‘rush down upon’ úpa áva √sr̥j- ‘reach over, give, bestow’ áva ā́ √gam- ‘undertake, begin’ ā́ Goal here (subject’s vicinity) (Goal) subject’s interest (+ mid ) to back reversative intensity telic ā́ áva √i- ‘rush down upon’ ā́ prá √dru- ‘run forth here’ sám ā́ √i- ‘come together, approach together’ ā́ nís √duh- ‘create out of ’ ā́ pári √vr̥- ‘surround with’ sám ā́ √vr̥t- ‘turn back, come back, return’ ádhi ā́ √dā- ‘take away from above’ ā́ ví √vid- ‘know by distinguishing’ ní ā́ √īr- ‘set someone down’ úd upward (Goal) upward+out of (+ abl ) (Source) práti úd √i- ‘rise and go toward’ úd ā́ √car- ‘rise out of ’ úpa toward (Path) beside (Location) beside (Goal) under- (as in understand) telic úpa prá √i- ‘march on, go toward’ úpa ní √pad- ‘lie down beside’ úpa ā́ √car- ‘come near to, attend upon’ úpa prá √vid- ‘understand’ úpa áva √sr̥j- ‘reach over, give, bestow’ <?page no="155"?> 4.4. The semantics of multiple preverbs 155 p reverb m eaning e xample ní downward (Path) down as lacking control down (lacking visibility) ní ā́ √īr- ‘set someone down’ ā́ ní √sad- ‘sit down on, cause to sit down, establish’ ní ā́ √kr̥- ‘hold back’ ápa ní √lī- ‘hide oneself, disappear completely’ nís away / forth (Source) out of (creation) division nís ā́ √i- ‘go off, depart’ ā́ nís √duh- ‘create out of ’ áti nís √tan- ‘penetrate with rays’ párā away (Source) back áchā párā √i- ‘go away toward’ ví párā √i- ‘go back (home) again’ pári around (Goal) around+telic lack of visibility / imprisonment from (+ abl ) pári prá √dhanv- ‘flow forth around’ ā́ pári √vr̥- ‘surround with’ antár pári √vyā- ‘hide in’ pári sám √hā 2 - ‘rise up from’ prá forth, forward (Path) metaphorical Path openly, loudly before (be pre-posed) in front of (Goal) in front of (Beneficiary) ingressive intensive telic abhí prá √i- ‘go near to, approach’ áti prá √vr̥dh- ‘outgrow’ abhí prá √arc- ‘sing loudly of ’ abhí prá √bhū- ‘assist, help’ áti prá √śr̥dh- ‘bring in front of in excess’ ánu prá √yaj- ‘win for oneself ’ abhí prá √gā 2 - (√gai-) ‘begin to praise’ abhí prá √mand- ‘feverishly await, confuse, infatuate’ abhí prá √han- ‘overpower’ práti toward (Goal) against (Goal) back, again (returning activity) práti úd √i- ‘rise and go toward’ práti ā́ √tan- ‘shine upon / against’ práti prá √yā- ‘go back, return’ ví toward different directions (Goal) from different directions (Source) covering through (Path) division (pseudoreversative) out (Source) precisely (in distinguishing) again (in two times) ví prá √i- ‘ go forth in different directions’ abhí ví √i- ‘come toward from different parts’ ví prá √sr̥- ‘spread’ ví ā́ √sr̥- ‘run through / over’ ví ā́ √ kr̥- ‘undo, sever, divide, separate from’ adhí ví √kṣar- ‘pour out, flow out’ ā́ ví √vid- ‘know by distinguishing’ ví párā √i- ‘go back (home) again’ <?page no="156"?> 156 4. Multiple preverbs in Vedic p reverb m eaning e xample sám with (Comitative) vicinity (Goal) mutually intensive telic abhí sám √i- ‘approach together, come together at’ ábhi sám √yā- ‘visit, approach to’ sám prá √yam- ‘offer together / mutually, give to’ pári sám √hā 2 - ‘rise up from’ sám ā́ √vr̥t- ‘turn back, come back, return’ Tab. 18: The meanings of Vedic preverbs in multiple preverb combinations 4.5. The syntax of multiple preverbs 4.5.1. Movable preverbs As noted in Section 4.1.2, Vedic preverbs are not morphologically bound to the verb. On the contrary, for syntactic or pragmatic reasons, preverbs can be displaced from the immediate preverbal position, that is, the unmarked position for preverbs with verbal orientation. Tmesis is very widespread in Vedic, and so little does it affect the semantic behavior of preverbs that dictionaries, such as Grassmann’s (1936[1873]), even register as lexical entries the composites that never attest to two preverbs in immediate preverbal position. For example, putative composites of this type are listed under the root √i- ‘walk, go’ To the combinations included in my sample, which feature the order #P_P_V#, Grassmann’s dictionary adds examples demonstrating that √ican be modified by the following multiple preverbs (after each composite, the attested preverb ordering is presented): (51) a. abhí ní √i- ‘come up, have sexual intercourse with’ #P …P_V# b. abhí úd √i- ‘rise over’ #P_V …P# c. abhí úpa √i- ‘come closer to’ #P_V …P# d. ápa prá √i- ‘pass over, distance oneself from’ #P …P_V# e. prá áti √i- ‘march past’ #P …P_V#, #P …P_V#, #P …P …V# f. sám abhí √i- ‘come near to’ #P …P_V# g. sám prá √i- ‘get ahead together’ #P …P_V# <?page no="157"?> 4.5. The syntax of multiple preverbs 157 h. úd ā́ √i- ‘come out, come upstairs’ #P …P …V# i. úpa ā́ √i- ‘come up, come close to, try to gain’ #P …P_V#, #P …P …V# Among composites in (51)a-i, the verb in (51)i is interesting, as it develops the non-compositional meaning of ‘try to gain’, besides retaining its basic use of ‘come up, come close to’. As shown in (52), úpa ā́ √idevelops a non-compositional meaning even though the EP úpa ‘to, unto, toward’ is displaced from the rest of the composite. Exceptionally, in (52), tmesis is possible even with a non-finite verbal form (cf. Section 4.1.2.5). (52) índur devā́nām úpa sakhyám ā-yán I. nom god. gen . pl to friendship. acc p -go. ptcp . prs . nom ‘Indu, trying to gain the friendship of the gods, …’ (R̥V 9.97.5a) For a number of occurrences with the composites of (51)a-i, it is difficult to determine whether a displaced preverb has nominal or verbal orientation. One such occurrence is exemplified in (53), containing the composite ápa prá √i- ‘pass over, distance oneself from’ (cf. (51)d): (53) ápāsmāt [ápa asmāt] préyān [prá iyāt] away dem . abl . n forward go. opt .3 sg ‘May (he) depart from here - (no home is that to rest in.)’ (R̥V 10.117.4c) In (53), the demonstrative pronoun in the ablative case, asmāt, which immediately follows and specifies the Source-preverb ápa, splits the composite ápa prá √i-. Though only rarely, the preverb ápa can also be used adnominally in combination with the ablative case, such as asmāt (Delbrück 1888: 446). Thus, the actual function of ápa is difficult to assess for (53), as its positioning allows for divergent readings (however, Hettrich and colleagues argue that orders such as that in (53) usually suggest adnominal orientation; cf. Section 4.1.2.4). Even composites with multiple preverbs in immediate preverbal position frequently allow for preverb displacement. Out of 116 composites, 73 occur in passages in which at least one of the preverbs is not immediately preverbal. Such freedom of position is exemplified by means of the composite úpa ā́ √yā- ‘come near, approach’. As shown in (54), the elements of the composite can be arranged in various ways within the Vedic sentence: <?page no="158"?> 158 4. Multiple preverbs in Vedic (54) a. upā́yātaṃ (R̥V 7.71.2a) #P_P_V# b. ā́ yātam úpa (R̥V 1.2.6c) #P_V_P# c. índrā́ yāhi tū́tujāna úpa (R̥V 1.3.6ab) #P_V …P# d. ā́ indra yāhi úpa naḥ parāváto (R̥V 1.130.1ab) #P … VP # e. úpem ā́ yāta mánasā juṣāṇā́ (R̥V 1.171.2c) #P … PV # f. ā́ na stutá úpa vā́jebhir ūtī́ índra yāhí (R̥V 4.29.1) #P …P …V# 4.5.2. Optional preverbs As explained in Section 4.1, the various publications by Hettrich and colleagues include a section in which the authors discuss the omissibility ( Weglassbarkeit in their terms) of a certain preverb. All in all, without affecting the syntax, preverbs-adpositions can be omitted in the vast majority of contexts, and their functions taken over by adpositionless cases. This is possible because Vedic cases retain many of their concrete meanings (cf. Section 4.1.2.7). As Hettrich and colleagues take into account all instances of a certain preverb, including those in which a preverb co-occurs with other preverbs, their findings also hold for multiple preverb composites, as I summarize in what follows. First, there are composites that do allow for the omission of the EP , though they undergo detectable semantic changes owing to the presence of that EP . In this respect, an instructive pair is práti prá yā- ‘go forth back’ and prá √yā- ‘go forth to’. (55) práti prá yāhīndra [yāhi indra] mīḷhúṣo nr̥ ̄ ́n in_reverse_direction forth go. imp .2 sg I. voc bountiful. acc . pl man. acc . pl ‘Drive forth back toward the men who grant rewards [= Maruts], Indra.’ (R̥V 1.169.6a) (56) pró[prá u] ayāsīd índur índrasya niṣkr̥táṃ forth ptc go. aor .3 sg drop. nom I. gen . sg place_of_rendezvous. acc ‘Indu has proceeded forth to Indra’s place of rendezvous.’ (R̥V 9.86.16a) Both composites in (55) and (56) take the accusative indicating the Goal, that is, mīḷhúṣo nr̥ ̄ ́n ‘bountiful men’ and niṣkr̥táṃ ‘place of randezvous’. The addition of <?page no="159"?> 4.5. The syntax of multiple preverbs 159 práti ‘in reversed direction, back to, against, in return’ thus does not affect the syntax of the composite in (55), though it does change its meaning: práti adds the notion of ‘in reversed direction, back’. The adpositionless accusative can also express Goal even in combination with composites containing only Source-preverbs. This is the case of párā √i- ‘go away toward’, as shown in (57). (57) párehi vígram ástr̥tam go_away. imp .2 sg strong. acc invincible. acc ‘Go away to the spirited and indestructible [= Indra].’ (R̥V 1.4.4a) In (57), the adpositionless accusative vígram ástr̥tam , Indra’s attributes, expresses the Goal of motion. This construction is possible even if the composite párā √ionly contains the Source-preverb párā ‘to a distance, away, forth’, without an additional Goal-preverb. Clearly, it is the concrete value of cases that imposes an orientation to the motion event. (58) √i- ‘go, walk’ + ACC kaníyeva [kanyā̀ iva] tanúvā śā́śadānām̐ éṣi devi devám girl. nom like body. ins fall. ptcp . prf . nom . f go. prs .2 sg goddess. nom god. acc ‘Like a girl exulting in her body, you go, o goddess, to the god (who seeks to attain you [= Sun]).’ (R̥V 1.123.10ab) (59) √i - + ABL prayuñjatī́ divá eti yoke_to. ptcp . prs . nom . f sky. abl go. prs .3 sg ‘Hitching up, she goes away from heaven.’ (R̥V 5.47.1a) The bare root √i- ‘walk, go’ can be combined both with the adpositionless accusative, devám (58), and with the adpositionless ablative (59), resulting in the opposite meanings ‘go toward’ and ‘go away’ respectively. 4.5.3. Vedic preverbs as transitivizing morphemes As discussed by Danesi (2009: 249) for single preverbation, Vedic preverbs also seem to function as applicatives (on this terminology, cf. Section 3.1.1; Austin 1997; Shibatani 2000; Petersen 2007). This transitivizing function is allegedly <?page no="160"?> 160 4. Multiple preverbs in Vedic connected to their ability to telicize stative and activity predicates. A case in point is the emission verb √bhā- ‘shine, emit light’, which is intransitive in its absolute usages (60), but can take the accusative in combination with the preverbs antár ‘between, among, within’ and ví ‘apart, asunder, away, out’ (61). (60) citrábhānur uṣásām bhāti ágre shining_with_light. nom dawn( f ). gen . pl shine. prs .3 sg foremost( n ). loc ‘Shining with light, he shines at the head among dawns.’ (R̥V 7.9.3c) (61) dyā́vā-kṣā́mā rukmó antár heaven-earth. nom . du golden. nom between ví bhāti asunder shine. prs .3 sg ‘Golden, (he) shines in between the heaven and the earth.’ (R̥V 1.96.5c = (37)) Thus, one might be inclined to think that the preverbs antár and ví allow the occurrence of the second argument in the accusative case in (61). However, a number of emission verb roots, including √cakṣ- ‘emit light, shine’ (62) and √tap- ‘give out heat’ (63), do not require preverbs to be employed as telic and transitive predicates. In fact, the addition of an adpositionless directional accusative / dative, such as áditiṃ dítiṃ ca ‘Aditi and Diti’ in (62) and nas ‘us’ in (63), can also result in the same effect. (62) Transitive √cakṣ- ‘see, look at’ ( √cakṣ- ‘emit light, shine’) átaś cakṣāthe áditiṃ dítiṃ ca from_this shine. prs .2 du A ( f ) . acc D ( f ) . acc and ‘And from that place, (O Varuṇa and Mitra,) look at Aditi and Diti.’ (R̥V 5.62.8d) (63) Transitive √tap- ‘warm up’ ( √tap- ‘give out heat’) śám agnír agníbhiḥ karact śáṃ together A. nom fire. ins . pl do. sbjv . aor .3 sg together nas tapatu sū́riyaḥ 1 pl . acc / dat / gen give_out_heat. imp .3 sg S. nom ‘Agni will behave auspiciously with his fires, and Surya warm us up pleasantly.’ (R̥V 8.18.9ab) <?page no="161"?> 4.5. The syntax of multiple preverbs 161 Moreover, somewhat like the verb √bhā- ‘shine, emit light’ in (60)-(61), there exist composites containing roots of location or motion verbs, in which preverbs seem to centralize the Goaland the Path-participants (i. e. they seem to function as applicatives). In (64), áti ā́ √yā- ‘drive by’ takes the Goal-accusative śáśvato[aḥ] ‘every’; in (65), áti prá √sr̥- ‘outstrip, surpass’ ( prá √sr̥- ‘extend) takes the Path-accusative jánima ‘generation’. (64) Transitive áti ā́ √yā- ‘drive by’ ( √yā- ‘go, move, drive’) atiā́yāhi śáśvato vayáṃ te áraṃ pass_by. imp .2 sg every. acc . pl 1 pl . nom 2 sg . dat properly sutébhiḥ kr̥ṇavāma sómaiḥ press_out. ptcp . pst . ins . pl . pass do. sbjv . prs .1 pl S. ins . pl ‘Drive beyond them, each and every one. We will prepare properly for you with pressed soma drinks.’ (R̥V 3.35.5cd) (65) Transitive áti prá √sr̥- ‘outstrip, surpass’ (√sr̥-) sá majmánā jánima mā́nuṣāṇām dem . nom greatness. ins generation. acc . pl human. gen . pl ámartiyena nā́mnā́ti [nā́mnā ati] prá sarsre immortal. ins . n name( n ). ins beyond forth run. intens . prs .3 sg ‘Through (his) greatness and (his) immortal name he has extended himself over the generations of humans.’ (R̥V 6.18.7ab) Like emission verbs, however, simplex motion or manner of motion verbs, such as √sr̥- ‘run’, can also be given an endpoint by adpositionless accusatives. In (66), the accusative sū́raṃ plays the role of the Goals of motion. (66) sū́raṃ cit sasrúṣīr íṣaḥ sun. acc even run. ptcp . prf . nom . pl . f drink( f ). nom . pl ‘(His are) the refreshing drinks that have run even to the sun.’ (R̥V 1.86.5c) However, not all (manner of) motion verbs behave this way: for example, √drā- ‘run’ , √dru- ‘run’ , and √vr̥t- ‘roll’ are not attested in combination with the adpositionless accusative of Goal, though √vr̥tcan instead take an adpositionless locative with the same function. All in all, the transitivization of an intransitive simplex verb can be caused by any linguistic element able to assign an inherent endpoint to the event - be <?page no="162"?> 162 4. Multiple preverbs in Vedic it a preverb, a multiple preverb combination, or a prepositionless accusative. Moreover, the transitivization of multiple preverb composites frequently seems to be a side-effect of the semantic changes brought about by preverbs: this is the case with the non-compositional transitive composites abhí prá √bhū- ‘assist, help’ ( √bhū- ‘be, become’), or abhí ā́ √tap- ‘torment, pain’ ( √tap- ‘give out heat’). The crucial role that idiomatic semantic changes of the verbal root play in the transitivization caused by preverbs is also assessed by Kulikov (2012: 732): all composites containing the putative applicative preverbs ádhi, abhí , úpa , pári , prá , and práti that pass Kulikov’s passivization test show non-compositional meanings (Kulikov 2012: 730 ff.). 44 4.5.4. Composites taking no second argument In passages in which multiple preverb composites take no second argument, preverbs seem able to activate particular spatial regions and relations, whose specifications (i. e. LM s) are constituted by certain known referents in the discourse sphere. Such known LM s can be of the following types: (a) referents that are active in the discourse sphere, as they have been mentioned in the previous discourse context (67); (b) referents that are active, as they belong to speakers’ encyclopedic knowledge (68). (67) a. ánu prá yanti vr̥ṣṭáyaḥ after forth go. prs .3 pl rain( f ). nom . pl ‘(Your troop of chariots, the glittering Marut flock of newer), do the rains come following after.’ (R̥V 5.53.10c) b. niraítu jīvó ákṣato go_off. imp .3 sg alive. nom not_crushed. nom ‘(Having lain for ten months within his mother,) let (the boy) come out, alive and unharmed.’ (R̥V 5.78.9c; cf. Jamison & Brereton 2014: 761) 44 The only reliable passivization test for Old Indo-Aryan is the ability of a verb to form -yápassives. However, such passives are virtually unattested in early Vedic (i. e. in the R̥ g-Veda ), and all Kulikov’s examples of these passives come from later stages of the Vedic language. Recently, Borchio (2018) tried to assess the transitivizing potential of Rigvedic preverbs by analyzing the preverbed past passive participles ( -taparticiples), putting forward similar results: an applicative function can hardly be ascribed to Rigvedic preverbs, as past passive participles of preverbed verbs usually carry resultative (and not passive) meaning. <?page no="163"?> 4.5. The syntax of multiple preverbs 163 (68) tuvā́ṃ hí agne sádam ít samanyávo 2 sg . acc for A. voc always indeed having_the_same_mind. nom . pl devā́so devám aratíṃ nierirá god. nom . pl god. acc assistant. acc set_down. prf .3 pl . mid íti krátvā nieriré thus plan. ins . abs set_down. prf .3 pl . mid ‘You, o Agni, the god, did the gods of equal fervor set down for always as the wheel (of the sacrifice) - with such a purpose did they set (you) down.’ (R̥V 4.1.1abc) In (67)a, the EP ánu ‘after, along, toward’ establishes a reference to the previously mentioned referents ( LM s), after which the TR starts its motion, the Maruts. In (67)b, the outward motion is that of a newborn, who abandons his mother’s womb (cf. also RIVELEX I 149 with fn. 4). In (68), instead, the downward motion is Agni’s, whom other gods send downward from heaven to earth. Thus, preverbs, by referring to certain spatial relations, are consequently able to recall previously mentioned entities (67) or entities belonging to speakers’ encyclopedic knowledge (68). In both cases, such entities can be regarded as topical, as they are either active in the discourse sphere or can be easily activated. This ability to recall topical entities may be one of the reasons preverbs underwent grammaticalization into actional markers. The fact that topical elements are conceptualized as entire in space and complete in time provides the link among topicality, telicity (and eventually perfectivity) (Viti 2008a, 2008b). A similar discourse-oriented grammaticalization has been previously assumed for Indo-European preverbs, in particular by Friedrich (1987: 134), Coleman (1994: 324), and Cuzzolin (1995: 137). 45 Such a development also has typological parallels: first, the preverbs of Rama (Chibchan, Nicaragua) are said to develop from previous postpositions that start gravitating toward verbs, when null anaphora of the noun phrase occurs (Craig 1991: 468). 46 Second, Severn Ojibwe (Algonquian, Canada) possesses a category 45 These authors however are inclined to think that Indo-European preverbs underwent grammaticalization from previous postpositions, and not from previous adverbs. 46 Rama is a language with no morphological cases which can express SRs. Thus, in Rama, it is undisputed that preverbs were postpositions at a preceding stage. This is not the case for Vedic, in which there is virtually no governing relation between preverbs (also called adpositions), and noun phrases (cf. Section 4.1.2.7) <?page no="164"?> 164 4. Multiple preverbs in Vedic of preverbs that are called “relational” by virtue of the fact that they require an antecendent in the previous context (cf. Section 3.3.2). Dunkel (1987), Boley (2004), and Klein (e. g. 2007) also describe preverbs as linguistic elements that fulfill pragmatic functions, by contributing to textual cohesion. This function of preverbs is clear when they are repeated outside the preverbal context, as in example (69), in which the EP of the multiple preverb composite pári prá √dhanv- ‘flow forth around’ also occurs at the beginning of the subsequent pāda . 47 (69) pári ū ṣú prá dhanva vā́jasātaye around ptc good forth run. imp .2 sg winning_of_a_prize( f ). dat pári vr̥trā́ṇi sakṣáṇiḥ around obstacle( n ). acc . pl vanquishing. nom ‘Around and forth run to the winning of the prize, around, winner of the obstacles! ’ (R̥V 9.110.1ab) 4.6. Preverb ordering A number of preverbs occur only rarely in multiple preverb constructions. For this reason, it is difficult to provide conclusive data on preverb ordering (cf. also Papke 2010: 102 on this issue). In particular, preverbs that occur less than 10 times in the designated corpus are the followings: ádhi ‘above, over, on, onto’ , áchā ‘to, unto’, antár ‘between, among, within’ , ápa ‘away, forth, off’ , áti ‘across, beyond, past, over’ , áva ‘down, off’ , ní ‘down, in, into’ , nís ‘out, forth’ , párā ‘to a distance, away, forth’ , práti ‘in reversed direction, back to, against, in return’ , and úd ‘up, up forth or out’ . Thus, one must look at Table 19, which summarizes the favourite positioning of Vedic preverbs, with this caveat in mind. Preverb Interior Exterior áchā 0 1 (100 %) áti 0 5 (100 %) ádhi 1 (20 %) 4 (80 %) 47 In (69), the composite pári prá √dhanv- ‘flow forth around’ is split by two indeclinables, i. e. ū and ṣú . The same composite occurs in initial position with no splits in the preceding hymn, R̥ V 9.109.1a. <?page no="165"?> 4.6. Preverb ordering 165 Preverb Interior Exterior ánu 0 18 (100 %) antár 0 3 (100 %) ápa 0 3 (100 %) abhí 1 (3 %) 32 (97 %) áva 3 (75 %) 1 (25 %) ā́ 38 (81 %) 9 (19 %) úd 2 (67 %) 1 (33 %) úpa 0 10 (100 %) ní 4 (67 %) 2 (33 %) nís 3 (75 %) 1 (25 %) párā 4 (100 %) 0 pári 2 (20 %) 8 (80 %) prá 41 (98 %) 1 (2 %) práti 0 4 (100 %) ví 7 (59 %) 5 (41 %) sám 12 (63 %) 7 (37 %) Tab. 19: The positioning of Vedic preverbs and their frequencies Not all preverbs can occur both in the interior and exterior positions: in particular, áchā, antár, ánu, ápa, áti, práti, and úpa ‘to, unto, toward’ are never interior; by contrast, párā is never exterior, and prá ‘forward, onward, forth, fore-’ hardly ever exterior. Other preverbs show clear, though not absolute, positional tendencies: ā́ ‘to, unto, at’, áva, ní, nís, and úd are inclined to be interior, whereas abhí ‘to, unto, against’, ádhi, and pári ‘round about, around’ tend to appear in the exterior position. Lastly, the positional tendencies of sám ‘along, with, together’ and ví ‘apart, asunder, away, out’ are unclear. These data confirm Macdonell’s (1910: § 593, fn. 7) observation that “when there are two [preverbs], párā always immediately precedes the verb; ā́ and áva nearly always; úd , ní , prá usually. On the other hand, abhí is all but invariably the first of the two; adhí and ánu are nearly always so, úpa and práti usually.” These comments are also consistent with Papke’s findings, summarized in Figure 3: <?page no="166"?> 166 4. Multiple preverbs in Vedic Fig. 3: Vedic token data: ordering (from Papke 2010: 101) This consistency is significant: both Macdonell’s and Papke’s data samples are wider than mine. Macdonell includes the entire Vedic (not only R̥g-Vedic) corpus. Papke’s investigation instead is limited to the R̥g-Veda , but she also takes into account those composites for which the order #P_P_V# is never attested (cf. fn. 8 in this Chapter). 4.6.1. Previous accounts of preverb ordering In her dissertation, Papke (2010: 104 ff.) remarks that the relative positioning of preverbs is very similar in Vedic and in Classical Sanskrit. In spite of the high productivity of verbal composition and its syntactic nature (Danesi 2013: 62; Section 4.1.2.1), these data suggest that Vedic multiple preverb composites should at least partially be regarded as conventionalized units, as they survive in the later stages of the language (cf. Table 12). Papke (2010) also considers conventionalization (i. e. lexicalization, in her terms) to be the reason that multiple preverb composites remain relatively stable from the Vedic to the Classical period. As for the motivations behind preverb ordering, Papke (2010: 108 ff.) dismisses any significant correlation between the order of preverbs and the resulting meaning of the composite verbs. In particular, she shows that composites allowing for more than one order are frequently synonyms. This also holds true for my data, in which the composites ā́ níṣ √i- ~ nís ā́ √i- ‘go off, depart’ and abhí prá √mr̥ś- ~ prá abhí √mr̥ś- ‘seize, grasp’ alternate without any significant semantic difference, as shown in examples (70)a-b. Crucially (cf. Section 4.6.2), <?page no="167"?> 4.6. Preverb ordering 167 however, the difference between (70)a and (70)b is the overt expression of the Goal-like-participant ( śriyé : dat ‘glory’), which specifies the generic Goal meaning of the exterior preverb ā́ ‘to, unto, at’ in (70)a. (70) a. ā́ níṣ √i- ‘go off, depart’ (overtly expressed Goal) śriyé jātáḥ śriyé ā́ nír iyāya light( f ). dat born. nom light( f ). dat to out go. prf .3 sg ‘Born for glory, he went off to glory.’ (R̥V 9.94.4a) b. nís ā́ √i- ‘go off, depart’ (omitted Goal) niraítu jīvó ákṣato go_off. imp .3 sg alive. nom not_crushed. nom ‘(Having lain for ten months within his mother,) let (the boy) come out, alive and unharmed.’ (R̥V 5.78.9c = (67)) Moreover, Papke (2010: 111-112) does not identify any correlation between the interior / exterior positions and the meanings of single preverbs: morphemes expressing Direction, Location, and Orientation (in Imbert’s 2008 terms; cf. Chapter 5) can equally select both positions. However, Papke’s approach raises a number of issues. First, it is not entirely clear which criteria Papke (2010) use to group preverbs according to their meanings. First, it is puzzling that Papke categorizes ní ‘down, in, into’ among Orientation preverbs, whereas she puts its antonym, i. e. úd ‘up, up forth, up out’, among Direction (i. e. Goal) preverbs. Second, Direction (i. e. Goal), Location, and Orientation are probably not the most useful distinctions to account for preverb ordering. As discussed in Section 4.6.2, among Direction preverbs, there is a difference between preverbs profiling different portions of the trajectory: on the one side are Sourceand Path-preverbs; on the other side Goal-preverbs. In addition to excluding any semantic motivation for preverb ordering, Papke (2010: 112-114) also rules out the possibility that formal factors, such as the meter, the phonological, or the syllabic shape of preverbs may account for preverb positioning. Ultimately, Papke argues that Vedic preverb ordering must be inherited (on issues relating to the inheritance of preverb ordering, cf. Section 8.2.1): multiple composition is a process of accretion (in McCone’s 2006 terms), whereby a new preverb stacks onto an already lexicalized composite. Thus, the question of preverb ordering can be approached as a question regarding the reasons for the lexicalization of the IP . <?page no="168"?> 168 4. Multiple preverbs in Vedic 4.6.2. An integrated account of preverb ordering A correlation exists between the interior position of preverbs and prevalent verbal orientation, and between the exterior position and prevalent nominal orientation. Put another way, preverbs tend to occur in interior position if they are unlikely to receive a further spatial specification by means of a noun phrase (cf. Table 8). Interestingly, these are mostly preverbs profiling portions of the trajectory that are frequently left unspecified in the discourse, i. e. Path (e. g. úd, ní, prá , cf. (71)a) or Source (e. g. párā, cf. (71)b): (71) a. abhí úd √car- ‘rise over’ práti úd √i- ‘ rise and go toward’ b. ánu párā √i- ‘go away after’ áchā párā √i- ‘go away toward’ ví párā √i ‘go back (home) again’ In (71)a, the Path-preverb úd ‘up, up forth, up out’ is interior with respect to the Goal-preverbs abhí ‘to, unto, against’ and práti ‘in reversed direction, back to, against, in return’. In (71)b, the Source-preverb párā ‘to a distance, away, forth’ occurs internally with respect to the Path-preverb ánu ‘after, along, toward’, the Goal-preverb áchā ‘to, unto’, and the Time-preverb ví ‘in two different times, again’ (cf. Section 4.4.2), which modify the preceding composite - párā √i- ‘go away, go back’ - as a whole. The preverb ánu , despite encoding Path, tends to show nominal, rather than verbal orientation (cf. Table 8). In fact, the composite ánu párā √i- ‘go away after’ takes a Path-participant ( páram …pánthāṃ ‘farther on the way’), which happens to be particularly salient in the context of R̥V 10.18.1: (72) páram mr̥tyo ánu pár-ehi pánthāṃ far. acc death. voc along away-go. imp .2 sg way.a cc . pl ‘O Death, go away farther along your own way.’ (R̥V 10.18.1a = (36)) As explained for ánu , the other preverbs that preferentially select the exterior position also have nominal orientation in the majority of the R̥g-Vedic passages. In particular, preverbs of this type are abhí , úpa , and práti, which all encode the Goal of motion. As is widely accepted (cf. Ikegami 1987; Ungerer & Schmidt 1996; Verspoor, Dirven & Radden 1999; various papers in Luraghi et al. 2017), the overt specification of the Goal of motion is more frequent than that of the Paths and Source. Goal is the result or the endpoint of the motion event, and as such, the most salient of its parts. Interestingly, the preverb ánu, which is <?page no="169"?> 4.6. Preverb ordering 169 exterior in Vedic though it encodes the Path of motion, tends to move to the interior position in Classical Sanskrit (Papke 2010: 105). This cognitive explanation for ordering is also supported by the behavior of those preverbs that do not show clear preference for either the interior or the exterior position. One such preverb is ví ‘apart, asunder, away, out’. It occurs internally when it expresses Source or adverbial meanings very close to the semantics of the verbal root itself (73)a. By contrast, it occurs externally when it indicates the Goal of motion, or when it reverses the meaning of a preceding composite treated as a whole (73)b (cf. also (71)b for another exterior usage of the preverb ví ). (73) a. Interior ví : - Source abhí ví √i- ‘come forward from different parts’ - ‘precisely’ abhí ví √spaś- ‘look at, look hither’ (with verbs of seeing) b. Exterior ví : - Goal ví prá √i- ‘go forth in different directions’, ví ā́ √sr̥- ‘run through’, ví prá √ sr̥- ‘spread’ - Reversative ví ā́ √kr̥- ‘undo, divide’ ( √kr̥- ‘do, make’) This principle of ordering might be called the “lower salience” principle: preverbs expressing less salient portions of spatial relations tend to occur in internal position, as they are less likely to be further specified by a noun phrase. Obviously, this principle is highly context dependent: as we have seen by means of example (72), even Path can constitute a salient piece of information in certain specific contexts. Another motivation for preverb positioning is the semantic solidarity between particular preverbs and particular verbal stems. That is to say, preverbs whose meaning is partially subsumed by the meaning of the verb occur internally. Examples are ní ‘down, in, into’ and prá ‘forward, onward, forth, fore-’, exemplified in (74) (see also examples in (49)): (74) a. ní ‘down’ + √pad- ‘fall’ , √lī- ‘melt, disappear’ , √sad- ‘sit’ b. prá ‘forward’ + √īr-, √i-, √dr̥ ̄-,√dru 2 -, √dhanv-, √yā-, √vah-, √vr̥t-, √sr̥- (motion, manner of motion, caused motion verbs) The downward motion indicated by ní is inherent in the events of falling ( √pad- ), of melting / disappearing ( √lī- ), and of sitting ( √sad- ). The forward Path expressed by prá is likewise implied in any motion, manner of motion, or caused motion verb of (74). <?page no="170"?> 170 4. Multiple preverbs in Vedic This principle ruling ordering, which one might call the “redundancy” principle, somewhat recalls the “order and relevance principle” of affix ordering, formulated by Bybee (1985). Bybee defines relevance as follows: “A meaning element is relevant to another meaning element if the semantic content of the first directly affects or modifies the semantic content of the second” (Bybee 1985: 13). However, in the case of preverbs, the meaning of the IP is often so close to the verb’s that it shows a high semantic solidarity with it. Thus, semantic closeness makes a certain preverb redundant rather than relevant. Such semantic redundancy or low saliency arguably triggers the reanalysis of preverbs as part of the verbal item to which they attach: this development results in the lexicalization of preverbs, which are reinterpreted by speakers as parts of verbal stems. Redundancy however can also activate the grammaticalization of preverbs as markers for actionality: once a lexical (i. e. spatial) contribution is perceived as irrelevant, speakers assign preverbs a new function of carrying more information about the event, i. e. notions such as those of ingressitivity, telicity, completion, and so on. In Slavic languages, in which the grammaticalization of preverbs has gone much farther than in other Indo-European languages, this phenomenon is called the Vey-Schooneveld effect or subsumption (cf. Chapter 6). <?page no="171"?> 5.1. Preverbs in Homeric Greek 171 5. Multiple preverbs in Homeric Greek 5.1. Preverbs in Homeric Greek 5.1.1. State of the art In traditional grammars of Ancient and Homeric Greek, such as Schwyzer & Debrunner (1950: 417 ff.) and Chantraine (1953: 82 ff.), the status, the development, and the meanings of preverbs are usually addressed together with those of their cognate prepositions. In her comprehensive monography on the expression of semantic roles in Ancient Greek, Luraghi (2003) also touches upon the grammaticalization undergone by preverbs, insofar as it follows a path similar to that of prepositions. Accordingly, Luraghi hints at the semantics of preverbs, before moving on to the semantics of the corresponding prepositions, which is the focus of her book. In fact, a systematic investigation of Ancient Greek preverbs is still a desideratum, though several papers are devoted to specific issues related to a single or a few specific preverbs (e. g. Dietrich 1909 and Papanastassiou 2011 on apo- ; Dunkel 1982b and Neri 2007 on the etymology of sún and xún ; Chantraine 1942 on en-; Citraro 2014 on epi- , anaand huper- ; Revuelta 2014 on peri- ; see also Shain 2011 on the values of eisin Koine Greek; and Charitonidis 2013 on apo- , ek(s)- , and ksein Modern Greek). Actually, Strömberg’s (1946) book does systematically address Ancient Greek preverbs, but it focuses only on their role in nominal composition. 1 Horrock’s (1980) book deals with preverbs as elements contributing to the expression of the conceptual domains of space and time in the Homeric language. In this work, space is the preferential domain for investigating how preverbs, prepositions, and cases interact and compete with each other to express certain meanings (e. g. Viti 2008a and 2008b; Zanchi 2017b; cf. Section 2.1 on Cognitive Grammar), for, to a certain extent, Homeric Greek attests to a linguistic stage at which constructions involving preverbs, prepositions, or both could express a certain spatial role (Section 5.1.2.2). Viti (2008a) highlights the contextual differences promoting the selection of either construction, showing that preverbs tend to encode spatial relations involving human, specific, and topical referents. 1 The literature on the prepositional counterparts of Ancient Greek preverbs is much more abundant: see, besides the above-mentioned, Luraghi (2003, 2006, 2009, 2012), Bortone (2010), and Hettrich (2012), among others. <?page no="172"?> 172 5. Multiple preverbs in Homeric Greek She then connects this tendency with the development of Ancient Greek preverbs into markers of completion (i. e. telicity, perfectivity), insofar as topicality implies that a referent is conceptualized as entire in space and complete in time. Viti’s (2008a, 2008b) papers lead us into another field of studies on Greek preverbs, i. e. their development from free-standing adverbs into affixes, and their parallel semantic shift from spatial indicators into actional markers - and, eventually, as Viti argues for en- , into transitive markers. On the so-called “tmesis” in Ancient Greek and in other Indo-European languages, see, among many others, Boley (2004), De Angelis (2004) and the traditional references cited therein, Haug (2009), and Bertrand (2014) (see also Duhoux 1998 and references therein on Mycenean; Chapter 3). The actional value of completion has been assigned to Ancient Greek preverbs by many scholars, including Brugmann (1885: 268), Meillet (1922: 352, 1948), Brunel (1939), Humbert (1960: § 586), Leroy (1958: 135), Roussel (1958), Grassi (1966: 158 ff.), Friedrich (1974: 5), Pernée (1983: 298), and Giannakis (1997: 55 ff.). 2 Caroline Imbert dedicated several studies to the encoding of spatial relations in Homeric Greek by means of multiple preverbs, to the historical sources of such a system, and to the semantic constraints ruling the orders of preverb stacking (Imbert 2006, 2007, 2008, 2009; Imbert & Grinevald 2004). Imbert’s theoretical framework is grounded in studies on grammaticalization, which she regards as a gradual process in diachrony, resulting in categorial gradience in synchrony (cf. Section 2.2). In addition, Imbert’s works take into account the studies on preverbs in non-Indo-European languages. In particular, Imbert applies the morphological and syntactic category of relational preverbs (i. e. preverbs that behave morphologically as preverbs, but syntactically as adpositions), which Craig & Hale (1988) individuated for preverbs in Rama (Nicaragua, Chibchan family; see Section 3.3) to Homeric Greek multiple preverbs. Accordingly, Imbert argues that Homeric Greek multiple preverbs have developed from previous postpositions, as Craig & Hale (1988) argued for preverbs in Rama. 5.1.2. The unclear categorial status of preverbs in Homeric Greek In a number of ancient IE languages including Homeric Greek, preverbs, besides preceding a verbal stem, may also occur before (pre-position) or after a noun (post-position), or in virtually any other position within the sentence (free-standing adverbs) (see, among many others, Chantraine 1953: 82 ff.; Watkins 1964; Luraghi 2010; Section 5.1.2.2). Such syntactic freedom results in a 2 As correctly pointed out by Viti (2008a: 395), “most of these scholars wrote in a period where aspect […] was often confounded with actionality […]” (cf. further Section 2.3.2). <?page no="173"?> 5.1. Preverbs in Homeric Greek 173 great deal of uncertainty regarding both the morphological status (clitics vs. independent words) and the parts of speech of preverbs (free-standing adverbs, adpositions, or preverbs proper). 5.1.2.1. Preverbs: clitics vs. independent words In Ancient Greek, preverbs can either carry a graphic accent (cf. amphí ‘on both sides’, aná ‘upward’, antí ‘in front of ’, apó ‘away from’, diá ‘through’, ení ‘in’, epí ‘on’, hupér ‘over’, hupó ‘under’, katá ‘downward’, metá ‘between’, pará ‘besides’, perí ‘around’, pró ‘before’, prós / potí ‘toward’, sún ‘with’) or lack it ( ek ‘out of ’, en ‘in’, eis / es ‘(in)to’). 3 However, accentuation has little bearing on the morphological status of preverbs for a number of reasons. To begin with, the invention of the Greek accentuation system dates back to the Hellenistic period (3 rd -2 nd centuries BC ) (Laum 1928; Pfeiffer 1968; Dickey 2006), which means at least about five centuries after the Homeric period (8 th century BC ; cf. Section 1.3.2): thus, Hellenistic philologists are likely to have projected their accentuation rules back onto the Greek of the Homeric period. Moreover, even though most scholars agree that the usual Greek diacritics mark lexically assigned pitch levels (Allen 1973: 249), the precise significance of acute, grave, and circumflex diacritics remains in question. Furthermore, accentuation practice has changed over time, as demonstrated in the papyrological tradition, on the one hand, and by scribes’ inconsistency in accent placement, on the other hand (see West 2000 for an analysis of some Homeric variants). Actually, in Homeric Greek, preverbs / prepositions sometimes show clitic behavior. For example, as suggested by Lupaş (1972: 174) and Probert (2003: 133-142; 2006: 69 n. 35), usually, they only bear a “conventional” and not linguistically real accent, as shown by example (1). (1) apó → ap’ (not * áp’ ) / _#vowel ap’ ouranó-then away_from heaven-from ‘from heaven’ ( Il. 8.365) (2) pollà → póll’ / _#vowel póll’ aekazoménē much constrained ‘much constrained’ ( Il .6.458) 3 The preverb en ‘in’ is a more frequent variant for ení ‘in’. <?page no="174"?> 174 5. Multiple preverbs in Homeric Greek When, as in (1), apó occurs before a vowel-initial word, it undergoes apocope, i. e. its tonic final vowel drops, and its accent does not retract onto the previous syllable, as is usual with clitics. Such behavior has to be compared with that of vowel-final oxytone proper words, shown in (2). In (2), after the drop of the tonic final vowel in front of a subsequent vowel, the accent moves one syllable to the left (see also Goldstein 2010: 48 ff.). Conversely, preverbs / prepositions may also behave as accented words in Homer. A case in point is when they bear an accent on the first syllable under certain syntactic conditions (for a thorough discussion of which, see Vendryes 1904: § 309 ff.), notably when in absolute (3), postnominal (4), and postverbal positions (5). 4 (3) órnuto d’ autík’ épeita ánax andrôn stir_up. impf .3 sg . mid ptc immediately then king. nom men. gen . pl Agamémnōn, àn d’ Oduseùs polúmētis A. nom upward ptc O. nom of_many_counsels. nom ‘Immediately then Agamemnon, king of men, rose up, and up (rose) Odysseus of many wiles as well.’ ( Il. 3.267-268) (4) Alkínoos dè tót’ êrkhe, theôn A. nom ptc at_times be_first. impf .3 sg god. gen . pl ápo mḗdea eidṓs away_from plan( pl ). acc know. ptcp . prf . nom ‘And Alcinous was now king, made wise in counsel by the gods.’ ( Od. 6.12) (5) olésant’ ápo pántas hetaírous destroy. ptcp . aor . acc away_from indf . acc . pl comrade. acc . pl ‘after losing all (his) comrades altogether’ ( Od .2.174) Hellenistic grammarians have named this apparent retraction of the accent anastrophḗ (< ana- ‘upside down’ + stréphō ‘turn’). However, the accentuation shown in (3), (4) and (5) is in fact etymological, as demonstrated by the comparison of Greek ápo ‘away from’, pára ‘beside’, péri ‘around’, húpo ‘under’, with their Sanskrit counterparts, such as ápa ‘away from’, pára ‘beyond’ , pári 4 A preposition-preverb occurs in absolute position when it is not linked to any noun or verb in the sentence, and ultimately “substitutes” for the verb (Chantraine 1953: 82; see further, in Section 5.1.2.2, the discussion of example (6); cf. also Vedic absolute preverbs in Chapter 4). <?page no="175"?> 5.1. Preverbs in Homeric Greek 175 ‘around’ , úpa ‘toward’, or with Greek etymologically related adverbs, such as ánō ‘upward’ and kátō ‘downward’. Thus, when preverbs-prepositions show the so-called anastrophḗ, they in fact retain their original accentuation. Only when employed as clitics, i. e. as unaccented words, do they lose their initial accent (Vendryes 1904: § 72). 5.1.2.2. Free-standings adverbs, adpositions, and preverbs proper Homeric Greek lacks both preverbal and prepositional configurations, which standard in post-Homeric Greek (cf. among many others Schwyzer & Debrunner 1950: 419; Chantraine 1953: 82 ff.; Fritz 1997; Hewson & Bubenik 2006: 4; Bortone 2010: 123 ff.; Luraghi 2010). The passages in (6)-(11) exemplify the quite free positioning of preverbs-prepositions in Homeric Greek (for this reason, and given their etymological meanings, preverbs / prepositions are sometimes called “local particles” or “place words” in order to leave unspecified their syntactic function; see Section 3.1.4 on terminology). (6) Absolute position ḕ apóeip’, epeì oú toi épi déos or deny. imp . aor .2 sg for neg 2 sg . dat on fear. nom ‘Or deny me, for (there is) no fear on you.’ ( Il .1.515) (7) Adverbial position kaì epì knéphas hieròn élthēi and on darkness. nom sacred. nom come. sbjv . aor .3 sg ‘And the sacred darkness closes in.’ ( Il. 11.209) (8) Prenominal position epì gaían ap’ ouranó-then protrápētai on earth. acc away_from heaven-from turn. sbjv . aor .3 sg . m / p ‘(He) turned (the sun) (on)to earth away from heaven.’ ( Od. 11.18) (9) Postnominal position hai mèn aletreúousi múlēis épi mḗlopa dem . nom . pl . f ptc grind. prs .3 pl millstone. dat on yellow. acc karpón grain. acc ‘They grind the yellow grain on the millstone.’ ( Od. 7.104) <?page no="176"?> 176 5. Multiple preverbs in Homeric Greek (10) Preverbal position tmḗdēn d’ aukhén’ ep-êlthe with_cutting ptc neck. acc upon-come. aor .3 sg ‘And (the spear) came upon his neck so as to cut.’ ( Il .7.262) (11) Postverbal position ḗluth’ épi psukhḗ Agamémnonos come. aor .3 sg on soul. nom A. gen ‘The soul of Agamemnon approached.’ ( Od. 24.20) Example (6) shows the absolute usage (in Chantraine’s 1953 terms): the preverb épi occurs alone instead of the whole composite verb ép-eimi ‘be upon, set upon’ ( epi - ‘on’ + eimí ‘be’): the copula eimí is omitted, and the preverb takes its place and carries out its function. Given the syntactic freedom of preverbs here, judging the actual function of preverbs / prepositions can be a tricky task. For example, one may wonder whether, in (10), ep- ‘upon’ selects the accusative aukhén’(a) ‘neck’, thus functioning morphologically as a preverb, but syntactically as a postposition (i. e. as a relational preverb, in Craig & Hale (1988) and Imbert’s (2008) terms). However, epiis not syntactically obligatory in (10): in similar contexts, the accusative case ( klisíēn ‘hut’ in (12)) can in fact express the Goal-participant in combination with the simplex verb érkhomai ‘go, come’, as shown by example (12). (12) érkhesthon klisíēn Pēlēïádeō Akhilêos go. prs . imp .2 du hut. acc of_ P . gen A . gen ‘Go to the hut of Achilles, Peleus’ son! ’ ( Il. 1.322) Therefore, in (10), the preverb episemantically contributes to specifying the spatial region in which the spatial relation expressed by the verb érkhomai ‘go, come’ and the accusative case takes place. However, it is not syntactically obligatory (see further Horrocks 1981; Fritz 1997; Luraghi 2003; Zanchi 2017b on the expression of spatial relations in Ancient Greek). 5 What the historical development of Greek tells us is that the most frequent positions for preverbs / prepositions, which later became conventionalized, are the prenominal (8) and preverbal (10) ones. By virtue of their frequency, such 5 Moreover, if epiin (10) were a postposition, it would have occurred between the noun and its adjectival modifier, as adpositions tend to be placed after the first word of a complex nominal phrase (Irigoin 1954). <?page no="177"?> 5.1. Preverbs in Homeric Greek 177 conventionalized positions then triggered the grammaticalization processes of local adverbs into prepositions or preverbs (cf., among many others, Horrocks 1981: 117 ff.; Hewson & Bubenik 2006; Bortone 2010: 109-170; Luraghi 2010). 5.1.2.2.1. The ongoing grammaticalization of prepositions So far, I have provided evidence that neither the postpositional nor the prepositional configurations are fully developed in Homeric Greek. To begin with, as shown by examples (8) and (9), local particles can occur both before and after a noun, even though the prenominal position is more frequent and later on became conventionalized. Moreover, case alternation is meaningful in Homeric Greek. A number of Greek adpositions, i. e. amphí ‘on both sides’, epí ‘on’, pará ‘beside’, perí ‘around’ prós ‘to’ hupó ‘under’, can be used in combination with all three cases that it is possible to combine with prepositions (genitive, dative, and accusative). The meanings of the adpositions in combination with the case endings to some extent correspond to the old concrete value of the respective case forms (Kuryłowicz 1964: 176; Luraghi 2003: 79-80; Cuzzolin et al. 2006). In addition, adpositionless cases occasionally retain their old concrete value in Homer, especially in association with inherent Goal-, or Source-oriented verbs, that is, composite or non-composite verbs requiring a specific spatial participant (Chantraine 1953: 38-81; Zanchi 2017b). As a result, the simultaneous occurrence of a verb requiring a specific spatial participant, of a preverb / preposition, and of a case ending, frequently results in a redundancy to express a certain meaning: the free combination of only two of these elements is sufficient to disambiguate a spatial relation (see Section 5.5.2, which contains a thorough discussion of relevant examples with multiple preverbs). However, Homeric Greek also offers evidence for the obligatoriness of preverbs / prepositions. In (13), the dative prṓtoisi : dat . pl ‘first’, in association with metà ‘between, among’, produces an optional participant expressing Location ‘among the foremost’. Without the preposition metà, however, the dative would function as second argument of the verb mákhomai, which, in the meaning of ‘fight against someone’, takes the dative case (Luraghi 2014: 34; see further Sausa 2015 on the encoding of second arguments in Homeric Greek). (13) epeì thoòs éske metà prṓtoisi mákhesthai for quick. nom be. impf .3 sg among first. dat . pl fight. inf . prs . m / p ‘For he used to be ready to fight among the foremost.’ ( Il .5.536) <?page no="178"?> 178 5. Multiple preverbs in Homeric Greek 5.1.2.2.2. The so-called tmesis in Homeric Greek Just as prepositions are not fully grammaticalized yet (see Section 5.1.2.2.1), the phenomenon of the so-called “tmesis” suggests that preverbs are not fully developed as such either. “Tmesis” is a misleading label applied by Hellenistic grammarians to the phenomenon in which a putative composite is split, and preverbs occur displaced from the verbal stem that they modify (cf. also Section 5.1.2.2). In Homeric Greek, tmesis is not a literary artifact. Rather, it provides further evidence for the indeterminate categorial status of local particles, which range from adverbial, to adpositional, and preverbal usages (cf. De Angelis 2004, and references therein). 6 Even though Ancient Greek appears to have gone further than any other IE language in the elaboration of a free word order (Dover 1960), the basic positional options for preverbs (and the allowed types of tmesis) are the following (Watkins 1964): 7 a) # PN(E) …V(…) #; b) # N(E) …VP(…) #; c) # N(E) …PV(…) #. Type a) is attested in Hittite, Vedic, Archaic Latin, and Archaic Old Irish, and was first described by Wackernagel (1924). Type b) is allegedly a Greek literary innovation, as it is rarely attested in other I. E. languages. Type c) is the most difficult to detect, as the preverb does precede the verb, though it is merely juxtaposed and not univerbated to it (see Section 5.1.2.2.4). 5.1.2.2.3. The interactions between preverbs and postpositives Postpositives are words that exhibit word order restrictions: a) roughly speaking, they either occur in second or peninitial position (Wackernagel’s law); 8 b) they cannot begin a clause; c) they never occur in isolation (Wackernagel 1892; Fraenkel 1964[1933]; Revuelta 2014: 128-130). The full list of Ancient Greek postpositives is provided by Dover (1960: 12-13). How does the position of postpositives interact with that of preverbs? Examples (14) and (15) give an initial answer. The placement of postpositives oc- 6 It is our flawed standpoint that regards the categorial status of these elements as uncertain. Indeed, we usually analyze the behavior of these elements in view of their subsequent functional and categorial split into adpositions and preverbs (cf. Haspelmath 2007b on the non-existence of pre-established linguistic categories; cf. Chapter 2). 7 Cf. fn. 9, Chapter 3. 8 For further discussion on the actual - syntactic, pragmatic, prosodic, or underspecified - nature of the so-called second position, see e. g. Fraenkel (1964[1933]), Dik (1995, 2007), and Goldstein (2010). <?page no="179"?> 5.1. Preverbs in Homeric Greek 179 casionally has the side effect of splitting a putative composite or a putative adpositional phrase. That is, postpositives sometimes intervene either between a preverb and a verb (14), or between an adposition and a noun (15). (14) es d’ êlthon mnēstêres agḗnores to ptc come. aor .3 pl wooer. nom . pl heroic. nom . pl ‘Then the heroic wooers came in.’ ( Od. 1.144) (15) en d’ autoîsi púlas poiḗsomen in ptc 3 sg . dat . pl gate. acc . pl build. sbjv . aor .1 pl ‘And let us build gates on them (ships).’ ( Il. 7.339) In (14), the particle d’(è) occurs in between the local particle es ‘(in)to’ and the verb êlthon ‘came’. In (15), the same particle separates en ‘in(to)’ from the dative autoîsi ‘to them’. In (14) and (15), Wackenagel’s law overrides the usual juxtaposition of locative particles and verbs (14), or of locative particles and nouns (15) (cf. further Section 3.1.2). 9 Passages (14) and (15) are examples of what Bertrand (2014) calls “non-lexical tmesis”, i. e. a split in which only non-lexical words intervene. Compare examples (14) and (15) with passages such as (16) below, in which a preverb is separated from the verbal stem by lexical words (Bertrand’s (2014) “lexical tmesis”). (16) en dè tà mêla labóntes in ptc dem . acc . pl . n sheep( n ). acc . pl take. ptcp . aor . nom . pl ebḗsamen walk. aor .3 pl ‘As we have taken the sheep, we stepped into (the ships).’ ( Od .11.4) In my view, examples (14) and (15), on the one hand, and (16), on the other, attest to two different stages in the grammaticalization process undergone by preverbs / prepositions. In particular, examples (14) and (15) show a more advanced stage than example (16). In (16), a particle ( dè ), a noun phrase ( tà mêla ‘the sheep’) and a conjunct participle ( labóntes ) split the putative composite 9 In addition, the elided particle d’(é) serves the meter in both examples: the consonant d lengthens the vowel of es and en , which can thus constitute the first long half of a spondaic foot. <?page no="180"?> 180 5. Multiple preverbs in Homeric Greek en …baínō ‘step into’. In contrast, in (14) and (15), only a postpositive particle intervenes, due to its tendency to occur in second position. 5.1.2.2.4. The status of preverbs: what philology and meter tell us Homeric poems were written down in the 8 th century BC, but attest to even more ancient stages of the Greek language. In addition, modern editions of the Homeric texts are based on those recorded by the Hellenistic grammarians (4 th -1 st centuries BC ), who to some extent restyled the language according to their own more recent variety of Greek. The Homeric text to which we have access today thus incorportates both older and newer elements, and Homeric tradition sometimes yields several textual variants for one single passage (cf. Section 1.3.2). Critically for the purposes of this analysis Hellenistic philologists’ work dates back to a period when Greek syntax no longer allowed tmesis. In fact, the above-mentioned textual variants frequently concern hesitations relating to word boundaries, and in particular to the univerbation of preverbs, as shown in examples (17)a-b: (17) a. Non-univerbated variant (Codex Venetus Marc. Gr. 454 (822),10 Rom. Bibl. Nat. Gr. 6,11 and Genav. 44: 12) mákhēs ex apo-díōmai struggle. gen out_of away_from-drive. prs .1 sg . m / p b. Univerbated variant (attested in all other manuscripts) mákhēs ex-apo-díōmai struggle. gen out_of-away_from-drive. prs .1 sg . m / p ‘I drive (Ares) out of the struggle.’ ( Il .5.763) In (17)a, the preverb ex ‘out of ’ occurs as a separate word from the composite apo-díōmai ‘drive away from’, whereas in (17)b the preverb is univerbated to it. Deciding between these two variants is particularly difficult: the composite ex-apo-díōmai is a hápax and the metrical structure of Il .5.763 does not provide clues to word boundaries. In what way might the meter be useful to detect word boundaries in other contexts? The metrical structure of Homeric poems has frequently been thought accountable for altering the Homeric Greek syntax (cf. Section 1.3.2). However, as noted by De Angelis (2004) and extensively investigated in Zanchi (2017a), the metrical structure can instead be exploited to reconstruct Homeric syntax, as well as word boundaries overridden by Hellenistic grammarians. As Sommer (1926: 257-261) and Chantraine (1953: 85 ff.) already pointed out, certain com- <?page no="181"?> 5.1. Preverbs in Homeric Greek 181 pound forms handed down by the manuscript tradition are likely to rely on the alteration of a group of words that did not constitute a single morphological unit. According to De Angelis (2004) and Zanchi (2017a), one must assume a split between a preverb and a verb whenever their univerbation violates a metrical pause (or incision, a place within the verse in which word boundaries occur more often than by chance; West 1982: 6). However, one cannot assume a pause when the meaning of a composite is non-compositional, when the elements making up the composite have undergone phonetic erosion, and when the remaining simplex verb or composite is not elsewhere attested in the Homeric texts. In fact, composites with multiple preverbs can occur concurrently with either a metrical pause (18), or a metrical bridge (19) (i. e. a place within the verse in which word boundaries tend to be avoided). (18) Composite in correspondence with a pause (so-called katà tríton trokhaîon ) all’ oú hoi kháris but neg 3 sg . dat grace. nom amphi-|-peri-stéphetai epéessin on_both_sides-|-around-put_as_a_crown. prs .3 sg . m / p word. dat . pl ‘But the grace is not put around his words as a crown.’ ( Od .8.175) (19) Composite in correspondence with a metrical bridge (so-called Hermann’s bridge) hoì d’ ou gignṓskontes apēnḗnanto dem . nom . pl ptc neg know. ptcp . prs . nom . pl deny. aor .3 pl . mid hékastos every_one. nom ‘But as they knew it not, everyone denied.’ ( Il .7.185) In (18), a metrical pause is assumed to be splitting amphifrom the remaining composite. This assumption is supported by linguistic evidence: first, the composite is semantically compositional and a Homeric hápax ; second, the composite containing only the IP ( peri-stéphō ‘surround’) is attested in Homer ( Od. 5.303); third, the presence of the preverbs is not syntactically compulsory, as there exists an equivalent passage, in which none of the preverbs occurs ( Od. 8.170). By contrast, in (19), ap-an-aínomai ‘refuse completely’ co-occurs with a metrical bridge. The composite shows non-compositional meaning and is attested in post-Homeric Greek. <?page no="182"?> 182 5. Multiple preverbs in Homeric Greek Obviously, an analysis such as that just outlined presupposes a fixed metrical structure for the Homeric verse, including a fixed number of metrical pauses. However, there is little agreement about the number of pauses to be assumed for the Homeric verse, ranging from one to three (for a thorough discussion of this issue, refer to Kahane 1994; Martinelli 2001; Zanchi 2017a, and references therein). Even given this caveat, metrical analysis, when suported by additional linguistic evidence, can be enlightening for identifying word boundaries. 5.2. Multiple preverbs in numbers 5.2.1. Composites with multiple preverbs In Homeric poems, multiple preverbs occur in 64 composites. These are attested in 138 Homeric passages, a number of them in a formulaic expression (cf. Section 1.3.2). In Table 20, attested composites and their meanings are displayed, together with their token and types frequencies (type-frequency counts each formula only once). <?page no="183"?> 5.2. Multiple preverbs in numbers 183 c omposites m eaning f requency Iliad Odyssey Total (token) Total (type) amphi-peri-stéphomai put round as a crown 0 1 1 1 amphi-peri-strōpháō keep turning about all ways 1 0 1 1 an-eph-állomai spring upon after 3 0 3 3 ap-an-aínomai refuse completely 1 1 2 2 ap-ek-lanthánomai forget entirely 0 1 1 1 apo-pro-airéō take away from 0 1 1 1 apo-pro-ḯēmi send forth away, let fall 0 3 3 3 apo-pro-témnō cut off from 0 1 1 1 di-éx-eimi go out through 1 0 1 1 di-ex-eréomai question closely, completely 1 0 1 1 eg-kata-pḗgnumi thrust firmly in 0 1 1 1 eg-kata-títhemi put upon / around, store up 2 2 4 4 eis-ana-baínō go upward to 7 6 13 13 eis-an-ágō lead upward to 0 1 1 1 eis-an-eîdon look upward to 2 0 2 1 eis-án-eimi go upward to 1 0 1 1 eis-aph-ikánō arrive at, come to 1 2 3 2 eis-aph-iknéomai arrive at, come to 2 8 10 7 <?page no="184"?> 184 5. Multiple preverbs in Homeric Greek c omposites m eaning f requency Iliad Odyssey Total (token) Total (type) eis-kata-baínō go down (in)to 0 1 1 1 ek-dia-baínō pass over 1 0 1 1 ek-kat-eîdon look down from 2 0 2 2 ek-kat-ep-állomai leap down from against 1 0 1 1 ek-pro-kaléomai call forth from 0 1 1 1 ek-pro-leípō forsake 0 1 1 1 ex-ana-baínō go upward out of 1 0 1 1 ex-ana-dúomai emerge from 0 2 2 2 ex-ana-lúō set quite free from 2 0 2 1 ex-an-íēmi send upward out, emit 1 0 1 1 ex-apo-baínō step out of 0 1 1 1 ex-apo-díomai chase out of 1 0 1 1 ex-apo-dúnō put off 0 1 1 1 ex-ap-óllumi perish utterly out of 2 1 3 3 ex-apo-néomai return back out of 2 0 2 2 ex-apo-nízō wash thoroughly 0 1 1 1 ex-apo-tínō pay back in full 1 0 1 1 <?page no="185"?> 5.2. Multiple preverbs in numbers 185 c omposites m eaning f requency Iliad Odyssey Total (token) Total (type) ex-aph-airéō take away from 0 1 1 1 ex-up-an-ístēmi stand up from under 1 0 1 1 ep-ana-títhemi Shut 1 0 1 1 ep-an-ístēmi stand up after 1 0 1 1 ep-em-baínō stand upon 1 0 1 1 ep-en-tanúō bind tightly to 0 1 1 1 epi-pro-ḯallō place forth before (set out) 1 0 1 1 epi-pro-ḯēmi send forth (to) 5 1 6 5 hup-ek-phérō carry out from under, carry away 3 1 4 4 hup-ek-pheúgō flee away secretly 8 7 15 11 hup-ek-pro-lúō loose from under 0 1 1 1 hup-ek-pro-théō run forth from behind, outstrip 2 1 3 3 hup-ek-pro-réō flow forth from beneath 0 1 1 1 hup-ek-pro-pheúgō flee away secretly 2 2 4 4 hup-ek-sṓizō save (by drawing) away from under the control of 1 0 1 1 hup-ex-ágō carry out from under (out of danger into safety) 0 1 1 1 hup-ex-aléasthai flee out from 1 0 1 1 hup-ex-ana-dúomai come up secretly 1 0 1 1 <?page no="186"?> 186 5. Multiple preverbs in Homeric Greek c omposites m eaning f requency Iliad Odyssey Total (token) Total (type) huper-kata-baínō go downward over 2 0 2 1 kat-eph-állomai leap down against 1 0 1 1 para-kata-bállō throw down beside 2 0 2 2 para-kata-lékhomai lie down beside 2 0 2 2 par-ek-pro-pheúgō flee away 1 0 1 1 par-ex-ágō lead past 0 1 1 1 par-ex-elaú ́nō drive past 1 2 3 3 par-ex-érkhomai slip past, pass by, overstep 1 3 4 3 peri-pro-khéomai be poured all around 1 0 1 1 pro-kath-ízō perch forth (of birds) 1 0 1 1 pro-pro-kulíndomai keep rolling in front of 1 1 2 2 t otal 77 61 138 Tab. 20: Homeric composites with multiple preverbs and their frequency <?page no="187"?> 5.2. Multiple preverbs in numbers 187 Table 21 shows those composites that are attested in post-Homeric prose, as well as their Homeric and post-Homeric meanings. 10 c omposite h omeric meaning p ost -h omeric meaning ap-an-aínomai refuse completely disown, reject di-éx-eimi go out through go out through, go through in detail eg-kata-pḗgnumi thrust firmly in thrust firmly in, sheathe eis-an-ágō lead upward to lead up to eis-aph-iknéomai arrive at, come to arrive at, come to eis-kata-baínō go down (in)to go down into ex-ana-dúomai emerge from emerge from, escape from ex-an-íēmi send upward out, emit send forth, loosen, slacken ex-ap-óllumi perish utterly out of perish utterily ep-ana-títhemi shut lay upon ep-an-ístēmi stand up after set up again, make to rise against ep-em-baínō stand upon step upon, mount on, approach epi-pro-ḯēmi send forth (to) send forth hup-ek-phérō carry out from under, carry away carry away, have a start by a day’s journey hup-ek-pheúgō flee away secretly escape from hup-ex-ágō carry out from under (out of danger into safety) carry out from, carry away, withdraw para-kata-bállō throw down beside make a claim to property together with a deposit par-ex-ágō lead past lead past, mislead par-ex-elaúnō drive past march by, ride past par-ex-érkhomai slip past, pass by, overstep pass by, go aside from pro-kath-ízō perch forth (of birds) sit in public, sit before Tab. 21: Composites attested in post-Homeric prose and their meanings 10 Only attestations in post-Homeric prose are relevant, as post-Homeric epic and poetic authors mostly try to imitate Homer’s poetic diction, by also using typically Homeric lexicon. <?page no="188"?> 188 5. Multiple preverbs in Homeric Greek As type-frequencies show, out of 64 composites, only six are attested in formulaic expressions, that is, eis-an-eîdon ‘look upward to’, eis-aph-iknéomai ‘arrive at, come to’, epi-pro-ḯēmi ‘send forth (to)’, hup-ek-pheúgō ‘flee away secretly’, huper-kata-baínō ‘go downward over’, par-ex-érkhomai ‘slip past, pass by, overstep’. Attestation in formulaic expressions is relevant for comparative purposes, as formulas frequently preserve a more ancient variety of the language (cf. Watkins 1976; Joseph & Janda 2003: 15; Sections 1.3.2 and 1.4). Thus, the six composites listed above are likely to be particularly old. As Table 21 shows, out of 64 composites, only 21 are attested in post-Homeric Greek. Attestation in post-Homeric Greek is also relevant: on the one hand, it offers clues to identify those composites which are likely to be conventionalized, on the basis of the fact that they are retained and univerbated in the later language. On the other hand, attestation in post-Homeric Greek allows for interesting comparisons. Predictably, a number of composites show more compositional meanings in Homeric than in post-Homeric prose. Two cases in point follow: ex-ap-óllumi, which means ‘perish entirely out of ’ in Homer, with the obligatory expression of the Source argument ( Il .6.60, 18.290, Od .20.357; see example (32)), but which allows for later absolute usages ‘perish entirely’ (e. g. Hdt.4.173); and para-kata-bállō,‘ throw down beside’ in Homer ( Il .23.127, 23.683), which later on develops the very specialized meaning of ‘make a claim to property together with a deposit’ in post-Homeric prose (e. g. in D.44.42). Unexpectedly, a few composites appear to show the opposite behavior. The composite pro-kath-ízō is particularly instructive in this respect. It means ‘perch forth (of birds)’ in Homer, and ‘sit in public, sit before’ in post-Homeric prose (e. g. Hdt.1.14, Plb.20.6.8). On the one hand, the meaning of the IP is clear, though redundant, both in Homeric and in post-Homeric usages: the IP kata- ‘downward’ contributes to expressing the downward movement implied by the act of sitting. On the other hand, the semantic contributions of the EP procan be better detected in Post-Homeric usages, regardless whether spatial (‘sit one before the other’ → ‘sit in public’) or temporal (‘sit before’), than in the Homeric ones. Indeed, the presence of proonly makes sense in the wider context of the passage shown in (20): In (20), the Achaeans, spreading forth (pro-khéō ) into the plain of Scamander, are compared to a whole flock of birds, which is driven forward by single birds that keep perching in front (pro-kath-ízō ) of the rest of birds. The preverb pro- , occurring both in pro-kath-izóntōn and in pro-khéonto, contributes to drawing the parallel between birds and Achaeans’ forward motion. <?page no="189"?> 5.2. Multiple preverbs in numbers 189 (20) klaggēdòn prokathizóntōn, smarageî dé with_a_clang_noise perch. ptcp . prs . gen . pl resound. prs .3 sg ptc te leimṓn hṑs tôn énthea pollà ptc meadow. nom so dem . gen . pl tribe. nom . pl many. nom . pl neôn ápo kaì klisiáōn es pedíon ship. gen . pl away_from and hut. gen . pl (in)to plain. acc prokhéonto Skamándrion pour_forth. impf .3 pl . m / p of_ S . acc ‘… while (various types of birds) are perching with a clang noise, the meadow resounds. In the same way, their many tribes were pouring forth from ships and huts into the plain of Scamander.’ ( Il .2.463) Furthermore, when working with inherently limited corpora, such as those of Ancient Greek, one must keep in mind that a gap in attestations does not necessarily correspond to an actual absence in the grammar or in the lexicon (cf. Section 1.4; Joseph & Janda 2003: 15-16). In particular, if a certain composite attests to compositional meanings in its post-Homeric usages, whereas it lacks them in the Homeric poems, this does not necessarily imply that compositional meanings are a later development. It could simply be the case that Homeric texts happen not to not preserve those compositional usages. 5.2.1.1. Imbert’s (2008) and Zanchi’s (2014) composites In her dissertation, Imbert (2008: 95, 196-198) collected 98 examples of multiple preverbs, whereas here I included as many as 138 passages containing multiple preverbs. Imbert’s corpus is smaller than mine for the following reasons. First, she did not include two composites, i. e. ep-en-tanúō ‘bind tightly to’ and para-kata-lékhomai ‘lie down beside’. Imbert also excluded the composite pro-pro-kulíndomai ‘keep rolling before’, simply because two different local particles must occur in front of a verbal stem according to her definition of multiple preverbation. Second, Imbert (2008: 232-236) left out four composites, because they do not conform to the semantic constraints on preverb ordering that she set up: an-eph-állomai ‘leap upon after’, ek-kat-eph-állomai ‘leap down against from’, kat-eph-állomai ‘leap down against’, and pro-kat-hízō ‘perch forth’. Imbert explains that a diachronic analysis is required to avoid such apparent inconsistency in preverb ordering, but that she limits her work to the synchronic level. My analysis includes all of these composites, first because Imbert’s exclusion <?page no="190"?> 190 5. Multiple preverbs in Homeric Greek appears to be an ad hoc solution for safeguarding her semantic constraints on preverb ordering, and second because a diachronic account is arguably essential to analyze Homeric texts, which constitute an inherently diachronic corpus (cf. Section 1.4). In fact, the composites an-eph-állomai ‘leap upon after’, ek-kat-eph-állomai ‘leap down against from’, and kat-eph-állomai ‘leap down against’ are problematic, but for reasons other than those that Imbert points out. There is no general agreement about whether to assign them to the verb hállomai ‘leap’ (< PIE *sel- ‘loosen, leap’) or to the verb pállomai ‘swing, dash oneself ’ (< PIE *pelh 1 - ‘knock (or whip) something into shape’). These two analyses result in two different segmentations: an-eph-állomai , kat-eph-állomai , and ek-kat-ephállomai, with two and three preverbs, respectively (Chantraine 1953: 63, 854; Beekes 2010: 1148) vs. ana-pállomai , kata-pállomai , and ek-kata-pállomai, with one and two preverbs, respectively (Fränkel 1923: 278 ff.; Laumann 1950: 60 ff.; Geiss 1957: 62 ff.). In agreement with Harðarson (1993: 196-198) and LIV 2 (257), I opted for hállomai ‘leap’, also in the light of some Homeric parallel passages containing hállomai ‘leap’ in combination with ex + gen ( Il .6.103; cf. ek-kat-ephállomai ), kat(á) + gen ( Il .18.616; cf. ek-kat-eph-állomai and kat-eph-állomai ), and epí + dat ( Il .20.353, 21.174, Od .22.80; cf. kat-eph-állomai and an-eph-állomai ). However, the ordering of such multiple preverbs raise some issues, as discussed in Section 5.6. Imbert (2008: 199) further chose to omit all composites containing two preverbs that elsewhere occur as double prepositions or double adverbs, i. e. ap-ex- ‘away out of ’ (< apo- + ex- ), di-ex- ‘out through’ (< dia- + ex- ), par-ex- ‘out beside’ (< para- + ex- ), and hup-ex- ‘out from under’ (< hupo- + ek- ), possibly because these composites fail to conform to her semantic constraints on preverb ordering. I include all of these formations, as all of them are still analyzable, and can tell us much about the ongoing processes of univerbation of multiple preverbs (see further Sections 5.2.3 and 5.4.3). 11 Differently from Imbert (2008) and Zanchi (2014), I excluded the composite epapeiléō ‘threaten’ , as the segmentation containing two preverbs (* ep-ap-eiléō ) turned out to be incorrect. This composite only contains one preverb, i. e. epi- . The remaining verb is a denominative formation from apeilḗ (mostly plural) ‘boastful promises, threats’, made up by an a-copulativum and the root *(s)pelH ‘speak in public’ ( LIV 2 576 fn. 2 on a-peiléō ‘promise, threaten’). The position of the augment is revealing in this case: the augmented forms of this verb ( epē- 11 It is revealing that Imbert (2008) chooses not to address the fact that amphi-peri- , apo-proand peri-proalso occur as double prepositions and / or double adverbs in Homer (see Section 5.2.3). Perhaps because these sequences do not contradict her semantic constraints on preverb ordering, she chooses not to discuss them. <?page no="191"?> 5.2. Multiple preverbs in numbers 191 peílēs’(en), Il .1.319, 14.45) highlight the augment (in bold, resulting from the contraction of e and a ) between the putative EP and IP , whereas elsewhere the augment fairly systematically occurs between the IP and the verbal stem (see Section 5.3.3). Similarly, I also omitted the verb ex-apatáō ‘deceive, beguile’, containing the preverb exand a denominative verb from apatḗ ‘trick, fraud’ of uncertain etymology ( DELG 95; Beekes 2010: 113-114): this composite also contains only one preverb. 12 5.2.2. Verbal roots modified by multiple preverbs Table 22 shows the 41 simplex verbs modified by multiple preverbs and their meanings, as well as their PIE roots and their meanings. Their frequencies are also provided, i. e. the number of composites containing each root. In addition, the rightmost column specifies the verb type. For my purposes, a coarse-grained semantic classification of verbs suffices: motion and location verbs are considered to include, besides motion or location verbs proper (e. g. érkhomai ‘come, go’), manner of motion verbs (e. g. baínō ‘walk’), posture verbs (e. g. lékhomai ‘ lie’), and verbs of caused motion (e. g. bállō ‘throw’). 13 12 I am aware that other non-univerbated sequences of the type #P …P …V# are also attested (cf. e. g. Il .12.213, 14.168, Od .17.139, 23.16). However, as noted in Section 5.1.2.2.4, Hellenistic editors were inclined to use univerbation, which was common in their variety of Greek. Thus, I see no reason to treat the preverbs occurring in those sequences as univerbated units, as even Hellenistic editors decided not to univerbate them with the following verb. 13 Cf. fn. 26, Chapter 4. <?page no="192"?> 192 5. Multiple preverbs in Homeric Greek s implex verbs m eaning PIE roots ( LIV 2 page) m eaning f requency v erb type ágō carry, bring *h 2 eĝ- (255) drive, carry 3 caused motion aléomai flee, avoid *h 2 leu̯- (278) stay away, keep away 1 motion (Source) -aínomai refuse ? uncertain ? uncertain 1 communication baínō walk, go, come *g u̯ em- (209) go, come 7 manner of motion bállō throw *g u̯ elh 1 - (208) meet, throw 1 caused motion díō put to flight *dei̯h 1 - (107) chase away 1 caused motion dúnō, dúō cause to sink, sink *du̯eH- (129) sink 3 motion/ caused motion eîdon see *u̯ei̯d- (665) see, catch sight of 2 perception eîmi go, come *h 1 ei̯- (232) go, walk 2 motion eíromai ask *h 1 reh 1 - (251) ask 1 communication elaúnō drive, set in motion ? *h 1 elh 2 - (235) drive to 1 caused motion érkhomai come, go, walk *h 1 er-, *h 1 erĝ h - (238-239) reach, climb 1 motion hairéō take *ser- (535) take, grasp 2 removing hállomai leap *sel- (527) loosen, leap 3 motion (h)ḯallō send forth *sel- (527) loosen, leap 1 caused motion hḯēmi send *Hi̯eh 1 - (225) throw 3 caused motion hízō sit *sed- (513) sit 1 posture <?page no="193"?> 5.2. Multiple preverbs in numbers 193 s implex verbs m eaning PIE roots ( LIV 2 page) m eaning f requency v erb type hikánō, hiknéomai come, reach *sei̯k- ( 522) reach, achieve 2 motion hístēmi stand *steh 2 - ( 590) step to, stand 2 posture hréō run, flow *sreu̯- (588) flow, stream 1 manner of motion kaléō call *kleh 1 - (361) call 1 communication khéō pour *ĝ h eu̯- (179) pour 1 caused motion kulíndō roll ? uncertain ? uncertain 1 manner of motion lanthánomai forget *leh 2 d h - (401) remain hidden 1 mental activity leípō leave *lei̯k u̯ - (406) leave behind, move away from 1 motion (Source) lékhomai lie *leg h - (398) lie down 1 posture lúō loosen *leu̯H- (417) loosen 2 caused motion néomai go, come, return *nes- (454) get away, return home 1 motion nízō wash *nei̯g u̯ - (450) wash 1 removing óllumi destroy *h 3 elh 1 - (298) go on the ground 1 change of state pḗgnumi fix in *peh 2 ĝ- (461) become fixed 1 caused motion phérō carry, bring *b h er- (76) bring 1 caused motion pheúgō flee, escape *b h eu̯g- (84) flee, escape 3 motion (Source) stéphō put round *(s)teg u̯h - (589) put round as a crown 1 caused motion strōpháō turn constantly *streb h - (603) spin 1 manner of motion <?page no="194"?> 194 5. Multiple preverbs in Homeric Greek s implex verbs m eaning PIE roots ( LIV 2 page) m eaning f requency v erb type sṓizō save ? *teut 2 - ( DELG 1084-1085; Frisk 1960: 844) be strong 1 helping tanúō stretch *ten- (626) stretch, tighten 1 caused motion témnō cut *temh 1 - (625) cut 1 removing théō run *d h eu̯- (147) run 1 motion tínō pay a price *k u̯ ei̯- (377) receive a penance, punish 1 transfer títhemi put *d h eh 1 - (136) put, make 2 putting Tab. 22: Homeric verbs and verb roots modified by multiple preverbs <?page no="195"?> 5.2. Multiple preverbs in numbers 195 Most composites with multiple preverbs (47 out of 64) contain a motion or location verb. Most others can also be assimilated to motion / location verbs. The verb of perception eîdon ‘see’ can be regarded as a verb of caused motion: one’s eyes can be directed toward or away from something. Verbs of putting and removing, such as hairéō ‘take’, can also be assimilated to verbs of caused motion (and thus to motion / location verbs). Verbs of washing, such as nízō ‘wash’, in turn, have a behavior similar to that of verbs of removing (cf. Luraghi & Zanchi 2018). The verb tínō ‘pay a price’ is a transfer verb, and can thus be paired with verbs of putting and removing, as well as to verbs of caused motion. Two remaining items are verbs of communication, i. e. eíromai ‘ask’ and kaléō ‘call’, which are similar to verbs of transfer in that they imply a transfer of information. Moreover, communication verbs, to some extent, show a behavior similar to that of verbs of putting and taking: for example, verbs of asking, like verbs of removing, allow for the ditransitive construction in Ancient Greek (Luraghi & Zanchi 2018).Interestingly, two of the remaining verbs go back to PIE roots showing meanings close to those of motion or location verbs: lanthánomai ‘forget’ (mental activity) < *leh 2 d h - ‘remain hidden’ (location verb) ( LIV 2 401); óllumi ‘destroy, perish’ (change of state) < *h 3 elh 1 - ‘go on the ground’ (motion verb) ( LIV 2 298). Lastly, an-aínomai ‘refuse’ and sṓizō ‘save’ have no clear etymologies. The former is a communication verb, connected to the noun aînos ‘tale, proverb’, which in turn derives the verbs ainéō ‘tell, approve’, ep-ainéō ‘approve’, and di-ainéō ‘decree’, and must go back to an unattested verb * aínomai ‘say yes, state’ ( DELG 35-36; Beekes 2010: 39-40; for the meaning of this composite, see Section 5.4). 14 The verb sṓizō ‘rescue’ is probably a denominative from sôs ‘safe’ ( DELG 1084-1085; Beekes 2010: 1440-1441). It can be assimilated to verbs of removing or of caused motion, in that the saved entity ( TR ) is removed by another entity from a dangerous situation ( LM ). Another verb from Table 22, kulíndō ‘roll’, has no clear etymology but might go back to a root *(s)kel- ‘bent, curved’ (cf. Gr. kullós ‘curved, lame’) combined with a suffix -nd- (cf. Gr. alindéō ‘make to roll’) ( DELG 599; Beekes 2010: 800). 14 In two papers on preverb iteration in IE languages, Dunkel (1981a, 1981b) suggested a different etymology for an-aínomai, which is allegedly built on the iteration of anaana- ‘upward-upward’+ the suffix -i̯ o- . The iteration of preverbs here allegedly has the intensive value of ‘tossing the head repeatedly upward to say no’. In the same vein, Dunkel also argued that the Homeric verb apaphískō ‘cheat, beguile’ is to be segmented as ap-aph-ískō, a composite made up by the iterated preverb apoand the root √iskh- ‘hold’ with metathesis of aspiration. Chantraine (1953: 96) does not meantion this possibility and Beekes (2010: 114) does not accept Dunkel’s proposal and invokes substrate origins for the verb apaphískō . Noting the convincing arguments refuting each of these disputed etymologies, I left out apaphískō from this investigation. <?page no="196"?> 196 5. Multiple preverbs in Homeric Greek 5.2.3. Attested combinations of preverbs Table 23 shows the 31 Homeric preverb combinations and their frequencies. In Table 23, frequency refers to the number of composites containing a certain combination. e xterior preverb m edial preverb i nterior preverb f requency amphi- peri- 2 ana- epi- 1 apo- ana- 1 apo- ex- 1 apo- pro- 3 dia- ex- 2 eis- ana- 4 eis- apo- 2 eis- kata- 1 en- kata- 2 ex- ana- 4 ex- apo- 8 exhupoana- 1 ex- dia- 1 ex- kata- 1 exkataepi- 1 ex- pro- 2 epi- ana- 2 epi- en- 2 epi- pro- 2 huperkata- 1 hupoexpro- 4 hupoex- 5 hupoexana- 1 kata- epi- 1 <?page no="197"?> 5.2. Multiple preverbs in numbers 197 e xterior preverb m edial preverb i nterior preverb f requency para- kata- 2 paraexpro- 1 para- ex- 10 peri- pro- 1 pro- kata- 1 pro- pro- 1 Tab. 23: Homeric combinations of preverbs and their frequencies The most frequent combinations are ex-apo- (8 composites), hupo-ex- (5 composites), and para-ex- (10 composites). Interestingly, the first combination contains two preverbs that originally expressed Source. This is consistent with the fact that Source-preverbs seem to have undergone earlier grammaticalization processes than Goal-preverbs (see Zanchi 2017b; Dickey 2012 for a similar view on Slavic preverbs). Thus, the IP apo- , after univerbating and undergoing semantic shifts (Source > marker of completion), is no longer capable of expressing Source and is thus later on supported by another Source-preverb, i. e. ex - (the opposition of ex-ap-óllumi ‘perish entirely out of ’, mentioned at Section 5.2.1, vs. ap-óllumi ‘perish entirely’ is revealing in this respect). 15 The combination hupo-exalso makes up a poetic double preposition, frequently used in Homer with the genitive, in the meaning of ‘from beneath’, but scarcely attested in Post-Homeric Greek. The highly lexicalized complex preposition paréx (< para- + ex- ) , instead, is frequently used in both Homeric and post-Homeric Greek, according to Chantraine (1953: 145 ff.). As a preposition, it takes the genitive (‘outside’) and the accusative (‘along the side of ’). As an adverb, it means ‘out beside, out along, excepting’. 16 Beside these, other combinations of preverbs are also attested as double prepositions or double adverbs in Homeric Greek. This is the case for amphiand peri- (double preposition: Il .2.305, 23.191, Od .11.609; double adverb: Il. 21.210; these are never recorded as single units and do not even constitute an entry in the LSJ ). As for the combinations with pro-, apo-procan also have adverbial (‘afar off’, Il .16.669) and prepositional functions (‘away from’, cf. Il .7.334) in 15 This combination also appears to be old in the light of its Latin cognate ex-po- , lexicalized in exponō ‘put out, set out’< *(ex-)po-znō < *(ex-)po-snō < *(ex-)po-sinō (Dunkel 1981b: 230 fn. 29; De Vaan 2008: 479). 16 All the non-univerbated sequences mentioned in fn. 12 show the complex adverb paréx . <?page no="198"?> 198 5. Multiple preverbs in Homeric Greek Homer, whereas peri-proonly occurs as an expressive particle, meaning ‘very, especially’ (its usage is declining in Attic prose, according to Chantraine 1953: 146). The repeated preverb prois also found in prepositional (‘before’) and adverbial usages (‘on and on’, ‘thoroughly’) in Post-Homeric Greek. According to Chantraine (1953: 144), the repetition of prois exceptional and archaic. In fact, the equivalents for this preverb are occasionally repeated in other IE languages, notably in Hittite parā parā ‘more and more’ and Vedic prápra ‘again and again’ (Dunkel 1981a: 214-219; on Vedic, cf. Section 4.2.3). The combination of epipró can also function as double preposition only in Post-Homeric Greek, with spatial meanings ‘right through, onward’. The combination di-ekmakes up a complex preposition, which occurs 13 times in Homer and retains its spatial value of ‘out of through’; after Homer, this preposition disappears. In contrast, the double preposition ap-ék is not attested in Homer, nor is it common in Post-Homeric Greek; when it is used, it shows the spatial meaning of ‘away out of ’. The existence of such complex prepositions / adverbs does not per se prevent preverbs from modifying verbal stems as independent units. One such case is the composite di-ex-eréomai ‘ask completely a number of questions’, shown in (21): (21) allà tíē emè taûta diexeréesthe but why 1 sg . acc dem . acc . pl . n question_completely. impf .2 pl hékasta each. acc . pl ‘But why did you question me completely regarding all these things one by one? ’ ( Il .10.432) In (21), the preverbs diaand ekcan hardly be considered to behave as a single unit: the double preposition diék always retains spatial meanings if used as a whole. Instead, it makes sense to analyse the preverbs separately: ekadds the actional meaning of ‘completion’ (cf. Chantraine 1953: 93), whereas diaexpresses the idea of arrangement, distinction, and distribution (cf. Chantraine 1953: 95, Section 5.4.4), which is also reinforced by the distributive adjective hékasta . <?page no="199"?> 5.3. The form of composites 199 5.3. The form of composites 5.3.1. Philological and metrical analyses of composites with multiple preverbs As mentioned in Section 5.1.2.2.4, certain compounds handed down by the textual tradition are likely to rely on the alteration of elements that did not in fact constitute a single morphological unit. This intuition is confirmed both by the scrutiny of the textual variants reported by Homeric critical editions (e. g. Monro & Allen 1920), and by a metrical analysis of passages containing multiple preverbs. Manuscripts attest to philological variants for 39 out of 138 relevant passages. In particular, scribes’ hesitations pertain to multiple preverbs in 24 out of 39 passages. More specifically, the textual tradition can vary in either of the following respects: (a) the univerbation of the EP , cf. example (17); (b) the replacement of one preverb with another one, cf. example (22); (c) the presence of two preverbs, cf. example (23); (d) the overall presence of preverbs, cf. example (24). (22) a. Variant hup-ek-leláthesthe ( Laurentianus 32, Ambrosianus 1015, Vaticanus 915, Vindoboniensis phil. 39, Baroccianus 203, etc.) b. Variant ap-ek-leláthesthe ( Parisiensis 2681, Venetus 455 ) ap-ek-leláthesthe dè thámbeus entirely-entirely-forget. aor .2 pl . mid ptc wonder. gen ‘(You) totally forget (your) wonder.’ ( Od. 24.394) (23) Variant huper-bebaṑs ( Laurentianus 32. 6 a. 1465, Ambrosianus 1057, Estensis 123, Parisiensis 2767, Vaticanus 27, Vaticanus 1318, etc.) oudoû ep-em-bebaṑs hupsērephéos threshold. gen on-in-stand. ptcp . prf . nom high_roofed. gen thalámoio bedroom. gen ‘Standing upon the threshold of the high-roofed bedroom …’ ( Il. 9.582) <?page no="200"?> 200 5. Multiple preverbs in Homeric Greek 24 Variant húpaith’(a) épheren ( Venetus 453, Mus. Brit. Burney 86 a. 1059 ) enth’ Aías mèn hup-ex-épheren sákos thence A . nom ptc under-out_of-carry. impf .3 sg shield. acc ‘Then Aias move his shield aside from (him).’ ( Il. 8.268) The metrical analysis reveals that a metrical pause may be be assumed in slightly fewer than half of the occurrences (62 out of 138). Such metrical pauses allegedly restore original word boundaries in any of the following positions: (a) between the EP and the remaining composite IP =V; (b) between the EP = IP and the simplex verb; (c) between the EP = MP and the remaining composite IP =V. The frequencies of (a), (b), and (c) are provided in Table 24, as well as the composites instantiating each type of assumed split. In Table 24, the symbol ## indicates the position of the assumed split, while = marks the univerbation of the remaining elements. p osition of w ord b oundaries t oken f requency c omposites (A) EP ## IP =V 37 amphi-peri-stéphomai, apo-pro-ḯēmi, eg-kata-pḗgnumi, eis-ana-baínō, eis-an-ágō, eis-an-eîdon, eis-án-eimi, eis-aph-ikánō, ek-dia-baínō, ek-kat-epállomai, ex-ana-baínō, ex-ana-dúomai, ex-apo-baínō, ex-apo-dúnō, ex-ap-óllumi, ex-apo-nízō, ex-aph-airéō, epi-pro-ḯallō, epi-pro-ḯēmi, huper-kata-baínō, para-kata-lékhomai, peri-pro-khéomai (B) EP = IP ##V 15 amphi-peri-strōpháō, di-ex-eréomai, hupek-phérō, hup-ek-sṓizō, hup-ex-aléasthai, par-ex-ágō, par-ex-elaúnō, par-ex-érkhomai (C) EP = MP ## IP =V 10 hup-ek-pro-théō, hup-ek-pro-réō, hup-ekpro-pheúgō, hup-ex-ana-dúomai, par-ekpro-pheúgō TOTAL 62 Tab. 24: Types of split brought about by assumed metrical pauses Interestingly, a split can be assumed between the EP and the IP=V for most composites. This restored word boundary suggests a path of formation whereby the <?page no="201"?> 5.3. The form of composites 201 EP has started gravitating toward a preceding composite with only one preverb. In contrast, positions (b-c) are only possible in the presence of an elsewhere attested double preposition, i. e. a stable collocation (e. g. amphí perí ) or a univerbated composite made up by two prepositions (e. g. diék, hupék, parék ). By contrast, positions (b-c) suggest that the EP and the MP / IP have stacked onto the remaining composite as a single unit. 5.3.2. Sandhi phenomena The juxtaposition of preverbs and verbal stems results in various consonant and vowel clusters. Their behavior follows the rules of external sandhi whenever they mismatch those of internal sandhi (cf. the Vedic data in Section 4.3.1). Therefore, for example, when the juxtaposition of preverbs and verbal stems produces the meeting of two vowels, those undergo elision, which is typical of external sandhi , rather than contraction, which characterizes internal sandhi (Alonso Déniz 2014a, 2014b, and references therein). This is shown in (25): (25) apo- + ana- + *aínomai → ap-an-aínomai ‘refuse completely’ (cf. example (19)) vs. contraction: a + o → ō e. g. aidóa → aidô a + a → ā e. g. géraa → gérā The only exception to this principle is instantiated by pro- ‘forth’. This preverb usually does not undergo elision: compare apo-pro-airéō ‘take away from’ vs. ap-an-aínomai ‘refuse completely’, in which the final o of apohas dropped. In addition, prooccasionally undergoes crasis, that is, the contraction applied to words belonging to the same phrase: e. g. prò ékhōn → proúkhōn ‘excelling’. However, one can also occasionally find elision in compounding, although it usually occurs word-externally: e. g. oudè heîs → oudeís ‘no one’ . 5.3.3. The position of preverbs with respect to inflectional affixes As happens in Vedic (cf. Section 4.3.2), Homeric preverbs interact with other pieces of preverbal morphology. In Homeric Greek, preverbal morphology comprises reduplication and augment. Preverbs tend to occur externally, rather than intervening between these more internal morphemes and the verbal stem (Schwyzer & Debrunner 1950: 646 ff.; Chantraine 1945: 309 ff.). <?page no="202"?> 202 5. Multiple preverbs in Homeric Greek Reduplication contributes to the formation of certain present stems - cf. arar-ískō ‘fit together’ with total reduplication (< *h 2 er-, LIV 2 269) and gí-gnomai ‘become’ (< *ĝenh 2 -, LIV 2 163-164) with partial reduplication - and of the perfect stems (cf. lé-luka , the perfect of lúō ‘loosen’). The phonological shape of reduplication is consistent with that of the reduplicated verbal root: either the entire verbal root is iterated, or reduplication consists of the initial consonant of the verbal root and of a front short vowel i or e. 17 Therefore, not surprisingly, reduplication is the innermost piece of verbal morphology: even in the pluperfect, which features both reduplication and augment, the latter occurs externally with respect to reduplication (e. g. e-lé-lukē , the pluperfect of lúō ‘loosen’). The augment instead marks distance in time in association with the secondary verbal endings and, in Homeric Greek, is optionally used in imperfect, indicative aorist and pluperfect tenses. The augment possibly goes back to an independent particle ( *h 1 e- ‘then, at that time’; e. g. Beekes 2011: 252). In Ancient Greek, it has the shape of a short e- , and usually occurs between the preverb and the verbal stem (Post-Homeric exceptions to this rule are found in Schwyzer & Debrunner: 656; Chantraine 1945: 313). In composites with multiple preverbs, the augment is the innermost element of preverbal morphology, as shown in example (26) (augment is highlighted in bold): (26) eis-an-é-bē-s-an to. ep -upward. ip pst -walkaor pst .3 pl ‘they went upward to’ ( Il .6.74) In my sample, there is only one exception to this otherwise strict rule: 18 (27) *ap-e-an-e-an-a-nto >apēnḗnanto ep pst ip pst -tellpfv pst .3 pl . mid ‘they rejected’ ( Il. 7.185; cf. example (19)) The form in (27) belongs to the verb ap-an-aínomai ‘refuse completely’, and seems to feature a double augment: the former eoccurs between the EP and the IP , while the latter between the IP and the verbal stem. However, although the double augment has been attested since ancient times, it still appears much later than the Homeric poems (400-350 BC +; Schwyzer & Debrunner 1950: 656). In fact, the form apēnḗnanto is better explained as a matter of poetic diction: only 10 verses above Il .7.185, a metrically equivalent and regular verbal form, that 17 If the initial consonant of the root is aspirated, reduplication contains the corresponding non-aspirated consonant (Grassmann’s Law). 18 The segmentation in (27) draws on the assumption that ap-an-aínomai relies on an unattested simple verb *aínomai ‘say yes, state’. This reconstruction is however disputed, as remarked in Section 5.2.2. <?page no="203"?> 5.4. The semantics of multiple preverbs 203 is, esēmḗnanto ( aor .3 pl . mid from sēmaínō ‘show by a sign’), occurs in the same colon as apēnḗnanto, and in equivalent metric position (so-called Hermann’s bridge). The two cola are compared in (28)a-b: What does the positioning of the augment suggest regarding the morphological status of preverbs? First, preverbs were probably not considered part of the verb, as they occur more externally than the usual outermost piece of verbal morphology, i. e. the augment. 19 Second, as prosodic (Section 5.1.2.1), philological (Section 5.1.2.2.4), and syntactic (Section 5.5) evidence confirms, multiple preverbs have the status of clitics in Homeric Greek (for similar considerations on Vedic and Classical Sanskrit, see Whitney 1955[1879]: 354 ff.; Papke 2010: 9, 94; Chapter 4). 5.4. The semantics of multiple preverbs 5.4.1. Preverbs with spatial, abstract and actional meanings From a semantic standpoint, (multiple) preverbs are polysemous elements: either they retain their original spatial meaning in some composites, or they develop further spatial meanings, abstract meanings, and actional meanings in other composites. As suggested by this semantic distinction, actional meanings, and especially telic ones, boast a special status among other abstract meanings developed from originally spatial preverbs. On the one hand, the spatial and abstract meanings of preverbs continue to contribute to the lexicon. On the other hand, preverbs developing actional meanings have a more grammatical than lexical character, and in this respect, they have undergone grammaticalization processes. The ability to add an endpoint to events makes possible the subsequent development of preverbs into markers of perfectivity (especially in Slavic languages, cf. Chapter 6) and / or into markers of transitivity (rarely, also in Homeric Greek, cf. Section 5.5.4; Viti 2008a, 2008b). Importantly, the basic 19 The robust cross-linguistic tendency according to which inflectional affixes are the farthest from the root is usually referred to as “relevance” or “scope principle”, on which see e. g. Bybee 1985; Rice 2000. <?page no="204"?> 204 5. Multiple preverbs in Homeric Greek semantic contribution of preverbs can undergo semantic bleaching, once they gain new meanings. Multiple preverbs modifying a verbal stem can behave similarly or differently in this respect. For example, in (29) the preverbs retain spatial meanings, and in (30), the preverbs develop actional meanings. (29) Spatial meanings: di-éx-eimi ‘go out through’ Skaiás, têi ár’ émelle S ( pl . f ). acc dem . dat . f ptc be_likely_to. impf .3 sg di-ex-ímenai pedíon=de through-out-go. inf . prs plain. acc =to ‘… the Scaean gate, and through that way (Hector) was likely to go out (of Troy) to the plain.’ ( Il. 6.393) (30) Actional meanings: ap-ek-lanthánomai ‘ forget entirely’ (cf. (22)) ap-ek-leláthesthe dè thámbeus entirely-entirely-forget. aor .2 pl . mid ptc wonder. gen ‘(You) totally forget (your) wonder.’ ( Od. 24.394) In (29), the two preverbs, ekand dia- , profile two different portions of the trajectory covered by Hector, who is going out ( ek- ) of Troy (Source), through ( dia- ) the Scaean gate (Path), forth to the open plain ( pedíon=de ) (Goal). The Source-participant referred to by ekis implicit, though inferable from the context. The Path-participant is instead made explicit by means of the accusative plural Skaiás and of the adverb têi ‘through that way’, to which the preverb diaexstablishes an anaphoric reference. 20 As for (30), the EP apoarguably reinforces the idea of completion already expressed by the IP ekin ek-lanthánomai ‘forget utterly’ (a single preverb composite that is also attested in Homer; cf. Il .2.600, Od .7.220, etc.). In contrast, there are composites in which one of the preverbs retains a lexical usage, while the other develops actional meanings. Neither the exterior nor the interior position is exclusively associated with either of such meanings, as demonstrated by examples (31) and (32): 20 As we will see in Section 5.5.3, preverbs can contribute to building textual cohesion. <?page no="205"?> 5.4. The semantics of multiple preverbs 205 (31) Actional EP vs. spatial IP : ek-dia-baínō ‘pass over’ táphron d’ ek-dia-bántes oruktḕn trench. acc ptc entirely-through-go. ptcp . aor . nom . pl dug. acc hedrióōnto seat. impf .3 pl ‘As (they) had passed over the dug trench, (they) seated.’ ( Il. 10.198) (32) Spatial EP vs. actional IP : ex-ap-óllumi ‘perish, vanish utterly out of ’ all’ háma pántes Ilíou exapoloíat’ but together indf . nom . pl I . gen out_of-utterly-perish. opt . aor .3 pl ‘But let all perish out of Ilios, (uncared for and blotted out).’ ( Il. 6.59-60) In (31), the IP diameans ‘through’, whereas the EP ekarguably expresses the idea of completion. This interpretation for ekis supported by the comparison between examples (31) and (33). (33) táphron epotrúnōn dia-bainémen trench. acc urge. ptcp . prs . nom through-walk. inf . prs ‘(Hector went through the throng and looked for his comrades), urging (them) to cross the trench.’( Il .12.50) In (31), the event of passing over is completed, so the composite with ekis used in a perfective context. By contrast, in (33), the composite lacking ekoccurs in an order which has an inherently imperfective context. In (32), the EP ekretains its Source meaning, which is further specified by the genitive Ilíou ‘Ilios’, whereas the IP apostrengthens the idea of completion implied by the event of dying. It is worth remembering that the reconstructed meaning for the root *h 3 elh 1 - (> Gr. óllumi ) is ‘go on the ground’ (cf. Table 22), i. e. that of a Goal-oriented motion verb. This original meaning possibly played a role in the early development of apo- ‘away from’ as a telic marker. 21 As leaving a Source ( apo- ‘away from’) is a sub-event implied by the event of reaching a Goal ( óllumi ‘go in the ground’), apowas reassigned a salient piece of information and thus reanalysed as a telic marker. Its semantic contribution as a 21 Neri (2007) assumed a telic value for the PIE preverb *pe / o- , based on the comparison between Gr. ap-óllumi , Lat. ab-oleō ‘destroy, kill’ (containing the preverb in the full grade), and OHG fallan ‘fall’ , Arm. p c lanim ‘fall’, Lith. puólu ‘fall’, and perhaps Hitt. (*) palla- ‘fall’ (containing the preverbs in the zero grade) . <?page no="206"?> 206 5. Multiple preverbs in Homeric Greek Source-marker would have been redundant (in the same vein, see Neri 2007: 80; in the literature on Slavic, this is the so-called Vey-Schooneveld effect or subsumption, on which see Chapter 6). Like their prepositional counterparts, Greek (multiple) preverbs are polysemous elements and can develop abstract meanings that are not actional (i. e. grammaticalized). The preverb apo- ‘away from’ provides one such example: besides the grammaticalized telic usages outlined above, it can mean ‘back’, as in the composite ex-apo-néomai ‘return back out of ’. In addition, it shows a pseudoreversative meaning in the composite ex-apo-dúnō ‘put off’, in which it reverses the meaning of the simple verb dúnō ‘put on’. The preverb apois regarded as pseudoreversative (and not as simply reversative), because it does not have the ability to build an opposite out of every simplex verb, as do the English prefix unand the Vedic preverb ví ‘apart, asunder’ (Delbrück 1888: 466). Rather, it creates a reversed composite when its “semantics happen to contradict the semantics of the base verb” (Sturm 2014: 9, who identified pseudoreversative meanings for the Vedic cognate of apo- , i. e. ápa ‘away, forth’, as well as for other Vedic preverbs): the Source-oriented preverb apoconflicts with the semantics of the base verb dúnō ‘put on’. 22 Interestingly, a single combination of preverbs may result in a specific meaning. In particular, iterated preverbs develop the actional meaning of iterativity. This development is iconic: moving more than once in the same way or toward the same direction implies iterating that movement (accordingly, reduplication is a typologically widespread means of intensification; Kajitani 2005; Fischer 2011a). In Homeric Greek, the only example of iteration proper is offered by pro-pro-kulíndomai ‘keep rolling in front of ’ (on pro-pro- , see Section 5.2.3). However, the composite amphi-peri-strōpháō ‘keep turning about all ways’ can also be seen as containing preverb iteration: as pointed out by Chantraine (1953: 129-130) and confirmed by Luraghi (2003: 256), the meaning of amphi- ‘on both sides’ eventually converged with that of peri- ‘around’. Therefore, from a semantic standpoint, around is iterated in amphi-peri-strōpháō ‘keep turning around all ways’, resulting in an iterative meaning. 5.4.2. Same (combinations of) preverbs, different meanings As anticipated for apo- (Section 5.4.1), preverbs are polysemous elements, which undergo multiple semantic shifts. Some interesting cases in point are ana- ‘upward’, hupo- ‘under’, and dia- ‘in two spaces’ (cf. Table 28). The preverb ana- 22 Papanastiossou (2011) offers a comprehensive semantic analysis of the preverb apo- , enhanced by numerous examples. <?page no="207"?> 5.4. The semantics of multiple preverbs 207 originally means ‘upward’ and implies motion along an upward trajectory. It further develops the spatial meaning of ‘upon’, which only profiles the endpoint (Goal) of an upward motion, as shown by the composite ex-up-an-ístēmi ‘stand upon (after being grown) from underneath’ in (34): (34) smôdix metaphrénou ex-up-an-éstē weal( f ). nom back. gen from-beneath-upon-stand. ptcp . aor . nom . f ‘a weal, (grown) from underneath the skin of the back, and standing upon’ ( Il .2.267) In (34), a weal is described, which, after growing from ( ex- ) underneath (( h ) up ( o ) - ) the skin of the back ( metaphrénou ), is standing upon ( an ( a ) - ) it. In addition, anashows the meaning of ‘refusal’ in the composite ap-anaínomai ‘disown, reject’. Despite some debate regarding the etymology of this composite (on which, see fn. 14 in this Chapter), as pointed out by Chantraine ( DELG 35-36), this usage is likely to be analogically established based on the following example: neúō ‘nod’ vs. ana-neúō ‘nod upward > refuse’. The development of ana-neúō is related to the fact that, from ancient times to the present day in Greece, as well as in the whole Balkans and in Southern Italy, the widespread gesture for refusal is nodding upward, sometimes just the raising of the eyebrows, with an optional dental or alveolar click ( Joseph 2000; Hauge 2002; Gil 2011; Friedman & Joseph forthc. 2019). In ap-an-aínomai ‘disown, reject’, the idea of refusal is then reinforced by the EP apo- , which implies complete, total rejection. 23 Furthermore, anashows a pseudoreversative meaning in combination with verbs of sinking: dúomai ‘go into, sink’ vs. ex-ana-dúomai ‘sink upward from > emerge from’. Lastly, in the composite ex-ana-lúō ‘set quite free from’, ana develops the meaning of ‘escaping out of control’, according to the following cluster of metaphors: having control or force is up , being subject to control or force is down (Lakoff & Johnson 1980: 15). The preverb hupo- ‘under, beneath’ develops one non-basic spatial meaning, two abstract meanings, and one actional meaning. The non-basic spatial meaning is exemplified by the composite hup-ek-pro-théō ‘run forth from behind, outstrip’, and results from the analogical equation ‘ above ’ : ‘ beneath ’ = ‘ before ’ : ‘ behind ’, which is also valid for the Homeric preposition hupó (Luraghi 2003: 226). As Luraghi puts it, “there are many similarities between the position ‘be- 23 In addition, as shown by Chantraine (1953: 91) and confirmed by Papanastassiou (2011: 101), the preverb apoin and of itself can mean ‘refusal’ in combination with verbs of saying (cf. eîpon ‘say’ vs. ap-eîpon ‘deny)’. In this light, the composite ap-an-aínomai would represent another instance of semantic iteration of preverbs (cf. Section 5.4.1 above). <?page no="208"?> 208 5. Multiple preverbs in Homeric Greek neath’ and the position ‘behind’ a referent, among others, the fact that both being beneath and being behind imply being away from the visible field.” This provides grounds for the further metaphorical shift into ‘secretly’ ( beneath → behind → inVisibly → secRetly ), undergone by hupoespecially in combination with verbs of escaping, such as hup-ek-pro-pheúgō ‘flee away secretly from’. In addition, hupoundergoes a semantic shift connected with the same set of metaphors outlined above for ana- , i. e. having control or force is up , being subject to control or force is down (Lakoff & Johnson 1980: 18): hup-ek-sṓizō ‘save (by drawing) away from under the control of ’. Once again, this is a shared development between the preverb hupoand the preposition hupó (Luraghi 2003: 227). Lastly, as shown by the composite hup-ex-ana-dúomai ‘start emerging gradually from’, the preverb hupocan also bring ingressive meanings and the notion of gradualness (cf. Chantraine 1953: 137), which can be explained by the following metaphor: beneath → infeRioR in quantity → at the beginning of . Another interesting development is that of the preverb dia-, whose etymology is related to the Proto-Indo-European numeral for ‘two’ ( DELG 276; LIPP II 145 ff.). Accordingly, this preverb has the original meaning of ‘in two pieces, in two places.’ As time is often described in terms of space (Lakoff & Johnson 1980: 267), diaalso came to mean ‘in two times’. This temporal meaning then offers the basis for the developing of the distributive meaning, shown in the composite di-ex-eréomai ‘ask completely a number of questions’ (see example (21) above). Interestingly, the same sequence of preverbs can also show different meanings, and these differences sometimes provide clues to the process of univerbation undergone by composites. A case in point is the sequence apo-pro- , featured in apo-pro-ḯēmi ‘send forth away’ (35), on the one hand, and in apo-pro-airéō ‘take away from’ and apo-pro-témnō ‘cut off from’ (36), on the other hand. 24 (35) tòn dè tétarton apo-pro-éēke pólin=de dem . acc ptc fourth. acc away-forth-send. aor .3 sg city. acc =to ‘Instead, (he) had sent the fourth away to the city.’ ( Od. 14.26) (36) oú tí moi étlēs sítou neg indf . acc 1 sg . dat dare. aor .2 sg bread. gen apo-pro-elṑn dómenai away-forth-take. ptcp . aor . nom give. inf . aor 24 As apo-pro-airéō ‘take away from’ and apo-pro-témnō ‘cut off from’ have similar behaviors, I only exemplify one of them in (36). <?page no="209"?> 5.4. The semantics of multiple preverbs 209 ‘(You who now, while sitting at another’s table,) do not dare to take away a piece of bread and give (it) to me.’ ( Od. 17.456-457) In (35) apo-pro-ḯēmi ‘send forth away’, the semantic contribution of each preverb is recognizable: apoindicates a generic separation from a Source, whereas promeans ‘forth’ indicating Path. The full expression of Goal is then left to the phrase pólin=de ‘to (the) city’. By contrast, in (36), the meaning of prois bleached: only the EP apois crucial to understand the meaning of the composite apo-pro-airéō ‘take away’, which is synonymous with the composite containing only apo- , i. e. aph-airéō ‘take away’ (cf. Od. 14.455). This suggests that the two preverbs are likely to modify the verbal base as a single unit. In support of this analysis, while the composite pro-airéō is not attested in Homeric Greek (it means ‘bring forth’ in later Greek), the double preposition apopró ‘away from’ is used in Homer, and a metrical pause can be assumed to be splitting the EP = IP from the simplex verb (cf. Table 24). Even the same composite can show various degrees of compositionality, if used in different contexts. This is the case of eg-kata-títhemi , which means ‘put downward inside’ in (37), but ‘put upon, put around’ in (38). (37) tḕn d’ átēn ou prósthen heôi dem . acc ptc guilt. acc neg before poss .3 sg . dat eg-kát-theto thumôi lugrḗn in-downward-put. aor . mid .3 sg soul. dat baneful. acc ‘But before, (Helen) did not put the baneful guilt downward inside her soul.’ ( Od. 23.223-224) (38) tê nûn toûton himánta teôi ptc now dem . acc strap. acc poss .2 sg . dat eg-kát-theo kólpōi poikílon in-downward-put. imp . aor .2 sg bosom. dat colored. acc ‘Here you are! Put around your bosom this colored strap! ’ ( Il. 14.219-220) In (37), the metaphorical trajectory covered by the emotion of guilt ( TR ) is described as going downward ( kata- ) into ( en- ) Helen’s soul ( LM ), which is a metaphorical Location conceptualized as a container. In (38), instead, the TR , a strap ( himánta ), does not cover any downward trajectory inside Hera’s bosom ( LM ); instead, it is simply put around it. Thus, the semantic contributions of enand kataare less clear in passage (38), in which the composite occurs probably <?page no="210"?> 210 5. Multiple preverbs in Homeric Greek echoing the passage in (37). Examples (37) and (38) in fact contain two quasi-formulaic expressions, i. e. heôi egkáttheto thumôi and teôi egkáttheo kólpōi, which both are metrically equivalent and equally made up of a possessive adjective, the composite, and a dative noun meaning either ‘soul’ or ‘bosom’ (the latter being conceived as the location of the former). 5.4.3. Different degrees of compositionality As anticipated in Section 5.4.2, composites that contain multiple preverbs show different degrees of semantic compositionality. The exact numbers for fully, partially and non-compositional composites are difficult to calculate. On one hand, the same composite can be more or less transparent, as explained for egkata-títhemi ‘put downward inside, put around’ above. On the other hand, the semantic contribution of certain preverbs, though still detectable, is redundant. Out of 64 composites, I regard 25 as being fully compositional, and 5 as non-compositional. These are shown in Table 25 and in Table 26, respectively. c omposite m eaning apo-pro-ḯēmi send forth away (cf. (35)) 1 di-éx-eimi go out through (cf. (29)) di-ex-eréomai ask completely a number of questions (cf. (21)) eis-ana-baínō go upward to eis-an-ágō lead upward to eis-an-eîdon look upward to eis-án-eimi go upward to eis-kata-baínō go downward to > pass over ek-kat-eîdon look downward from ek-kat-ep-állomai leap down against from ek-pro-kaléomai call forth from ex-ana-baínō go upward out of ex-an-íēmi send upward out, emit ex-apo-tínō pay back in full ex-up-an-ístēmi stand up from under (see example (34)) epi-pro-ḯēmi send forth (to) <?page no="211"?> 5.4. The semantics of multiple preverbs 211 c omposite m eaning hup-ek-phérō carry out from under hup-ek-pro-théō run forth from behind hup-ek-pro-réō flow forth from beneath hup-ek-sṓizō save (by drawing) away from under the control of hup-ex-ágō carry out from under (out of danger into safety) hup-ex-aléasthai flee out from under kat-eph-állomai leap down against huper-kata-baínō go downward over para-kata-bállō throw down beside Tab. 25: Homeric fully compositional composites c omposite m eaning amphi-peri-strōpháō keep turning about all ways ap-an-aínomai disown, reject ek-dia-baínō pass over (see example (31)) ep-ana-títhēmi shut pro-pro-kulíndomai keep rolling in front of Tab. 26: Homeric non-compositional composites As one might expect, fully compositional composites contain motion verbs or verbs that can be assimilated to motion verbs. Most of the non-compositional composites included in Table 26 have been discussed earlier in this work (see Section 5.4.1 and Section 5.4.2). The role of preverbs in the remaining non-compositional composite, ep-anatíthēmi ‘shut’, is challenging to figure out, and can be understood only by means of a scrutiny of the Homeric texts. The composite ep-ana-títhēmi is made up by epi- ‘on’, ana- ‘upward’, and títhēmi ‘put’. The combination of these, however, does not result in the compositional meaning of putting upon, but in that of shutting, as shown in (39). This semantic shift only makes sense in the light of the passage in (40). 25 This composite can also mean ‘let fall’. <?page no="212"?> 212 5. Multiple preverbs in Homeric Greek (39) aûtis epanthémenai sanídas pukinôs again shut. inf . aor shutter. acc . pl firmly araruías fit.together. ptcp . prf . acc . pl ‘Shut again the shutters firmly fit together! ’ ( Il. 21.535) (40) líthon d’ ep-éthēke thúrēisi stone. acc ptc on-put. aor .3 sg door. dat ‘(Athena) put a stone against the entrance (of the cave).’ ( Od .13.370) In (40), the composite epi-títhēmi ‘put on’, which lacks the IP ana- , takes the direct object ( líthon ‘stone’) of the entity (TR) that Athena puts ( éthēke ) against the entrance of a cave ( LM ) to shut it. The prepositionless dative ( thúrēisi ‘to (the) entrance’) encodes the Goal of this caused motion. In (39), instead, the closing entity is omitted, and the direct object of the entity being closed, “shutters”, only occurs in the accusative case. Originally, both epiand anain ep-ana-títhēmi possibly contributed to profiling the endpoint of the motion event of putting something against something else to be shut. 26 Later on, once ep-ana-títhēmi undergoes lexicalization, it spreads to events of closing that imply no caused motion. I consider the majority of Homeric composites with multiple preverbs to be partially compositional (34 out of 64), which are displayed in Table 27. c omposite m eaning amphi-peri-stéphomai put around as a crown an-eph-állomai leap upon after ap-ek-lanthánomai forget entirely apo-pro-airéō take away from apo-pro-témnō cut off from eg-kata-pḗgnumi thrust firmly in eg-kata-títhemi put upon / around, store up eis-aph-ikánō arrive at, come to 26 This motion implies a rotation from the vertical to the horizontal axis, which Luraghi (2003: 299-300, 2006) also observed for the preposition epí and for other prepositions expressing verticality, such as katá ‘downward’ and hupér ‘over’ (Luraghi 2003: 206, 222; on hupér , see also Zanchi 2016). <?page no="213"?> 5.4. The semantics of multiple preverbs 213 c omposite m eaning eis-aph-iknéomai arrive at, come to ek-pro-leípō forsake ex-ana-dúomai emerge from ex-ana-lúō set quite free from ex-apo-baínō step out of ex-apo-díomai chase away out of ex-apo-dúnō put off ex-ap-óllumi perish utterly out of ex-apo-néomai return back out of ex-apo-nízō wash thoroughly ex-aph-airéō take away from ep-an-ístēmi stand up after ep-em-baínō stand upon ep-en-tanúō bind tightly to epi-pro-ḯallō place forth before (set out) hup-ek-pheúgō flee away from under hup-ek-pro-lúō loose from under hup-ek-pro-pheúgō flee forth away secretly hup-ex-ana-dúomai emerge from under para-kata-lékhomai lie down beside par-ek-pro-pheúgō flee away from beside par-ex-ágō lead past par-ex-elaúnō drive past par-ex-érkhomai slip past, pass by, overstep peri-pro-khéomai be poured all around pro-kath-ízō perch forth (of birds) Tab. 27: Homeric partially compositional composites A number of such composites show redundancy of some kind. Either the meanings of the preverbs overlap with one another (41); or the meaning of one of <?page no="214"?> 214 5. Multiple preverbs in Homeric Greek the preverbs, usually the IP , shows semantic solidarity with the verbal stem to which it attaches (42). (41) Composites containing preverbs with overlapping meanings amphi-peri-stéphomai around-around-put_as a crown ap-ek-lanthánomai totally-totally-forget (42) Composites containing a redundant preverb a. an-eph-állomai up-after-leap b. ep-an-ístēmi after-up-stand c. ex-apo-néomai return back out of d. pro-kath-ízō forth-downward-sit > perch forth (of birds) e. peri-pro-khéomai around-all-be_poured f. hup-ek-pro-pheúgō secretly-away-forth-flee g. hup-ek-pheúgō under-away-flee h. eis-aph-iknéomai to-arrive The semantic overlap between amphiand perihas already been addressed (cf. Section 5.4.1). As for ap-ek-lanthánomai ‘forget entirely’, the two preverbs here share the telic and grammaticalized meaning of ‘totally’. Examples (42)a-h display composites containing a preverb that shows semantic solidarity with the verb. In (42)a-b, the meaning of ana- ‘upward’ is implied by the events of leaping ( hállomai ) and standing (up) ( hístēmi ), as the meaning of kata- ‘downward’ is implied by the event of sitting, shown in (42)d (cf. also para-kata-lékhomai ‘lie down beside’). 27 In addition, the meaning of “way back” conveyed by apois inherent to the verb néomai ‘return’, as shown in (42)c. Similarly, in (42)e, the idea of covering and intensification brought about by prois inherent to the act of pouring (cf. also ek-pro-leípō (literally) ‘out_offorth-leave’, the composite describing Achean soldiers’ overwhelming ( pro- ) run out of ( ek- ) the Trojan horse against their enemies). The preverb proexpressing Path is also redundant in combination with the verb pheúgō ‘escape’ (42)f, as the event of escaping implies a Path to follow on the run. In fact , prois omitted in hup-ek-pheúgō (42)g, in which the preverb ekcan be in turn seen as redundant: the event of escaping also implies an entity to escape from. 27 The combination of kata- + hízō started being treated as a non-composite unit at an early stage of written Greek, as shown by forms such as e-káthize: impf .3 sg (Xen.+), whereby the augment eoccurs externally with respect to the preverb ( Joseph 2017). <?page no="215"?> 5.4. The semantics of multiple preverbs 215 Interestingly, in eis-aph-iknéomai ‘arrive at, come to’ (42)h, the EP eisindicating Goal is added to a motion verb already implying a Goal-participant. The composite aph-iknéomai ‘arrive’ contains a preverb, aph- ( apo- ) ‘away from’, that underwent semantic bleaching at a preceding stage. The root for hiknéomai , that is, *sei̯k- ‘reach, achieve’ ( LIV 2 522), has an inherent Goal component. As reaching a Goal subsumes the previous event of leaving a Source, the Source-preverb apoearly on became redundant and was reanalyzed as a marker for telicity (cf. the discussion on ex-ap-óllumi in (32)). Non-compositionality does not always originate from redundancy. In (43), for example, the EP exis added to a preceding composite, whereby the semantic contribution of the IP ana reverses the meaning of the simplex verb ( dúomai ‘sink’ vs. ana-dúomai ‘emerge’): (43) [ ex- [ ana-dúomai ]] ‘[from-[emerge]]’ poliês halòs ex-anadûsai grey. gen sea. gen from-emerge_from. ptcp . aor . nom . pl . f ‘(The seals sleep close together,) as they had emerged from the grey sea.’ ( Od. 4.405) Intriguingly, ex-ana-dúomai ‘emerge from’ , shown in (43) is likely to have influenced the formation of the composite ex-apo-dúnō ‘put off’, shown in (44). (44) ex-apo-dúnō ‘put off’ heímata d’ exapédune clothing. acc . pl ptc put_off. impf .3 sg ‘(He) put off the clothes, (which heavenly Calypso had given him.)’ ( Od. 5.372) The simplex verb dúnō can also mean ‘put on’ and is reversed by the addition of apo- ‘away from’, resulting in ‘put off’. As far as the EP exis concerned, however, in (43), it clearly contributes to pointing to a Source-participant, which is further specified by the genitive case ( poliês halòs ‘of (the) grey sea’), whereas it has a bleached meaning in (44). A telic interpretation is difficult for it, as the composite occurs in the imperfect. Although in principle ex might signify generic separation, my guess is that the exin ex-apo-dúnō is added by analogy with ex-ana-dúomai : both composites contain the same verbal root, an interior pseudoreversative preverb, and the exterior ex-. In addition, ex-apoin itself is the most frequent combination of preverbs (cf. Table 23). <?page no="216"?> 216 5. Multiple preverbs in Homeric Greek 5.4.4. Summarizing the meanings of preverbs in multiple preverb combinations Table 28 summarizes the meanings of Homeric multiple preverbs. Each meaning is exemplified by a composite. p reverb m eaning e xample amphiaround (Goal) amphi-peri-stéphomai ‘put round as a crown’ iteration amphi-peri-strōpháō ‘keep turning around all ways’ anaupward (Path) eis-ana-baínō ‘ go upward to’ upon (Goal) (implied movement) ex-upan-ístēmi ‘stand upon (being grown) from underneath’ refusal ap-an-aínomai ‘disown, reject’ pseudoreversative ex-ana-dúomai ‘emerge from’ up as having control ex-ana-lúō ‘set free from’ apoaway from (Source) apo-pro-ḯēmi ‘send forth away, let fall’ back (again) ex-apo-néomai ‘return back out of ’ pseudoreversative ex-apo-dúnō ‘put off’ completion ap-ek-lanthánomai ‘ forget entirely’ diathrough (Path) di-éx-eimi ‘go out through’ distributive di-ex-eréomai ‘ask completely a number of questions’ eisto (Goal) eis-ana-baínō ‘go upward to’ eninto (Goal) eg-kata-pḗgnumi ‘thrust firmly in’ (in)to (Goal) eg-kata-títhemi ‘put upon / around, store up’ tightly ep-en-tanúō ‘bind tightly to’ exout of (Source) di-éx-eimi ‘go out through’ from (Source) ek-kat-eîdon ‘look down from’ pseudoreversative ex-apo-dúnō ‘put off’ completion di-ex-eréomai ‘ask completely a number of questions’ epiupon (Goal) an-eph-állomai ‘leap upon after’ to (Goal) ep-en-tanúō ‘bind tightly to’ <?page no="217"?> 5.5. The syntactic status of multiple preverbs 217 p reverb m eaning e xample against (Goal) ek-kat-eph-állomai ‘leap down from’ after (Stimulus) ep-an-ístēmi ‘stand up after (someone’s words)’ huperover (Resultative) huper-kata-baínō ‘go downward over’ hupounder, beneath (Souce) ex-up-an-ístēmi ‘stand up from under’ behind (Source) hup-ek-pro-théō ‘run forth from behind, outstrip’ secretly hup-ek-pro-pheúgō ‘flee away secretly from’ down as lacking control hup-ek-sṓizō ‘save (by drawing) away from the control of ’ ingressive hup-ex-ana-dúomai ‘start emerging gradually from’ katadownward (Path) eis-kata-baínō ‘go down to’ along (Path) huper-kata-baínō ‘go downward over’ parabeside (Goal) para-kata-bállō ‘throw down beside’ past (Goal) par-ex-ágō ‘lead past’ periaround (Goal) amphi-peri-stéphomai ‘put round as a crown’ iteration amphi-peri-strōpháō ‘keep turning around all ways’ proforth, forward (Path) ek-pro-kaléomai ‘call forth from’ covering peri-pro-khéomai ‘be poured all around’ iteration pro-pro-kulíndomai ‘keep rolling in front of ’ Tab. 28: The meanings of Homeric multiple preverbs 5.5. The syntactic status of multiple preverbs This Section addresses the issue of whether multiple preverbs can select the case taken by the verbs onto which they attach, thus affecting their argument structure. According to Imbert (2008), EP s are relational in nature: they are clitics that syntactically function as adpositions. Imbert further argues that multiple preverbs developed from previous post-positions. Imbert’s analysis is supported by a substantial number of Homeric occurrences (53 out of 138), in which the second argument of the composite (ARG in (45) below) immediately precedes it. What is more, <?page no="218"?> 218 5. Multiple preverbs in Homeric Greek this argument is inflected in the same case as that required by the EP when it functions as a preposition. This results in the following ambiguous construction: (45) [ ARG ] [ EP IP V] vs. [ ARG EP ] [ IP V] By contrast, as I argued in Zanchi (2014), my account follows Boley’s position (2004: 52) on this issue. In Homeric Greek, one can find some evidence suggesting that preverbs (Boley’s “place words”) were originally clear additions to what we regard as the basic sentence, whose meaning could be also expressed by means of a verb and a concrete case. Therefore, if we think of grammaticalization as a process involving increasing obligatorification (cf. Lehmann 1995[1982]), then Homeric multiple preverbs are not - or at least not always - fully grammaticalized yet, as they are not always obligatory. In what follows, I will show that preverbs are still movable (Section 5.5.1, and sometimes even syntactically optional (Section 5.5.2). Furthermore, I will highlight passages that do not raise ambiguous interpretations, as the argument occurs in a position or has a form that cannot be selected by the EP (Section 5.5.3). In Section 5.5.4, I will show that the addition of preverbs occasionally has the side-effect of centralizing a verbal argument (i. e.making it possible for preverbs to function as applicatives). Lastly (Section 5.5.5), I will discuss examples in which preverbs developed from previous adverbs. In this case, there is no ambiguity as the composites take no arguments. 5.5.1. Movable preverbs Constructions involving multiple preverbs are often semantically equivalent to constructions involving a preverb (the IP ) and an adverb or an adposition. (46) a. Construction with multiple preverbs ēélios dè ouranoû ex-ap-ólōle sun. nom ptc heaven. gen out_of-utterly-perish. prf .3 sg ‘The sun has perished out of heaven.’ ( Od. 20.356) b. Construction with the IP and (quasi-) prepositional phrases hōs éris ék=te theôn ék=t’ oh_that strife. nom out_of=and god. gen . pl out_of=and anthrṓpōn ap-óloito man. gen . pl utterly-perish. opt . fut .3 sg ‘Oh! May strife perish utterly from among gods and men! ’ ( Il. 18.107) <?page no="219"?> 5.5. The syntactic status of multiple preverbs 219 In (46)b, ék does not occur in preverbal position, but it is repeated in front of its genitive modifiers, namely theôn ‘gods: gen ’ and anthrṓpōn ‘men: gen ’. It is separated from the genitives by the enclitic coordinative particle =t(e) ‘and’, which suggests its free-standing status. Also, the adverbial status of ék is backed up by the fact that it bears the accent and hosts an (en)clitic coordinative particle. Arguably, in passages such as (46), the IP apois not sufficient to assign a spatial orientation to the composites. 28 Thus, in cases like these, a further spatial modification - in the form of a preverb, an adposition, or an adverb - is mandatory. The comparison of examples (47)a-b is also instructive in this respect: (47) a. Passage containing eknemésēse d’ Apóllōn Pergámou feel_resentment. aor .3 sg ptc A . nom P . gen ek-kat-idṓn out_of-downward-look. ptcp . aor . nom ‘But Apollo, looking down from Pergamus, had indignation.’ ( Il. 4.507) b. Passage lacking ek- (and containing eph’(í) instead) eph’ hippopólōn Thrēikôn on herding_horses. gen . pl T . gen . pl kath-orṓmenos aîan downward-look. ptcp . prs . nom . m / p land. acc ‘[…] as he looked upon the land of the Thracian horsemen.’ ( Il. 13.4) In (47)a, the EP ekimposes a Source-orientation onto the verb kat-eîdon ‘look downward’. By contrast, in (47)b, the composite kath-oráō ‘look downward’ has a Goal-orientation, suggested by the adverb epí ‘on(to)’, occurring in initial position. 29 Apart from ek-ap-óllumi ‘perish utterily out of ’ and ek-kat-eîdon ‘look downward from’, other composites behave in a similar fashion: eis-án-eimi ‘go upward to’, eis-an-ágō ‘lead upward to’, ex-ana-lúō ‘set quite free from’, ek-kateph-állomai ‘leap down from against’, kat-eph-állomai ‘leap down against’, and huper-kata-baínō ‘go downward over’ (Zanchi 2014: 121-133). 28 I discussed the early semantic bleaching undergone by apoin combination with óllumi in Section 5.4.1. 29 The composites kat-eîdon and kath-oráō contain two different roots for seeing (i. e. *u̯ ei̯ d- ‘see, catch sight of ’ and *ser ‘keep an eye on’; LIV 2 665-667, 532), which in Ancient Greek merged into the same paradigm. <?page no="220"?> 220 5. Multiple preverbs in Homeric Greek 5.5.2. Optional preverbs With other composites, such as ex-ana-dúomai ‘emerge from’, EP s can even be omitted completely without violating the grammar or altering the meaning of the sentence: (48) a. Construction with multiple preverbs (cf. (43)) poliês halòs ex-anadûsai grey. gen sea. gen out_of-emerge. ptcp . aor . nom . pl . f ‘(The seals sleep close together), after emerging out of from the grey sea.’ ( Od. 4.405) b. Construction without the EP karpalímōs d’ anédu poliês halòs quickly ptc emerge. aor .3 sg grey. gen sea. gen “And (Thetis) quickly arose from the grey sea.” ( Il. 1.359) Arguably, the construction in (48)b is allowed, as the orientation of ana-dúomai ‘emerge’ in combination with a noun phrase referring to the ‘grey sea’ and inflected in the genitive case ( poliês halòs ) is unambiguous: it is common knowledge that entities can emerge out of the sea, whereas they cannot emerge ? into the sea. Homeric poems offer more of such passages: See Zanchi (2014: 123-132) for a thorough discussion of the composites eis-aph-iknéomai ‘arrive at, come to’, eis-ana-baínō ‘go upward to’, eis-kata-baínō ‘go downward to’, eg-kata-pḗgnumi ‘thrust firmly in’, and epi-pro-ḯēmi ‘send forth (to)’ (cf. also Zanchi 2017b). 5.5.3. Non ambiguous constructions In other occurrences, such as (49)a-b, the lack of adjacency between the EP and the argument makes it difficult to assume any binding syntactic relation between them. (49) a. Post-verbal argument kaì ex-ap-ébēsan hetaîroi nēós and ouf_of-away_from-walk. aor .3 pl comrade. nom . pl ship. gen ‘And (the) comrades stepped out of the ship.’ ( Od. 12.306-307) <?page no="221"?> 5.5. The syntactic status of multiple preverbs 221 b. Non-immediately preverbal argument tôi d’ ára par-kat-élekto gunḗ dem . dat ptc ptc beside-down-lie. aor .3 sg . mid woman. nom ‘A woman, (whom he brought from Lesbos,) lay next to him.’ ( Il. 9.664) According to Imbert’s (2008: 209 ff.) explanation, in occurrences such as (49)a-b, the argument and the EP are no longer syntactically related. However, such an explanation commits Imbert to assuming either postpositions for Proto-Indo-European, a debatable point (cf. Section 3.1.3), or a non-economic path of development, including an intermediate postpositional stage between two distinct adverbial stages: (50) ( PIE ) *adverb > postposition > adverb > preverb / preposition In the light of (49)a-b, it is simpler to assume no postpositional stage between the adverbial origin and the preverbal development. Similarly, in passages such as (51), a syntactic relation holding between the EP and the argument can be excluded, as the form of the argument and the form required by the EP do not match: (51) ou gár pṓ poté m’ hôde theâs éros neg ptc yet ever 1 sg . acc thus goddess. gem desire. nom oudè gunaikòs thumòn enì stḗthessi neg woman. gen spirit. acc in breast. dat . pl peri-pro-khutheìs edámassen around-forth-be_poured. ptcp . aor . pass . nom overpower. aor .3 sg ‘For never such a desire for goddess or mortal woman has so been poured all round my breast and overwhelmed my heart.’ ( Il. 14.315-316) In (51), the prepositional phrase enì stḗthessi ‘in my breast(s)’ expresses the Goal-participant taken by the composite peri-pro-khéō ‘be poured all around’, but cannot be selected by the EP peri-. Notably, these occurrences challenge Imbert’s claims that EP s are relational in nature, as not all multiple preverb composites take prepositional arguments ( contra Imbert 2008: 212). <?page no="222"?> 222 5. Multiple preverbs in Homeric Greek 5.5.4. Multiple preverbs as transitivizing morphemes The composite ex-apo-nízō ‘wash thoroughly’ (a Homeric hápax ) features the following construction: (52) toû pódas exapénizen dem . gen foot. acc . pl wash_thoroughly. impf .3 sg ‘(And the old woman took a bright-shining pot,) thoroughly washed his feet, (and poured much water in.)’ ( Od. 19.387) The passage in (52) tells about Odysseus’ old nurse who recognized him, while washing his feet. The composite ex-apo-nízō is transitive and takes the direct object pódas ‘feet’. The EP ekhighlights the actional notion of completeness, already inherent to apo-nízō ‘wash off completely’ (Chantraine 1953: 97), rather than being an applicative marker: the composite apo-nízō ‘wash off’, which only contains the IP apo- , is also transitive. As shown in (53)a-b, apo-nízō (and its Homeric and Attic-Ionic variant apo-níptō, analogically constructed from the future and the aorist forms; cf. DELG 754) can take the accusative either of the washed entity (53)a, or of the removed substance (53)b: (53) a. Accusative of the washed entity (cf. (52)) tḕn apo-nízousa phrasámēn dem . acc away_from.wash. ptcp . prs . nom perceive. aor .1 sg . mid ‘While washing it [Odysseus’ scar], I recognized it.’ ( Od. 23.75-76) b. Accusative of the removed substance apo-nípsantes mélana bróton ex away_from.wash. ptcp . aor . nom . pl black. acc blood. acc out_of ōteiléōn wound. gen . pl ‘As (they) had washed the black blood from the wounds …’ ( Od. 24.189) The example (53)b contains the same elements as the composite ep-apo-nízō, though arranged in a different order: in (53)b, the EP ekfunctions as a preposition, and retains its Source-meaning, which is bleached in (52), resulting in a lexicalized composite. 30 30 To be more precise, in (53), the preposition ek also partially deviates from its etymological meaning: it no longer has elative value, but only a generic ablative meaning. <?page no="223"?> 5.5. The syntactic status of multiple preverbs 223 Furthermore, the same construction alternation shown for apo-nízō is also featured by the simplex verb nízō ‘wash’, which may take the accusative of the thing washed (e. g. Il. 16.230; cf. (53)a) or of the removed substance (e. g. Il. 11.830; cf. (53)b). Hence, the transive construction featured by ex-apo-nízō (52) is arguably not the final step of the lexicalization process undergone by the composite. In other words, the addition of exand apohas nothing to do with the extension of the transitive construction to ex-apo-nízō ( contra Imbert 2008: 206 ff.). By contrast, there are composites in which the addition of preverbs has the effect of centralizing a previously peripheral argument (cf. Chapter 3; Peterson 2007 on applicatives). In examples (54)a-b, the addition of proand diabuilds transitive verbs out of intransitive manner of motion verbs, i. e. théō ‘run’ and baínō ‘walk’. The centralized participants are the Goaland the Path-participants, respectively. (54) a. hup-ek-pro-théō ( Il .9.505) behind-out-forth-run ‘run forth from behind’ > ‘outstrip’+ acc [pro-théō+ acc ] b. ek-dia-baínō ( Il. 10.198; cf. (31)) entirely-through-walk ‘walk through entirely’ > ‘pass over, cross’+ acc [dia-baínō + acc ] in (54)a-b, the Homeric preverbs proand diaseem to function as applicatives, i. e. ‘a means some languages have for structuring clauses which allow the coding of a thematically peripheral argument or adjunct as a core-object argument. Such constructions are signaled by overt verbal morphology’ (Peterson 2007: 1). 31 However, the passivization test, which would unambiguously prove the applicative function of preverbs in (54)a-b, is not available: these two composites are not attested in the passive. 31 Horrocks (1981: 44) assigns a similar function to the preverb prosin pros-eîpon ‘address’ (< ‘to-say’); his claim however is challenged by counterexamples (cf. Chapter 4, examples (9)-(10)). Viti (2008a) thoroughly investigates the similarities between Homeric preverbs and applicatives, focusing on their common ability to introduce topical arguments. Viti (2008a) regards topicality as the key for understanding the subsequent development of preverbs into markers of telicity (cf. Section 5.5.5). <?page no="224"?> 224 5. Multiple preverbs in Homeric Greek 5.5.5. Composites taking no second argument Several Homeric composites containing multiple preverbs take no second arguments. This does not necessarily imply that preverbs have developed into actional markers but simply suggests that preverbs in principle might not call for further spatial specifications. In fact, some composites that retain their compositional and spatial meaning, such as hup-ek-pro-réō ‘flow forth from beneath’, take no second argument: (55) polù d’ húdōr kalòn hup-ek-pró-reen much ptc water. nom fair. nom under-from-forth-flow. impf .3 sg ‘And much and clear water was flowing forth from beneath.’ ( Od. 6.86-87) In (55), the text tells about the abundant water of a spring, welling up from beneath ( hup-ek- ) and flowing all around ( pro- ). The preverbs profile each component of the trajectory covered by the flowing water. Arguably, they do not need further specifications, as how water springs from beneath the ground is part of the common encyclopedic knowledge presumably shared by speakers. Preverbs in (55) above and (56) below both draw a link to common, non-salient (i. e.topical) pieces of information. (56) hoì d’ ep-an-éstēsan peíthontó te dem . nom . pl ptc at-up-stand. aor .3 pl obey. impf .3 pl . m / p and poiméni laôn herdsman. dat people. gen . pl ‘(As soon as he spoke in this manner, he was the first going away from the assembly,) they stood up at (his words) and obeyed the herdsman of people.’ ( Il. 2.84-85) In Section 5.4.2, I already discussed the redundancy of the IP anain this composite. Interestingly, the addition of the EP epíestablishes an anaphoric reference to the previous context: the Achaeans did not stand up on their own initiative, but only after Nestor’s words. As Chantraine (1953: 106) highlights, epioften underlines the feeling that comes from an event, which is also the function that episeems to have in (56). Thus, in (56), epicontributes to textual cohesion, a function that has been assigned to Hittite and Homeric preverbs by Boley (2004: 56-58) and to Vedic preverb repetitions by Klein (e. g. 1987, 2007). In other composites from my sample, multiple preverbs show similar behaviors: e. g. an-ep-állomai ‘leap upon after’ (see example (58)), di-éx-eimi ‘go out <?page no="225"?> 5.6. Preverb ordering 225 through’(29), and ep-en-tanúō ‘bind tightly to’. Another clear context in which a preverb contributes to textual cohesion is example (20), containing the composite pro-kath-ízō ‘perch forth’. As I pointed out in Section 5.2.1, prohas the function of providing precise textual cues for the comparison between Achaeans’ and birds’ forward movement. 5.6. Preverb ordering Table 29 summarizes the positioning of Homeric preverbs. Not all preverbs allow for both interior and exterior positioning; moreover, not all possible combinations of preverbs are actually attested. This might be a matter of gap in the attestations or suggest that there are constraints, or tendencies, driving the univerbation of preverbs. p reverb e xterior m edial i nterior amphi- 2 (100 %) - ana- 1 (7 %) - 13 (93 %) apo- 3 (23 %) - 10 (77 %) dia- 1 (50 %) - 1 (50 %) eis- 7 (100 %) - ek- 18 (44 %) 6 (15 %) 17 (41 %) en- 1 (33 %) - 2 (67 %) epi- 6 (67 %) - 3 (33 %) huper- 1 (100 %) - hupo- 10 (91 %) 1 (9 %) kata- 1 (11 %) 1 (11 %) 7 (78 %) para- 13 (100 %) - peri- 1 (33 %) - 2 (67 %) pro- 2 (14 %) - 12 (86 %) Tab. 29: The positioning of Homeric Greek preverbs and their frequencies <?page no="226"?> 226 5. Multiple preverbs in Homeric Greek 5.6.1. Previous accounts of preverb ordering Bybee (1985: 33-35) emphasizes that an “order” and “relevance principle” rules the order of affixes on stems. Such a principle implies that the more relevant an affix is to the stem, the closer it will be to that stem (cf. Section 5.3.3). Taking inspiration both from Bybee’s work and from Craig’s (1993) and Grinevald’s (2003) studies on the affix order of directionals in Jakaltek Popti’ (Mayan family, Guatemala), Imbert (2008: 236 ff.) posited synchronic semantic constraints determining the ordering of preverbs in Homeric composites. Imbert arranged the Homeric preverbs in three slots, the leftward numbering starting from the verb stem, as shown in Table 30. These slots are not interchangeable: a preverb in the slot [-3] cannot be more internal than a preverb in the slot [-1]; nor overlapping: two preverbs belonging in the same slot cannot co-occur. In line with Papke (2010: 68), I rename the three slots as follows: - [-1] Orientation / Trajectory . Preverbs belonging in this slot determine how the TR is oriented on its trajectory (vertical or horizontal orientation) and express median types of motion (Path). - [-2] Location . Preverbs belonging in this slot localize the TR with respect to the LM during a motion event; they establish the action at a point in space. - [-3] Direction . Preverbs belonging in this slot determine the relation of the trajectory with respect to the LM . -3 Direction -2 Location -1 Orientation / Trajectory VVerb eis- ‘(in)to’ ek- ‘out of ’ epi- ‘at, onto’ amphi- ‘on both sides’ apo- ‘off’ en- ‘in,into’ para- ‘beside’ peri- ‘around’ huper- ‘above’ hupo- ‘under’ parek- ‘out beside’ hupek- ‘out from under’ ana- ‘up, back’ kata- ‘down’ pro- ‘forth’ dia- ‘through’ Tab. 30: Imbert’s synchronic semantic constraints on preverb ordering Overall, the closer the preverb is to the verbal stem, the higher its relevance is to the determination of the trajectory and of the TR . Conversely, the farther the preverb is from the verbal stem, the higher its relevance is to the determination of the LM and of how trajectory relates to it. <?page no="227"?> 5.6. Preverb ordering 227 5.6.2. Issues with previous approaches Imbert’s approach to preverb ordering raises a number of issues. To begin with, a purely synchronic account is not appropriate for analyzing Homeric poems, which constitute an inherently diachronic corpus (cf. Section 1.3.2). Moreover, her commitment to this synchronic account forces her a priori to get rid of a number of composites, i. e. those containing double prepositions whose second member is ek- . When they are univerbated to verbal stems, these combinations result in an interior ek- , whereas, according to Imbert, this preverb is supposed to occupy the outermost slot [-3]. Imbert’s treatment of double prepositions, however, is not consistent: she neglects to discuss composites containing apo-pro- , amphi-peri- , and peri-pro- , although they can also occur as double prepositions in Homer, simply because these sequences comply with her constraints (see fn. 11 in this Chapter, Section 5.2.3). In addition, as argued in Section 5.2.3, the existence of double prepositions does not imply that the preverbs univerbated as a double unit: for example, we have seen that ap-ek-lanthánomai (30), and apo-pro-ḯēmi (35), di-éx-eimi (29), and di-ex-eréomai (21) can be better analyzed as containing an EP and an IP , than as containing a double preverb. Imbert also a priori excludes pro-kat-hízō ‘perch forth’, as it contains two preverbs belonging to the [-1 Figure Orientation and Median Path] slot. This analysis underestimates the polysemy of Greek preverbs: in pro-kat-hízō, prodoes not indicate Path, but the final Location in which the birds complete their trajectory (cf. example (20)). In this respect, it is more consistent to Imbert’s [-3 Path / Ground relation] slot. Furthermore, Imbert (2008) does not sufficiently take into account the interaction between verbs, preverbs, and cases in describing spatial relations. For example, in (57), the EP epídoes not establish a Path / Ground relation (slot [-3]), as it is supposed to do. Rather, the described spatial relation is static and does not imply any trajectory at all (i. e. Path in Imbert’s terms): (57) oudoû ep-em-bebaṑs hupsērephéos threshold. gen on-in-walk. ptcp . prf . nom high_roofed. gen thalámoio bedroom. gen ‘Standing upon the threshold of the high-roofed bedroom.’ ( Il. 9.582; cf. (23)) In (57), the composite ep-em-baínō is in the perfect and has the resultative meaning of ‘stand upon’. Thus, a resultative stative verb takes the adpositionless <?page no="228"?> 228 5. Multiple preverbs in Homeric Greek genitive ( oudoû ) of Location. The spatial relation in (57) does not imply motion, and thus epíindicates Location (as preverbs of the [-2] slot do), rather than Motion. 32 Imbert (2008) also excluded the composites an-eph-állomai ‘leap upon after’ (58), kat-eph-állomai ‘leap down against’ (59), and ek-kat-eph-állomai ‘leap down from against from’ (60) for methodological reasons. In fact, if the segmentation resulting in hállomai is correct (see Section 5.2.1.1), then the interior position of epiposes problems for Imbert’s constraints. (58) hōs eîd’, hṑs an-ep-âlto, kaì as see. aor .3 sg so up-after-spring. aor . sg . mid and eukhómenos épos ēúda pray. ptcp . nom . m / p word. acc speak. impf .3 sg ‘As soon as (Achilles) saw (him), so (he) leapt up at (his sight) and, praying, said a word.’ ( Il. 20.424) (59) ḗtoi hó g’ ex híppōn truly dem . nom ptc out_of horse. gen . pl kat-ep-álmenos antíos éstē down-against-leap. ptcp . aor . nom . mid set_against stand. aor .3 sg ‘(Oïleus) verily leapt down from his chariot and stood and faced him.’ ( Il. 11.94) (60) ouranoû ek-kat-ep-âlto di’ aithéros heaven. gen from-down-against-leap. aor .3 sg . mid through ether. gen ‘(Athena) leapt down from heaven through ether.’ ( Il. 19.351) In (58), epiarguably refers to the Stimulus-participant, who was seen by Hector and triggered Hector’s hostile leaping. In (59) and (60), epialso carries a hostile sense, pointing to the enemy to be faced by the leaping hero or deity. In (58)-(60), episeems to contradict Bybee’s order and relevance principle, as the interior preverb is possibly less consistent with the semantics of the verb ‘leap’ than the exterior one. However, Bybee’s principle by definition applies to affixes, whereas Homeric preverbs partially retain their clitic status (cf. Section 5.1.2). Thus, anaand katamight simply retain much of their original adverbial usage, 32 In line with this interpretation, the prepositional phrase ep’oudoû ‘upon the threshold’ ( epí + gen ) always occurs with stative verbs such as hístēmi ‘stand’ (e. g. Od .1.104) or hézomai / hízō ‘sit’ ( Od .4.718, 10.62, 17.339). <?page no="229"?> 5.6. Preverb ordering 229 which is in fact documented for both elsewhere in Homer (Chantraine 1953: 90, 112). Lastly, in none of the above examples, is epispecified by a further nounor prepositional phrase indicating Stimulus or Goal. Conversely, in (60), the semantic contribution of kata- (Path) is spelled out by the prepositional phrase di’aithéros ‘through ether’, whereas that of ek- (Source) is specified by the adpositionless genitive ouranoû ‘from the heaven’. In this respect, the preverb ordering is not surprising for (60). 5.6.3. An integrated account of preverb ordering My account of preverb ordering integrates Bybee’s order and relevant principle with other cognitive and historical kinds of explanations. The “redundancy” principle. As far as developments of preverbs are concerned, Bybee’s relevance could more aptly be renamed and described as semantic overlapping or redundancy. Those preverbs whose meaning is subsumed by the semantics of the verbal stem that they modify are likely to attach more closely to it. Ultimately, they might also be re-analyzed as part of the verb or as actional markers. The “lower salience” principle. Preverbs whose semantic contribution is not further specified by a noun or prepositional phrase tend to attach more internally. These are likely to be Pathand Source-preverbs, as the Path and the Source components are arguably less salient to the expression of spatial relations. 33 Their low salience, or high topicality, also explains their tendency to be re-analyzed as markers for completion (cf. Viti 2008a, 2008b). Both these two principles contribute to clarifying why apo- (Source) , kata-, ana-, pro- (Path) preferably select the interior position. Idiosyncratic developments of certain preverbs. The two principles outlined above do not override the fact that each preverb’s meaning or path of development can be relevant to its positioning. Although seemingly obvious, it is worth stating this principle since it reminds us to resist the temptation to overgeneralize (see also Boley 2004: 23). For example, why does the Source-preverb ekhave a weaker tendency to select the interior position with respect to apo- ? A possible answer is that ekbears heavier semantic content than apo- : whereas apoindicates a generic separation, ekalso carries an elative nuance. The preverb eknevertheless develops telic usages on grounds relating to metaphor. events can be thought of as locations , and in particular as containers : both events 33 On the marginal status of Path and perlatives within case systems, see Luraghi (2003: 22) and Malchucov & Spencer (2009: 614). On the lower saliency of Source with respect to Goal, see Ikegami (1987), Ungerer & Schmidt (1996) and Verspoor, Dirven & Radden (1999). <?page no="230"?> 230 5. Multiple preverbs in Homeric Greek and containers imply physical or temporal boundaries. Furthermore, moving out of a container event implies that such an event is over. As a telic marker, then, ekcan modify composites containing another preverb attached at a preceeding stage: e. g. apo-nízō ‘wash off’ vs. ex-apo-nízō ‘wash off thoroughly’; apo-tínō ‘pay back’ vs. ex-apo-tínō ‘pay back in full’; dia-baínō ‘go through’ vs. ek-diabaínō ‘pass over’. Two other preverbs select the exterior position: eis- ‘(in)to’ and para- ‘besides’. The former is a relatively recent formation, not spread to all Greek dialects ( *en- ‘in’+ s ‘modal ending’ > eis with compensatory lenghtening; DELG 326; LIPP II 226 and references therein). As such, eisretains a clear spatial meaning. The preverb para- , instead, is known for its exceptional ‘adverbial’ character (Luraghi 2003: 131), which is consistent with its preference for the exterior positioning. <?page no="231"?> 6.1. Preverbs in Old Church Slavic 231 6. Multiple preverbs in Old Church Slavic 6.1. Preverbs in Old Church Slavic 6.1.1. State of the art: the Slavic prefixes Preverbs are a major topic in Slavic linguistics. Within this field of study, such uninflected items are usually called “prefixes” due to their affixal status (cf. Sections 3.1.4.1 and 6.1.2). Slavic preverbs have been widely investigated both from a diachronic and from a synchronic perspective, and within both cognitive and formal-oriented theoretical frameworks. From a diachronic viewpoint, most studies devoted to Slavic preverbs are concerned with understanding the steps and the reasons for their development from spatial adverbs into “bounder perfectives” in modern Slavic languages (in Bybee & Dahl’s terms 1989; cf. Section 6.1.2.2). Up to now, many scholars have attempted to show whether, in which manner, at which diachronic stage, and to which extent Slavic preverbs retained their original spatial meanings, gained new lexical, abstract, and actional meanings, and possibly underwent fully grammaticalization processes into purely aspectual markers of perfectivity. Another major topic of investigation is the interaction of the new derivational “Slavic-style aspect” (in Dahl’s 1985 terms), coded by means of perfectivizing preverbs and imperfectivizing suffixes, with the aspectual system inherited from Proto-Indo-European, which was based on verbal stem alternations (cf., among many others, Meillet 1924; van Wijk 1929; Kuryłowicz 1929; Vaillant 1939, 1946; more recently, see Dickey 2012; Eckhoff & Haug 2015; Ruvoletto 2016 on Old Russian; Wiemer & Seržant 2017, and references therein). From a synchronic perspective, the preverbs of modern Slavic languages have been dealt with in different ways. On the one hand, a good number of cognitively-oriented studies aim to build semantic maps for Slavic preverbs, that is, organized networks (also called “schemas”) in which all concrete, abstract, and aspectual meanings of linguistic polysemous items are directly or indirectly connected to a prototypical meaning in a motivated way (Lakoff 1987; Langacker 1987; Luraghi 2003; Tyler & Evans 2003). Most work in this regard has been done on East and West Slavic: see Janda & Šarić (2009), who report all the relevant references published before 2009; Le Blanc (2010); Nesset et al. (2011); Janda & Lyashevskaya (2013). As regards South Slavic languages, cognitively-oriented studies on preverbs are also available (cf. e. g. Klikovac 2004, 2006; <?page no="232"?> 232 6. Multiple preverbs in Old Church Slavic Lazarevska-Stančevska 2004; Tchizmarova 2005, 2006; Belaj 2008; Šarić 2008, 2010, 2012 and references therein). Most importantly, a number of these studies try to identify lexical content in the so-called “empty prefixes” (named as such, e. g. in Avilova 1959, 1976; Tixonov 1964, 1998; Forsyth 1970; Švedova et al. 1980; Čertkova 1996), which are usually regarded as being purely aspectual, i. e. grammatical (e. g. Le Blanc 2010; Janda 2012; Janda & Lyashevskaya 2013; Janda et al. 2013; Dickey & Janda 2015). The idea that the putative purely perfectivizing prefixes might not be lexically empty dates back to Vey and van Schooneveld’s work (Vey 1952; van Schooneveld 1958), and has recently been supported by new statistical evidence. In particular, works by Janda and her associates show that certain Russian preverbs have unique semantic profiles and exhibit strong inclinations to combine with verbs that belong to specific semantic classes. Furthermore, Janda argues that preverbs’ semantic profiles usually make reference to a spatial path, thus emphasizing the linkage between preverbs and their cognate prepositions, which occasionally retain concrete meanings lost by preverbs. 1 Nevertheless, most scholars, especially if formally-oriented, still support the view that certain preverbs simply function as purely perfectivizing morphemes. On the other hand, many formally-oriented works on modern Slavic preverbs are concerned with classifying them, according to their semantic and syntactic behaviors, into (a) “lexical/ internal” and (b) “super-lexical/ external” preverbs, and with identifying their combination rules in multiple preverbation or stacking. 2 Besides lexical and super-lexical preverbs, some scholars suggest that there exists another separate class of preverbs, (c) the “empty” preverbs mentioned above (cf., among others, Babko-Malaya 1999; Filip 1999, 2003; Ramchand 2004; Romanova 2004; Svenonius 2004a, 2004b; Tatevosov 2008, 2009). Briefly, these classes of preverbs show the following distributional and semantic 1 An exception to this otherwise valid rule seems to be the preverb po-, which has lost its spatial profile in Russian . This preverb productively only adds the abstract meanings of perfectivity, as in po-xudet’ ‘lose weight’, and delimitation, as in po-pisat’ ‘write for a while’, whereas its cognate preposition po still exhibits the spatial meanings of ‘on, over, along’. Scanty remnants of the original spatial surface contact (Path) meaning of poare found in the composites po-kryvat’ - po-kryt’ ‘cover’, po-sejat’ ‘sow’, po-xromirovat’ ‘chrome’, and po-nikelirovat’ ‘nickel’ (see Shull 2003: 147-172 for further details). Preverbs such as Russian po- , which synchronically have little to do with the spatial meanings of their cognate prepositions, are named “orphan prefixes” (i. e. orphan preverbs) by Dickey (2012). Interestingly, however, pocontinues spatial meanings in both West (e. g. Polish) and South Slavic (e. g. Slovenian, Croatian). By contrast, pois also orphan in Bulgarian. 2 Tatevosov (2008), on the basis of the distributional behavior of Russian completive doand repetitive pere- , argues for the existence of a third group of preverbs, which he names “intermediate prefixes”. <?page no="233"?> 6.1. Preverbs in Old Church Slavic 233 differences. (a) The lexical preverbs display directional and non-compositional meanings. In addition, they are able to derive a new lexical item, and to modify the argument structure of the simplex verb onto which they attach, usually, but not exclusively, by adding new arguments (e. g. Blg. dam ‘give’ vs. pro-dam ‘sell’). (b) By contrast, the super-lexical preverbs are considered to have predictable and actional (or quantizing) meanings, such as ‘begin’ (ingressive), ‘finish’ (egressive), ‘for a while’ (delimitative), ‘for many times’ (accumulative, distributive), and they do not modify the argument structure of the simplex verb onto which they attach (Blg. pre-glăštam ‘swallow up’, pre-pročitam ‘read again’). (c) Lastly, the preverbs with a pure perfectivizing role perfectivize an imperfective verb without adding any semantic modification (Rus. na-pisat’ ( pfv ) ‘write’ vs. pisat’ ( ipfv ) ‘write’). 3 In contrast with cognitively-oriented scholars, who regard preverbs as polysemous items, these authors claim that different preverbs with the same phonological content show unrelated lexical, super-lexical and pure perfectivizing usages. As a combinatory rule in multiple preverbation or preverb stacking, it is stated that whenever two preverbs stack onto a single verbal stem, the innermost should be a lexical one, whereas the outmost a super-lexical one, as in (1)a-b: (1) a. Lexical usage of the preverb otkry-t’ ot-kry-t’ ot-kry-va-t’ coverinf p fv away-coverinf p fv > open away-coveripfv inf i pfv > open ‘cover’ ‘open’ ‘open’ b. Super-lexical usage of the preverb popo-ot-kry-va-t’ distr -away-openipfv inf p fv ‘open one after another’ (adapted from Romanova 2004: 255) As mentioned, multiple preverbs in modern Slavic languages have received some attention: to the above-cited studies, the paper by Istratkova (2004) should be added. Istratkova (2004) focuses on Bulgarian, which is a language allowing for the exceptional stacking of as many as seven preverbs, as shown by example (2): 3 Janda (2007) suggested the labels (a) “Specialized perfectives” for verbs combining with lexical preverbs, (b) “Complex Act Perfectives” for verbs combining with super-lexical preverbs, and (c) “Natural Perfectives” for verbs combining with pure perfectivizing preverbs. <?page no="234"?> 234 6. Multiple preverbs in Old Church Slavic (2) za-iz-po-na-raz-pre-prodavam ( ipfv ) vs. prodam ‘sell’ (< dam ‘give’) start-completely-little_by_little-cumulation-to_many-again-sell ‘start selling again everything to many little by little’ (adapted from Istratkova 2004: 311) Multiple preverbs are far less productive in other Slavic languages than they are in Bulgarian: Russian, for instance, usually combines two preverbs, although combinations of three preverbs are also attested (Istratkova 2004: 306). By contrast, investigations devoted to multiple preverbs in ancient Slavic languages are scarce and usually rely only on data taken from dictionaries (e. g. Fil’ 2011). The recent paper by Zanchi & Naccarato (2016) aims to expand the lexicographic source of data: it examines the semantics of Old Church Slavic and Old Russian multiple preverbs by using corpus-based data semi-automatically extracted from the TOROT Treebank (Haug & Jøhndal 2008; Eckhoff & Berdicevskis 2015). 6.1.2. The status of preverbs in Old Church Slavic 6.1.2.1. The morphological status of preverbs and their meanings Especially since Kuryłowicz (1964: 171-178), it has been generally acknowledged that Indo-European preverbs developed from previously free-standing adverbs with spatial meanings, which were originally able to modify both nouns and verbs (cf. Section 3.1.3; on Slavic in particular, see Vaillant 1966: 467 ff.; 1977: 109 ff.; Arkadiev 2015: 201 ff.). Later on, these adverbial items increasingly became bound to the verbs or nouns that they modified and thus underwent the well-known functional bifurcation into preverbs and adpositions. Some ancient Indo-European languages such as Vedic (cf. Chapter 4), Homeric Greek (cf. Chapter 5), and possibly Archaic Old Irish (cf. Chapter 7), transmit to us examples attesting to a linguistic stage in which preverbs still showed proclitic behavior (cf. further Section 3.1.3 with references). In other words, preverbs could be separated from the modified verbal stem by means of non-lexical and lexical material (this is the so-called “tmesis”). As far as we know, proclitic behavior is not documented for preverbs in any Slavic text, and thus Old Slavic preverbs are generally said to exhibit the morphological status of prefixes, that is, of bound morphemes (Vaillant 1966: 467; Wiemer & Seržant 2017). By contrast, Baltic, the most closely related branch to Slavic, preserves scanty traces of the preceding proclitic behavior, as shown in (3): <?page no="235"?> 6.1. Preverbs in Old Church Slavic 235 (3) a. Standard Lithuanian b. Old / non-standard Lithuanian per-si-kel-ti ap-mi-šviesk akis throughrefl -raiseinf up/ on / to-me-illuminate eyes ‘to move (to another place)’ ‘illuminate my eyes (literally, the eyes on / to me)’ (adapted from Wiemer & Seržant 2017; Rosinas 1995: 10 f.) c. Multiple preverbs and reflexive pronouns in Lithuanian su-si-pa-žìn-ti pri-si-pa-žìn-ti withrefl -alongv inf atrefl -alongv inf ‘to become acquainted with’ ‘to confess, avow’ (adapted from Nevis & Joseph 1992: 96) In (3)a-c, the reflexive and the first person enclitic pronouns intervene between the preverb(s) and the verbal stem, thus splitting the EP s from the remaining elements of the composite (this is similar to what happens in Old Irish, on which see Chapter 7). To sum up, the free positioning that we have observed for Vedic and Homeric preverbs (Chapters 4 and 5) is no longer allowed in the earliest attested Slavic languages, in which preverbs are advanced in their univerbation process and seem to show distinct behaviors from those of the corresponding prepositions. However, as Vaillant (1977: 30) points out, Old Church Slavic only occasionally contains passages in which a construction with a preverbed verb and a bare case freely alternates with an equivalent construction containing a simplex verb and a prepositional phrase. Vaillant notes the constructions do-iti+ gen (preverbed verb + bare case) and iti do+ gen (simplex verb + prepositional phrase). As discussed in Section 6.5.1, even in the relatively small sample of multiple preverbs, one can find such examples, in particular with the composites prěd( ъ )po-lagati ‘distribute to’ and v ъ s-pri-imati ‘receive in return’. Though the univerbation process of preverbs was advanced in Old Church Slavic, the primarily meaning brought about by preverbs was still detectable (cf. Vaillant 1966: 470 ff.), as exemplified in (4). <?page no="236"?> 236 6. Multiple preverbs in Old Church Slavic (4) Old Church Slavic preverbs with spatial meanings iti ‘go’ iz(ъ)-iti ‘go out’ ob-iti ‘go around’ ot(ъ)-iti ‘go away’ po-iti ‘go along a surface’, ‘depart from’ prě-iti ‘go over, across’ vъn-iti ‘go into’ However, there are also several composites to which preverbs add lexical, but non-spatial, meanings (Aitzetmüller 1991: 150 ff.), as shown in (5)a-e: (5) Old Church Slavic preverbs with lexical but non-spatial meanings Preverb ‘basic meaning’ Simple verb Composite a. na- ‘onto’ rešti ‘say, tell’ na-rešti ‘announce, designate’ b. ot ъ - ‘away from’ dati ‘give’ otъ-dati ‘give back’ c. s ъ - ‘with, down from’ tvoriti ‘make’ sъ-tvoriti ‘prepare’ d. v ъ z- ‘upward’ dati ‘give’ vъz-dati ‘give back’ e. za- ‘behind’ byti ‘be, become’ za-byti ‘forget’ In a number of the above composites, the semantic contributions of preverbs are quite clear: both ot ъ and v ъ zmean ‘back, in return’ in examples (5)b and (5)d. Similarly, it is easy to understand how the lexicalized meaning of za-byti ‘forget’ developed from ‘behind-be’ (5)e. In contrast, the semantic analysis is less straightforward for other composites: in (5)a and (5)c, the exact semantic contributions brought about by naand s ъ are more difficult to figure out (possibly, narefers to the Area-participant, whereas s ъ brings about the idea of togetherness, linked to the event of preparing something). In other composites, where the meanings of preverbs are particularly difficult to identify, traditional grammars usually describe preverbs as mere completion or perfectivity markers (Vaillant 1966: 471 ff.; Aitzetmüller 1991: 154 ff.). 4 4 As correctly pointed out by Viti (2008a: 395-396) and frequently noted throughout this work, the terms completion / telicity and perfectivity were interchangeably used by less recent scholars who discussed the actional values of preverbs. Nowadays, however, the terminology has been fixed, and it is generally agreed that the so-called completion / telicity pertains to the lexical aspect, whereas perfectivity belongs to the category of grammatical aspect (cf. Section 2.3.2.1). As discussed in Section 6.1.2.2, the issue of whether in Old Church Slavonic the system of “bounder perfectives” was at its beginning, under construction, or fully developed, is still being debated. Thus, for the sake of prudence, I speak here of completion / telicity markers, and not of perfectivity markers. <?page no="237"?> 6.1. Preverbs in Old Church Slavic 237 Preverbs that are explicitly ascribed a telic / perfective function follow: na- (na-pl ъ niti ‘fill up’ vs. pl ъ niti ‘fill’), ot ъ - (otъ-pěti ‘finish a song’ vs. pěti ‘sing’) , pro- (pro-sl ъ ziti ‘burst into tears’ vs. sl ъ ziti ‘cry’) , u- (u-biti ‘kill’ vs. biti ‘hit’) , and za- (za-klati ‘slaughter’ vs. klati ‘kill’) . Along with telicity, other types of actional meanings can be expressed by preverbs: for example, both proand v ъ zshow an ingressive meaning in pro-glagolati and vъz-glagolati ‘start talking’, as well as uin u-z ь rěti ‘catch sight of ’. Delimitative meanings were also attested: pofunctions as a delimitative marker in the composite s ъ -po-žiti ‘live for a while with’ (cf. Section 6.4.2). 6.1.2.2. The development of Slavic preverbs 6.1.2.2.1. A glimpse into the contemporary system of “bounder perfectives” Modern Slavic languages exhibit a fully developed system of grammatical or viewpoint aspect (for further discussion on the distinction between grammatical and lexical aspect, see Section 2.3.2). This grammatical category is regularly expressed, in Slavic languages, by means of both preverbal (6)a and postverbal (6)b morphology: (6) a. Simple verb: ipfv → preverbation: pfv Polish łowi-ć → z-łowi-ć catchinf p -catchinf b. Preverbation: pfv → secondary suffixation: ipfv Polish na-mówi-ć → na-mówi-a-ć p -persuadeinf p -persuadesfx inf ( ∅-mówić ‘say, tell’) (adapted from Wiemer & Seržant 2017) As exemplified in (6)a, preverbs, besides adding new lexical meaning to the simplex verbs onto which they attach, are systematically employed to perfectivize imperfective simplex verbs. In some cases, as shown in (6)b, the meaning of the new compound verb turns out to differ greatly from that of the corresponding simple verb after the addition of the preverb, and thus the latter no longer constitutes an adequate imperfective counterpart. Therefore, a new imperfective verb is built via secondary suffixation. As mentioned in Section 6.1.2.1, preverbs were originally free-standing spatial adverbs, which later on underwent grammaticalization processes into bounder perfectives. On the other hand, imperfectivizing suffixes mostly go back to ancient Indo-European suffixes used to build various actional formations semanti- <?page no="238"?> 238 6. Multiple preverbs in Old Church Slavic cally close to imperfectivity, such as iteratives, habituals, or duratives (as a cover term for these, Wiemer & Seržant 2017 adopt the label “marked unbounded verbs”). Later on, the semantic markedness of these suffixes gradually bleached, and such suffixes ended up functioning as unmarked options for expressing any type of unbounded events, including progressives, iteratives, habituals, and other values related to imperfectivity (cf. Section 6.3). 5 Typologically speaking, the aspectual system outlined above complies with the so-called “bounder perfective” type, whereby perfective markers arise from adverbs such as up , down , over , and through , which attach onto verbs to create a sense of completion (Bybee & Dahl 1989; Bybee et al. 1994). Bounder perfectives are not typologically unusual. Many languages employ previously spatial adverbs to express telicity: cf. English eat up, German aufessen ‘eat up’. Italian also exploits particle-verb locutions to convey telic meanings, especially with the Source-particle via ‘away’: e. g. volare ‘fly’ vs. volare via ‘fly away’; passare ‘pass’ vs. passare via ‘fade away’ (Iacobini & Masini 2006). Interestingly, Breu (1992) and Arkadiev (2015) described Slavic, Hungarian, Yiddish, and Caucasian languages as belonging to a linguistic area in which preverbs are employed to develop a rudimentary aspectual system. What is typologically exceptional with Slavic is the paradigmaticization of the system of bounder perfectives (Bybee & Dahl 1989: 86). Moreover, the so-called Slavic-style aspect is rare for three further reasons: (i) its relative independence from time reference; (ii) its derivational character; (iii) its usual, though not absolute, association with telicity (Dahl 1985: 84-85; Eckhoff & Haug 2015: 191). 6.1.2.2.2. The reasons for and timings of the grammaticalization of Slavic preverbs The link between the lexical, the subsequent actional, and eventually aspectual usages of preverbs lies primarily in the fact that preverbs, as spatial markers, are able to add an inherent endpoint to the spatial events expressed by verbs (Maslov 1959; Bermel 1997: 466; Shull 2003; Wiemer & Seržant 2017, among others; cf. also Viti 2008a, 2008b on Homeric Greek preverbs). So, for example, whereas iti ‘go’ can have both telic and atelic readings, its composite counterparts (cf. (4)) can only be understood as telic. Later on, preverbs also came to function as bounder markers for more abstract events: their spatial meanings bleached, and in parallel, their distribution broadened. The addition of an inherent endpoint to spatial and then to non-spatial events is a straightforward explanation for the development of Goal-preverbs into ac- 5 The only surviving postverbal affixes related to bounded events are the nasal suffix -nu and the archaic infix *-n-. In Polish, for example, the cognate -ną suffix is used for semelfactive formations. <?page no="239"?> 6.1. Preverbs in Old Church Slavic 239 tional markers. However, in a number of languages, preverbs etymologically related to Source seem to have formerly acquired a telic function, as shown by Dickey (2012) for Slavic, and by Zanchi (2017b) for Ancient Greek (in addition, as Iacobini & Masini 2006 show, a special actional function is ascribed to the Italian Source-particle via ‘away’). An additional explanation for the developments of Source-preverbs comes into play. e vents can be metaphorically thought of as locations : departing from an event (i. e. from a location) implies that such an event has been completed. Dickey (2012: 84) attempted to provide an alternative, but compatible, explanation, speculating that “the combination of the original source meaning combined with a new abstract goal meaning produces a semantic potential including both the beginning of an action and its endpoint, i. e. the complete profile of an action.” A crucial factor contributing to the reanalysis of spatial preverbs as actional markers is the overlap between their meaning and the meaning of the verbal stem onto which they stack, which makes the spatial reading redundant. For example, compare iti ‘go’ with po-iti ‘go along a surface’: the spatial contribution brought about by po- ‘along a surface’ to iti is redundant, as the act of going already implies the presence of a surface along which the TR moves (Dickey 2007). This overlap, known either as “Vey-Schooneveld effect” (Vey 1952; van Schooneveld 1958) or as “subsumption” (Poldauf 1954), arguably triggered the reinterpretation of the preverb as a default telic marker, as the actional reading is the only possible piece of salient information added by the preverb, given the redundancy of the spatial addition (Zaliznjak & Šmelev 2000; Dickey 2007; Plungjan 2011: 319; Janda et al. 2013; Arkadiev 2015; Wiemer & Seržant 2017). 6 Later on, the telic reading triggered by the preverb became conventionalized, as an effect of the opposition between the composite and its simplex counterpart. Subsequently, the extension of some preverbs to verbs denoting atelic activities, i. e. events that do not entail an inherent endpoint, represents a further step toward their grammaticalization, and crucially maximizes the distribution of preverbs as bounders (Lehmann 2004). Occasionally, preverbs add boundaries to atelic events in Old Church Slavic, for example in the composite s ъ -po-žiti ‘live for a while with someone’, in which the prefix poseems to have the only function of establishing temporal limits to the activity of living with someone (see Section 6.4.2). Table 31, adapted from (Wiemer & Seržant 2017), summarizes the stages of the development just described: 6 The Vey-Schooneveld effect is briefly mentioned within studies dealing with languages different from Slavic (cf. Rovinskaja 2001; Panov 2012 on Latin, cited after Ruvoletto 2016). <?page no="240"?> 240 6. Multiple preverbs in Old Church Slavic (0) spatial / non-spatial meaning lexical modification (1) telic meaning actional modification (2) conventionalized telic meaning (3) limitation (perfective reading on atelic events) aspectual modification Tab. 31: Grammaticalization of Slavic prefixes as markers of perfectivity This grammaticalization process did not give rise to a single marker of perfectivity in modern Slavic languages. In Russian, for example, pro-, za-, s- (and other preverbs to a lesser extent) mark telic perfective verbs, whereas poperfectivizes atelic verbs; in Bulgarian, iz-, o-, na-, s- (ordered on a frequency scale) mark perfectivity on telic predicates, whereas podoes the same for atelic ones (Dickey 2012). Such an abundance of markers of perfectivity is one of the reasons a number of scholars doubt, though from different perspectives, whether the development of Slavic preverbs into bounder perfectives should be regarded as a proper grammaticalization process (cf. Section 2.2.4; Campbell 2001; Newmeyer 2001; Joseph 2004). In their view, the grammaticalization of Slavic preverbs results as an epiphenomenon from a series independently motivated semantic changes. Another issue relating to the development of the Slavic-style aspect is the exact timing of the steps displayed in Table 31. Several scholars believe that the Slavic system of grammatical aspect had already developed in the Old Church Slavic period, and verbs were already organized in pairs of perfectives ~ imperfectives (van Schooneveld 1951; Dostál 1954; Eckhoff & Janda 2014). According to other scholars, instead, the Slavic-style aspect of aspectual pairs did not develop earlier than the Old Russian period (Borodič 1953; Bermel 1997; Lazorczyk 2010), thus implicitly arguing that the Old Church Slavic affixational system expressed lexical rather than grammatical aspect. Other investigations take an intermediate position: the aspectual system was established, but only in its beginning stages and not widespread across all verb classes (Amse-De-Jong 1974; Forsyth 1972; see also Růžička 1957 on Old Russian). Similarly, in Kamphuis’ (2016) view, Old Church Slavic exhibited a fully fledged verbal aspect system, in which overt morphological elements, such as prefixes and suffixes, functioned as indicators of verbal aspect. Accordingly, Kamphuis (2016) regards verbs that do not carry such outer morphological characteristics as “anaspectual”. Anaspectual verbs allegedly constituted a well-defined large group of verbs in Old Church Slavic: they were able to occur in the aorist and past participle (perfective contexts), and in the imperfective and present participle (imperfective <?page no="241"?> 6.1. Preverbs in Old Church Slavic 241 contexts); furthermore, they could fulfill the functions of both perfective and imperfective verbs. According to Wiemer & Seržant (2017), the steps outlined in Table 31 occurred with the following timings: step (0) is assumed to have come about in Early Common Slavic (before 300 AD ), steps (1) and (2) in Common (300-700 AD ) and Early Slavic (Old Church Slavic and Old Russian times), and step (3) in Late Slavic (= Modern Slavic). The above-mentioned Old Church Slavic compound s ъ -po-žiti ‘live for a while with someone’, containing a delimitative po-, could be itself a timid signal of an early beginning of step (3), which then developed more fully during the 16 th century (Dickey 2007). In parallel, Eckhoff & Haug (2015) contributed to a description of the advancement of stages (1-2) with their corpus-based study on Codex Marianus and Codex Zographensis . In these texts, they observed a significant correlation between preverbed verbs (without imperfectivizing suffixes) and perfective contexts, and between suffixed stems and imperfective contexts (see Section 6.3). 6.1.3. The ongoing development of Slavic prepositions Old Church Slavic is a prepositional language. 7 However, secondary postpositions are also sporadically attested (Vaillant 1977: 109, 134 ff.). These are derived from nouns, and accordingly take the genitive case: gen + radi, gen + dělja , and gen + cěšta ‘because of ’ (cf. Lat. gen + causā and gen + gratiā; in Modern Slavic languages, these postpositions came to be preposed: cf. Rus. dlja togo , BCS radi toga ‘because of than’). Old Church Slavic primary prepositions derive from Proto-Indo-European spatial adverbs, which later on developed into preverbs or adpositions (Section 3.1.3; on Slavic in particular, Vaillant 1966: 467, V: 109; Hewson & Bubenik 2006: 178-204). Old Church Slavic also exhibits secondary prepositions. They partly derive from Proto-Indo-European adverbial roots with the Slavic addition of innovative linguistic material. The prepositions nad ъ ‘up’, pod ъ ‘down’, and prěd ъ ‘before, in front of ’ are such: they go back to Proto-Indo-European *h 2 en- , *h 2 (é)po- , and *preh 2 -i- , extended by means of the suffix -d ъ of uncertain etymology (Machek 1997: 466; Hewson & Bubenik 2006: 182-183). Despite their secondary derivation, nad ъ ‘up’, pod ъ ‘down’, and prěd ъ ‘before, in front of ’ can also be used as preverbs: nadъ-ležati ‘be laid upon’, podъ-imati ‘take upon, support, lift’, and prědъ-po-lagati ‘distribute to’. Secondary prepositions may also derive from adjectival, adverbial, or nominal bases (so-called “relator nouns”; cf. Engl. in front of ) (Vaillant 1977: 128 ff.; Hewson & Bubenik 2006: 184-187). 7 The present Section is primarily based on Vaillant (1977: 21 ff.), and Lunt (1965: 143-153). <?page no="242"?> 242 6. Multiple preverbs in Old Church Slavic 6.1.3.1. The scanty vestiges of the former adverbial status of prepositions As mentioned in Section 6.1.2.1, Old Church Slavic prepositions are advanced in their grammaticalization process. Nevertheless, they still exhibit a number of features that point to their former adverbial status. To begin with, though prepositions usually immediately precede the noun that they take, and with which they constitute a single accentual unit, one can find exceptions to this rule (Vaillant 1977: 110). One such exception is the locution v ъ …město ‘in …(the) place’, in which the noun město regularly occurs displaced from the preposition v ъ (Vaillant 1977: 135; e. g. v ъ ryby : gen město : acc ‘in the place of a fish’). In addition, prepositional phrases could be split by relative, anaphoric and demonstrative pronouns ( i prěbyvajǫ : ind . prs .1 sg v ъ nego: 3 sg . gen ljub ь ve : loc ‘I remain in his (of him) love’, Mar. Jn 15.10), as well as by genitive or by other types of nominal modifiers ( v ъ efremъ : acc naricaemъ : ptcp . prs . p . acc grad ъ: acc ‘in the village called Ephraim’, Mar. Jn 11.54). However, rather than constituting actual examples of discontinuous syntax, such putative splits can arguably be attributed to scribes’ desire to retain the Greek word order and are especially frequent in Codex Suprasliensis. Interestingly, in a few passages, Slavic scribes restored the usual Slavic word order, and thus different variants for the same locution may be attested, as shown in (7)a-b: (7) a. Continuous prepositional phrase (Mar. Lc 9.27) ot ъ stojęštiich ъ sъde from stand. ptcp . prs . gen . pl here b. Discontinuous prepositional phrase (Mar. Mk 9.1) ot ъ sъde stojęštiich ъ from here stand. ptcp . prs . gen . pl ‘among those standing here’ (adapted from Vaillant 1977: 110) Furthermore, as exemplified in (8)a-b, with coordinated nouns, the repetition of the preposition is frequent but not compulsory in Old Church Slavic, as well as in most ancient Indo-European languages (Vaillant 1977: 11): <?page no="243"?> 6.1. Preverbs in Old Church Slavic 243 (8) a. Repetition (Mar. Lc. 2.44) i iskaašete ego vъ roždenii i vъ znanii and seek. impf .3 du 3 sg . gen in birth. loc and in knowledge. loc ‘Then they began looking for him among relatives and among friends.’ b. Coordination reduction (Mar. Mt 5.45) ěko sl ъ n ь ce svoe s ь ěat ъ na z ъ ly i so_that sun( n ). acc poss .3 sg . acc . n rise. prs .3 sg on evil. acc . pl and blagy good. acc . pl ‘For he makes his sun rise on the evil and (on) the good.’ Example (8)b instantiates a type of the so-called coordination reduction, i. e. a coordination between two coordinants one of which is not a constituent (in other words, a type of coordination involving ellipsis; see Haspelmath 2007a; Luraghi forthc. d). In (8)b, the ellipsis of the preposition na ‘on’ occurs. Both discontinuity (7)b and coordination reduction (8)b are typical features related to non-configurationality (Luraghi 2010), contributing to suggesting that prepositions were quite advanced, but not yet fully developed, in Old Church Slavic. In addition, prepositions could take different cases. The Old Church Slavic cases that could accompany verbs were the accusative, genitive, dative, instrumental, and locative cases. A number of prepositions could take as many as three cases (Vaillant 1977: 145 ff.): e. g. za ‘behind’, and s ъ ‘with, downward’ take the accusative, the genitive, and the instrumental cases; po (etymologically) ‘surface-contact (Path), ablativity (Source)’ takes the accusative, the dative, and the locative cases. 8 There are contexts in which case alternation is undoubtedly meaningful: the accusative ~ locative alternation expresses Goal ~ Location in combination with na ‘up’ and v ъ ‘in’, and so does the accusative ~ instrumental 8 Originally, pocarried the meaning ‘surface contact’ and ‘ablativity’ (Dickey 2012). This preverb goes back to Proto-Indo-European *h 2 po , cf. Goth. afar ‘away from’, AG apó ‘away from’, Lat. ab ‘away from’, pōnere ‘put, place’ (< *po-sinere ), Ved. ápa ‘away’ (LIPP II 66 ff.). It is still productive with spatial meanings in Croatian, Slovene and West-Slavic (Dickey 2011, 2012). Elsewhere, it developed actional meanings including resultative, delimitative, ingressive, distributive and attenuative. Considering its complex semantic development, I chose to assign poits etymological meaning ‘surface-contact, ablativity’ within this work. <?page no="244"?> 244 6. Multiple preverbs in Old Church Slavic alternation with nad ъ ‘above’, pod ъ ‘under, from under’, and prěd ъ ‘before, in front of ’. 9 6.1.3.2. The residual usages of the prepositionless cases Prepositionless cases were still able to express spatial and non-spatial meanings only to a very limited extent in Old Church Slavic. 10 The prepositionless accusative, besides marking the direct object of transitive verbs (unless negated), can express extension in time (Duration) or in space (Measure, rather than Path). The prepositionless accusative is also taken by a number of composite motion verbs, preverbed with do- , na-, o(b ъ )-, po-, prě-, pro- (e. g. do-iti ‘arrive at’, more often + gen ; na-xoditi ‘come upon’; o(bъ)-xoditi ‘go around’; po-xoditi ‘walk onto (a surface)’; prě-xoditi ‘go around’, pro-iti ‘walk through’). Prepositionless genitive can also express different spatial roles with several preverbed verbs, containing do-, iz-, ot ъ -, prěd ъ -, s ъ -, u-, za- (do-iti ‘arrive at’ (Goal); iz-iti ‘go out of ’ (Source); ot-iti ‘go to’ (Goal); prědъ-xoditi ‘walk in front’ (Goal); sъ-vlěšti ‘strip of ’ (Source-like participant); u-běžati ‘escape from’ (Source); za-blǫditi ‘be far from’ (Location)). Moreover, it systematically expresses the direct object with negated verbs or in partitive contexts, as opposed to the accusative case, which in its turn expresses the complete affectedness of the object participant (Vaillant 1977: 74 ff.). The ablative usage of the genitive is extremely limited: the genitive case only indicates Source with a number of quasi-adverbial expressions taken by a few verbs, such as běžati ‘run’ and osvoboditi ‘liberate’ (Hewson & Bubenik 2006: 179). Source is usually expressed by iz + gen , s ъ+ gen , or ot ъ+ gen (Thomason 2006: 127 ff., 135 ff.). The dative case is employed to express the indirect object, as well as the direct object with certain verbs, such as pomošti ‘help’, zaviděti ‘envy’, and dosaditi ‘annoy’, involving two human participants (on the linkage between PIE dative case and animacy, cf. Luraghi 2003: 63 ff. with references). Its use as a Goal marker is only residual in Old Church Slavic (cf. Mar. Mt 14.11, 17.19), but flourishing until the 15 th century in Old Russian. The dative case is more often accompanied by the preposition k ъ to express the Goal-participant (Vaillant 1977: 86; Lunt 1965: 148; Thomason 2006: 138 ff.). 9 For a thorough overview of the usages of prepositions in Old Church Slavic, I refer to the traditional grammars by Leskien (1922[1971]: 115 ff.), Vaillant (1977: 22 ff.), Aitzetmüller (1991: 154 ff.), Lunt (1965: 151 ff.), as well as to the comparative study by Thomason (2006). 10 For an exhaustive examination of the usages of Old Church Slavic cases, see Vaillant (1977: 22-108) and Lunt (1965: 143 ff.). <?page no="245"?> 6.2. Multiple preverbs in numbers 245 The instrumental case can express Instrument, Means, Cause and Agent (with passive verbs). It is not usually employed for the Comitative-participant, which is usually encoded by s ъ + ins (Vaillant 1977: 91-93). Besides functioning as second argument with a very limited number of verbs, such as obilovati ‘be glad of ’, it occasionally expresses Path and Duration. The prepositionless locative is only residual in Old Church Slavic, but more widespread in Old Russian. It possibly survives in a few Old Church Slavic quasi-adverbial locative and Time expressions, such as zimě ‘in winter’ and polu nošti ‘at midnight’ (Lunt 1965: 147). The prepositionless locative is taken as a second argument by a number of simple verbs (e. g. kosnǫti sę ‘touch’), and by several composites preverbed with pri- (pri-ložiti ‘add’, also + dat , na + acc ; pri-bližiti ‘approach’) and other preverbs, such as naand s ъ - ( na-ležati ‘be upon, press’; s ъ byti sę ‘take place’) (Vaillant 1977: 102 ff.; Lunt 1965; 147-148). 6.2. Multiple preverbs in numbers 6.2.1. Composites with multiple preverbs Old Church Slavic composites with multiple preverbs were extracted automatically from the TOROT Treebank, as the Old Church Slavic section of this treebank already provides with word-formation annotation. 11 From the extracted data, I excluded composites containing the negative prefix ne- , as the negation does not belong to the category of Indo-European ‘adverbs / preverbs / adpositions’ (Cuzzolin et al. 2006), whose members show at least one of the following features: i. Proto-Indo-European etymology going back to a deictic / local adverb; ii. basic spatial meaning; iii. subsequent functional bifurcation into adpositions / preverbs in the daughter languages. The composites excluded in this way are the following: iz-ne-mošti ‘become unable, weak, ill’, ne-do-kon ь čati ‘not complete’, ne-do-mysliti , ne-do-myšljati ‘be in doubt’, ne-do-stati ‘lack’, and ne-na-viděti ‘hate’. These composites are nevertheless interesting: (a) in iz-ne-mošti, the negative prefix occurs internally, while the exterior preverb izseems to provide the composite with an ingressive meaning ( iz- ‘ingressive’ + ne ‘negation’ + √ mog- ‘be able’). (b) The composites ne-do-myšljati ‘be in doubt’ and ne-do-stati ‘lack’ are only used in combination with the negative prefix (but cf. the -ipresent do-mysliti ‘consider, understand’ from the 11 Courtesy of Hanne M. Eckhoff, to whom I express my deep gratitude. <?page no="246"?> 246 6. Multiple preverbs in Old Church Slavic same root √ mysl - ‘think’). (c) The composite ne-na-viděti ‘hate’ shows a non-compositional meaning that can be compared with that of its ingressive counterpart v ъ z-n е -n а -viděti ‘conceive a hatred, come to hate’ (cf. Sections 6.3.4 and 6.4.2). This selection process yielded 23 lemmas and 363 occurrences for Old Church Slavic composites, which are displayed in Table 32, together with their frequencies in each Codex contained in the TOROT Treebank (as discussed in Section 6.3, however, a number of such composites are organized in aspectual pairs or triplets). c omposite m eaning c odex Marianus Zographensis Suprasliensis Total is-po-vědati confess, explain 6 6 50 62 is-po-věděti confess, explain 6 9 36 51 is-po-vědovati confess, explain 0 0 2 2 is-pro-vrěšti overturn, destroy 2 2 2 6 iz-ob-rěsti find out 0 0 4 4 iz-ob-rětati find out 0 0 1 1 o-pro-vrěšti overturn 1 1 0 2 prěd ъ -po-lagati distribute to 0 1 0 1 pri-ižditi (-iz-žiti) 12 spend in addition 1 1 0 2 pri-ob-rěsti acquire 8 6 17 31 pro-po-vědati proclaim, predict 19 22 40 81 pro-po-věděti proclaim, predict 10 3 6 19 pro-po-vědovati announce, proclaim 0 1 0 1 s ъ -po-žiti live for a while with 0 0 1 1 s ъ -prě-byvati remain together with 0 0 1 1 s ъ -v ъ -kupiti gather together 1 0 3 4 s ъ -v ъ -kupljati unify, copulate 0 0 3 3 v ъ s-po-męnǫti start remembering, remind 2 2 2 6 <?page no="247"?> 6.2. Multiple preverbs in numbers 247 c omposite m eaning c odex Marianus Zographensis Suprasliensis Total v ъ s-pri-im а ti receive in return 3 1 2 6 v ъ s-pri-jęti receive in return 6 8 10 24 v ъ z-n е -n а -viděti come to hate 10 8 1 19 za-po-vědati order 7 3 11 21 za-po-věděti order 6 7 2 15 t otal 88 81 193 363 Tab. 32: Old Church Slavic composites with multiple preverbs As Table 32 shows, Codex Suprasliensis alone attests to more than the half of the occurrences of multiple preverbs (193 out of 363). Moreover, it contains six composites that do not occur in the other two manuscripts. These are is-po-vědovati ‘explain, tell’, iz-ob-rěsti ‘find out’, iz-ob-rětati ‘find out’, s ъ -po-žiti ‘live for a while with someone’, s ъ -prě-byvati ‘remain together with’, and s ъ -v ъ -kupljati ‘unify, copulate’. As the language of the Codex Suprasliensis is usually regarded as being more recent than the Slavic variety attested by the translations of the Gospels (Lunt 1965: 9), these data might suggest that multiple preverbs have become more widespread in the later variety of language, and that the composites listed above are more recent than the remaining ones (with the caveat that, when working with inherently limited corpora, such as the Old Church Slavic Codices , a gap in the attestation does not necessarily correspond to an actual absence in the grammar or in the lexicon; Joseph & Janda 2003: 15-16). Interestingly, in cases of polysemous composites, Codex Suprasliensis usually attests to a more lexicalized meaning, as one may expect from its more recent dating. This is shown by examples (9) and (10): (9) i d ъ sky tr ъ ž ъ nik ъ i sědališta and table( f ) acc . pl merchant. gen . pl and bench( n ) acc . pl prodajǫštix ъ golǫbi isprovrъže sell. ptcp . prs . gen . pl dove. acc . pl overturn. aor .3 sg ‘(He) overturned the tables of the merchants and the benches of those selling doves.’ (Mar. Mk 11.15) 12 The following phonological rule comes into play here : z + ž → žd (Lunt 1965: 44). <?page no="248"?> 248 6. Multiple preverbs in Old Church Slavic (10) ô glasa silo ad ъ isprovrъgъši ptc sound. gen power. voc hell. acc destroy. ptcp . pst . voc ‘o power of the word, (you) destroying death, …’ (Supr. 27.200) 13 While in (9) from Codex Marianus, the composite is-pro-vrěšti is used in a semi-compositional way, in example (10) from the Codex Suprasliensis, its usage is more lexicalized (cf. Section 6.4.3). In (9), Jesus destroys merchants’ tables by literally throwing (√ vrěg- ) them forward ( pro- ) completely ( iz( ъ )- ), i. e. by overturning them. By contrast, in (10), the event of destroying ( is-pro-vrěšti ) affects a metaphorical Patient ( ad ъ: acc ‘hell > death’), and is performed by a metaphorical Agent ( glasa : gen silo : voc ‘power of word’). Table 33 contains those composites that are attested in later (South) Slavic languages (here, I reported only Bulgarian), as well as their Old Church Slavic and Bulgarian meanings. 14 c omposite m eaning b ulgarian m eaning is-po-vědati confess, explain iz-po-vjadam (pfv) confess, profess is-po-vědovati confess, explain iz-po-vjadvam (ipfv) confess, profess iz-ob-rěsti find out iz-ob-retja (pfv) invent, devise iz-ob-rětati find out iz-ob-retjavam (ipfv) invent, devise o-pro-vrěšti overturn o-pro-vergavam (ipfv) o-pro-vergaja (pfv) refute, disprove prěd ъ -po-lagati distribute to pred-po-lagam (ipfv) suppose, assume pri-ižditi (iz-žiti) spend in addition pri-iždam (ipfv) arrive, rise pro-po-vědovati predict, proclaim pro-po-vjadvam (ipfv) Preach v ъ s-po-męnǫti start remembering, remind v ӑ z-po-minavam (ipfv) Remember v ъ s-pri-im а ti receive in return v ӑ z-pri-emam (ipfv) perceive, apprehend v ъ s-pri-jęti receive in return v ӑ z-pri-ema (pfv) perceive, apprehend 13 Examples from Codex Suprasliensis are numbered as in the TOROT Treebank. 14 The time gap that divides Old Church Slavic from Bulgarian is wider than that separating R̥ g-Vedic from Classical Sanskrit (Chapter 4) and Homeric Greek from Classical Greek (Chapter 5). Nevertheless, Bulgarian composites are reported here, as they allow for enlightening comparisons, as shown in what follows. <?page no="249"?> 6.2. Multiple preverbs in numbers 249 c omposite m eaning b ulgarian m eaning v ъ z-n е -n а -viděti come to hate v ӑ z-ne-na-vidja (pfv) come to hate za-po-vědati order za-po-vjadam (pfv) order, command Tab. 33: Composites attested in later (South) Slavic languages and their meanings As expected, a number of composites retaining a concrete meaning in Old Church Slavic develop more abstract meanings in Bulgarian. Three cases in point are o-pro-vrěšti ‘overturn’ (11), prěd ъ -po-lagati ‘distribute to’ (12), and v ъ s-pri-jęti ‘take, receive, have back’ (13). (11) i d ъ sky o-pro-vrъže and table( f ) acc . pl over-forth-throw. aor .3 sg ‘(And he poured out the coins of the money-changers) and overturned (their) tables.’ (Mar. Jn 2.15) (12) i daěše oučenikom ъ svoim ъ da and give. impf .3 sg disciple. dat . pl poss .3 sg . dat so_that prědь-po-lagaj.tъ distribute_to. prs .3 pl ‘And (he) gave (them) to his disciples to distribute (them) to (the people).’ (Zogr. Mk 8.6) (13) i ašte v ъ zaim ъ daate ot ъ nich ъ že čaate and if in loan. acc give. prs .2 pl from rel . gen . pl hope. prs .2 pl vъs-prijęti. kaě vam ъ chvala est ъ in_return-receive. inf . prs what. nom . f 2 pl . dat credit( f ) nom be. prs .3 sg ‘And if you give a loan (to those) from whom you hope to receive back, which credit do you have? ’ (Mar. Lc 6.34) The context of (11) is similar to that of (9) above. The effects of Jesus’ anger are described: Jesus overthrows the merchants’ tables in front of the temple. The elements building the composite o-pro-vrěšti profile different components of tables’ movement: o(b ъ )describes its endpoint (‘over’), whereas prothe medial Path of movement (‘forward’); √vrěgmeans ‘throw’. In Bulgarian, the same elements make up a lexicalized composite meaning ‘refute, disprove’. The semantic shift can be easily explained: e. g. refuting / disproving an argument can be seen as metaphorically overthrowing it. <?page no="250"?> 250 6. Multiple preverbs in Old Church Slavic In (12), Jesus gives food to his disciples, so that they can distribute it to the people around him. The elements of the composite are prěd ъ - ‘in front of ’ + po- ‘distributive’ + √ lag-/ log- ‘put’. As only occasionally happens in Old Church Slavic, in Bulgarian, the same composite predpolagam has lexicalized the shifted meaning of supposing, assuming (cf. also Rus. predpolagat’ ‘suppose, assume’): its meaning is no longer spatial, nor compositional. Once more, this semantic shift is not surprising: for example, the English verb to put also shows a similar development in such expressions as As Wackernagel puts it (on prěd ъ -po-lagati , see also Sections 6.4.2 and 6.5.1). In (13), the composite v ъ s-pri-jęti ‘receive in return’ is employed to describe the event of getting a repayment: the elements building the composite are v ъ z( ъ )- ‘in return’ + pri- ‘resultative’ + √ь m- ‘take’. In Bulgarian, the idea of repayment is no longer implied by the composite, which simply means ‘perceive, apprehend’: the semantic contribution of v ъ z( ъ )has bleached. 6.2.2. Verbal roots modified by multiple preverbs In my sample, only 10 Old Church Slavic verbal roots allow for multiple preverbs. Table 34 shows these Old Church Slavic roots and their meanings, as well as their Proto-Indo-European roots and their meanings. Moreover, Table 34 provides their frequencies, i. e. the number of composites containing each root, and the verb types in the rightmost column (verbs’ classification is a simplified version of Levin’s (1993), which suffices for my purposes). ocs m eaning pie root (liv 2 page ) m eaning f requency v erb t ype √bybe *b h u̯eh 2 - (98) grow, come into being, become 1 location √kupbuy ? uncertain ? uncertain 2 transfer √loglay *leg h - (398) lie down 1 posture √minthink *men- (435) come to think 1 mental activity √rětmeet, find *reh 1 t- (501) ? ret- (Vaillant 1966: 184-185) meet, find\run 3 obtaining\ motion √vědknow *u̯ei̯d- (665) see, catch sight of 8 mental activity √vidsee *u̯ei̯d- (665) see, catch sight of 1 perception <?page no="251"?> 6.2. Multiple preverbs in numbers 251 ocs m eaning pie root (liv 2 page ) m eaning f requency v erb t ype √vrěgthrow *u̯erg u̯ - (689) throw 2 caused motion √ ь mtake *h 1 em- (236) take away 2 removing √žilive *g u̯ i̯eh 3 - (215) live 2 existence Tab. 34: Old Church Slavic verbal roots modified by multiple preverbs As Table 34 shows, composites containing a motion or a location verb proper are infrequent ( s ъ -prě-byvati ‘remain together with’, o-pro-vrěšti ‘overthrow’, and is-pro-vrěšti ’overturn, turn upside down, destroy’). This in itself suggests an advanced lexicalization for Old Church Slavic composites. As Arkadiev (2015: 217) remarks, the pattern of multiple preverbs was not productive in ancient Slavic languages but it was employed when the IP was lexicalized (cf. Sections 6.4.3 and 6.6). Some verbs of Table 34 can be assimilated to location verbs, including posture verbs ( √log-/ lag- ‘lay’), and verbs of existence ( √ži- ‘live’). Also, other verbs can be treated as motion verbs, such as perception verbs (√ vid- ‘see’), given that eyes can be directed toward or away from a certain LM , and verbs of removing, such as √ь m- ‘take away’. As for the two roots indicating a mental state, i. e. √ věd- ‘know’ and √ min- ‘think’, the former goes back to the same Proto-Indo-European root as √ vid- ‘see’, that is, *u̯ei̯d- ‘see, catch the sight of ’ (perception verb > verb of mental state) ( LIV 2 665); the latter traces back to a Proto-Indo-European root indicating a mental state itself, i. e. * men- ‘come to think’ ( LIV 2 435). The root √ rět- ‘meet, find’ has no direct reflexes even in Baltic and is difficult to provide an etymology for. LIV 2 (501) rules out a connection with *ret- ‘run’ (cf. Old Irish reithid ‘run’, as well as noun reflexes in Lithuanian r-tas ‘wheel’, Latin rota ‘wheel’, Sanskrit ráthah͎ ‘wagon’, and Old Church Slavic rota ‘oath’; contra Vaillant 1966: 184-185) and instead assumes a root *reh 1 twith the meaning of ‘meeting’ and ‘finding’ and with reflexes only in Slavic preverbed verbs. Thus, √ rětis allegedly a verb of obtaining (< *reh 1 t- ), whose semantics are similar to that of verbs of taking; as such, it can be assimilated to a motion verb. Last, there is no agreed upon Indo-European etymology for the root √kup- ‘buy’. The Old Church Slavic verb kupiti ‘buy’ is a probable borrowing from German (Germ. kaufen ‘buy’ < OHG koufōn ), which in turn is likely to be a borrowing from the Latin caupō ‘tradesman’. The origin of the Latin word is dubious itself, as well as its connection with the Greek kápēlos ‘retail dealer’ (the Greek vocalism does not match with that of its Latin putative counterpart; kápēlos is said to have a Mediterranean origin by DELG 494 and De Vaan 2008: <?page no="252"?> 252 6. Multiple preverbs in Old Church Slavic 100). In any case, the Old Church Slavic lemma kupiti is a transfer verb and, as such, it can be associated to verbs of caused motion. 6.2.3. Attested combinations of preverbs Old Church Slavic preverbs that occur in multiple preverbation are the following: iz( ъ )- ‘out of ’, na- ‘on(to)’, o(b ъ )- ‘around’, po- ‘surface contact, ablativity’, prě- ‘across’, prěd ъ - ‘in front of ’, pri- ‘at’, pro- ‘through’, s ъ - ‘with’, v ъ - ‘in’, v ъ z( ъ )- ‘up’, za- ‘behind’. Table 35 shows the attested combinations of preverbs. In the rightmost column, frequency refers to the number of composites containing each combination. None among these combinations has an overwhelming frequency with respect to the others, nor does it make up a double preposition in Old Church Slavic. No preverb is iterated. e xterior preverb i nterior preverb f requency iz( ъ )o(b ъ )- 2 iz( ъ )p о - 3 iz( ъ )pr о - 1 o(b ъ )pr о - 1 prěd ъ p о - 1 priiz( ъ )- 1 prio(b ъ )- 1 prop о - 3 s ъ p о - 1 s ъ prě- 1 s ъ v ъ - 2 v ъ z( ъ )na- 1 15 v ъ z( ъ )p о - 1 v ъ z( ъ )pri- 2 z а p о - 2 Tab. 35: Old Church Slavic preverb combinations and their frequencies 15 The composite attesting to this combination, i. e. v ъ z-n е -n а -viděti ‘come to hate’, also contains the negative prefix ne . <?page no="253"?> 6.2. Multiple preverbs in numbers 253 6.2.4. The Greek counterparts of Old Church Slavic composites Table 36 displays the Greek counterparts of Old Church Slavic composites with multiple preverbs. c omposite m eaning g reek equivalent is-po-vědati confess, explain omologéō, ex-omologéō, di-ēgéomai is-po-věděti confess, explain omologéō, ex-omologéō, ex-agoreúō, ex-ēgéomai, an-aggéllō is-po-vědovati confess, explain omologéō is-pro-vrěšti overturn, destroy ana-stréphō,kata-stréphō, diarrḗgnumi iz-ob-rěsti find out ex-eurískō iz-ob-rětati find out ex-eurískō o-pro-vrěšti overturn ana-trépō prěd ъ -po-lagati distribute to para-títhēmi pri-ižditi (iz-žiti) spend in addition dapanáō, pros-dapanáō pri-ob-rěsti acquire (ana)-ktáomai, kerdaínō, pro-xenízō pro-po-vědati predict, proclaim kērúttō, pro-ana-kērúttō, euaggelízomai pro-po-věděti predict, proclaim kērúttō pro-po-vědovati predict, proclaim kērúttō s ъ -po-žiti live for a while with sun-ana-stréphomai s ъ -prě-byvati remain together with sun-ana-stréphomai s ъ -v ъ -kupiti gather together epi-sun-ágō, sun-áptō, enóō s ъ -v ъ -kupljati unify, copulate sun-áptō, sum-meígnumi, meígnumi v ъ s-po-męnǫti start remembering, remind ana-mimnḗskō, hupo-mimnḗskō, hupómnēsin lambánō v ъ s-pri-im а ti receive in return apo-lambánō, ap-ékhō v ъ s-pri-jęti receive in return lambánō, ana-lambánō, apo-lambánō, ap-ékhō v ъ z-n е -n а -viděti come to hate miséō za-po-vědati order en-téllomai, dia-tássō, pros-títhēmi, ep-aggéllomai za-po-věděti order en-téllomai Tab. 36: The Greek counterparts of Old Church Slavic composites <?page no="254"?> 254 6. Multiple preverbs in Old Church Slavic The term “counterpart” must be used with caution, since it holds a different significance for Codices Marianus and Zographensis , on the one hand, and for Codex Suprasliensis , on the other hand. The Gospels of the Codices Marianus and Zographensis allow for a more direct comparison with their Greek parallels, though the original Greek source text has not survived the ravages of time. A good approximation of the Greek text is provided by Von Tischendorf (1869-1872) and is available in the TOROT (here referred to as GNT , Greek New Testament ). By contrast, the Greek sources of the Codex Suprasliensis either are reflected to a lesser extent in their Slavic translation or are missing altogether. Zaimov & Capaldo (1982) nevertheless attempted to collect these sources, based on a single Greek manuscript, and to amend it, either by removing certain Greek passages with no Slavic counterparts, or by integrating other passages from different Greek manuscripts (see also Capaldo 1984). Their effort resulted in a combined Greek text, which can be now accessed on the website of the The Codex Suprasliensis Project. Despite this issue of the Greek sources, the language of the Codex Suprasliensis is usually said to have been influenced by the Greek original to a greater extent than the refined translations of the Gospels contained in the Codices Marianus and Zographensis (Lunt 1965: 7-9). As emphasized by the bold print in Table 36, only four Old Church Slavic composites exhibit Greek counterparts containing multiple preverbs: i. e. pro-po-vědati ‘proclaim, predict’ ~ pro-ana-kērúttō (not exclusive) , s ъ -po-žiti ‘live, stay for a while with someone’ ~ sun-ana-stréphomai , s ъ -prě-byvati ‘remain together with’ ~ sun-ana-stréphomai , and s ъ -v ъ -kupiti ‘gather, tie together’ ~ epi-sun-ágō (not exclusive). The composites s ъ -po-žiti and s ъ -prě-byvati only occur in Suprasliensis; the Greek multiple preverb counterparts for pro-po-vědati and s ъ -v ъ -kupiti, that is, pro-ana-kērúttō and sun-ana-stréphomai , are only attested in the Greek sources of Suprasliensis (in passages corresponding to Supr. 337.2, 341.27 ff.). In fact, the Greek composite epi-sun-ágō does occur in the GNT , but it corresponds to Slavic composites different from s ъ -v ъ -kupiti, such as s ъ -b ь rati ‘gather’. These data might support the general assumption that Codex Suprasliensis has undergone Greek interference to a greater extent than Marianus and Zographensis : only in Codex Suprasliensis , one finds Greek equivalents containing multiple preverbs. All in all, Old Church Slavic composites with multiple preverbs cannot be regarded as calques from Greek (see also Section 6.6.2). Nor does the Slavic usage of preverbs point to a single translation equivalent. For example, the EP v ъ zmeans ‘in return, in exchange, back’ in the pair v ъ s-pri-im а ti and v ъ s-pri-jęti ‘receive in return’, as shown in (14), which directly continues the passage in (13). <?page no="255"?> 6.2. Multiple preverbs in numbers 255 (14) ibo i grěš ъ nici grěš ъ nikom ъ v ъ zaim ъ indeed and sinner. nom . pl sinner. dat . pl in loan. acc dajǫt ъ . da vьs-priimǫt ъ rav ъ no give. prs .3 pl so_that in_return-receive. prs .3 pl equal. acc . n ‘Even sinners lend to sinners, to get back the same amount.’ (Mar. Lc 6.34) The preverb v ъ zcan either correspond to apo- (etymologically) ‘away from’ or ana- (etymologically) ‘upward’ in the Greek sources. Attested equivalents for these Old Church Slavic composites are apo-lambánō, ap-ékhō , on the one hand, and ana-lambánō, on the other hand. It is also unlikely that v ъ zacquired the meaning of ‘in return’ as a calque from the Greek preverb ana- , on the basis of the fact that both v ъ zand anamean ‘upward’ in their spatial usages. The preverb v ъ zshows this meaning in several composites, whose Greek equivalent does not contain ana- : e. g. OCS vъz-dati ~ Gr. apo-dídōmi ‘give back’; OCS vъz-vratiti ~ Gr. apo-stréphō ‘turn back’. By contrast, there are Old Church Slavic usages of preverbs that suggest Greek influence. One case in point is izin the composites iz-ob-rěsti, iz-obrětati ‘find out’, which translate the Greek ex-eurískō ‘find out’ (only in Codex Suprasliensis ). As discussed in Section 6.2.2, the bare root √ rět- ‘meet, find’ is not attested without preverbs, thus the IP -verb altogether corresponds to the Greek simple verb eurískō ‘find’ (see also Section 6.6.2). Then, the addition of the preverbs iz- ( eurískō ) and ex- (Ancient Greek) further brings about the actional nuance of completion (cf. also Engl. find ~ find out, in which Engl. out originally had the same spatial usage as OCS izand Gr. ex- ‘out of ’). In fact, Old Church Slavic izand Gr. exare also etymologically related ( IEW 292-293; LIPP II 204 ff.). 16 However, cognacy is not necessary for two preverbs to develop parallel uses. The phonetically similar (but not etymologically related) preverbs s ъ - (OCS) and sun- ( AG ) are similarly used to convey the idea of togetherness in sъ-v ъ -kupiti , sъ-v ъ -kupljati ~ sun-áptō, sum-meígnumi ‘gather together’, and in sъ-po-žiti, sъ-prě-byvati ~ sun-ana-stréphreō ‘live for a while with’, ‘remain together with’. 16 The Bulgarian preference for perfectivizing izover po- , which is the favorite perfectivizing preverb in Russian instead (Dickey 2007), may well have been driven by Greek influence. Since Homeric times (cf. Section 5.4), exfrequently occurs in exterior position and carries telic meanings. <?page no="256"?> 256 6. Multiple preverbs in Old Church Slavic 6.3. The form of composites Unlike their Vedic and Homeric Greek counterparts (cf. Chapters 4 and 5), Old Church Slavic preverbs are undoubtedly bound morphemes, and thus their morphological status does not need to be discussed. For this reason, this Section does not aim to assess the univerbation / non-univerbation of composites. Rather, it deals with their form from another angle: the interactions between preverbs and secondary imperfectivizing suffixes, which is a particularly relevant matter for the development of Slavic-style aspect (cf. Section 6.1.2). 6.3.1. The actional suffixes of multiple preverb verbs As displayed in Table 37, a number of multiple preverb composites can be grouped into two- (or even three-) element sets of verbs (rows from 10 to 15). Suffixes (a)-(b) are explicit markers for bounded events. By contrast, suffixes from (e) to (h) are primary and secondary markers for unboundedness and more or less directly go back to a Proto-Indo-European suffix * -a- , marking various types of durative events (for a thorough discussion of the origins and the developments of such suffixes, see Wiemer & Seržant 2017 and references therein). The zero in (c) and the -ěsuffix in (d) do not express (un)boundedness: (c) constitutes an unproductive verb class including about 50 verbs of everyday activities (Lunt 1965: 131-135); the (e) class mostly includes intransitives expressing a state (Lunt 1965: 116-118). As demonstrated by Eckhoff & Haug (2015), the system of viewpoint aspect, which we know from Modern Slavic, was already quite developed in the Old Church Slavic of Codex Marianus and Codex Zographensis . In particular, Eckhoff & Haug (2015) showed what follows: (i) preverbed and unsuffixed verbs rarely occur as imperfectives, that is, in the imperfect, in the present infinitive, and in the present participle; (ii) verbs occurring in the imperfective contexts listed in (i) are explicitly imperfective: they contain one suffix among (e)-(h). Thus, as Eckhoff & Haug (2015) argue, in Old Church Slavic, there is already evidence for secondary imperfectivization. This new system, already developed, though not spread throughout the whole lexicon, coexisted along with the inherited alternations of Proto-Indo-European, in which different verbal stems were employed to express different aspectual values. In Sections 6.3.2 and 6.3.3, I examine Old Church Slavic multiple preverb verbs in light of Eckhoff & Haug’s (2015) study. <?page no="257"?> 6.3. The form of composites 257 (a) -i- (b) -nǫ- (c) -∅- (d) -ě- (e) -(j)a-/ -jaj- (f) -aj- (g) -ova-/ (h) -vabound, causative bounded everyday activities intransitive, stative unbounded, ipfv unbounded, iterative, durative unbounded, ipfv unbounded, ipfv, iterative 1. - - v ъ z-n е -n а -viděti - - - - 2. - v ъ s-po-męnǫti - - - - - - 3. - pri-ižditi - - - - - 4. - s ъ -po-žiti - - - - - 5. - is-pro-vrěšti - - - - - 6. - o-pro-vrěšti - - - - - 7. - - - prěd ъ -po-lagati - - 8. - - - - - - s ъ -prě-byvati 9. - - za-po-věděti za-po-vědati - - 10. - - is-po-věděti is-po-vědati is-po-vědovati - 11. - - pro-po-věděti pro-po-vědati pro-po-vědovati - 12. - pri-ob-rěsti - - - - - 13. - iz-ob-rěsti iz-ob-rětati - - - 14. - v ъ s-pri-jęti v ъ s-pri-im а ti - - - 15. s ъ -v ъ -kupiti - - s ъ -v ъ -kupljati - - - Tab. 37: The actional suffixes of multiply preverbed verbs <?page no="258"?> 258 6. Multiple preverbs in Old Church Slavic 6.3.2. The alternations involving the suffix -(j)a- Within multiple preverb verbs, there are three pairs involving the suffix -(j)a- : iz-ob-rěsti ~ iz-ob-rětati ‘find out’, s ъ -v ъ -kupiti ~ s ъ -v ъ -kupljati ‘gather together’, and v ъ s-pri-jęti ~ v ъ s-pri-im а ti ‘receive in return’. The former two pairs work as predicted by Eckhoff & Haug (2015): iz-ob-rěsti and s ъ -v ъ -kupiti attest no imperfective forms, whereas iz-ob-rětati and s ъ -v ъ -kupljati (with the -(j)asuffix) have no perfective forms. 17 Data are more complicated for the latter pair, i. e. v ъ s-pri-jęti ~ v ъ s-pri-im а ti ‘receive in return’. The -jęti composite, which is supposed to be the perfective counterpart of the pair, does occur in the infinite present (Mar., Zogr. Lc 6.34, Supr. 1.297, 43.166, 48.650). The Slavic infinite present translates the Greek infinite present, which is undoubtedly imperfective. Thus, a non-explicitly imperfective form occurs in an imperfective context, though an explicitly imperfective form is available. In addition, the -im а ti composite, which is instead explicitly marked for imperfectivity via the -(j)asuffix, occurs once in the past participle. As Slavic past participles translate Greek aorist participles, this means that an imperfective form occurs in a perfective context. The latter mismatches between imperfective form / perfective context and perfective form / imperfective context are attested only in Codex Suprasliensis : this manuscript is more recent than Marianus and Zographensis. Thus, such forms possibly represent an innovation, whereby the emergent viewpoint aspect has partially been emancipated from its actional origins lying in telic bounders. However, the -jęti and -im а ti forms are also used interchangeably in the translations of the Gospels in a couple of passages. In Mt 6-5-6 and Mt 6.16, Codex Marianus attests to the -(j)aform and Zographensis to the suffixless form. The context for one of these passages is shown in (15)a: (15) Mt 6.5-6 a. (And when you pray, do not be like the hypocrites, for they love to pray standing in the synagogues and on the street corners to be seen by others.) amin ь gljǫ vam ъ . ěko vъspriemlj ǫ tъ [Mar.] vъsprim ǫ tъ [Zogr.] truly say. prs .1 sg 2 pl . dat that receive_in_return. prs .3 pl m ъ zdǫ svojǫ. reward( f ). acc poss . refl .3 pl . acc . f ‘Truly I tell you that they have received back their reward already.’ 17 It must be mentioned that we are dealing here with very low frequencies: iz-ob-rěsti x4, iz-ob-rětati x1, s ъ -v ъ -kupiti x4, and s ъ -v ъ -kupljati x2 (cf. Table 32). <?page no="259"?> 6.3. The form of composites 259 b. (But when you pray, go into your room, close the door and pray to your Father, who is unseen.) i otc ъ tvoi vidęi v ъ taině and father. nom poss .2 sg . nom see. ptcp . prs . nom in secret. loc vъz-dastъ tebě avě back-give. prs .3 sg dat .2 sg openly ‘And your Father, the one seeing in secret, will reward you openly.’ The present of v ъ s-pri-jęti or v ъ s-pri-im а ti translates the Greek present ap-ékhousin : prs .3 pl ‘get in return’. Here (and in similar passages), Jesus is expressing an atemporal precept: he is urging his followers not to perform good actions in public just to be rewarded by people’s admiration. Rather - Jesus preaches - one should perform good acts privately to receive God’s compensation in the future. In a similar context, Mt 6.2, Marianus also hands down the -jęti form, unmarked for the imperfective aspect. To sum up, four passages (i. e. Mar . Mt 6.2, Zogr. Mt 6.2, 6.6, 6.5) contain a present form that is not explicitly marked for imperfectivity and yet has no future meaning, as perfective presents usually do. Interestingly, going ahead to Mt 6.6 (15)b, one notes that the subsequent verb of giving back, i. e. v ъ z-dati in the present tense, unmarked for imperfectivity, does have a future value. Accordingly, the present v ъ z-dati always translates the future apo-dṓsei : fut .3 sg ‘give back’ of the GNT . In light of this, it might be the case that Slavic translators were unsure about how to render the Greek opposition between the present and the future in Old Church Slavic. These translator uncertainties possibly have produced the inconsistencies outlined above. 6.3.3. Triplets containing speech verbs Further interesting cases are the couple / triplets za-po-věděti ~ za-po-vědati ‘order’, is-po-věděti ~ is-po-vědati ~ is-po-vědovati ‘confess, explain’, and pro-pověděti ~ pro-po-vědati ~ pro-po-vědovati ‘predict, proclaim’. The verbs containing the -ovasuffix are extremely rare and occur as variants for the -ajforms in Zographensis, as shown in (16): (16) Mk 5.20 i načęt ъ propovědovati [Zogr.] propovědati [Mar.] and begin. aor .3 sg tell_openly. inf . prs ‘And (he) bagan to tell …’ <?page no="260"?> 260 6. Multiple preverbs in Old Church Slavic The -vědati forms, though explicitly unbounded via the -ajsuffix, are occasionally used in the aorist and in the past participle, i. e. in perfective contexts. 18 By contrast, the -ěforms, unmarked for unboundedness, are never used in the imperfect, in the present participle and in the infinite, i. e. in imperfective contexts. The aorists and the past participles of the unbounded forms are mostly attested in Suprasliensis (Supr. 3.510, 4.481, 7.120, 10.3, 16.93, 16.97, 16.402, 23.214, 25.440, 28.289, 29.162, 31.11, 31.239, 31.343, 32.90, 32.377, 45.12, 45.206, 46.189, 46.465). However, similar aorists of imperfectives are not unknown to Codex Marianus either ( Mt 15.4, 28.20). In the same passages, Zographensis instead uses the -ěverb, as shown in (17): (17) Mt 28.20 oučęšte ję bljusti v ь sě eliko teach. ptcp . prs . nom . pl 3 pl . acc watch. inf . prs indf . acc . pl . n rel . indf zapovědachъ [Mar.] vam ъ zapověděchъ [Zogr.] order. aor .1 sg 2 pl . dat ‘Teaching them to observe all things, whatever I commanded you.’ Note further that the composites po-věděti ~ po-vědati ‘tell’, lacking the EP , attest to the same mismatches outlined above: po-vědati unexpectedly occurs in perfective contexts in Zographensis ( Lc 8.47, 14.21, 18.37), where Marianus instead presents po-vědati. To sum up, the composites containing -vědati, though explicitly unbounded, seem not to have become specialized for imperfective contexts yet. 6.3.4. The perfectivizing value of v ъ z- It is much harder to detect aspectual pairs based on the alternation of a preverbed verb (perfective) with a non preverbed one (imperfective) than to detect pairs and triplets of the types in Sections 6.3.2 and 6.3.3 (Lunt 1965: 92). In fact, in Old Church Slavic, the addition of any preverb virtually also produces alterations in meaning (cf. Section 6.4). In contrast, actual aspectual pairs feature both similarity in form and identity in meaning. However, even in the small sample of multiple preverb verbs, such pairs do exist. The composite v ъ z-n е -n а -viděti ‘come to hate’, in which v ъ zhas an in- 18 Eckhoff & Haug (2015: 218-221) also found these, as well as other speech and thought verbs occurring in the aorist, though explicitly imperfect. <?page no="261"?> 6.4. The semantics of multiple preverbs 261 gressive value (cf. Section 6.4.2), and which is explicitly perfective due to the presence of this preverb, does not occur in imperfective contexts. In parallel, n е -n а -viděti ‘hate’ is unmarked for boundedness, and accordingly can be used both in perfective and imperfective contexts. Another composite containing the ingressive v ъ zonly occurs in perfective contexts, i. e. v ъ s-po-męnǫti ‘start remembering, remind’. This verb however also contains the suffix -nǫ-, which explicitly marks bounded events, and thus perfectivity. Indeed, the composite po-męnǫti ‘remember’, which lacks v ъ z-, also selects only perfective contexts. Accordingly, it has an unbounded counterpart in po-minati ‘remember’, which only appears in imperfective contexts. 6.4. The semantics of multiple preverbs 6.4.1. Preverbs with spatial, abstract and actional meanings From a semantic standpoint, Old Church Slavic preverbs show a two-fold development. Either they become polysemous elements, gaining new lexical meanings more or less straightforwardly connected with their basic spatial meaning; or they lose part of their lexical content, and in parallel acquire actional meanings, which are ultimately responsible for their subsequent grammaticalization into bounder perfectives (cf. Section 6.1.2). In Old Church Slavic, however, the system of the viewpoint aspect was still evolving, as discussed in Sections 6.1.2.2 and 6.3: thus, one can easily observe the semantic linkage between the lexical and the actional meanings of preverbs (cf. Section 6.4.2; see also Ruvoletto 2016 for a comparable analysis on the Old Russian preverbs of the Povest’ vremennych let ). A preverb can attest both developments outlined above, as shown by means of poin what follows. The preverb po- ‘surface-contact (Path), ablativity (Source)’ lexically modifies the simplex verbal stem in pro-po-věděti : věděti ‘know’ ~ po-věděti ‘tell’ (< ‘cause to know’) ~ pro-po-věděti ‘proclaim’ (< ‘tell openly’), ‘predict’ (< ‘tell in advance’). By contrast, poshows an actional delimitative meaning in s ъ -po-žiti ‘live for a while with’. Neither lexical nor actional meanings are associated with a specific position with respect to the verbal base: both EP s and IP s can exhibit lexical and actional meanings. In contrast, in formally-oriented studies on multiple preverbs, the so-called “internal” and “external” preverbs are distinguished based on various syntactic and semantic criteria, including preverbs’ ability to develop actional or quantizing meanings. In particular, external preverbs are said to be associated with predictable actional meanings, whereas internal preverbs are connected <?page no="262"?> 262 6. Multiple preverbs in Old Church Slavic with unpredictable lexical meanings. In Old Church Slavic, however, multiple preverbs appear to constitute a quite different phenomenon (cf. Arkadiev 2015: 217; Section 6.6). For example, going back to pro-po-věděti ‘announce, foretell’ and s ъ -po-žiti ‘live for a while with’, one can observe that pooccurs internally within both composites. Nevertheless, this preverb brings lexical content to the former composite, but actional content to the latter. The comparison between pro-po-věděti ‘predict, proclaim’ and similar composites, i. e. is-po-věděti ‘confess, explain’ and za-po-věděti ‘order’, provides further pieces of evidence supporting the fact that the EPs can carry lexical content, as shown in (18): Musterdatei NFA_Basis_A.dot 254 The comparison between pro-po-věděti ‘predict, proclaim’ and similar composites, i.e. is-po-věděti ‘confess, explain’ and za-po-věděti ‘order’, provides further pieces of evidence supporting the fact that the EPs can carry lexical content, as shown in (18): a. iz-po-věděti ‘confess, explain’ (18) věděti → po-věděti → b. pro-po-věděti ‘proclaim, predict’ ‘know’ ‘tell’ c. za-po-věděti ‘order’ The contribution of each EP is undoubtedly lexical in (18)a-c. The EP iz- (18)a metaphorically means ‘out of’ in iz-po-věděti : ‘tell out of’ > ‘confess, declare’ (interestingly, this composite also occasionally takes the prepositionless genitive in Euch. 68a13; cf. Vaillant 1977: 71). The preverb pro- ‘in front of’ (18)b provides the temporal indication ‘before’, or the meaning ‘openly’, due to the following metaphorical shifts: BEFORE IS IN FRONT OF and OPENLY IS IN FRONT OF , respectively (cf. also Section 6.4.2). The semantics of za- (18)c is more difficult to discern but still detectable: in passages where za-po-věděti is used in the sense of ‘forbid’, the EP zabrings the idea of a metaphorical limit that cannot be overcome. In this respect, it is quite revealing that, in Mk 7.36, Codex Zographensis employs the composite za-prěščati ‘impose bans’, whereas Codex Marianus attests to za-po-vědati . Accordingly with the sense of ‘metaphorical obstacle’ just outlined, zais also preverbed to several more concrete simplex verbs. In these formations, its obstacle meaning is clearer: cf. za-tvoriti, zaklěpsti and za-ložiti ‘close’ (‘put an obstacle against (a door)’), as well as zakryti ‘cover, hide’ (‘put an obstacle against (one’s sight’)) (on the role of zain Old Russian, cf. Böttger 2004; Tomelleri 2012; Ruvoletto 2016: 108 ff.). The behavior of izalso shows that both EPs and IPs can carry lexical and actional content. As an EP, izshows a lexical non-compositional meaning in the above-mentioned iz-po-věděti : ‘tell out of’ > ‘confess, declare’. As an actional preverb, it can give a sense of completeness occurring both externally and internally: compare pri-iž-diti ‘spend in addition’ (< ‘in_addition-completely-live on/ by’; see Section 6.4.3) and is-pro-vrěšti ‘overturn, destroy’ (< ‘completely-forth-throw’; see Section 6.4.3). 132 132 Further pieces of evidence supporting a telic reading for izin is-pro-vrěšti are provided by the comparison between this composite and the following verbs: (a) iz-vrěšti ‘cast out of’, in which The contribution of each EP is undoubtedly lexical in (18)a-c. The EP iz- (18)a metaphorically means ‘out of ’ in iz-po-věděti : ‘tell out of’ → ‘confess, declare’ (interestingly, this composite also occasionally takes the prepositionless genitive in Euch. 68a13; cf. Vaillant 1977: 71). The preverb pro- ‘in front of ’ (18)b provides the temporal indication ‘before’, or the meaning ‘openly’, due to the following metaphorical shifts: before is in front of and openly is in front of , respectively (cf. also Section 6.4.2). The semantics of za- (18)c is more difficult to discern but still detectable: in passages where za-po-věděti is used in the sense of ‘forbid’, the EP zabrings the idea of a metaphorical limit that cannot be overcome. In this respect, it is quite revealing that, in Mk 7.36, Codex Zographensis employs the composite za-prěščati ‘impose bans’, whereas Codex Marianus attests to za-po-vědati . Accordingly with the sense of ‘metaphorical obstacle’ just outlined, zais also preverbed to several more concrete simplex verbs. In these formations, its obstacle meaning is clearer: cf. za-tvoriti, za-klěpsti and za-ložiti ‘close’ (‘put an obstacle against (a door)’), as well as za-kryti ‘cover, hide’ (‘put an obstacle against (one’s sight’)) (on the role of zain Old Russian, cf. Böttger 2004; Tomelleri 2012; Ruvoletto 2016: 108 ff.). The behavior of izalso shows that both EP s and IP s can carry lexical and actional content. As an EP, izshows a lexical non-compositional meaning in the above-mentioned iz-po-věděti : ‘tell out of’ > ‘confess, declare’. As an actional preverb, it can give a sense of completeness occurring both externally and internally: compare pri-iž-diti ‘spend in addition’ (< ‘in_addition-completely-live <?page no="263"?> 6.4. The semantics of multiple preverbs 263 on / by’; see Section 6.4.3) and is-pro-vrěšti ‘overturn, destroy’ (< ‘completely-forth-throw’; see Section 6.4.3). 19 Surprisingly, given the the later development of modern Slavic languages, including Bulgarian, in which several actional preverbs can stack onto the same verb (cf. example (2)), in my sample, no composites contain two actional preverbs stacked onto the same verbal stem. In the composite v ъ s-po-męnǫti, which means ‘start remembering’, both preverbs might be interpreted as ingressive, and thus be suspected of being actional. A closer investigation, however, reveals that they in fact have a different function. The ingressive IP polexically modifies the root √ min- ‘think’: po-√minmeans ‘start thinking > remember’ ( poelsewhere shows ingressive meanings: po-iti, besides ‘go along a surface’, can also mean ‘start going > depart from’). 20 The EP v ъ zinstead focuses one’s attention on the starting point of the event of remembering; hence, it is purely actional: po-męnǫti ‘remember’ ~ vъs-po-męnǫti ‘start remembering’, ‘remind’. 6.4.2. Same preverbs, different meanings As anticipated for poand izin Section 6.4.1, preverbs are polysemous morphemes which have undergone multiple semantic shifts. In this Section, I discuss a number of interesting cases in point, i. e. po- ‘surface-contact (Path), ablativity (Source)’, pri- ‘beside’, pro- ‘in front of ’, and v ъ z- ‘upward’ (cf. Table 39 for a summary). The preverb pooriginally indicates Path (‘surface-contact’) and ablativity (Source) and is etymologically related to Vedic ápa ‘away from’, Ancient Greek apó ‘away from’, and Old Irish ó, úa ‘from’ (REW 292-293; Watkins 2000: 5; LIPP II 66 ff.; fn. 1 and 8 in this Chapter). In combination with other preverbs, ponever retains its basic spatial usages, though it retains these usages elsewhere 19 Further pieces of evidence supporting a telic reading for izin is-pro-vrěšti are provided by the comparison between this composite and the following verbs: (a) iz-vrěšti ‘cast out of’, in which izclearly retains its basic spatial usage in most contexts; (b) o-pro-vrěšti ‘overturn’, which occurs in the same context as is-pro-vrěšti in Mar. Zogr. Jn 2.15 (cf. (11)). As such composites contain two different EPs, iz- ‘out of ’ and o(b ъ )- ‘around’, that have two different basic spatial meanings, and nevertheless occur in the same context, the EPs must be bleached into telic markers. In Bes. 34, 238aα10 and 238bβ4, another composite occurs, i. e. vъs-pro-vrěšti ‘overturn’, with a different EP v ъ s- (etymologically) ‘upward’ but the same meaning as is-pro-vrěšti and o-pro-vrěšti. 20 In po-męnǫti , the preverb pocontributes, with the suffix -nǫ- , which marks bound events, to assigning limits to the action of thinking (√ min-, unbound event). The fact that it also carries lexical content is backed up by the existence of the secondary imperfective po-minati being built on this verb (see Section 6.3.2). <?page no="264"?> 264 6. Multiple preverbs in Old Church Slavic in Old Church Slavic (cf. po-iti ‘go along a surface, depart from’). The preverb podevelops the following abstract meanings: i. ingressive ( v ъ s-po-męnǫti ‘start remembering, remind’); ii. delimitative ( s ъ -po-žiti ‘live for a while with’); iii. distributive ( prěd ъ -po-lagati ‘distribute to’); iv. causative ( pro-po-věděti ‘predict, proclaim’). On meaning (i), discussed in Section 6.4.1, it should be added that this meaning goes back to the ablative spatial meaning of po- , according to the following metaphor: events can be thought of as locations ; the starting point of a non-spatial event (ingressive) can be thought of as a starting point of a spatial event (Source). The delimitative meaning of po- , on which see Section 6.4.1, is instead related to the Path component of the spatial meaning of po- : focusing on a Path can carry the implication of focusing on its startingand end-points (i. e. its limits). Then, the spatial Path is reinterpreted as a metaphorical Path, i. e. as an event . The distributive sense of pois shown in example (19) (cf. also Sections 6.1.2.1, 6.1.3, and 6.2.1): (19) i daěše oučenikom ъ svoim ъ da and give. impf .3 sg disciple. dat . pl poss .3 sg . dat so_that prěd ь -po-lagajǫt ъ todistr -put. prs .3 pl ‘And (he) gave (them) to his disciples to distribute (them) to (the people).’ (Zogr. Mk 8.6 = (12)) In (19), pofocuses reader’s attention upon the movement of disciples, who go through the crowd to distribute, food to people, person by person. This abstract meaning is also grounded on the Path meaning of po- : covering a Path can imply covering all intermediate steps that make up the Path itself, as represented in Figure 4 (on the origin of distributive po- , see also Dickey 2012: 92; cf. the usages of the Ancient Greek katá , Luraghi 2003: 197 ff.). 21 21 It is unlikely that posimply corresponds to AG para- (cf. Table 36): first, a more direct correspondence holds between paraand prěd ь in this context. Second, there are only a few Old Church Slavic preverbs which straightforwardly mirror their Ancient Greek counterparts (cf. Section 6.2.4). Old Church Slavic employment of preverbs is indeed creative. Third, the preverb poextensively develops distributive meanings in South Slavic, especially in Croatian (Dickey 2012). As a preposition, po has distributive meanings in other varieties of Slavic as well. <?page no="265"?> 6.4. The semantics of multiple preverbs 265 Fig. 4: The link between Path and distributive meanings of po- Finally, the preverb poseems to function as a quasi-causative formation in po-věděti ‘tell’ and related composites (cf. (18)). Compare examples (20) and (21) below: (20) blǫdite ne věd ǫ šte k ъ nig ъ be_wrong. prs .2 pl neg know_ ptcp . prs . nom . pl scripture. gen . pl . f ni sil bžiję (= božiję) neither force( f ). gen of_ G . gen . f ‘You are in error because you do not know the Scriptures or the power of God.’ (Mar. Mt 22.29) (21) povědešę že emu ěko is ъ (= Isus ъ ) nazarěnin ъ tell. aor .3 pl ptc 3 sg . dat that J. nom of_ N . nom mimochodit ъ pass_by. prs .3 sg ‘(They) told him that Jesus of Nazareth was passing by.’ (Mar. Lc 18.37) In (20), the simple verb vědeti ‘know’ takes the direct object k ъ nig ъ ‘of the scriptures’, playing the semantic role of Theme, which is in the genitive case because of the occurrence of the negation ne (Lunt 1965: 146). Example (21), instead, contains the speech verb po-vědeti ‘tell’, which is a trivalent verb taking a direct object playing the semantic role of Theme (the completive clause introduced by ěko ‘that’), and an indirect object playing the role of Addressee (the dative emu ‘to him’). Thus, the preverb poseems to function as a causative derivation: po- + vědeti ‘know’ results in po-vědeti ‘make one know > tell’. The preverb poarguably adds a participant (the Addressee) to the described event. This function is semantically consistent with the distributive meaning of po- , which can be understood as introducing Recipients (the preposition po + loc can also express Recipient in a distributive sense; cf. Thomason 2006: 123). Then, the link between Addressees and Recipients is easy to draw: an Addressee, as a <?page no="266"?> 266 6. Multiple preverbs in Old Church Slavic Recipient, receives certain non-concrete entities, such as news, words, or pieces of information. 22 The preverb prialso displays interesting semantic shifts from its basic spatial meaning of ‘beside’. First, it has acquired the spatial + actional meaning that I named “beside + resultative” based on Ruvoletto (2016: 72 ff.). This meaning entails gaining something through an active effort (hence, the resultative component), as shown by the comparison between pri-ob-rěsti ‘acquire, earn’ (22)a ~ ob-rěsti ‘find’ (22)b: (22) a. šed ъ že priemy. d (= пѧть ) go. ptcp . pst . nom but receive. ptcp . pst . nom five. acc . f talan ъ t ъ děla o nich ъ . i priobrěte talent. gen . pl do. aor .3 sg about 3 pl . loc and acquire. aor .3 sg drougǫjǫ d (= пѧть ) talant ъ . other. acc . f five. acc . f talent. gen . pl ‘He who had received the five talents went at once and traded with them, and he made five talents more.’ (Mar. Mt 25.16) b . i abie v ъ xodęšta v ь nǫjǫ and suddenly enter. ptcp . prs . nom . du in 3 sg . acc 22 The causative-like ability of postill remains unexplored, and deserves further investigation, given that there are other pairs of verbs in which poappears to have a similar function (e. g. mošti ‘be able’ ~ po-mošti ‘help’ (< ‘cause to be able’)). It can be observed that there are other languages in which the addition of a Recipient-like participant results in a causative-like formation, as shown in examples (a) and (b): (a) non-standard Engl. learn someone something = teach someone something (Morris 1981: 744) cf. also the one who first noticed it to me (cognitition verb → communication verb; Croft 2017) (b) archaic (i) / current non-standard (b) Italian i. E dolce un canto le imparava and sweet indf song 3 sg . dat learn. impf .3 sg ‘And, she (Aphrodite), sweet, was teaching her (Sappho) a poem.’ (G. Carducci, Juvenilia , 19th cent.) ii. Chi ti ha imparato a rispondere così? who 2 sg . dat aux learn. ptcp . pst . pass to answer. inf . prs that_way ‘Who taught you to answer that way? ’ (Google search) <?page no="267"?> 6.4. The semantics of multiple preverbs 267 obręšteta žrěbec ъ privęzan ъ . find. prs .2 du foal. acc tie_down. ptcp . pst . acc . pass ‘(Go to the village ahead of you,) and just as you enter it, you will find a colt tied (there).’ (Mar. Mk 11.2) The event described in (22)a implies active participation of the Agent, who invests money in order to gain some more. By contrast, in (22)b, the Agent finds the colt ( žrěbec ъ: acc ) by simply entering the village, with no additional effort. The meaning just outlined typically occurs with verbs entailing an idea of approaching, such as the mentioned pri-ob-rěsti ‘acquire, earn’, and v ъ s-pri-im а ti/ v ъ s-pri-jęti ‘receive in return’, as well as many others (cf. e. g. pri-zvati ‘invite, call’, pri-vleči ‘attract, conquer’). As remarked by Ruvoletto (2016: 72 ff.) for Old Russian, pri- ‘beside’ instead develops the meaning of ‘in addition’ with verbs lacking such an idea of approaching. This happens with pri-iž-diti ‘spend in addition’, as shown in (23): (23) prileži em ь . i eže ašte take_care_of. imp . prs .2 sg loc .3 sg and rel . acc . n ever pri-iždiveši az ъ egda v ъ z-vraštǫ in_addition-spend. prs .2 sg 1 sg . nom when back-turn. prs .1 sg sę v ъ zdam ь ti refl .3 sg . acc back-give. prs .1 sg 2 sg . dat ‘Take care of him. And whatever you spend in addition, I will give you back when I come back.’ (Mar. Lc 10.35) In this context, the Greek source text shows pros-dapanáō ‘spend besides, spend in addition’, in which the preverb prosmeans ‘in addition’, as it frequently does in other composites as well (e. g. pros-ktáomai ‘gain and add’, pros-dídōmi ‘give in addition’, which interestingly has an exact Slavic counterpart in pri-dati ‘give in addition’). Another good example for the correspondence OCS pri- ~ Gr. pros- ‘in addition’ is provided by the couple pri-ložiti/ pri-lagati ~ pros-títhēmi ‘put beside, add’. In (23), the EP priof pri-iž-diti ‘spend in addition’ might also draw an anaphoric reference to the locative em ь ‘him’, taken by the preceding composite pri-leži : imp , which also contains pri- (cf. Section 6.1.3.2 on the usages of prepositionless locatives). In this respect, primay also act as an instrument of textual cohesion. <?page no="268"?> 268 6. Multiple preverbs in Old Church Slavic In Old Church Slavic, multiple preverb composites, the preverb proshows only lexical meanings. In the composite is-pro-vrěšti , when used in the meaning of ‘overturn’, prostill retains a spatial meaning: (24) i d ъ sky tr ъ ž ъ nik ъ i sědališta and table( f ). acc . pl merchant. gen . pl and bench( f ). acc . pl prodajǫštix ъ golǫbi isprovrъže sell. ptcp . prs . gen . pl dove. acc . pl overturn. aor .3 sg ‘(He) overturned the tables of the merchants and the benches of those selling doves.’ (Mar. Mk 11.15 = (9)) The lexical contribution brought about by prois discernible, though not completely clear: this preverb etymologically means ‘forward’ (cf. LIPP II 636), but in Old Church Slavic it acquires the spatial meaning of ‘through’ (Path, perlative). In (24), proarguably emphasizes the movement of the tables, overturned by Jesus. Thus, to some extent, the preverb procontributes to describing the effect of the motion on the tables, even though the direction of this motion is not completely clarified by the addition of the preverb. 23 Interestingly, the preverb progains different meanings in different contexts that contain the composite pro-po-vědati, as shown in (25)a-b: (25) a. pro- ‘before’ (Mar. Mk 1.7) i pro-povědaaše glę grędet ъ and fore-tell. impf .3 sg say. ptcp . prs . nom come. prs .3 sg krěplei mene v ъ slěd ъ mene … stronger. nom 1 sg . gen in trace. acc 1 sg . gen ‘And ( John) predicted, saying: “After me (one who is) greater than I comes, …’ b. pro- ‘openly’ (Mar. Mk 1.45) on ъ že iš-ed ъ načęt ъ dem . nom but away-go. ptcp . pst . nom begin. aor .3 sg 23 Old Russian also attests to is-pro-vrěšti , as well as a very similar composite, i. e. iz( ъ )pro-metati ‘throw out, devastate’, containing the same preverbs ( iz( ъ )-, pro- ) and the root √met- , which is semantically close to √vrěg- ‘throw’ (Zanchi & Naccarato 2016). <?page no="269"?> 6.4. The semantics of multiple preverbs 269 pro-povědati m ъ nogo. i pronositi slovo. openly-tell. inf . prs much and make_known. inf . prs word( n ). acc ‘(“See that you say nothing to anyone, but go, show yourself to the priest and offer for your cleansing what Moses commanded, for a proof to them.”) But he went out and began to talk openly about it, and to spread the word.’ In (25)a, John the Baptist is foretelling ( pro- ‘fore-’) Jesus’ future coming. In (25)b, a leper, after being healed by Jesus’ hand, is strongly advised not to tell anyone about this miracle. But ( že ) the leper disobeys Jesus’ advice and proclaims (pro-po-vědati ) and divulges (pro-nositi ) the miracle that Jesus has performed. 24 The multiple preverb v ъ z- ‘upward’ shows both lexical and actional semantic developments. Its meaning of ‘in return, in exchange’ has been addressed in example (14). This semantic shift can be connected with the usage of the corresponding preposition v ъ z+ acc ‘in exchange for’ (Vaillant 1977: 114 ff.; Thomason 2006: 144 ff.). The preverb v ъ z- ‘upward’ can further develop an ingressive meaning, as is clearly shown by comparing n е -n а -viděti ‘hate’ and vъz-n е -n а -viděti ‘come to hate’ in example (26): (26) ašte mir ъ vas ъ nenavidit ъ . vědite ěko if world. nom 2 pl . gen hate. prs .3 sg know. imp . prs .2 pl that mene prěžde vas ъ vъz-nenavidě 1 sg . gen prior_to 2 pl . gen start-hate. aor .3 sg ‘If the world hates you, know that it came to hate me before you.’ (Mar. Jn 15.18) The same passage shows the two composites occurring side by side: n е -n а -viděti ‘hate’ lacks the ingressive component brought about by v ъ z- The ingressive v ъ zis particularly frequent with mental verbs (e. g. vъs-po-męnǫti ‘start remembering, remind’), and verbs of emotion (e. g. vъz-ljubiti ‘start loving’, vъs-chotěti ‘start wishing’, vъs-tužiti ‘start suffering’), but is also attested for other types of verbs (e. g. vъz-glagolati ‘start speaking’). Though v ъ zhas problematic etymological origins (in LIPP II 823 ff., REW 333), its basic meaning is ‘upward’. 24 In both examples, the corresponding Greek text contains the plain verb kērússō . However, this does not prevent the preverb profrom expressing two distinct meaning in two different passages. Preverbs and verbs are both polysemous linguistic elements: kērússō itself can mean ‘be a herald, proclaim, invoke, announce, declare’ (LSJ). <?page no="270"?> 270 6. Multiple preverbs in Old Church Slavic The link between its basic spatial usage and its actional ingressive meaning can be easily drawn through the interplay of the following metaphors (cf. Figure 5). (a) more is up , less is down (Lakoff & Johnson 1980: 15-16), based on the fact that by adding more of a certain substance to a recipient or of physical objects to a pile, the level of the substance or the height of the pile goes up. (b) events can the thought of as piles . (c) g oing upward ( v ъ z- ) along a pile means going from the starting point toward the culmination of an event . Fig. 5: The link between the spatial and ingressive meanings of v ъ z- Interestingly, in Old Church Slavic, not only can different preverbs be associated with different meanings, but the converse situation is also attested: for example, as shown by (27) and (28), the delimitative meaning of Duration can be brought about by both poand prě-. Arguably, the same metaphor outlined above for the delimitative pocomes into play with prě- : the duration of an event is the exten sion of a path . The preverb prě- , like po- , also has an original spatial meaning indicating Path ‘across’. (27) I tretii і d ь n ь v ъ stav ъ iz mr ъ tvyich ъ . And third. acc day. acc arise. ptcp . pst . nom out_of dead. gen . pl jav svoim ъ oučenikom ъ . rekʼše appear. aor .3 sg his. dat . pl disciple. dat . pl say. ptcp . pst . nom styim ъ (= svjętim ъ ) apslom ъ (= apostolom ъ ) m ъ nogom ъ iže holy. dat . pl apostle. dat . pl many. dat . pl rel . nom . pl po istině věrovav ъ ši і m ъ v ь ń ь . after truth. dat believe. ptcp . pst . dat . pl in 3 sg . acc s ь -po-živ ъ s ь ńimi with-for_while_live. ptcp . pst . nom with 3 sg . ins <?page no="271"?> 6.4. The semantics of multiple preverbs 271 ‘And, after rising on the third day from among the dead, (he) appeared to his own disciples (and) spoke with the holy apostles and with many (others) believing in him in truth, and (he) lived for a while with them.’ (Supr.1.52) (28) s ь -prě-byvaat ъ s ъ člky (= člověky) aky člověk ъ with-for_a_while-be. prs .3 sg with man. ins . pl as man. nom vęšte triі desętъ lětъ more three. gen . pl ten. gen . pl year( n ). gen . pl ‘And (he) lived among men as a man for more than thirty years.’ (Supr.1.331) The Duration of the event of ‘remaining with’ is further specified by means of the locution vęšte tri і desęt ъ lět ъ ‘for more than thirty years’ in (28). In addition, the durative reading of prěis supported by the fact that byvati contains the suffix -va-, which marks durative and iterative events (cf. Section 6.3.1). In (27), instead, the Duration is omitted. Given that the Greek equivalent for s ь -po-žiti is sun-anastréphomai (cf. Table 36), one might wonder whether pomeans ‘back’ ( iz mr ъ tvyich ъ: gen ‘from among the dead’), as anadoes in the Greek composite. However, this interpretation is unlikely, given the occurrence of passages such as (29): (29) iže malo po-živ ъ oumirajet ъ rel . nom a_few for_a_while-live. ptcp . pst . nom die. prs .3 sg ‘… (one) who dies after having lived for such a little while.’ (Supr. 258.12) In (29), the meaning of pois clearly delimitative: the composite po-žiti ‘live for a while’ is further specified by the adverb malo ‘for a little while’. 6.4.3. Different degrees of compositionality Table 38 displays the semantic analysis of Old Church Slavic composites. To determine whether the composites are fully (+), partially (-/ +) or non-compositional (-), I take into account the meaning of the simple verbal root, as well as the meaning of each preverb attaching to it. Occasionally, this semantic analysis can be very hard to perform: the same composite can display different degrees of compositionality in different contexts. For example, the composite is-pro-vrěšti means ‘overturn’ in (24) above but ‘destroy’ in (30): <?page no="272"?> 272 6. Multiple preverbs in Old Church Slavic (30) ô glasa silo ad ъ isprovrъgъši ptc sound( f ). gen power. voc hell. acc destroy. ptcp . pst . voc . f ‘o power of the word, (you) destroying death, …’ (Supr. 27.200 = (10)) c omposite m eaning c ompositionality is-po-vědati confess, explain is-po-věděti confess, explain is-po-vědovati confess, explain is-pro-vrěšti overturn, destroy -/ + or - iz-ob-rěsti find out -/ + iz-ob-rětati find out -/ + o-pro-vrěšti overturn -/ + prěd ъ -po-lagati distribute to + pri-iž-diti spend in addition -/ + pri-ob-rěsti acquire pro-po-vědati predict, proclaim pro-po-věděti predict, proclaim pro-po-vědovati predict, proclaim s ъ -po-žiti live for a while with + s ъ -prě-byvati remain together with + s ъ -v ъ -kupiti gather together -/ + s ъ -v ъ -kupljati unify, copulate v ъ s-po-męnǫti start remembering, remind -/ + v ъ s-pri-im а ti receive in return -/ + v ъ s-pri-jęti receive in return -/ + v ъ z-n е -n а -viděti come to hate -/ + za-po-vědati order za-po-věděti order - Tab. 38: The compositionality of Old Church Slavic composites <?page no="273"?> 6.4. The semantics of multiple preverbs 273 In (30), a metaphorical Agent, the ‘power of the word’, causes the destruction, or ‘death’, of a metaphorical Patient: such a Patient, unlike the tables in example (24), cannot be overturned. The Greek equivalents for iz( ъ )-pro-vrěšti are either ana-stréphō or kata-stréphō ‘turn upside down, overturn’, for passages in which the Slavic composite is employed in its partially compositional usage, as in (24). 25 The Greek source-text instead contains dia-rrḗgnumi ‘break through/ asunder’, when iz( ъ )-pro-vrěšti is non-compositional. 26 Another case in point is the composite za-po-věděti, which shows two slightly different meanings, that is, ‘forbid’ in (31)a and ‘order’ in (31)b: (31) a. ‘impose bans, forbid’ zapovědě im ъ is ъ glę. nikomouže forbid. aor .3 sg 3 pl . dat J. nom say. ptcp . prs . nom nobody. dat ne povědite viděniě. neg tell. imp .2 pl vision. gen . n ‘Jesus forbade them: “Tell no one the vision”.’ (Mar. Mt 17.9) b. ‘order’ n ъ da razouměat ъ mir ъ ěko ljublju but may undestand. prs .3 sg world. nom that love. prs .1 sg otca. i ěkože zapovědě m ь ně otcъ tako father. gen and as order. aor .3 sg 1 sg . dat father. nom so tvorjǫ 25 The Greek composites ana-stréphō and kata-stréphō both include a preverb expressing telicity, i. e. ana- (etymologically) ‘upward’ and kata- (etymologically) ‘downward’. Importantly, the etymological spatial usages of these preverbs have opposite meanings. As they are nevertheless translated by means of the same Slavic composite, they must be semantically bleached in these contexts, and have mere actional properties. Interestingly, such preverbs already show actional - but opposite - meanings in Homeric Greek: ana can indicate the beginning of an event, whereas kataprofiles its completion (Chantraine 1953: 90, 112). 26 In Old Russian, besides displaying both the partially compositional and the non-compositional usages outlined above, this composite is also frequently used in a fixed expression with the meaning of ‘killing oneself ’, as shown in (i). (i) i tu isprovrъže životъ svoi zъlě. and there transform. aor .3 sg life. acc his. acc miserably ‘And there (he) killed himself miserably.’ (Usp. Sbor. The Tale of Boris and Gleb 659) Within the idiom shown in (i), the composite iz( ъ )-pro-vrěšti seems to retain a less lexicalized meaning with respect to example (29), i. e. ‘transform, take out of ’. It is the whole expression that is lexicalized (Zanchi & Naccarato 2016: 374; Dmitrij Sičinava, p.c.). <?page no="274"?> 274 6. Multiple preverbs in Old Church Slavic make. prs .1 sg ‘But the world may learn that I love the Father and do exactly what my Father has commanded me.’ (Mar. Jn 14.31) In (31)a, Jesus provides his disciples with instructions on what should not be done. In (31)b, instead, Jesus says to follow his father’s instructions (literally ‘what his Father ordered’). As mentioned in Section 6.4.1, the semantics of zais less discernible in (31)b than it is in (31)a, where the preverb provides the idea of adding a metaphorical obstacle, that is, of posing a limit to one’s behavior. The meaning of zain (31)b can still be understood, however, if one also thinks of giving orders as the act of dictating certain behavioral limitations that cannot be surpassed. As shown in Table 38, only a few composites with multiple preverbs retain their fully compositional meanings: these are prěd ъ -po-lagati ‘distribute to’ (> Blg. prědpolagam ‘suppose, assume’; cf. Table 33), s ъ -po-žiti ‘live for a while with, and s ъ -prě-byvati ‘remain together with’. The vast majority of composites are either only partially compositional or entirely non-compositional. In partially compositional composites, it is often the case that the EP retains more distinct semantics with respect to the remaining composite, as illustrated through examples in (32)a-f: (32) a. ‘spend in addition’: pri-[iž-diti] ‘in addition-[spend]’ → iž-diti ‘out_of-live’ b. ‘gather together’: sъ-[v ъ -kupiti] ‘with-[gather]’ → v ъ -kupiti ‘in-buy’ c. ‘find out’: iz-[ob-rěsti] ‘out-[find]’ → ob-rěsti ‘around-meet’ d. ‘receive in return’: vъs-[pri-im а ti] ‘in return-[receive]’ → pri-im а ti ‘result-take’ e. ‘start remembering’: vъs-[po-męnǫti] ‘start-[remember]’ → po-męnǫti ‘start-think’ f. ‘come to hate’: vъz-[n е -n а -viděti] ‘start-[hate]’ → n е -n а -viděti ‘ neg against-see’ The semantic contribution of the IP varies greatly in these examples. In (32) a-b, the semantic input is unclear, in (32)c-d, it is redundant, and in (32)e-f or lexicalized (and redundant). In (32)a, for example, the telic meaning provided by the IP izand the semantics of the whole composite can only be understood by interpreting the simplex verb žiti as ‘live on / by’, rather than as simply ‘live’ (cf. žiti ‘live on / by’ in Bes. 36.273bβ; 274bα; cf. OI r. ar-√ber- ‘live, eat, use, employ’ for a verb showing a similar polysemy). The fact that izassigns the act of living by / on a culmination also emerges from example (33): <?page no="275"?> 6.4. The semantics of multiple preverbs 275 (33) iž-div ъ šju že emou v ь sě … completely-live_on. ptcp . pst . dat but 3 sg . dat indf . acc . pl ‘After he had spent everything, (there was a severe famine in that whole country and he began to be in need).’ (Mar. Lc 15.14) In (33), the composite iž-diti ‘completely-live on / by’ > ‘spend’ takes the direct object v ь sě : acc , which backs up the interpretation assigned to izin this context. The fact that the subject participant uses up his resources is also supported by how the passage continues: ‘… there was a severe famine in that whole country and he began to be in need’. In (32)c-d, the spatial meanings of the IP s overlap with the meaning of the simple verbs: ‘meet’ (√ rět- ) can be said to imply the spatial component of ‘around’ ( ob ъ - ), as well as ‘take’ (√ь m- ) the spatial component of vicinity or approximation ( pri- ). Due to this overlap - i. e. the so-called Vey-Schooneveld effect (cf. Section 6.1.2.2.2) - the preverbs are reanalyzed as telic and resultative markers. As for v ъ z-ne-na-viděti ‘come to hate’ (32)f, the semantic contribution brought about by the IP na- ‘against’ is both lexicalized (‘look against > hate’) and redundant, as the negatively oriented meaning of ‘against’ might also be provided by the negative prefix ne-. The preverb nacan also mean ‘against’ as a preposition: na + acc can mean ‘against’ in Stimulus expressions, such as na mę li gněvajete sę? ‘Are you angry with me? ’ (Mar. Jn 7.23) (Thomason 2006: 132) (Frigione 2015: 33). 27 In other partially compositional composites, the meaning of completeness brought about by the EP also results in a lexicalized formation: 27 Interestingly, Old Church Slavonic attests to another composite with the same root that is used to express negative feelings, i. e. za-viděti ‘be envious’. Consider also the Latin compound in-videō, containing the prefix in- ‘against’ and the same root for seeing, which is metaphorically employed for ‘having negative feelings’, ranging from being envious to being hostile. In Latin, there is another composite containing a different root for seeing (PIE *spek̑ - ‘see, look at’ > Gr. sképtomai ‘look about carefully’, Lat. speciō ‘look, look at’, Ved. páśyati ‘(he) sees, looks at’; cf. LIV 2 575-576) and developing a meaning connected with negative feelings, i. e. dē-spiciō ‘look down upon > despise, disdain, disregard’ (Prof. Pierluigi Cuzzolin, p.c.). Hittite also shows a similar compound: the root au(s)-, u(wa)- ‘see, look, watch, behold, observe, inspect, read’ (PIE *h 1 eu̯ - ‘see, catch sight of ’; LIV 2 243), when modified by the adverb / preverb parā ‘(as a preverb) forth, ahead, along; away, off, out, over’, results in parā au(s)- ‘overlook, disregard, pay no attention to’ (Puhvel 1984: 234 ff., 2011: 106; Prof. Silvia Luraghi, p.c.). <?page no="276"?> 276 6. Multiple preverbs in Old Church Slavic (34) isprovrěšti o(b ъ )provrěšti ‘completely’ ‘forth’ ‘throw’ → ‘overturn’, ‘destroy’ The composites in (34) are still analyzable as partially compositional, as they retain the spatial component of movement, which pertains both to the IP pro- ‘forth’ and to the root √vrěg- ‘throw’. The semantic developments of non-compositional composites have been discussed in Sections 6.4.1-6.4.2. 6.4.4. Summarizing the meanings of preverbs in multiple preverb combinations Table 39 summarizes the different meanings of Old Church Slavic multiple preverbs. The most relevant semantic shifts have been discussed through Sections 6.4.1-6.4.3. Each meaning is exemplified by a composite. p reverb m eaning e xample izmetaphorical ‘out of ’ (Source) is-po-věděti ‘confess, explain’ telic is-pro-vrěšti ‘overturn, destroy’ pri-iž-diti ‘spend in addition’ n а upon - against (Goal) v ъ z-n е -n а -viděti ‘come to hate’ o(b ъ )spatial ‘around’ (Goal) o-pro-vrěšti ‘overturn’ telic iz-ob-rěsti ‘find out’ podelimitative (Duration) s ъ -po-žiti ‘live for a while with’ ingressive v ъ s-po-męnǫti ‘start remembering, remind’ distributive prěd ъ -po-lagati ‘distribute to’ causative (< distributive) is-po-věděti ‘confess, explain’ prěextension in time (Duration) s ъ -prě-byvati ‘remain together with’ prěd ъ beside, in front of (Recipient) prěd ъ -po-lagati ‘distribute to’ priin addition, in excess pri-iž-diti ‘spend in addition’ resultative (< ‘beside’ position) pri-ob-rěsti ‘acquire’ resultative (smth. generated through an action) v ъ s-pri-jęti ‘receive in return’ <?page no="277"?> 6.5. The syntax of multiple preverb composites 277 p reverb m eaning e xample proin front of (Path) is-pro-vrěšti ‘overturn, destroy’ before, in advance (Time) pro-po-vědati ‘predict’ openly (< in front of) pro-po-věděti ‘proclaim’ s ъ with (Comitative) s ъ -po-žiti ‘live for a while with’ togetherness s ъ -v ъ -kupiti ‘gather together’ v ъ in s ъ -v ъ -kupiti ‘gather together’ v ъ z( ъ )back, in return v ъ z( ъ )-pri-jęti ‘receive in return’ ingressive v ъ z-n е -n а -viděti ‘come to hate’ causative v ъ s-po-męnǫti ‘make one remember’ zametaphorical obstacle za-po-věděti ‘forbid, order’ Tab. 39: The meanings of Old Church Slavic multiple preverbs 6.5. The syntax of multiple preverb composites As discussed in Sections 6.1.2 and 6.4.3, in Old Church Slavic, preverbs are much more advanced either in their grammaticalization process into markers of actionality or in their lexicalization process within non compositional composites than they are either in Vedic (Chapter 4) or in Homeric Greek (Chapter 5). Accordingly, their meanings and behavior differ widely from those of the corresponding prepositions (cf. Dickey’s 2012 “orphan prefixes”). In parallel, prepositions are also quite grammaticalized and prepositionless cases have already lost their ability to express semantic roles to a great extent (cf. Section 6.1.3). These issues are addressed here with respect to multiple preverbs. In next Section (6.5.1), I discuss a few relics of a previous more fluid situation, in which preverbs and prepositions used to behave similarly. Section 6.5.2 shows passages in which multiple preverbs are repeated outside the preverbal context as prepositions. Finally, in Section 6.5.3, I present data suggesting that the semantic modifications brought about by preverbs occasionally have the side effect of altering verb argument structures. <?page no="278"?> 278 6. Multiple preverbs in Old Church Slavic 6.5.1. The alternative constructions to multiple preverbs: scanty relics of a preceding stage In a few passages, composites showing no lexicalization or a low degree of lexicalization can be replaced by an equivalent construction in which the compound contains only the IP , whereas the EP occurs outside the composite as a preposition. Compare, for instance, examples (35) and (36): (35) i daěše oučenikom ъ svoim ъ da and give. impf .3 sg disciple. dat . pl poss .3 sg . dat so_that prědь-polagaj ǫ tъ to-distribute. prs .3 pl ‘And (he) gave (them) to his disciples to distribute (them) to (the people).’ (Zogr. Mk 8.6 = (12)) (36) i daěše oučenikom ъ svoim ъ da and give. impf .3 sg disciple. dat . pl his. dat . pl so_that polagaj ǫ tъ prědъ nimi distribute. prs .3 pl in_front_of 3 pl . ins ‘Then (he) gave them to his disciples to distribute to them.’ (Mar. Mk 6.41) The situation described in (35) and (36) is almost the same: Jesus is giving his disciples some food to be distributed to people nearby. In (35) from Zographensis, prěd ъ - ‘in front of ’ functions as a preverb, and the Recipient is not explicitly mentioned, though recoverable from the previous context. By contrast, in (36) from Marianus , prěd ъ functions as a preposition and explicitly expresses the Recipient together with the instrumental case ( nimi ). Another case in point is the composite v ъ s-pri-im а ti/ v ъ s-pri-jęti ‘receive in return’, as shown by example (37) in contrast with (38): <?page no="279"?> 6.5. The syntax of multiple preverb composites 279 (37) ibo i grěš ъ nici grěš ъ nikom ъ v ъ zaim ъ indeed and sinner. nom . pl sinner. dat . pl in loan. acc . pl dajǫt ъ . da vьs-priimǫt ъ rav ъ no give. prs .3 pl so_that in_return-receive. prs .3 pl equal. acc . n ‘Even sinners lend to sinners, to get back the same amount.’ (Mar. Lc 6.34 = (14)) (38) i ot ъ ispl ъ neniě ego my v ь si and from fullness. gen 3 sg . gen 1 pl . nom indf . nom prijęchom ъ blagodět ь vъz blagodětь receive. aor .1 pl grace. acc in_exchange_of grace. acc ‘From his abundance we have all received one gracious blessing in exchange of another.’ (Mar. Jn 1.6) Here, as was the case in (35) and (36), a construction with multiple preverbs alternates with a construction containing a composite with only one preverb and a prepositional phrase, i. e. v ъ z + acc (differently from (35) and (36), though, the meanings of the two passages are not equivalent). The multiple preverb composite of (37) occurs 30 times (cf. Table 32), whereas the construction in (38) is also attested in Codex Suprasliensis 411.18 and 446.3. As for (35) and (36), the participant recovered by the EP is usually omitted in case of multiple preverbs, but expressed in the construction with the prepositional phrase. To be sure, there is one exception to this tendency. In (39) below, containing the multiple preverb composite v ъ s-pri-im а ti ‘receive in return’ , the Substitute is explicit, and expressed by the prepositionless dative dělom ъ: dat ‘deeds’. Crucially, however, it is not expressed by means of the prepositionless accusative, as one might expect given the presence of v ъ z-. (39) dostoinaa bo dělom ъ naju worthy. acc . pl . n because deed. dat . pl . n 1 du . gen v ъ s-priemlevě. in_return-receive. prs .1 du ‘For we are receiving the due reward of our deeds.’ (Mar. Lc 23.41) <?page no="280"?> 280 6. Multiple preverbs in Old Church Slavic In (39), the prepositionless dative dělom ъ functions as a Substitute expression (cf. Vaillant 1977: 83 ff.) and translates a heavy Greek construction that contains a relative clause ( áxia gàr hôn epráxamen apolambánomen , GNT Lc 23.41). What do examples (35)-(36) and (37)-(38) have in common? Out of the two variants above, the one with the EP usually omits a participant, i. e. the Recipient and the Substitute, respectively (but cf. (39)). These participants are, however, recoverable from the previous context. As Viti (2008a, 2008b) argued for Ancient Greek preverbs, one reason that preverbs have developed into markers of telicity is their ability to introduce topical referents. These are typically known participants and as such are likely to be conceptualized as entire in space and complete in time. Here lies the link between topicality and telicity (cf. Section 2.3.2.3). 6.5.2. Preverb repetition As briefly discussed in Section 6.1.3, preverbed verbs can occasionally take the case required by the corresponding prepositions. This never happens with composites containing multiple preverbs. Instead, in a number of examples, the EP shows up both inside and outside the composite, i. e. it is repeated as a preposition, though not being semantically bleached. For example, this phenomenon often occurs in the presence of s ь - ‘with’ in Old Church Slavic. The composites s ъ -po-žiti ‘live for a while with’ (Supr.1.52), s ъ -prě-byvati ‘remain with’, and s ъ -v ъ -kupljati ‘gather, unite’ (Supr.1.208) take prepositional phrases constituted by s ь ‘with’ and the instrumental case, expressing the Comitative. An example with s ъ -v ъ -kupljati ‘gather, unite’ is shown in (40): (40) i sъv ъ koupľěę s ⱕ sь͗ ńimi … and gather. ptcp . prs . nom refl . acc with 3 pl . ins . f ‘And after gathering together with them …’ (Supr.1.208) The repetition in (40) is probably due to the fact that, in Old Church Slavic, prepositionless cases retain their concrete usages only to a limited extent and thus can only express semantic roles under specific conditions. In particular, the prepositionless instrumental is rarely used to express a Comitative-like participant, e. g. in expressions such as ženęi ( ptcp . prs . nom ) sę ( refl . acc ) puštenojǫ ( ins ) ( Lc 16.18) ‘he who married a divorced woman’ (Lunt 1965: 150-151; Hewson & Bubenik 2006: 179). Preverb repetition of the type in (40) however is not obligatory in Old Church Slavic. With composites containing only a single preverb, for instance, two con- <?page no="281"?> 6.5. The syntax of multiple preverb composites 281 structions are occasionally allowed: the one, containing a preverb ( ot ъ - ‘from’) repeated outside the composite as a preposition ( ot ъ sebe ‘from yourself ’) (41) a; the other, showing the corresponding simplex verb ( vr ъ zi ‘throw’) and the expected prepositional phrase ( ot ъ tebe ‘from you’) (41)b. 28 (41) a. Preverb repetition (Zogr. Mt 5.29) i otъ-vr ъ zi otъ sebe and away-throw. imp .2 sg from refl .2 sg . gen b. Simplex verb + prepositional phrase (Mar. Mt 5.29) i vr ъ zi ot ъ tebe and throw. imp .2 sg from 2 sg . gen ‘(And if your right eye causes you to stumble, pluck it out) and cast it from you.’ Alternations of this type are not uncommon: other relevant examples with the same verbal base vrěšti ‘throw’ are e. g. Mar. ~ Zogr. Lc 17.2; Mar. ~ Zogr. Mk 11.23. These pieces of evidence, again, support the contention that preverbs used to have a freer positioning than that we observe in the earliest Old Slavic texts documented by the textual traditions. 6.5.3. Preverbs as modifiers of verb argument structure Old Church Slavic preverbs occasionally modify the meaning of simplex verbs so as to result in composites with a different argument structure or taking a different case. These argument structure alterations, however, are better regarded as byproduct of the lexical modifications brought about by preverbs (Vaillant 1977: 35). As discussed in Section 6.4.2, this function is particularly remarkable for the preverb po-. This preverb, in its causative-like sense (cf. Section 6.4.2), is able to build a trivalent communication verb ( po-věděti ‘tell’) out of a bivalent verb of mental state ( věděti ‘know’). Elsewhere, the semantic modifications brought about by preverbs do not alter the cases taken by the simplex verbs. Compare the two Old Church Slavic verbs viděti ‘see’ and v ъ z( ъ )-n е -n а -viděti ‘begin to hate’. Both these verbs can take the accusative or the genitive case as a second argument (cf. Section 6.1.3.2); exam- 28 In the Greek version of the Gospels, the preposition apó only occurs in the text. <?page no="282"?> 282 6. Multiple preverbs in Old Church Slavic ples (42) and (43) show a second argument in the genitive case - česo ‘what? ’ and vraga svoego ‘his enemy’. 29 (42) česo vidětъ izidete v ъ poustynjǫ int . gen . n see. sup . acc go_out. aor .2 pl in desert( f ). acc ‘What did (you) go out into the desert to see? ’ (Mar. Mt 11.7) (43) v ъ zljubiši iskr ь něgo svoego i vьznenavidiši love. prs .2 sg neighbor. gen poss . refl .3 sg . gen and hate. prs .2 sg vraga svoego enemy. gen poss . refl .3 sg . gen ‘You will love your neighbor and hate your enemy.’ (Mar. Mt 5.43) As expected (Vaillant 1977: 60; Lunt 1965: 145), in (42) the perception verb viděti ‘see’ takes the genitive case expressing Stimulus (occasionally, this verb also takes the accusative case). The addition of preverbs in this context does not change the cases required by the verb (i. e. genitive or accusative), as the resulting compound v ъ z( ъ )-n е -n а viděti ‘begin to hate’ (43) also requires a Stimulus-participant. 6.6. Preverb ordering Table 40 summarizes the positioning of Old Church Slavic preverbs. As is also the case for Vedic (Section 4.6), Homeric Greek (Section 5.6), and Old Irish (Section 7.6), not all possible preverb orders are attested. This Section offers some observations about why certain sequences of preverbs occur in the data and why some others are lacking. 29 In the present Old Church Slavonic corpus, v ъ z( ъ )-n е -n а -viděti ‘come to hate’ only takes masculine animate direct objects in the genitive case. However, in Old Church Slavic, masculine animate nouns show genitive-accusative syncretism. Therefore, one may wonder whether vraga svoego and other similar forms should be better interpreted as accusatives, rather than as genitives. The interpretation as accusatives however is unlikely, as n е n а viděti ‘hate’ can take neuter genitive direct objects (e. g. Mar. Mk 13.13, Mt 10.22, etc.). One may also wonder whether the genitive object taken by v ъ z( ъ )-n е -n а -viděti and n е n а viděti cannot be due to the presence of the negation ne as a prefix (cf. Section 6.1.3.2). However, this scenario is also unlikely, as n е n а viděti ‘hate’ is a lexicalized compound, in which the semantic addition given by the negation is not detectable anymore (accordingly, *n а viděti is not attested). Furthermore, other verbs of emotion (and more generally, verbs requiring a Stimulus-participant) take the genitive case (see Vaillant 1977: 56-65). <?page no="283"?> 6.6. Preverb ordering 283 p reverbs e xterior i nterior iz( ъ )- 6 (85, 7) 1(14, 3) na- - 1 (100 %) o(b ъ )- 1 (25 %) 3 (75 %) po- - 12 (100 %) prě- - 1 (100 %) prěd ъ - 1 (100 %) pri- 2 (66, 7 %) 1 (33,3 %) pro- 3 (60 %) 2 (40 %) s ъ - 4 (100 %) v ъ - - 2 (100 %) v ъ z( ъ )- 3 (100 %) za- 2 (100 %) - Tab. 40: The positioning of Old Church Slavic preverbs and their frequencies 6.6.1. Preverb ordering: the account of Modern Slavic A number of formally-oriented works on Modern Slavic preverbs aim to identify rules of preverb stacking (cf. Section 6.1.1). In particular, the following predictions have been made: i. whenever two preverbs modify a single verbal stem, the innermost should be the lexical one (Babko-Malaya 1999); ii. whenever more than two preverbs stack onto the same verbal stem, only the innermost is lexical (e. g. Istratkova 2004: 306 on Bulgarian). These authors generally do not acknowledge the presence of two lexical preverbs stacked onto the same verbal stem, and even composites that have survived till modern times are usually not cited in these works (cf. Table 33). 30 The Old Church Slavic multiple preverbs that I describe throughout this Chapter seem to represent a different phenomenon from the multiple prefixation in Modern Slavic languages. To begin with, in Old Church Slavic, actional preverbs can occur internally, as do delimitative poand prěin the composites s ъ -po-žiti 30 See however Svenonius (2004b: 242), who cites the Slovenian verb iz-pod-riniti ‘drive from under’, which contains two lexical preverbs, and as such is regarded as problematic. <?page no="284"?> 284 6. Multiple preverbs in Old Church Slavic ‘live for a while with’ and s ъ -prě-byvati ‘remain together with’. In parallel, in the same composites, the position of the lexical preverb s ъ - ‘with’ is external. Admittedly, the composites that show this anomalous ordering did not survive in Bulgarian. Also, they are suspected of being calques of the Greek composite sun-ana-stréphomai ‘live together with’ (cf. Section 6.2.4). Thus, what differentiates multiple preverbation in Old Church Slavic from that in Modern Slavic, is that in Old Church Slavic, two lexical preverbs are allowed to stack onto the same verbal base, as shown in (44) (=(18)): Musterdatei NFA_Basis_A.dot 275 a. iz-po-věděti ‘confess, explain’ (44) věděti → po-věděti → b. pro-po-věděti ‘proclaim, predict’ ‘know’ ‘tell’ c. za-po-věděti ‘order’ Besides the composites in (44), in which the two preverbs are clearly lexical, there are also verbs in which one of the two preverbs seems to be actional, but nevertheless able to modify the lexical content of the simplex verb. One such instance is vъs-po-męnǫti ‘start remembering’, in which the IP pohas an ingressive value that changes the meaning of the bare root √min- ‘think’: po- + √ min- ‘ingressive’ + ‘think’ → ‘remember’ (and not ‘start thinking’). Another relevant case is pri-iz-žiti ‘spend in addition’: iz- + √ ži- ‘completely’ + ‘live on/ by’ → ‘spend’ (and not ‘live completely on/ by’). Note further that, in priiz-žiti, the EP is clearly lexical: it means ‘in addition’. Nevertheless, it occurs more externally than iz-. These ambiguous cases, in which a quasi-actional preverb still modifies at a lexical level the simplex verb onto which it attaches, can shed light on the possible link between the lexical (and synchronically idiosyncratic), and the quantizing (and synchronically predictable) meanings of preverbs. Furthermore, there are composites for which it is hard to draw a clear-cut distinction between lexical and actional usages. For example, does prědъ-po-lagati owe its meaning ‘distribute to’ to pobeing a lexical preverb, or to ‘put ( lagati ) repeatedly ( po- ) being in front of ( prědъ- )? The fuzziness of the type just described for pobacks up the assumption that the homophonous lexical and super-lexical preverbs are not distinct linguistic items; rather, preverbs are polysemous morphemes that have undergone one of two parallel paths of development: either lexicalization or grammaticalization into bounder perfectives. In fact, a few Old Church Slavic composites already seem to instantiate a preverb ordering that resembles the modern pattern: accordingly, they display purely actional preverbs in the outermost position. In vъz-nе-nа-viděti ‘come to hate’, for example, vъzfunctions as an ingressive actional marker, and occurs externally (possibly, vъzfunctions as a perfectivizing preverb; cf. Section 6.3.4). Besides the composites in (44), in which the two preverbs are clearly lexical, there are also verbs in which one of the two preverbs seems to be actional, but nevertheless able to modify the lexical content of the simplex verb. One such instance is v ъ s-po-męnǫti ‘start remembering’, in which the IP pohas an ingressive value that changes the meaning of the bare root √min- ‘think’: po- + √ min- ‘ingressive’ + ‘think’ → ‘remember’ (and not ‘start thinking’). Another relevant case is pri-izžiti ‘spend in addition’: iz- + √ ži- ‘completely’ + ‘live on / by’ → ‘spend’ (and not ‘live completely on / by’). Note further that, in pri-iz-žiti, the EP is clearly lexical: it means ‘in addition’. Nevertheless, it occurs more externally than iz-. These ambiguous cases, in which a quasi-actional preverb still modifies at a lexical level the simplex verb onto which it attaches, can shed light on the possible link between the lexical (and synchronically idiosyncratic), and the quantizing (and synchronically predictable) meanings of preverbs. Furthermore, there are composites for which it is hard to draw a clear-cut distinction between lexical and actional usages. For example, does prěd ъ -po-lagati owe its meaning ‘distribute to’ to pobeing a lexical preverb, or to ‘put ( lagati ) repeatedly ( po- ) being in front of ( prěd ъ - )? The fuzziness of the type just described for pobacks up the assumption that the homophonous lexical and super-lexical preverbs are not distinct linguistic items; rather, preverbs are polysemous morphemes that have undergone one of two parallel paths of development: either lexicalization or grammaticalization into bounder perfectives. In fact, a few Old Church Slavic composites already seem to instantiate a preverb ordering that resembles the modern pattern: accordingly, they display purely actional preverbs in the outermost position. In v ъ z-n е -n а -viděti ‘come to hate’, for example, v ъ zfunctions as an ingressive actional marker, and occurs externally (possibly, v ъ zfunctions as a perfectivizing preverb; cf. Section 6.3.4). <?page no="285"?> 6.6. Preverb ordering 285 6.6.2. An integrated account of preverb ordering My theory of preverb ordering takes into account both diachronic observations and the critical fact that Old Church Slavic texts have been translated from Greek sources. Diachrony reveals the fact that, for most composites, the EP seems to constitute a later addition to an existing composite lexicalized at a preceding stage. This explanation is consistent with the fact that Old Church Slavic EP s seem to retain more distinct semantics than IP s (cf. Section 6.4.3). In addition, it is consistent with the “redundancy principle”, according to which preverbs whose semantics is closer, i. e. subsumed, by the semantics of the verbal stem are likely to attach closer to it. This intuition is backed up by the fact that the Slavic sequences IP - VB frequently translate a Greek simplex verb. This is clear at least for the composites displayed in Table 41. In addition, given that the simplex verb *rěsti is not attested, and thus that ob-rěsti is lexicalized, the composite pri-ob-rěsti ‘acquire’ can also be included in the list in Table 41 below. The EP prihas been possibly attached at a later stage to ob-rěsti ‘find’, providing it with a resultative semantics: pri- + ob-rěsti ‘at, in addition, attachment’ + ‘find’ → ‘acquire’. In this case, however, pri-ob-rěsti ‘acquire’ seems to be lexicalized as a whole: it corresponds to Greek ktáomai or kerdáinō ‘obtain’ in most passages. In addition, the semantics of pro-po-vědeti ‘proclaim, predict’, za-po-vědeti ‘order’, and iz-po-vědeti ‘confess, explain’ also suggests that all these composites are derived from the preverbed speech verb po-vědeti ‘tell’, rather than from the cognition verb vědeti ‘know’. Semantically, the generic speech verb po-vědeti ‘tell’ arguably constitutes the intermediate stage between the cognition verb vědeti ‘know’, and the specific speech verbs pro-povědeti ‘proclaim, predict’, za-po-vědeti ‘order’, and iz-po-vědeti ‘confess, explain’. Composite Meaning Greek equivalent is-po-věděti confess, explain ex-omologéō ex-agoreúō ex-ēgéomai an-aggéllō iz-ob-rěsti find out ex-eurískō prěd ъ -po-lagati distribute to para-títhēmi pri-ižditi (iz-žiti) spend in addition pros-dapanáō s ъ -vъ-kupiti gather together sun-áptō v ъ s-po-męn ǫ ti start remembering, remind ana-mimnḗskō hupo-mimnḗskō <?page no="286"?> 286 6. Multiple preverbs in Old Church Slavic Composite Meaning Greek equivalent v ъ s-pri-jęti receive in return ana-lambánō apo-lambánō ap-ékhō za-po-vědati order en-téllomai dia-tássō pros-títhēmi, ep-aggéllomai Tab. 41: Lexicalized IP -V sequences and their Greek counterparts <?page no="287"?> 7.1. Preverbs in Old Irish 287 7. Multiple preverbs in Old Irish 7.1. Preverbs in Old Irish 7.1.1. State of the art Traditional grammars of Old Irish include a section devoted to preverbs, which are called either “preverbs” or “prepositions” ( VKG II 242 ff.; GOI 495 ff.; Vendryes 1923: 231 ff.; Lewis & Pedersen 1961[1937]: 245 ff.). Notably, because of their preverbal position, traditional grammars also refer to other preverbal morphemes, including the negation, the interrogative particles, and the relative preverbs, as “preverbs” . A major concern of such grammars is discussing preverbs’ accentual and positional properties (cf. Section 7.1.2.1.1), and the related formal changes that they undergo (cf. Section 7.3.1). In addition, traditional grammars usually offer an exhaustive catalogue of Old Irish lexical (i. e. word-forming) preverbs, their allomorphs associated with their basic meanings, and examples illustrating their usages in nominal and verbal composition. Verbal bases are frequently modified by one or more preverbs in Old Irish, in which a single verbal root can constitute the basis for several composites. For example, along with the simplex verb gairid ‘call’, the root √ garifunctions as a basis for the following composites: ad·gair (ad-√gari-) ‘summon’, ar·gair (air- √gari-) ‘forbid’, con·gair (com-√gari-) ‘cry out’, fris·gair (frith-√gari-) ‘answer, reply’, in·gair (in(de)-√gari-) ‘call in’, do·gair (to-√gari-) ‘summon’, ad·togair (adto-√gari-) ‘recall’, and others (for a total of 20 composites; cf. KPV 331-332; McCone 2006: 177). 1 Given this relative abundance of multiple preverbs, most grammars also touch upon the issue of preverb ordering and its motivations, as well as the complex matters of the semantic contributions brought about by accumulated preverbs, and of the difficulty of separating Old Irish multiple preverb composites (cf. Sections 7.3, 7.4, and 7.6; VKG II 302 ff.; GOI 495; Lewis & Pedersen 1961[1937]: 267). Notably, multiple composition must have been a very ancient Celtic phenomenon: there is evidence for sequences of two preverbs in 1 In morphological segmentations, I consistently used preverbs’ first allomorph cited in GOI (495 ff.), though the Milan and the Priscian Glosses databases occasionally employ different forms (e. g. they alternate aith- ~ athi-, air- ~ ar- ~ are- ). The databases also contain inconsistencies as for the underlying form of certain verbal roots (e. g. √ben- ~ √bina- ~ √bena-, √swizd- ~ √seth- ). In this case, I chose and consistently kept the most frequent allomorph occurring in the databases (to facilitate the reader, all allomorphs are displayed in Table 45). <?page no="288"?> 288 7. Multiple preverbs in Old Irish British ( dyrllyddu (to-ro-√sel-) ‘merit’), as well as sporadic hints of the same in Gaulish (Rossiter 2004: 9). Moreover, a specific section of the mentioned grammars is dedicated to non-lexical preverbs that play a grammatical role within the Old Irish verbal system: these are the so-called perfective preverbs (occasionally also referred to as “augments” or “temporal preverbs”, e. g. by McCone 2006; and by Lewis & Pedersen 1961[1937]: 251 ff.), that is, ro-, ad- , com-, and less widespread others. The preverb rois paradigmaticized to indicate completion in the past, and potentiality in the future (cf. Section 7.1.2.3; VKG II 261 ff.; GOI 339 ff.; Vendryes 1923: 241 ff.; Lewis & Pedersen 1961[1937]: 251 ff.). Within the Old Irish catalogue, another grammaticalized preverb is the reportedly empty (i. e. meaningless) no- , which serves the function of contributing to forming certain Old Irish verbal forms or filling certain positional gaps under particular morphosyntactic conditions (cf. e. g. GOI 348; Lewis & Pedersen 1961[1937]: 259). As in other Indo-European languages, (a number of) the same morphemes that function as preverbs can also behave as prepositions (‘position C’ in GOI ): accordingly, traditional grammars also contains dedicated to the prepositional usages of Old Irish preverbs in combination with nominal morphological cases (cf. Section 7.1.2.2; VKG II 72 ff.; GOI 496 ff.; Vendryes 1923: 142 ff.; Lewis & Pedersen 1961[1937]: 162 ff.). To sum up, preverbs play a prominent role within the Old Irish verbal system. On the one hand, they productively build new composite verbs; on the other hand, a number of preverbs are grammaticalized as markers of perfectivity. In addition, they are crucial to the morphological distinction between absolute and conjunct personal endings (cf. e. g. GOI 350). Therefore, a number of works also deal with preverbs and accumulation of preverbs, while investigating wider issues such as the rise of the Old Irish verbal complex and its diachronic syntax. McCone’s (1997, 2006) monographies belong with this group, in that they explore the role of multiple preverbs in the origins and development of the Old Irish and Insular Celtic verbal complex. McCone (1997: 89-90) is notorious for his attempt at drawing a positional hierarchy ruling the relative order of preverbs that may be simultaneously added to a verb in primary (i. e. the first layer of) composition. To primary composition, which follows this hierarchy, Mc- Cone proposes adding a more recent stage of composition, that is, the so-called “secondary composition”, which consists of adding one more preverb among the still productive ones (i. e. com-, eter-, fo-, frith-, and imm- ) to an already existing and lexicalized composite, essentially disregarding the hierarchy. Though McCone’s hierarchy works reasonably well, once one puts aside calques from Latin (cf. Section 7.6), McCone (2006) himself later discarded the idea of a clear-cut distinction between primary and secondary composition, <?page no="289"?> 7.1. Preverbs in Old Irish 289 especially in the light of Rossiter’s (2004) results. Rossiter’s (2004) dissertation shows that the removal of the EP largely results in actually attested Old Irish composites. These data contributed support to the hypothesis of a single process in the formation of composites in Old Irish, and specifically a step by step accretion or recomposition (cf. Section 7.2.3 and 7.6). McCone (1997, 2006) also pursues the broader goal of investigating the rise of the VSO pattern in Old Irish, as well as the rise of the differentiation between absolute and conjunct inflections in the light of both comparative and areal considerations. 2 In fact, the development of preverbs is related to both processes. Though there is no general agreement on the exact mechanism(s) behind the generalization of the VSO pattern in Old Irish, this development, divergent from the rest of Indo-European, possibly relates to the following co-factors: (a) the tendency of Indo-European clitics to occupy the second position (Wackernagel’s Law); (b) Vendryes’ restriction (Vendryes 1911, 1912; Dillon 1943), whereby Old Irish second-position clitics tend to be hosted by specific elements, i. e. the simplex verbs, the first preverb of a composite, and the so-called conjunct particles. As a consequence, simplex verbs and preverbs were allegedly attracted toward the first position by second-position clitics. Later on, the verbal bases of the composites also underwent a leftward shift, due to the tendency toward univerbation between preverbs and verbal bases (Watkins 1963). Lastly, the initial position was generalized even in clauses without second-position clitics (cf. also Eska 1994; Adger 2000; Doherty 2000; McCone 2006: 61 ff.; Eska 2007; and references therein). The growth of the unmarked VSO pattern relates to other open issues regarding the Old Irish verbal complex, notably the dichotomy between the absolute and the conjunct endings. Conjunct endings always occur with composite verbs, after the preverbs roand no- , and after a number of other preverbal particles; absolute endings are employed elsewhere. What does this dichotomy have to do with preverbs? A number of scholars from Boling (1972) and Cowgill (1975) onward argue that the absolute flexion could only be explained by assuming a second-position particle with the shape *=es and of uncertain etymology, which prevented the apocope of -i that instead characterizes the conjunct flexion. Nowadays, some scholars either accept the asseverative particle *=es but reject the apocope of -i (Kortlandt 1979, 1982, 1994), or viceversa (McCone 1979; 1982, 1985a, 2006, 2007; Sims-Williams 1984; Koch 1987; Isaac 1993, 2000). Others accept both parts of Cowgill’s theory, but assume a different shape for the 2 The VSO character of Celtic languages is explored in a language contact perspective in Morris-Jones (1899), Pokorny (1949), Wagner (1959, 1964), Hewitt (2009), Matasović (2012a, 2012b), Mikhailova (2012). In recent decades, VSO languages also raised typologists’ interest (Carnie & Guilfoyle 2000; Carnie et al. 2005). <?page no="290"?> 290 7. Multiple preverbs in Old Irish reconstructed particle, i. e. *et(i) ‘and’ (Schrijver 1994, 1997; Schumacher 1999; KPV ; Eska 2012). More recently, an innovative hypothesis has been put forward by Budassi (2017), which also comprises an explanation for the rise of relative endings (which are also morphologically distinct from declarative ones in Old Irish). Instead of assuming a single discourse particle, Budassi assumes a whole set of clitic pronouns either with deictic or with anaphorical function. Budassi speculates that these pronouns later on underwent univerbation and grammaticalization as absolute and relative endings, respectively (cf. also Budassi & Roma 2018). Whatever the solution to this puzzle is, it must take into account the following facts related to the phonology and the syntax of preverbs: (a) the lack of a regular lenition occurring after the proclitic first preverb of initial composite verbs (cf. Section 7.3); possibly, lenition is blocked by the presence of the assumed 2P clitic particle / pronoun; (b) verbal forms occurring in tmesis (# PE …V#) and in Bergin’s (# … PV #) construction show the conjunct flexion (cf. Section 7.1.2.1.3). Along with these comprehensive works, a number of papers touch upon different aspects of a single Old Irish preverb. These papers mainly focus on etymological, phonetic, or syntactic aspects, and usually only cursorily address the semantic and functional properties of Old Irish preverbs (e. g. Russell 1988, and references therein, on uss- ‘up, off’; Stifter 2014, and references therein on to- ‘to, toward’; Dedio & Widmer (2017), and references therein, on imm- ‘about, mutually’). 3 7.1.2. The status and functions of Old Irish preverbs 7.1.2.1. The morphosyntactic status of Old Irish preverbs 7.1.2.1.1. The accentual properties of Old Irish preverbs Usually, composite verbs bear the accent on the second element of the composition: these are the so-called “deuterotonic” forms of composites. This means that, with composites containing only one preverb, such a preverb occurs before the accent, that is, in “pretonic” or “prenuclear” (in Anderson’s 2016 terms) position, and the verb base is accented. With multiple preverb composites, instead, the EP still remains before the accent, but it is the first of the MP s or the IP that bears the accent. With deuterotonic forms, the EP occurring before the accent behaves as a separate unit. Accordingly, it can be split from the rest of the composite by a personal pronoun (Section 7.1.2.1.2; GOI 27 ff., 351 ff., 534 ff.; Vendryes 1923: 232; Lewis & Pedersen 1961[1937]: 69, 245 ff.; McCone 1997: 1-8). 3 Within the main text, Old Irish preverbs are consistently translated as in GOI (495 ff.). <?page no="291"?> 7.1. Preverbs in Old Irish 291 However, though the EP is usually pretonic, it does bear the accent under certain morphosyntactic conditions, e. g. in the imperative, after certain conjunctions and particles, (occasionally) in relative clauses (cf. (1)), and in the archaic constructions in which the verb occurs in the last position (Section 7.1.2.1.3; GOI 28 ff.). The accented position is called “nuclear“ by Anderson (2016). The form of composites showing these behavioral properties are called “prototonic” . In prototonic forms, the EP bearing the accent becomes a fullyfledged part of the rest of the verb. In (1), the same composite do·indnaig ( to-in(de)-√aneg- ) ‘give, bestow, grant’ is shown in deuterotonic and prototonic forms: accordingly, the EP tosurfaces as duand t- (cf. Section 7.3): (1) a. The composite do·indnaig in deuterotonic position 7 duindainsed da [dam] inna-huli-se and give. sbjv . pst .3 sg to.1 sg . dat art . acc . pl indf . acc . pl . n dem ‘… and who could give me all these things.’ (Ml.78b18) b. The composite do·indnaig in prototonic position tinnagat ní ass-a-nucht di give. prs .3 pl indf . acc out_ofposs .3 pl -bosom. dat of mainib possession. dat . pl ‘… who give some treasures out of their bosom.’ (Ml.93a20) 7.1.2.1.2. The morphological status of preverbs In Old Irish, the lexical (i. e. word-forming) preverbs stacked onto a single verbal base show different statuses according to their position relative to the verbal base. In particular, the EP is usually a clitic, whereas the MP s and the IP are affixes (cf. further Kuryłowicz 1964: 174). These divergent morphological statuses emerge from the position of the accent and from the fact that second position clitics can intervene in between the EP and the remaining composite. In the socalled deuterotonic forms (cf. Section 7.1.2.1.1) with two or more preverbs, the accent falls onto the first preverb after the EP (2). <?page no="292"?> 292 7. Multiple preverbs in Old Irish (2) The position of the accent in deuterotonic multiple preverb composites Old Irish text Translation Segmentation Locus ad déicider : sbjv . prs .2 sg ‘you may look’ ad-di-in(de)-√kwis- Ml.43a19 dufórban : prs .3 sg ‘it comes’ to-for-√ben- Ml.61a22 fuácbat : prs .3 pl ‘they leave’ fo-ad-√gabi- Ml.80a10 b. The composite do·indnaig in prototonic position tinnagat ní ass-a-nucht di give. prs .3 pl indf . acc out_ofposs .3 pl -bosom. dat of mainib possession. dat . pl ‘… who give some treasures out of their bosom.’ (Ml.93a20) In addition, Old Irish syntax allows for infixed pronouns with different functions including the following: (a) direct objects; (b) various types of dative participants with the verb ‘to be’; (c) first and second person subjects with passive verbs; (d) Goal after motion verbs; (e) (rarely) indirect objects in dative ( GOI 255 ff.; Lewis & Pedersen 1961[1937]: 196 ff.). These infixed pronouns are always attached to the element that immediately precedes the accent, be it a lexical preverb (3) or a preverbal element of another type (4). (3) at-[t]=chom-aing fri= agitofel p -3 sg . n = p -strike. prs .3 sg against= A . acc ‘He struck it against Achitophel.’ (Ml.24c16) (4) dia-ndam=chon-delc frit-su if-1 sg = p -compare. sbjv . prs .1 sg against-2 sg . acc ‘If I compare myself to you …’ (Ml.91d8) Kuryłowicz (1964: 174 ff.) relates the status of Old Irish EP to the particular productivity of multiple composition in Old Irish in the following way. Multiple preverbs are especially productive, as they do not constitute an ambiguous structure: the preverb farthest from the verbal stem (the EP ) is clearly separated from the rest of the composite. Thus, Kuryłowicz argues in favor of a process of formation that McCone later called “recomposition” or “accretion” (McCone 2006: 180): “a process of step by step accretion entailing the prefixing of a single extra preverb to an already existing simple or compound verbal form” (cf. further Sections <?page no="293"?> 7.1. Preverbs in Old Irish 293 7.2.3 and 7.6). Accordingly, for example, on gairid ‘call’, the composites in·gair (in(de)-√gari-) ‘herd, tend, protect’, do·ingair (to-in(de)-√gari-) ‘call (by a name)’, and for·dingair (for-di(de)-in-gari-) ‘signify, express’ are successively built. Deviations from the pattern outlined above either contain the preverbs imm- ‘about, mutually’ or roin their grammaticalized function, or can be explained through Latin influence ( GOI 30, 256). Example (5) shows the composite imm·aig ‘drive around’, which contains the preverb imm- ‘about, mutually’, and is preceded by the conjunction an ‘when’. Accordingly, one would expect to infix the personal pronoun after the conjunct particle (cf. example (4)). Surprisingly, however, the infix pronoun occurs after both the conjunctive particle and the preverb imm-. GOI (256) points out that this anomalous positioning of immoccurs only in passages in which this preverb serves a reciprocal construction and means ‘mutually’. Thus, the pattern exemplified in (5) is possibly analogical from the examples in which immfunctions as a reciprocal. 4 (5) [a]n-im-da=[a]ig dia when-around-3 pl =drive. prs .3 sg God. nom ‘… when God drives them around’ (Ml.66d18) In (6), instead, the infix pronoun follows the grammaticalized preverb roin its perfectivizing function (glossed as aug ), as is also described in GOI (256): (6) acht ni-ru-m=chom-ar-leicis se namma but neg aug -1 sg = p p -permit. prf .2 sg em .1 sg only ‘But only, you have not allowed me (to be captured)! ’ (Ml.76d5) The Milan Glosses allow for other anomalous formations. In (7), two lexical preverbs show the status of clitics: (7) Two preverbs in pretonic position: ad·cuimtig ‘build up, build to’ (ad-com-uss-ding-) ol ad-com=rótaig because p p =build_to. prf .3 sg ‘Because it built up.’ (Ml.35b13) 4 The corresponding Latin text presents the expression deo inminente : abl ‘with God’s command’. Though imm·aig ‘drive around’ is not a literal translation of Latin in-mineo (literally) ‘bend / lean toward’, the Old Irish preverb immpossibly echoes Latin in-m-. <?page no="294"?> 294 7. Multiple preverbs in Old Irish In (7), the presence of adis arguably influenced by the corresponding Latin source text, which contains ad-strueret : sbjv . impf .3 sg ‘build near, add’ (cf. also e DIL .ie / 406). Note that the composite con·utaing (com-uss-ding-) ‘build, construct’, containing only the MP s and the IP , is also attested in the Milan Glosses and is almost equivalent to ad·cuimtig ‘build up, build to’ (cf. Section 7.2.4). 7.1.2.1.3. Syntactic patterns with the verb in non-initial position As touched upon in Section 7.1.1, the usual position for the Old Irish verbal complex is clause-initial (8): (8) ni= guid digail du= thabairt foraib neg =pray. prs .3 sg punishment. acc to= inflict. dat . sg on.3 pl ‘He does not pray that punishment is inflicted upon them.’ (Ml.42a4) Old Irish is quite consistently a VSO language. However, verb-final / verb-medial patterns also occasionally occur in Old Irish poetry and rhythmic prose, both involving and not involving tmesis, i. e. the displacement of (a) preverb(s) from the verb it(they) modify(ies) (cf. Bergin 1938; Greene 1977; Binchy 1979-1980; Watkins 1963; Eska 2007 and references therein). Quite strikingly, these non-initial verbal complexes take conjunct endings when simplex (cf. Section 7.1.1), but prototonic stress when composite (cf. examples (10)a-b; Section 7.1.2.1.1). The tmesis pattern is exemplified in (9), the so-called Bergin’s Rule pattern in (10): (9) a. Tmesis pattern ( ad-√kwis- ‘look at’) (from Watkins 1963: 32) adcruth caín -cichither p form. nom fair. nom see. fut .3 sg . pass ‘Fair form will be seen.’ b. Usual preverbal position ( ad-√kwis- ‘look at’) noch ní ac-cam i-sint-saltair in however neg p -see. prs .1 pl indat . n - P. dat art . acc fers n-isin verse. acc that. acc ‘However we do not see that verse in the Psalter.’ (Ml.111d1) (10) a. Bergin’s Rule with verb in final position (from Eska 2007: 255) ar= mind n-axal n-acallad 1 pl . gen = hero. nom apostle. acc converse. impf .3 sg . prot <?page no="295"?> 7.1. Preverbs in Old Irish 295 ‘Our hero used to converse with the apostle.’ ( ACC § 82) (corresponding deuterotonic form = ad·gládad ) b. Bergin’s Rule with verb in medial position (from Eska 2007: 255) lāithe gailēoin gabsat inna= warrior. nom . pl G . gen take. pret .3 pl in.3 pl . gen lāmaib lāigne hand. dat . pl spear. acc . pl ‘The warriors of the Galēon took spears in their hands.’ ( CGH 1.9) (corresponding absolute form = gabsait ) In (9)a, the composite ad-√kwis- ‘look at’ is split by the subject cruth caín ‘fair form’, whereas in (9)b ad- (i. e. its allomorph ac- ) occurs close to the verbal base. In (10)a, the composite ad·gládathar ‘address, speak to’ is not ‘split’, but it does not occur in its usual position: it is preceded by its subject ( ar=mind ) and its second argument ( n-axal ). In (10)b, the simple verb gaibid ‘take’ stands in medial position: it is preceded by the subject participant but followed by the second argument and the Goal-participant. A controversial issue is the actual significance of the orders shown in (9) and (10) for the reconstruction of the prehistoric clausal configuration of Irish. Tmesis and Bergin’s Rule constructions can be regarded as literary artifacts, constructed to produce alliterations or particular cadences in poetry and rhythmic prose (Wagner 1976; Greene 1977; Breatnach 1984). Indeed, for example, splitting adfrom -cichither produces a sequence of three allitterating words in (9)a. Alternatively, these orders can be considered as petrified relics of a pre-literary stage of the language, in which the positional properties of the Old Irish verbal complex were closer to the orders - mainly verb-final, but also verb-medial - found in other ancient Indo-European languages such as Hittite, Vedic, and Homeric Greek (e. g. Watkins 1963; McCone 1979; Ahlqvist 1980; Hamp 1982; MacCoisdealbha 1998; Eska 2007). A few scholars, who support the antiquity of tmesis and Bergin’s Rule constructions, have not very convincingly attempted to reduce these patterns to verb-second constructions (Koch 1987; Doherty 1999, 2000). They accordingly regard Old Irish as a residual V2 language, and not a residual verb-final / verb-medial language. This reconstruction is however unlikely, as several Old Irish passages contain more than a single clausal constituent occurring before the verb (cf. examples at (10)a in Eska 2007: 258 ff. and the related discussion). <?page no="296"?> 296 7. Multiple preverbs in Old Irish 7.1.2.2. The prepositional function of Old Irish preverbs A number of Old Irish preverbs can also occur outside the preverbal complex as prepositions. GOI (495 ff.) assigns to preverbs the following positions: A. under or after the accent (cf. Section 7.1.2.1.1), that is, in close composition with nouns and verbs (preverbs have the morphological status of affixes); B. before the accent in front of a verb or another preverb (clitic status, cf. 7.1.2.1.1); C. before the accent in front of an inflected noun (clitic status); D. before a suffixed personal pronoun (cf. example (11)). Positions C and D above are those in which preverbs function as prepositions. Not all preverbs can occur in all the above positions. According to GOI , a number of them only occur in A and B, that is, they cannot function as prepositions: ad- ‘to, toward’ ( co is used instead of ad in C and D); aith- ‘re-, ex-’; cét- ‘with’ (only in close composition, i. e. position A); -ne- ‘down’ (only A); -uss- ‘up, off’; ro- ‘forth’ (cf. Section 7.1.2.3); to- ‘to, toward’. 5 Others are only employed in positions C and D, that is, they can only function as prepositions: al ‘beyond’ (obsolescent according to GOI 500); amal ‘as, like’; cen ‘without’; co ‘to, till’ (replaced by ad in A and B); do, du ‘to’; 6 fíad ‘in the presence of ’; inge ‘except’ (only C); (h)is ‘underneath’; la ‘with, along’; ó, úa ‘from, by’; oc ‘at’; ós ‘above, over’. Moreover, echtar ‘outside, without’ never participates in verbal composition (i. e. it cannot occur in position B). These data are summarized in Table 42 and in Table 43. p reverb m eaning a b c d 1. ad (aud, as) to, toward, up to + + - - 2. air (er, ir, ar,are, aur, etc.) before, for, on account of, around + + + l ( acc , dat ) + 3. aith- (aid, ath, ad, aud, ed, id) re-, ex- + + - - 4. com (cum, coím(m), cot, co, cu) with + + + n ( dat ) + 5. di (de, dī, dí, do ) of, from, between + + + l ( dat ) + 6. ess (es, é, a, as, ass, as(s)a, ad, at) out of + + + ( dat ) + 7. eter (etar) between, among + + + ( acc ) + 8. fo (fu, fa, -f-, -b-) under + + + l ( acc , dat ) + 5 On the etymology of the preverb toand its etymological relation with the preposition do ‘to’, see Stifter (2014). 6 Cf. fn. 5 in this Chapter. <?page no="297"?> 7.1. Preverbs in Old Irish 297 p reverb m eaning a b c d 9. for (fur, far) on, over + + + (l) ( acc , dat ) + 10. frith (frid, fres, fris(s),fri) against + + + ( acc ) + 11. íarm (íar, íarmi) after + + + n ( dat ) + 12. imm (imb, im, imp) about, mutually + + + l ( acc ) + 13. in (ind, en, ini, inde) in, into + + + n ( acc , dat ) + 14. ne down + - - - 15. os(s) (uss) up, off + + + ( dat ) - 16. re (ri, rem) before, pre- + + + n ( dat ) + 17. ro (ru) *forth, intensive + + - - 18. sech past, beyond + + + ( acc ) + 19. tar (dar, tairm, tarm) across, over + + + ( acc ) + 20. to (do, ta, t, te, tu, t) to, toward + + - - 21. tri, tre, trem through + + + l ( acc ) + Tab. 42: Old Irish preverbs and their positions ( GOI ) 7 1. Often mixed with aith- , ess- , and in(de)- . It often replaces in(de)- (Lewis & Pedersen 1961[1937]: 260). 5. di+fo → dú (Pokorny 1914: 120; GOI 504), di+in → din , di+ess → dé- , di+uss → diu- ( GOI 504-505; Lewis & Pedersen 1961[1937]: 261). 6. The preverb adoccasionally substitutes ess- (Pokorny 1914: 120-121). 14. The preverb neonly combines with a few verbal roots (cf. air-ne-√guid- ‘pray’ in Ml.61b1). 19. The preverbal usage of sechmo- ‘past, beyond’ is limited to motion verbs ( GOI 530); in sechmo·ella ( sechmo-in(de)-√ell- ) ‘pass by, pass, neglect’, it is combined with √ell- ‘put in motion’, resulting in compositional and spatial (‘pass by’), as well as in non-compositional and abstract (‘neglect’) meanings. 21. Cf. Stifter (2014); Table 43, n. 5. In combination with other preverbs: to+fo → tó , túa ; to+for → tór , túar (Pokorny 1914: 124). a dnominal p reposition m eaning 1. al (ol < oll) (+ acc ) beyond 2. amal l (+ acc ) as, like 7 The apices l and n indicate initial mutations. Rare mutations are reported between brackets (cf. n. 9. for ). Preverbs are cited as in GOI. <?page no="298"?> 298 7. Multiple preverbs in Old Irish a dnominal p reposition m eaning 3. cen l (+ acc ) beyond, without 4. co, cu l (+ acc ) to, till 5. do, du l (+ dat ) to, for 6. fíad (+ dat ) in the presence of 7. inge (+ acc ) except 8. (h)ís (+ dat ) underneath 9. la (+ acc ) beside, with, among 10. ó, úa from, by 11. oc (+ dat ) at 12. ós (+ dat ) above, over 13. sechtar (+ acc ) out of, outside Tab. 43: Preverbs featuring positions C and D only 4. In A and B, co is replaced by ad- (Table 42, n.1). It is also used as a conjunction co l ‘so that’ ( GOI 502). 5. In A and B, it is replaced by to- (Table 42, n. 21, GOI 506 after Holmer cited therein; McCone 1997); do and to are discussed together in Pokorny (1914: 124) and in Lewis & Pedersen (1961[1937]: 266). 13. Cf. also echtar ‘outside, without’ occurring as a prefix in nominal compounds, and as a preposition+acc. In (11), the two-fold usage of preverbs is exemplified by means of frith ‘against’: in (11)a, the form fris contains a suffixed personal pronoun in the accusative case, and thus functions as a prepositional phrase; (11)b, frithfunctions as EP of the composite fris·tabair ‘set aganst, oppose’ . (11) The two-fold usage of frith- , fris- , fri ‘against’ a. Prepositional function: do·beir ( to-√ber- )+fri. ACC con-ducthar nomen fri-s. until-bring. sbjv . prs . sg . pass name. nom againstacc .3 sg . m / n ‘(The possession is indefinite …) until a name is put against it.’ (Sg.200b13) b. Preverbal function: fris·tabair ( frith-to-√ber- ) hua-nd-í fris-tarat fromart . dat deict against-set. prf .3 sg ‘In that he opposed …’ (Ml.51d3) <?page no="299"?> 7.1. Preverbs in Old Irish 299 On the diachrony and usage of Old Irish cases and prepositions, I refer to Windisch (1879), Moore (1882: 65 ff.), GOI (495 ff.), Vendryes (1923: 136 ff.), Lewis & Pedersen (1961[1937]: 161 ff.), and Hewson & Bubenik (2006: 228 ff.). Very briefly, Old Irish retains the nominative, accusative, genitive, dative, and vocative cases out of the eight-fold Proto-Indo-European reconstructed case system. Old Irish cases can mostly be observed as resulting in phonetic shifts that affect vowels and syllable structure, along with palatalization and mutation of consonants. Proto-Indo-European instrumental, ablative, and locative all merged into the dative case. Accordingly, the dative is combined with various prepositions to express the senses previously assigned to the three merging Proto-Indo-European cases. Prepositionless dative is obsolescent in Old Irish, in which only a few petrified instrumental datives can be found as adverbs (cf. (20); aithieriuch ‘again’; cf. Vendryes 1923: 141). The accusative, combined with different prepositions, is instead employed to indicate Goal. A few motion verbs take a prepositionless accusative expressing Goal: e. g. the simplex verb téit (√tēg-) ‘go’ (Ml.62b20), and the composites do·icc (to-√icc-) ‘come to, approach’ (Ml.41d9, 123c3) and ro·icc (ro-√icc-) ‘come, attain, reach’ (Ml.9d20, 55d2) (Vendryes 1923: 139). 8 The genitive, as expected, being a strictly adnominal case, is used with prepositions or prepositional locutions stemming from noun phrases (i. e. secondary prepositions of nominal origin such as ar cuit ‘on account of ’, in degaid ‘beside’, fri lorg ‘behind’; cf. Vendryes 1923: 137). Old Irish prepositions govern the case that they precede; accordingly, they usually select only one case. However, case alternation marginally retains its significance with prepositions governing both the accusative and the dative case: accusative-dative alternation distinguishes Goal from Location (cf. Section 6.1.3.2 on OCS ). Peculiar to Old Irish are the so-called conjugated prepositions ( GOI 272 ff.; Lewis & Pedersen 1961[1937]: 199 ff.; Vendryes 1923: 277 ff.), whose endings stand for personal pronouns. One relevant example is provided in (11): the preposition fri is followed by a suffixed accusative singular third person pronoun -s . 7.1.2.3. The preverbs ro-, no-, and the other grammatical preverbs of Old Irish Along with their lexical, or word-forming capabilities, Old Irish preverbal morphemes can also serve more grammatical functions ( GOI 339 ff.; Vendryes 1923: 241 ff; Lewis & Pedersen 1961[1937]: 251 ff.). A few preverbs, variably called “verbal particles” ( GOI ), “temporal preverbs” (Lewis & Pedersen 1961[1937]), or “augments” (McCone 2006: 190) work as aspectual markers. These temporal preverbs can be prefixed to express various meanings connected with perfectivity or potentiality. For example, they can be added (i) to the preterite to express the 8 The simplex verb téit (√tēg-) ‘go’ can also take a Path-participant expressed by a prepositionless accusative (Ml.28c19). <?page no="300"?> 300 7. Multiple preverbs in Old Irish perfect (the most frequent use); (ii) to the imperfect to denote an action completed multiple times in the past; (iii) to the present indicative and subjunctive to describe general actions to be completed before another action can take place; (iv) to the present subjunctive with potential value. Among the perfectivizing preverbs, the most paradigmaticized is ro- (etymologically) ‘forward, forth’ (< *pr-ó , cf., among others, AG pró , OCS pro-, Ved. prá ; see LIPP II 637), which can be added to nearly all simplex and composite Old Irish verbs. Other reasonably frequent preverbs supplying the function of roare ad- (etymologically) ‘to, toward’ and com- (etymologically) ‘with’. The preverb adis mostly used with composite verbs containing lexical com- ‘with’ (e. g. conaitecht : prf .3 sg ‘(he) asked’ from com-di-√sag- ‘ask, seek, demand’), whereas commostly, but by no means exclusively, is used with roots ending in -g- (e. g. √org- ‘slay, kill’). 9 Furthermore, a few perfectivizing preverbs ( di-, ess-, in(de)-, to- ) only occur in combination with specific verbal roots, possibly on account of their basic meaning. For example, ess- ‘out of ’ (<* éĝh-s , cf., among others, AG ex- , Lat. ex- , OCS iz- ; see LIPP II 204 f.) perfectivizes two verbs for drinking ( √ib-, √lu- ), as ‘drink out of ’ can easily be understood as ‘drink completely’: e. g. ibis : pret .3 sg ‘drank’ vs. as·ib : prf .3 sg ‘has drunk’ (cf. Germ. trinken ‘drink’ vs. aus-trinken (literally) ‘out of-drink’ → ‘in_full-drink’) (for other examples, cf. GOI 345). The preverb no- (<* nú ‘now’; cf. among others Hitt. nu, Ved. nú , Cypr. nu ), always unaccented, works as a host (traditionally called “auxiliary” , e. g. in GOI) in a number of contexts: it functions as a dummy preverb used with simplex verbs to host infix pronouns and to construct relative clauses. In addition, it is added to simplex verbs in those verbal tenses that require the conjunct forms, and consequently a prenuclear constituent, including the imperfect, the secondary future, and the past subjunctive. In three verbal forms, specifically ro·cluinethar ‘hear’ , ro·finnadar ‘discover’, and ro·laimethar ‘dare’, roseems to be as empty as no-, in that it adds neither grammatical nor lexical meanings to the verbal bases. These developments are reminiscent of the fully grammaticalized Slavic-style aspect system of “bounder perfectives” (Bybee & Dahl 1989; Bybee et al. 1994; see further Chapter 6). What are the reasons for these similar developments? One motivation is semantic broadening: spatial preverbs are able to add an inherent endpoint to spatial - and then also to non-spatial - events (e. g. Shull 2003; Wiemer & Seržant 2017): thus, rois particularly appropriate to add telic meanings, given its basic spatial semantics ‘through to the end’ (McCone 1997: 9 The perfectivizing rocan very occasionally be used with composites containing lexical com- (cf. in Ml.102d5, co[n]runes : prf .3 sg from con·nessa ‘condemn, spurn, trample under foot’). <?page no="301"?> 7.2. Multiple preverbs in numbers 301 117; Rossiter 2004: 22; Dalle Ceste 2014: 145 ff.). However, this explanation does not work for Source-preverbs such as Old Irish ess- ‘out of ’ or for preverbs with an original Comitative meaning such as com- ‘with’ . 10 Here, conceptual metaphor comes into play: (a) events can be conceptualized as locations ; as a consequence, departing from an event implies that the event is completed (Zanchi 2017b); (b) completion can be thought of as togetherness . Metaphor (a) explains the development of ess- ‘out of ’ , whereas metaphor (b) accounts for the employment of comas a marker of perfectivity. In addition, in the case of essperfectiziving verbs of drinking, it is the specific combination of these two elements that might result in a telic reading: one usually drinks a liquid out of a container, thus the meaning of the preverb is implied, or subsumed, by the meaning of drinking. Therefore, the redundancy of the expression ‘drink out of ’ might trigger a new salient reading, that is, the actional meaning of completeness (so-called Vey-Schooneveld effect; cf. Chapter 6; Section 8.2.3). 11 7.2. Multiple preverbs in numbers 7.2.1. Composites with multiple preverbs As mentioned in Section 1.3.4, this study of Old Irish multiple preverb composites is limited to the Milan and the Priscian Glosses . The choice is motivated both by methodological and by practical considerations. On the one hand, given the overabundance of multiple preverb composites in Old Irish, these two corpora suffice for my comparative purposes; in addition, these Old Irish collections of glosses offer an obvious Latin text as a counterpart, which is crucial for understanding the process of formation of multiple preverb composites (Section 7.2.4). On the other hand, these two collections of Glosses are available online in electronic format (Griffith & Stifter 2007-2013; Bauer & Schumacher 2014). The text provided online can be easily downloaded and queried, as the contained glosses are tagged for morphology and syntax, as well as fully translated and paired to the Latin text that they explain, comment, or translate. These corpora allow for queries starting both from specific lemmas and from specific morphemes, including preverbs and verbal bases. Multiple preverb composites have been manually extracted, starting from Anderson’s (2016) full catalogue of Old Irish 10 Source-preverbs are also common telic markers in Slavic and other IE languages (cf. Dickey 2012; Zanchi 2017b). 11 A number of verbs express perfective meanings by means of suppletive stems. In most cases, these suppletive stems contain a preverb, frequently do- , a Goal-preverb (GOI 345; McCone 1997: 92; Rossiter 2004: 18). <?page no="302"?> 302 7. Multiple preverbs in Old Irish composites. 12 Out of Anderson’s list, I selected those verbs occurring in the relevant texts, using the Milan and the Priscian Glosses databases. This selection process yielded 178 composites, which are displayed in Table 44 together with their segmentation and their frequency in the Milan and in the Priscian Glosses. 13 The segmentation is necessary for Old Irish composites, as the surface form of Irish preverbs undergoes considerable changes according to the position that preverbs occupy (cf. Section 7.3). Segmentations of Table 44 follow the morphological analyses contained in the Milan and Priscian Glosses databases; I signal with an asterisk those composites for which the e DIL and / or Anderson (2016) suggest a different segmentation. The number of Old Irish composites attested in this corpus (178), if compared to Homeric Greek (64) and to Old Church Slavic (23), is quite high: this confirms Lewis & Pedersen’s (1961[1937]: 267) observation that Old Irish has gone further in the development of multiple composition than any other ancient Indo-European language. A limited corpus of Old Irish per se reveals a number of composites that even surpasses that of Vedic ( OI r. 178 vs. Ved.114). In addition, Old Irish attests to a number of composites with a relatively high frequency (e. g. a frequency higher than 10 is shown by 30 out of 178 composites): these data are quite different from those of Vedic (Section 4.2), in that Old Irish composites frequently seem to constitute conventionalized formations. Furthermore, the number of occurrences of Old Irish composites, i. e. their token-frequency, is far higher than the number of composites, i. e. their type-frequency (1240 vs. 178): this also backs up the assumption that multiple preverb composites occupy an established position within the Old Irish lexicon. This analysis also offers confirmation in the accentual properties (cf. Section 7.1.2.1.1) and in the non-compositional semantics (cf. Section 7.4.3) of Old Irish composites. 12 My deep gratitude goes to Cormac Anderson, who provided me with the source data of his dissertation during my stay in Jena, before his dissertation was completed. 13 In the entire corpus of multiple preverb composites compiled from DIL and contained in her dissertation (Rossiter 2004: 172 ff.), Rossiter included as many as 483 multiple preverb composites. <?page no="303"?> 7.2. Multiple preverbs in numbers 303 c omposite s egmentation m eaning m ilan p riscian t otal do·futhraccair di-fo-tre-√acc-* desire, wish 4 0 4 con·osna com-uss-√anā-* cease, stop, desist, remain, end in 6 3 9 fo·fúasna fo-uss-√anāperturb, disturb 2 1 3 do·indnaig to-in(de)-√aneggive, bestow, grant, hand over 16 1 17 imm·comairc imm-com-√arcquestion, ask, inquire of 5 4 9 ar·díbdai air-di-√bādīsubmerge, drown, sink, wreck 2 0 2 do·aithbig to-aith-√begdissolve, break up 0 1 1 airdbidi air-di-√benbe destroyed, cut off 1 0 1 foindarbaide fo-in(de)-ad-ro-uss-√benbe relegated, be subjected 1 0 1 in·árban in(de)-ad-ro-uss-√ben-* drive out, expel 8 2 10 do·eipen to-ess-√benexcise, cut (out of, off) 0 1 1 do·forban to-for-√bencome, arrive, happen to, reach 11 2 13 ad·cuimben aith-com-√bencut, strike, wound, lacerate 1 0 1 ar·díben air-di-√bencut off, slay, destroy 0 1 1 etar·diben eter-di-√bendestroy 8 1 9 imm·díben imm-di-√benexcise, circumcise 0 1 1 do·fuiben to-fo-√bencut, cut down, cut out, destroy 6 2 8 do·immdiben to-imm-di-√bencut away, shorten 1 0 1 <?page no="304"?> 304 7. Multiple preverbs in Old Irish c omposite s egmentation m eaning m ilan p riscian t otal ad·tairbir ad-to-air-√berbring back, deliver again 0 1 1 ad·opair ad-uss-√bersacrifice, offer up 6 0 6 fo·tabair fo-to-√berplace under 4 0 4 fris·tabair frith-to-√ber-* set against, oppose 1 1 2 imm·tabair imm-to-√bercarry around, surround 1 0 1 remi·epir rem-ess-√bersay beforehand, say previously 11 1 12 do·adbair to-ad-uss-√berdisplay, show, bring forward, offer 4 0 4 do·airbir to-air-√berbend, bend down, incline, lower 3 2 5 do·opair to-uss-√ber-* take away, deprive, defraud 0 1 1 do·eprainn to-ess-√brennflow, trickle, gush 1 2 3 do·airchain to-air-√can-* prophesy, foretell 11 1 12 do·inchain to-in(de)-√canchant, utter 1 0 1 fo·acain fo-ad-√cansing to, accompany in song 0 1 1 ar·foichlea air-fo-√cēllālook after, take care of, attend to 1 0 1 imm·timchella imm-to-imm-√cēllāsurround 15 0 15 do·imchella to-imm-√cēllāsurround, encompass 4 0 4 do·fuinchid to-fo-in(de)-√cid-* descend 2 0 2 remi·escaid rem-ess-√cidstretch forth 1 0 1 con·érchloí com-air-√clowstir up, disturb, drive away, agitate 3 0 3 <?page no="305"?> 7.2. Multiple preverbs in numbers 305 c omposite s egmentation m eaning m ilan p riscian t otal con·imchloí com-imm-√clowchange 0 1 1 as·rochoíli ess-ro-√coildefine, determine 7 0 7 do·rochoíni di-ro-√coindespair of 8 0 8 do·aithchuiredar to-aith-√corireturn 3 1 4 do·accradi to-ad-√crādexasperate, provoke 5 0 5 do·aithchren to-aith-√crinabuy back, redeem 3 0 3 ar·condla air-com-√dālīshare 0 1 1 con·fodlai com-fo-√dālīshare jointly, divide, apportion 1 0 1 ad·cuimtig ad-com-uss-√dingbuild to, build up 1 1 2 ar·utaing air-uss-√dingbuild up, restore, refresh 4 0 4 con·utaing com-uss-√dingbuild, construct, build up, embellish 4 2 6 do·aidlea to-ad-√ellcome to, approach, visit, touch 1 0 1 ad·comla ad-com-√elljoint, unite 2 0 2 do·ella di-en-√ellturn aside, deviate, bend, decline 7 7 14 fo·accomla fo-ad-com-√ellsubjoin 1 0 1 fo·indlea fo-in(de)-√ellwander, rove 1 0 1 sechmo·ella sechmo-in(de)-√ellpass by, pass, neglect 5 2 7 do·inóla to-in(de)-uss-√ellgather, collect, assemble 4 1 5 ar·foím air-fo-√emaccept, receive, assume, take 26 10 36 <?page no="306"?> 306 7. Multiple preverbs in Old Irish c omposite s egmentation m eaning m ilan p riscian t otal as·toasci ess-to-√fāscīexpress 1 2 3 ar·coat air-com-√fēdprevent, injure 4 0 4 as·indet ess-in(de)-√fēddeclare, relate, tell, set forth 45 4 49 fo·tuidchet fo-to-di-com-√fēdsubdue 1 0 1 remi·aisndet rem-ess-in(de)-√fēdtell beforehand, predict 1 0 1 do·adbat to-ad-√fēdshow, manifest, set forth 46 13 59 do·diat to-di-√fēdlead down, lead, bring 4 3 7 con·foíra com-fo-√ferāprovide 2 0 2 remi·foírea rem-fo-√ferāprovide previously 1 0 1 ar·ingaib air-in(de)-√gabiavoid, flee from 1 0 1 con·ocaib com-uss-√gabilift up, raise, exalt 16 0 16 as·ingaib ess-in(de)-√gabiexceed, surpass, go beyond 13 4 17 fo·acaib fo-ad-√gabileave 6 0 6 imm·imgaib imm-imm-√gabiavoid, shun, evade 17 2 19 do·furgaib to-air-uss-√gabi-* raise up, produce, cause 6 4 10 do·fúarascaib to-for-ess-√gabiexpress, characterize 0 1 1 do·rogaib to-ro-√gabicommit, transgress 12 0 12 do·focaib to-uss-√gabirise 1 0 1 as·congair ess-com-√gariproclaim, give notice 4 0 4 <?page no="307"?> 7.2. Multiple preverbs in numbers 307 c omposite s egmentation m eaning m ilan p riscian t otal for·congair for-com-√garicommand, order 12 6 18 for·díngair for-di-en-√garisignify, express 1 3 4 imm·freccair imm-frith-√garicorrespond, answer, encounter 2 0 2 in·togair in(de)-to-√garicall on, invoke 1 0 1 do·accair to-ad-√gari-* declare, tell 1 0 1 do·airngir to-air-in(de)-√garipromise 17 1 18 do·oggell dē-uss-√gellpurchase 1 0 1 imm·accaldathar imm-ad-√glādīconverse together 1 0 1 do·eclainn to-ess-√glennpick out, select 2 0 2 fodéinti fo-di-√gnībe sufficient 1 0 1 imm·fogni imm-fo-√gnībe construed with (grammar) 2 1 3 ar·neget air-ne-√guidpray 1 0 1 remi·uicsed rem-uss-√guschoose beforehand, pre-elect 1 0 1 con·ricc com-ro-√iccmeet, encounter, join 6 0 6 con·táirci com-to-ad-ro-√icc-* confer 1 0 1 fo·tairci fo-to-ad-ro-√icc-* substitute, supply 2 0 2 for·cumaing for-com-√icchappen, occur, be made, be brought about 8 4 12 imm·airicc imm-air-√iccbe appropriate to 23 0 23 ad·cumaing in(de)-com-√iccstrike, cut, happen 6 0 6 <?page no="308"?> 308 7. Multiple preverbs in Old Irish c omposite s egmentation m eaning m ilan p riscian t otal do·áirci to-ad-ro-√icc-* cause, effect, induce, bring about 22 0 22 do·ecmaing to-in(de)-com-√iccstrike, hit a mark 2 9 11 ad·déici ad-di-en-√kwisregard, look at 1 0 1 do·écai dē-en-√kwislook at, behold, see 9 2 11 etar·décai eter-di-en-√kwisintrospect 1 0 1 imm·accai imm-ad-√kwislook after, regard, examine, consider 4 0 4 remi·décai rem-di-en-√kwisprovide for, foresee 3 0 3 do·incai to-in(de)-ad-√kwislook, gaze at 1 0 1 do·farcai to-for-ad-√kwislook down on, guard, fence around 0 1 1 fris·accai frith-ad-√kwislook forward to, expect, hope 23 2 25 ar·osailci air-uss-√lēcībe opened 20 1 21 con·airléici com-air-√lēcīpermit, allow, let go 28 0 28 do·atailci to-ad-to-ad-√lēcīfondle, caress, pacify 3 0 3 do·fúasailci to-fo-uss-√lēcīloosen, relax 5 4 9 as·cuindligi ess-com-di-√logīdisrupt 1 0 1 fo·álgi fo-ad-√logīlay low, prostrate, throw down 3 3 6 imm·folngai imm-fo-√longīcause, produce, make, affect 45 8 53 remi·folngi rem-fo-√longībear beforehand, anticipate 3 0 3 as·comlai ess-com-√ludepart, escape, set out, start 1 2 3 <?page no="309"?> 7.2. Multiple preverbs in numbers 309 c omposite s egmentation m eaning m ilan p riscian t otal comforaithmiti com-for-aith-√manibe commemorated 1 0 1 do·romnathar di-ro-√maniforget 8 0 8 for·aithminedar for-aith-√mani-* call to mind, commemorate, remember 9 1 10 do·aithmenadar to-aith-√manicall to mind, commemorate, remember 2 2 4 imm·ruimdethar imm-ro-√medisin, transgress 8 0 8 do·inmlig to-en-uss-√mligpromulgate, proclaim, announce 2 0 2 do·fonaig di-fo-√nigwash, wash away 2 3 5 do·immna to-imm-ad-√nocommand, enjoin 1 0 1 con·erairg com-air-√orggo astray 1 1 2 con·túairc com-to-√orgstrike, pound 4 0 4 do·essuirg di-ess-√orgsmite, slay 1 0 1 fris·túairc frith-to-fo-√orgthump, blunt 1 0 1 do·fúairc to-fo-√org-* crush, grin, beat, pound 18 1 19 do·immoirc to-imm-√orgpress, compress, chastise 9 2 11 for·comai for-com-√owkeep, preserve, retain 1 10 11 con·tetarrat com-to-eter-√rethcomprise, comprehend 0 2 2 fo·timmthirid fo-to-imm-di-√rethsubminister, fumigate 0 3 3 for·deret for-di-√rethpass through, go over 2 0 2 for·díurat for-di-uss-√rethremain, remain over 1 0 1 <?page no="310"?> 310 7. Multiple preverbs in Old Irish c omposite s egmentation m eaning m ilan p riscian t otal do·airndret to-air-in(de)-√rethrun about, roam over 1 0 1 do·etarrat to-eter-√rethcomprehend, grasp, overtake 2 0 2 do·íarmórat toíarm-fo-√rethfollow, come after, go after 1 1 2 ad·éirrig aith-ess-√rigrepeat, reiterate, change, emend 5 0 5 do·érig di-ess-√rig abandon, forsake 9 1 10 as·éirig ess-ess-√rigrise again, arise 2 0 2 du·dúrig to-di-uss-√rigbe enticed, be excited 1 0 1 ad·tóirndea ad-to-fo-√rindā-* prick again 0 1 1 fris·tóirndea frith-to-fo-√rindāmark off, trace (a limit) 1 0 1 do·foirndea to-fo-√rindāexpress, signify, denote 5 18 23 con·dieig com-di-√sagask, seek, demand 20 0 20 íarmi·foich íarm-fo-√sagseek after, inquire about 3 1 4 imm·tascra imm-to-√scarāstruggle together 1 0 1 do·foscart(a) di-uss-√scart(ā)remove, put aside 2 0 2 con·oscaigi com-uss-√scochīmove, change, remove, shake, upset 15 10 25 do·róscai di-ro-uss-√scochīstand forth, distinguish oneself 31 11 42 remderscaigthi rem-di-ro-uss-√scochībe pre-distinguished 1 0 1 di·fíuschi di-uss-√secharouse, excite, call into 0 1 1 in·coisig in(de)-com-√sechsignify beforehand 13 0 13 <?page no="311"?> 7.2. Multiple preverbs in numbers 311 c omposite s egmentation m eaning m ilan p riscian t otal do·díuschi to-di-uss-√sechawake, arouse 0 1 1 ar·neät air-ni-√sedexpect, await 10 0 10 fo·coislea fo-com-√selātake away, remove 2 1 3 do·fochsla to-fo-com-√selā-* bring, drag over, draw over 1 0 1 do·aissilbi to-ad-√selbīassign, ascribe 14 2 16 fo·teissim fo-to-ess-√sempour down, pour out 1 0 1 fo·uisim fo-uss-√sembe stored, be put away 1 0 1 do·eissim to-ess-√semshed, pour out 8 0 8 do·fuissim to-uss-√sembring, bring forth 11 8 19 con·tairissedar com-to-air-√sistāremain constant, consist 1 0 1 do·airissedar to-air-√sistāstand, stay, remain 5 0 5 fris·tairissedar frith-to-air-√sistāmake a stand against, oppose 3 0 3 ad·roilli ad-ro-√slīdeserve, be entitled to 25 1 26 ad·cosnai ad-com-√snīmake for, strive, seek 5 0 5 as·roinni ess-ro-√snīescape 1 0 1 fo·cosnai fo-com-√snī steal, snatch away 1 0 1 imm·fresnai imm-frith-√snīcontend, dispute, disagree, gainsay 2 0 2 con·toí com-to-√sowturn, convert, change 4 0 4 do·intai to-in(de)-√sowturn back, return, translate 6 2 8 <?page no="312"?> 312 7. Multiple preverbs in Old Irish c omposite s egmentation m eaning m ilan p riscian t otal fris·tinfet frith-to-in(de)-√swizd-* blow against 1 0 1 do·infet to-in(de)-√swizd-* blow, breathe, blow on, breathe on 2 0 2 ad·cota en-com-√tā-* get, obtain, procure 31 3 33 do·esta to-ess-√tābe absent, lacking, wanting 5 1 6 con·éitet com-en-√tēggo with, accompany, agree with 1 2 3 fris·taít frith-to-√tēgcome against, be at variance with, oppose 9 0 9 in·otat in(de)-uss-√tēgenter into 10 0 10 fo·éitsi fo-in(de)-√tōsīunderstand 1 0 1 t otal 1014 226 1240 Tab. 44: Old Irish composites with multiple preverbs ( Milan and Priscian Glosses ) 141 14 The segmentations included in Table 44 rely on those of the Milan and the Priscian Glosses databases. If eDIL and / or Anderson (2016) suggest a different morphological analysis, this is marked, row by row, with an asterisk. Composites are sorted first by root, and then by EP, MPs, and IP. <?page no="313"?> 7.2. Multiple preverbs in numbers 313 The composites usually have similar meanings in the two series of glosses, which roughly date back to the 9 th century (cf. Section 1.3.4). However, a number of composites exhibit more specialized meanings in the Priscian Glosses, due to the grammatical character of the Latin text, Priscian’s Institutiones Grammaticae , commented on by the Old Irish glosses . Some cases in point are shown in (12). Relevant passages are provided for con·osna ‘stops, ends in’ in (13): (12) Composite Meaning in Ml. Meaning in Sg. con·osna stops end in do·ella turns aside inflect, decline do· fúasailci sets free lose into constituent elements, solve a question (13) a. The meaning of con·osna in Ml. (Lat. cessare ) conosnai-ssiu stop. sbjv . prs .2 sg em . cl .2 sg ‘may you cease’ (Ml.70c3) b. The meaning of con·osna in Sg. (no Latin equivalent) air cheso in .us. conosna són for although in -us ends_in. prs .3 sg em . ana . cl .3 sg . n ‘for although this ends in -us .’ (Sg.206a3) 7.2.2. Verbal roots modified by multiple preverbs Table 45 contains the 43 Old Irish verbal roots modified by multiple preverbs. The simplex verb is provided for the roots that are also attested without preverb(s). 15 In addition, Table 45 displays the meaning of these roots, and their PIE etymologies paired with the corresponding reconstructed meaning. Moreover, the number of composites containing each root is tabulated (i. e. their frequency). In the rightmost column of Table 45, the verb type is also indicated; as with the other languages of the sample, a coarse-grained semantic classification of verb types suffices for the purposes of this study. 15 Among the primary verbs collected in the KPV, only 10 % do not allow for composition, while about 17 % are attested only in composition (cf. also McCone 2006: 177). <?page no="314"?> 314 7. Multiple preverbs in Old Irish v erbal root s imple verb (vkg page ) m eaning pie root (l iv 2 page ) m eaning f requency v erb type accin composition ( KPV 207) be inclined to *h 2 enk- (268) bend, incline 1 mental state anāanaid (455) stay *h 2 enh 1 - (267) breathe 2 bodily process aneg- (anag-) aingid (456) save, protect ? *h 1 egH- ? *h 1 eĝH- (231, IEW 45-47; Vend. A-77) reach, lead 1 helping arcin composition (457) ask, beg *prek̂- (490) ask 1 communication bád- (báid-) báidid (458) sink, destroy *g u̯ eh 2 - (205) step 1 caused motion beg- (bong-) bongid (477) break, reap *b h eu̯g- (84-85) use, benefit 1 contact / impact ben- (bina-, bena-) benaid (461) cut, beat *b h ei̯H- (72) hit 11 removing berbeirid (463) bring, bear *b h er- (76) carry, bring 9 caused motion brennbebernatar : prf .3 pl pour out, spring *b h rend- (95) pour out 1 manner of motion cancanaid (479) sing *k̂an- (342) sing, sound 2 communication cēllā- (cial-) ceilid (482) go around hide, conceal *k̂el- (322) *k u̯ elh 1 - (342) go around turn 3 motion <?page no="315"?> 7.2. Multiple preverbs in numbers 315 v erbal root s imple verb (vkg page ) m eaning pie root (l iv 2 page ) m eaning f requency v erb type cid- (cis-) in composition (490) create ? *k̂ēi̯s- ( IEW 538-539; KPV 391) ? *sed- ( IEW 887; Vend. C-97) move sit 2 creation clow- (clo-) cloïd (493) turn, conquer *k u̯ elh 1 - (386, Vend. C-122; IEW 639-640) turn around 2 manner of motion coildenominative from cáel (495) thin, clear sign *koi-lo- ( IEW 610, Vend. C-6) naked, miserable 1 coincaínid (495) lament ? kan- (342 f., IEW 535-536); ? borrowing of Goth. qainōn ‘laments’ (Vend. C-18) sing howl 1 communication cuir- (cori-) in composition (498) put, throw *kerH- (353) scatter 1 caused motion crādī- (crād- , cráid-) cráidid denom. from crád torment torture ? uncertain (Vend. C-221) ? uncertain 1 mental activity crina- (cren-) crenaid (496) buy, obtain *k u̯ rei̯h 2 - (395) exchange 1 change of possession dālī- (dál-) dáilid (502) portion out *deh 2 (i)- (103) divide 2 removing dingdingid (505) press, thrust *d h ei̯ĝ h - (381) smear, model 3 change of state ell- (lā-, elnā-) in composition (509) go, put in motion *pelh 2 - (470) approach 7 (caused) motion <?page no="316"?> 316 7. Multiple preverbs in Old Irish v erbal root s imple verb (vkg page ) m eaning pie root (l iv 2 page ) m eaning f requency v erb type em- (eim-) in composition (511) take *h 1 em- (236) take 1 removing fāisc- (fāscī-) faiscid (515) press, squeeze ? *u̯ed h h 1 - (660) ? push 1 emission fēdfeidid (515) bring, refer *u̯ed h - (659) drive 6 caused motion ferāferaid (518) grant, supply noun u̯er- ( IEW 1166) friendship 2 transfer gabi- (gaib-) gaibid (527) grasp, reach, go *ĝ h eHb- (195) grasp, take 9 removing gair- (gari-) gairid (533) call *ĝar- (161) sound, call 7 communication gellgellaid (573) gell : noun pledges, promise, deposit *g h ei̯d h - (196, IEW 426) yearn for, wait for 1 communication glādī- (glád- ) in composition (537) talk ? *ĝ h lād- ( IEW 451; Tichy: 64-65) sound, call 1 communication glennin composition (539) examine *g h lend h - (200) see, catch sight of 1 perception gnī- (ghí-) gniid (540) do, make *ĝenh 1 - (163) generate 2 creation guidguidid (550) pray, ask *g u̯h ed h - (271) ask, wish 1 communication <?page no="317"?> 7.2. Multiple preverbs in numbers 317 v erbal root s imple verb (vkg page ) m eaning pie root (l iv 2 page ) m eaning f requency v erb type gusin composition (539) choose *ĝeu̯s- (166) cost 1 removing iccin composition (552) reach *h 2 nek̂- (282) reach 8 motion verb kwis- (cī-) in composition (487) see *k u̯ ei̯s- (381) perceive, look after 8 perception lēcīléicid (562) let go, release *lei̯k u̯ - (406) leave, take away 4 caused motion logī- (luig-) logaid (572) concede, obtain *leg h - (398) put, place 2 change of possession longin composition (568) cause *leu̯ĝ- (416) bend 2 creation luluid, lotar (571) goe *h 1 leu̯d h - (248) climb, grow 1 caused motion mani- (muin-) muinithir (580) meditate *men- (435) think 4 mental activity medi- (mid) midithir (577) weight, measure *med- (423) measure, look after 1 measure mlig- (melg-) mligid (580) milk *h 2 melĝ- (279) milk 1 emission nignegar (585) wash *nei̯g u̯ - (450) wash 1 removing noin composition (585) ? nod ? *neu̯- (455; KPV 491) nod 1 communication <?page no="318"?> 318 7. Multiple preverbs in Old Irish v erbal root s imple verb (vkg page ) m eaning pie root (l iv 2 page ) m eaning f requency v erb type orgoirgid (587) kill, slay, strike *h 2 erg- (301) die 6 contact / impact ow- (o-) in composition (586) ? have ? *h 2 eu̯- (274) enjoy 1 possessing rethrethid (597) runs *ret- (274) run 7 manner of motion rig- (reg-) rigid (593) streches *reiĝ- (503) stretch 4 change of state rindā- (rind-) rindaid, denominative from rind (603) cut, engrave peak, mocks *h 3 er- ( IEW 326-332) rise, move 3 removing sag- (saig-) saigid (606) approach, seek out *seh 2 ĝ- (520) follow a trail 2 motion scarā- (scar-) scaraid (613) separate *(s)k̂erH- (558) separate 1 removing scartāsceirtid (616) strip, scrape *(s)kerd h - (558) ? cut 1 removing scochī-(scuich-) scuchaid, scuichid (617) moves, starts *skek- (558) move quickly, leap 3 motion sechin composition (619) say *sek u̯ - (526) say 3 communication sedsaidid (604) sit *sed- (513) sit 1 posture <?page no="319"?> 7.2. Multiple preverbs in numbers 319 v erbal root s imple verb (vkg page ) m eaning pie root (l iv 2 page ) m eaning f requency v erb type selā- (sel-) in composition; cf. sel ‘ turn, while‘ silid (621) go, leave, carry, take off turn, a while drops, flows *su̯ei̯- (Vend. S-78) balance, turn 2 motion selbī- (selb-) selbaid denominative from selb has property *selh 1 - (Vend. S-80) take, grab 1 possessing semin composition (624) pour, create, give birth *semH- (531) dig, scoop 4 creation sistā- (siss-) sessaid (< sessam ‘act of standing’; 628) stand *steh 2 - (590) step, stand 3 posture slī- (slí-) in composition (630) gain, win ? *selh 1 - (590; Vend. S-130) take, grab 1 change of possession snī- (sní-) sníïd (633) twist, bind, tie *sneh 1 - (571) spin 4 manner of motion sow- (so-) soïd (635) turn *seu̯h 1 - (538) drive, keep in movement 2 manner of motion swizd- (seth-) in composition (627) blow, inspire *su̯ei̯sd- ( KPV 611; Vend.: S-100) breathe 2 bodily experience <?page no="320"?> 320 7. Multiple preverbs in Old Irish v erbal root s imple verb (vkg page ) m eaning pie root (l iv 2 page ) m eaning f requency v erb type tā- (tá-) suppletive tá- (638) exist, be *steh 2 - (590; Vend. T-2) step, stand 2 existence tēg- (tét-, tiag-) téit (639) go *stei̯g- (593) climb 3 motion tōsī- (tóis-, tuais-) denom. from tó (651) be still *th 2 eu̯s- (642) be still 1 posture Tab. 45: Old Irish verbal roots modified by multiple preverbs 16 16 Verbal roots are sorted as in VKG II (441-658). <?page no="321"?> 7.2. Multiple preverbs in numbers 321 In comparison with the other languages under investigation, Old Irish allows for multiple preverbs in combination with a wider range of verbal roots that also show a wider range of meanings, as is discussed in what follows. Location and motion verbs (e. g. √tēg- ‘go’) are considered to include manner of motion verbs (e. g. √reth- ‘run’), verbs of caused motion (e. g. √ber- ‘bring’), and posture verbs (e. g. √sed- ‘sit’). In Old Irish, only 21 out of 66 verbal bases belong to this group. These data per se suggest an advanced lexicalization, which in turn implies the loss of spatial meanings and the semantic bleaching of Old Irish preverbs. The majority of the remaining verbal roots can be assimilated to motion and location verbs, including the following: (a) transfer verbs and verbs of putting / removing, which are similar to verbs of caused motion (e. g. √ferā- ‘grant, supply’; √em- ‘take’); (b) verbs of existence and of possession, which can be assimilated to location verbs (e. g. √tā- ‘exist’; √selbī- ‘have’); 17 (c) change of state and change of possession verbs, which can be paired with verbs of caused motion via verbs of putting / removing; this closeness is occasionally mirrored by their etymologies (e. g. PIE *leg h - ‘put, place’ > OI r. √lōgī- ‘concede, obtain’ = verb of putting > change of possession verb; PIE *seh 1 - ‘take, grab’ > OI r. √slī- ‘earn’ = verb of removing > change of possession verb); verbs of contact / impact, which can be regarded as a subtype of change of state verbs, in that the impact causes a change of state on a certain entity (e. g. √beg- ‘break’); (d) creation verbs, which can also be associated with change of state verbs, in that the act of creating can be seen as turning a certain entity ( TR ) from non-existence into existence (e. g. √long- ‘cause’); (e) perception verbs, which can be treated as caused motion verbs, in that eyes, as moving entities, can be directed toward or away from certain entities (e. g. √kwis- ‘see’); (f) emission verbs, in which a substance is extracted / goes out of a certain entity ( LM ) (e. g. √mlig- ‘milk’), and communication verbs, in which word or utterances are seen as moving entities going from one speaker to another (e. g. √sech- ‘say’). The remaining verbal bases include measure verbs, verbs indicating bodily processes or mental states / activities, and verbs of helping. As its etymology suggests, the only measure verb contained in Table 45 (i. e. PIE *med- ‘measure, look after’ > √med- ) can be grouped together with perception verbs: the act of measuring can be thought of as the act of looking carefully at a certain entity. 17 The semantic closeness of Location and possession can also be observed from the fact that, among the sources for Possessor expressions, Locations play a prominent role (Narrog 2014: 77, 80). Notably, in Old Irish, the possessive construction is based on a Location expression containing the existential verb attá and the preposition la (literally) ‘beside’ (cf. Nuti 2010). <?page no="322"?> 322 7. Multiple preverbs in Old Irish The two roots expressing bodily processes both mean ‘breathe’: one of them, √anā- , develops the meaning of a posture verb, and comes to mean ‘stay, stay calm’; the other one instead can be seen as an emission verb with breath as a TR going out of a body ( LM ) ( √swizd- ‘blow’). Two verbal bases indicate mental states. The root √acc- ‘be inclined to’ can be regarded as indicating a metaphorical location verb, as is also suggested by its probable etymology: √acc- ‘be inclined to’ possibly go back to PIE *h 2 enk- ‘bend, incline’, whose reconstructed meaning is more concrete (see the discussion contained in KPV 207). In fact, the composite do·futhraccair ‘desire, wish’, is segmented differently by different authors. Different morphological analyses can also result in segmentations including different verbal roots, as shown in (14): (14) The segmentations of do·futhraccair ‘desire, wish’ a. di-fo-√traccb. di-fo-*tre-√acc- VKG II 653 KPV 207 e DIL (dil.ie / 17 715) Milan Glosses database < PIE *trenk- ( LIV 2 649) < PIE *h 2 enk- ( LIV 2 268) According to Pedersen ( VKG II 653), this composite belongs to a root √traccthat goes back to PIE *trenk- ( LIV 2 649), related to Gothic þreihan ‘push, force’. However, this analysis is problematic, as the attested form of the perfect does not match the expected perfect for *trenkin Proto-Irish ( GOI 112; McCone 1996: 124; KPV 208). Thus, Schumacher ( KPV 207-208) suggests further splitting -thracinto -thrand -ac- , in which -thris allegedly the elided form of an additional preverb *tri- ‘through’. The other verb indicating a mental activity, √crādi- ‘torment’, instead can be considered a metaphorical contact / impact verb, though this analysis does not have an etymological confirmation, as the PIE root for this verb is uncertain. The last mental verb, √mani- , indicates the activity of meditating: this verb cannot easily be boiled down to a motion / location verb, though the activity of thinking may be seen as the act of remaining in a certain mental state (that is in a certain metaphorical location). The root √aneg- ‘save, protect’, the only verb of helping in my Old Irish sample, can be assimilated to a transfer verb: saving or protecting is the act of bringing aid, protection, and salvation (accordingly, one of its probable PIE etymologies goes back to *h 1 eĝH- ‘lead’). Generally, multiple preverbs seem to attach frequently to roots that do not only express motion or location events proper (cf. Table 45): <?page no="323"?> 7.2. Multiple preverbs in numbers 323 - √ben- ‘cut, beat’ (x11) , √gabi- ‘grasp’ (x8) → removing; 18 - √ber- ‘bring’ (x9), √ell- ‘put in motion’ (x7), √fēd- ‘bring, lead’ (x6) → caused motion; - √gari- ‘call’ (x7) → communication ; - √kwis- ‘see’ (x8) → perception ; - √org- ‘kill, slay’ (x6) → contact / impact; - √reth- ‘run’ (x7) → motion. 7.2.3. Attested combinations of preverbs In Old Irish, there are as many as 116 combinations of multiple preverbs (more than twice the number of Vedic combinations; cf. Section 4.2.3). These are displayed in Table 46, together with their frequencies, that is, the number of composites instantiating each combination. p reverbs m eanings f requency ad+com toward+with 2 ad+com+uss toward+with+up / off 1 ad+di+en toward+of / from+in(to) 1 ad+uss toward+up / off 1 ad+ro toward+forth 2 ad+to+air toward+to+before / for 1 ad+to+fo toward+to+under 1 air+com before / for+with 2 air+di before / for+of / from 3 air+fo before / for+under 3 air+in before / for+in(to) 1 air+ne before / for+down 1 air+uss before / for+up / off 4 aith+com re-+with 1 aith+ess re-+out of 1 18 The root √gabi-, which gives the simplex verb gaibid, can mean ‘grasp’, but also ‘reach, go’. Cf. the semantic change undergone by the PIE root *sei̯ k- (LIV 2 522) ‘grasp, reach, achieve’ > AG hiknéomai ‘come’. <?page no="324"?> 324 7. Multiple preverbs in Old Irish p reverbs m eanings f requency com+air with+before / for 3 com+di with+of / from 1 com+fo with+under 2 com+imm with+about 1 com+en with+in(to) 1 com+uss with+up / off 4 com+ro with+forth 1 com+to with+to 2 com+to+ad+ro with+to+toward+forth 1 com+to+air with+to+before / for 1 com+to+eter with+to+between 1 com+for+aith with+over+re- 1 di+ess of / from+out of 3 di+fo of / from+under 1 di+fo+treof / from+under+through 1 di+en of / from+in(to) 3 di+uss of+up / off 4 di+ro of / from+forth 2 di+ro+uss of / from+forth+up / off 1 ess+com out of+with 3 ess+com+di out of+with+of 1 ess+ess out of+out of 1 ess+in(de) out of+in(to) 3 ess+ro out of+forth 2 ess+to out of+to 1 eter+di+en between+of / from+in(to) 1 eter+di between+of / from 1 fo+ad under+toward 3 fo+ad+com under+toward+with 1 <?page no="325"?> 7.2. Multiple preverbs in numbers 325 p reverbs m eanings f requency fo+com under+with 2 fo+di under+of / from 1 fo+in(de) under+in(to) 2 fo+in(de)+ad+ro+uss under+in(to)+toward+forth+up / off 1 fo+uss under+up / off 2 fo+to under+to 1 fo+to+ad+ro under+to+toward+forth 1 fo+to+di+com under+to+of / from+with 1 fo+to+ess under+to+out of 1 fo+to+imm+di under+to+about+of / from 1 for+aith over+re- 1 for+com over+with 3 for+di over+of / from 1 for+di+in(de) over+of / from+in(to) 1 for+di+uss over+of / from+up / off 1 frith+ad against+toward 1 frith+to against+to 2 frith+to+air against+to+before / for 1 frith+to+fo against+to+under 2 frith+to+in(de) against+to+in(to) 1 íarm+fo before+under 1 imm+ad about+toward 3 imm+air about+before / for 1 imm+com about+with 1 imm-di about+of / from 1 imm+fo about+under 2 imm+frith about+against 2 imm+imm about+about 2 imm+ro about+forth 1 <?page no="326"?> 326 7. Multiple preverbs in Old Irish p reverbs m eanings f requency imm+to about+to 2 imm+to+imm about+to+about 1 in(de)+ad+ro+uss in(to)+toward+forth+up / off 1 in(de)+com in+with 3 in(de)+uss in(to)+up / off 1 in(de)+to in(to)+to 1 rem+di+en pre-+of / from+in(to) 1 rem+di+ro+uss pre-+of / from+forth+up / off 1 rem+ess pre-+out of 2 rem+ess+in(de) pre-+out of+in(to) 1 rem+fo pre-+under 2 rem+uss pre-+up / off 1 sechmo+in(de) beyond+in(to) 1 to+ad to+toward 6 to+ad+uss to+toward+up / off 2 to+ad+ro to+toward+forth 2 to+ad+to+ad to+toward+to+toward 1 to+air to+before / for 2 to+air+ro to+before / for+forth 1 to+air+in(de) to+before / for+in(to) 2 to+air+uss to+before / for+up / off 1 to+aith to+re- 4 to+di to+of / from 1 to+di+uss to+of / from+up / off 2 to+ess to+out of 5 to+eter to+between 1 to+fo to+under 4 to+fo+com to+under+with 1 to+fo+in(de) to+under+in(to) 1 to+fo+uss to+under+up / off 1 <?page no="327"?> 7.2. Multiple preverbs in numbers 327 p reverbs m eanings f requency to+for to+over 1 to+for+ess to+over-out_of 1 to+for+ad to+over+toward 1 to+íarm+fo to+after+under 1 to+imm to+about 2 to+imm+ad to+about+toward 1 to+imm+di to+about+of / from 1 to+in(de) to+in(to) 4 to+in+ad to+in(to) 1 to+in+com to+in(to)+with 1 to+in(de)+uss to+in(to)+up / off 2 to+uss to+up / off 3 to+ro to+forth 1 Tab. 46: Old Irish combinations of preverbs and their frequency Out of 116 combinations, the vast majority (72) contains two preverbs; 37 contain three preverbs, and only seven as many as four preverbs. These data back up Lewis & Pedersen’s remark (1961[1937]: 267) that “accumulation of preverbs is exceedingly common in Celtic […] but groups of more than three preverbs are rare.” In the Milan and Priscian Glosses , for example, only one composite contains five preverbs, whereas the e DIL attests to three composites with five preverbs (cf. fn. 156 on Rossiter’s 2004 data sample). As shown in Table 46, most combinations (71 out of 115) are only instantiated in a single composite. Many fewer combinations (25 out of 115) are attested in two composites, and 12 combinations in three composites. Six combinations are contained in four composites, and only two combinations are instantiated in five and six composites. Most combinations can be reduced to an actual Old Irish composite by removing preverbs in sequence (Rossiter 2004; McCone 2006: 179), as shown in (15): Musterdatei NFA_Basis_A.dot 310 to+for to+over 1 to+for+ess to+over-out_of 1 to+for+ad to+over+toward 1 to+íarm+fo to+after+under 1 to+imm to+about 2 to+imm+ad to+about+toward 1 to+imm+di to+about+of/ from 1 to+in(de) to+in(to) 4 to+in+ad to+in(to) 1 to+in+com to+in(to)+with 1 to+in(de)+uss to+in(to)+up/ off 2 to+uss to+up/ off 3 to+ro to+forth 1 Tab. 46: Old Irish combinations of preverbs and their frequency Out of 116 combinations, the vast majority (72) contains two preverbs; 37 contain three preverbs, and only seven as many as four preverbs. These data back up Lewis & Pedersen’s remark (1961[1937]: 267) that “accumulation of preverbs is exceedingly common in Celtic […] but groups of more than three preverbs are rare.” In the Milan and Priscian Glosses , for example, only one composite contains five preverbs, whereas the eDIL attests to three composites with five preverbs (cf. fn. 156 on Rossiter’s 2004 data sample). As shown in Table 46, most combinations (71 out of 115) are only instantiated in a single composite. Many fewer combinations (25 out of 115) are attested in two composites, and 12 combinations in three composites. Six combinations are contained in four composites, and only two combinations are instantiated in five and six composites. Most combinations can be reduced to an actual Old Irish composite by removing preverbs in sequence (Rossiter 2004; McCone 2006: 179), as shown in (15): (15) do·immdiben ‘cut away, shorten’ > imm·díben ‘excise, circumcise’ > do· ben ‘cut away’ > benaid accretion reduction <?page no="328"?> 328 7. Multiple preverbs in Old Irish However, the removal of (one of the) preverbs does not always result in an actual composite verb, but rather in a nominal formation that points to the previous existence of the corresponding composite verb (cf. also VGK II 450-658). As suggested by Rossiter (2004: 123) and confirmed by McCone (2006: 180), one such case is Old Irish comarc ‘outcry, invocation’, which arguably functioned as the base for the composites in (16), and in turn relies on the lost verbal composite *con·airc- . (16) Composites based on a lost *con·aircimm·comairc ‘question, ask, inquire of ’ (Ml.27d4, 18a1, 18a3, 20b18, 63c9; Sg.197b10, 27a2, 197b10, 138a4) ad·comairc ‘ask, question’ do·comairc ‘ask, request’ fo·comairc ‘inquire’ for·comairc ‘question, inquire’ fris·comairc ‘ask questions of ’ However, there are exceptions to the principle of reducibility outlined above. In particular, the following composites cannot be reduced: (sorted by root) con·osna (com-uss-√anā-) ‘cease, stop, desist, remain, end in’, fo·fúasna (fo-uss- √anā-) ‘perturb, disturb’, ad·opair (ad-uss-√ber-) ‘sacrifice, offer up’, do·opair (touss-√ber-) ‘take away, deprive, defraud’, ad·cuimtig (ad-com-uss-√ding-) ‘build to, build up’, ar·utaing (air-uss-√ding-) ‘build up, restore, refresh’, con·utaing (com-uss-√ding-) ‘build, construct, build up, embellish’, con·ocaib (com-uss-√gabi-) ‘lift up, raise, exalt’, do·focaib (to-uss-√gabi-) ‘rise’, do·oggell (dē-uss-√gell-) ‘purchase’, remi·uicsed (rem-uss-√gus-) ‘choose beforehand, pre-elect’, ar·osailci (air-uss-√lēcī-) ‘be opened’ , do·airndret (to-air-in(de)-√reth-) ‘run about, roam over’, do·íarmórat (to-íarm-fo-√reth-) ‘follow, come after, goes after’, du·dúrig (to-di-uss-√rig-) ‘be enticed, be excited’, ar·neät (air-ni-√sed-) ‘expect, await’, con·oscaigi (com-uss-√scochī-) ‘move, change, remove, shake, upset’, di·fíuschi (di-uss-√sech-) ‘arouse, excite, call into’, fo·uisim (fo-uss-√sem-) ‘be stored, be put away’, do·fuissim (to-uss-√sem-) ‘bring, bring forth’, and in·otat (in(de)-uss- √tēg-) ‘enter into’. Notably, all mentioned composites belong to one of the three categories of exceptions to reducibility identified by Rossiter (2004: 89, 154-161) and McCone (2006: 181 ff.): (a) composites containing the IP -ne- ‘down’; (b) composites containing the IP -uss- ‘up, out’; (c) composites containing the EP to-. Cases (a)-(b) involve a Path-preverb whose meaning frequently comes to show semantic solidarity with the meaning of the verbal base to which it attaches (e. g. -ne- ‘down’ overlaps with √sed- ‘sit’ in ar·neät (air-ni-√sed-) ‘expect, await’). As I discuss in Section 7.6.3, both Path-meaning and semantic solidarity are good <?page no="329"?> 7.2. Multiple preverbs in numbers 329 motivations for internal placement and for the consequent reanalysis as part of the verbal base. The case (c) is of a different sort: it involves the preverb to- , whose etymology and meaning motivate its non-reducibility. Etymologically, tois probably the outcome of the merging of two Celtic homophonous preverbs: a preposition *to 1 - ‘back, re-’, and a clausal connector *to 2 - (Stifter 2014). The prehistory of toas a clausal connector explains the fact that it strongly selects the exterior position (cf. Section 7.6), as well as its tendency to retain this position even after the addition of further preverbs. As shown by Rossiter (2004: 138), a further preverb occasionally appears to be infixed after rather than prefixed before to- : e. g. do·aithbeir (to-aith-√ber-) ‘take back’, which is possibly based upon the extremely widespread do·beir ‘give’. In addition, as shown by Gillon (1962: 121-122), the semantic contribution brought about by tois bleached in most composites in which it occurs. Consequently, the composites containing toand those lacking it happened to become near synonyms: see, for example, from the Milan and Priscian Glosses , ad·opair (ad-uss-√ber-) ‘sacrifice, offer up’ and do·adbair (to-ad-uss-√ber-) ‘display, show, bring forward, offer’; ad·cumaing (in- (de)-com-√icc-) ‘strike, cut, happen’ and do·ecmaing (to-in(de)-com-√icc-) ‘strike, hit a mark’; ad-muinethar (aith-√mani-) ‘remember, call to mind’ and do·aithmenadar (to-aith-√mani-) ‘call to mind, commemorate, remember’; di·fíuschi (di-uss-√sech-) ‘arouse, excite, call into’ and do·díuschi (to-di-uss-√sech-) ‘awake, arouse’. This might have made the competing composites without toobsolete, and thus particularly prone to disappearing: e. g. the composites do·indnaig (to-in(de)-√aneg-) ‘give, bestow, grant, hand over’, do·aithbig (to-aith-√beg-) ‘dissolve, break up’, do·eipen (to-ess-√ben-) ‘excise, cut (out of, off)’, do·inchain (to-in(de)-√can-) ‘chant, utter’, do·aithchuiredar (to-aith-√cori-) ‘return’, do·aithchren (to-aith-√crina-) ‘buy back, redeem’, do·incai (to-in(de)-ad-√kwis-) ‘look, gaze (at)’, do·foirndea (to-fo-√rindā-) ‘express, signify, denote’, do·intai (to-in- (de)-√sow-) ‘translate, turn back, return’, and do·infet (to-in(de)-√swizd-) ‘blow, breathe, blow on, breathe on’ cannot be reduced to the correspondent composites lacking to- . 19 Within the frame of an overall abundance of verbal composition in Old Irish, preverb iteration is also relatively more widespread than in the other languages under investigation. As expected, preverb iteration usually results in iterative or intensive meanings (cf. Rossiter 2004: 7): 19 Needless to say, one can always find counterexamples to tendencies such as that outlined above: for example, by reducing the composite do·forban (to-for-√ben-) ‘come, arrive, happen to, reach’, one obtains for-√benthat has the entirely different meaning of ‘strike, smite, cut’ (eDIL.ie / 23 375). <?page no="330"?> 330 7. Multiple preverbs in Old Irish (17) a. as·éirig (ess-ess-√rig-) ‘rise again’ (Lat. re-surgere ) (iterative) b. imm·imgaib (imm-imm-√gabi-) ‘go around around’ → ‘avoid’ (intensive) 20 In other composites, however, the meaning of preverb iteration is not so easy to assess: (18) a. imm·timchella (imm-to-imm-√cēllā-) ‘surround’ (Lat., among others, circum-dare, amb-ire ); b. do·atailci (to-ad-to-ad-√lēcī-) ‘fondle, caress, pacify’. In (18)a, the occurrence of the EP immis possibly a calque of the Latin preverbs circumor ambi- , which are sometimes contained in the Latin verbs glossed through imm·timchella (cf. Table 47, which contains all Latin counterparts of Old Irish composites). The semantic contribution brought about by the double repetition of to-adis obscure in (18)b, especially in the light of the much more compositional meaning of do·léici (to-ad-lēcī-) ‘let go, release’. In fact, preverb repetition is occasionally described as a meaningless way to avoid the stress shift whenever the deuterotonic form is required ( GOI 351). One such case is fo·fúasna ‘perturb, disturb’, which alternates with ·fúasna (fo-uss-√anā- ; cf. Table 44). As relevant examples of meaningless preverb repetition, GOI also quotes the mentioned composites imm·imgaib (17)b and imm·timchella (18)a . All in all, preverb iteration appears to constitute a recent phenomenon in Old Irish: it emerges from calquing from Latin; it serves mere morphosyntactic purposes; it indicates concepts that arguably became common after the introduction of Christianity in Ireland, i. e. 5 th century AD (cf. Rossiter 2004: 7). 7.2.4. The Latin counterparts of Old Irish composites Table 47 displays the Latin counterparts of Old Irish multiple preverb composites. Importantly, not all Old Irish composites have an obvious Latin equivalent (158 out of 166 do so). Both the Milan and Priscian Glosses contain annotations of different types: some glosses only consist of a single word and seem to be direct translations of the Latin text; other glosses, however, either contain longer comments or elucidations of the main text or provide background information necessary to understand the main text. In the latter cases, the one-to-one correspondence to Latin can be impossible to draw. 20 In (17)b, the usage of imm- ‘around’ is similar to that of its English and Italian equivalents around and intorno in collocations such as dance around and girarci intorno (literally) ‘turn around’ , both developing the non-compositional meaning of ‘avoid’. <?page no="331"?> 7.2. Multiple preverbs in numbers 331 c omposite m eaning l atin counterpart ( s ) com-uss-√anācease, stop, desist, rest, remain cessare, de-sistere, con-quiescere fo-uss-√anāperturb, disturb per-turbare to-in(de)-√aneggive, bestow, grant, hand over dis-tribuere, con-tribuere imm-com-√arcquestion, ask, inquire of per-contari, in-terrogare air-di-√bādīsubmerge, drown, sink, wreck ex-tinguere to-aith-√begdissolve, break up ab-rogare air-di-√benbe destroyed, cut off inter-ficere fo-in(de)-ad-ro-uss-√benbe relegated, be subjected sub-iacere in(de)-ad-ro-uss-√bendrive out, expel iacere, de-pellere, di-vertere, ex-cludere to-for-√bencome, arrive, happen to, reach venire, per-venire, pro-desse, pro-ficiscere to-ess-√benexcise, cut (out of, off) con-cidere air-di-√bencut off, slay, destroy inter-imere eter-di-√bendestroy inter-imere, per-imere, ex-terminare, inter-ficere imm-di-√benexcise, circumcise ab-scidere, circum-cidere to-fo-√bencut, cut down, cut out, destroy pytissare, suc-cidere, in-cidere, abs-cidere, ex-cidere, truncare ad-to-air-√berbring back, deliver again re-digere ad-uss-√bersacrifice, offer up ob-ferre, con-secrare fo-to-√berplace under sub-ducere <?page no="332"?> 332 7. Multiple preverbs in Old Irish c omposite m eaning l atin counterpart ( s ) frith-to-√berset against, oppose ob-dare imm-to-√bercarry round, surround circum-dare rem-ess-√bersay beforehand, say previously prae-dicere to-ad-uss-√berdisplay, show, bring forward, offer ad-hibere to-air-√berbend, bend down, incline, lower re-digere to-uss-√bertake away, deprive, defraud privare to-ess-√brennflow, trickle, gush ad-fluere fo-ad-√cansing to, accompany in song suc-cinere to-air-√canprophesize, foretell prae-cinere, pro-fiteri, vaticinari, ad-nuntiare to-in(de)-√canchant, utter (spell) in-cantare imm-to-imm-√cēllāsurround cingere, ac-cingere, circum-dare, amb-ire, tegere, vallare to-imm-√cēllāsurround, encompass amb-ire, vallare to-fo-in(de)-√ciddescends discendere rem-ess-√cidstretch forth prae-tendere com-air-√clowstir up, disturb, drive away, agitate ex-agere, agree com-imm-√clowchange cambiare ess-ro-√coildefine, determine de-cernere di-ro-√coindespair of dif-fidere, di-sperare, nutare to-aith-√corireturn re-dire, re-vertere <?page no="333"?> 7.2. Multiple preverbs in numbers 333 c omposite m eaning l atin counterpart ( s ) to-ad-√crádexasperate, provoke ex-asperare, ex-acerbare, asperare to-aith-√crinabuy back, redeem re-dimere com-fo-√dālīshare jointly, divide, apportion participare ad-com-uss-√dingbuild to, build up ad-struere air-uss-√dingbuild up, restore, refresh re-ficere, ad-ficere com-uss-√dingbuild, build up, embellish ad-struere ad-com-√elljoint, unite con-iungere, iungere di-en-√ellturn aside, deviate, bend, decline de-viare, di-versari, de-clinare fo-ad-com-√ellsubjoin sub-icere fo-in(de)-√ellwander, rove evagari sechmo-in(de)-√ellpass by, pass, neglect praeter-ire, oblivisci, omittere, vacuare to-in(de)-uss-√ellgather, collect, assemble colligere, ad-plicare, locare, occupare air-fo-√emaccept, receive, assume, take in-venire, ex-cipere, ac-cipere, sus-cipere ess-to-√fāscīexpress ex-primere air-com-√fēdhinder, prevent, injure nocere, im-pedire ess-in(de)-√fēddeclare, relate, tells, set forth ex-ponere, ex-plicare, ob-serere, ferre, al-ligare, ad-serere, in-ducere, re-ferre, re-plicare, con-serere, am-plicare, ex-primere, per-stringere, ad-ferre, e-licere, de-scribere, re-texere fo-to-di-com-√fēdsubdue sub-ducere <?page no="334"?> 334 7. Multiple preverbs in Old Irish c omposite m eaning l atin counterpart ( s ) rem-ess-in(de)-√fēdtell beforehand, previously, predict prae-fari to-ad-√fēdshow, manifest, set forth, demonstrate o-stendere, o-stentare to-di-√fēdlead down, leads, bring in-ducere, de-ducere com-fo-√ferāprovide com-parare rem-fo-√ferāprepare, provide previously prae-stare air-in(de)-√gabiavoid, flee from re-fugere com-uss-√gabilift up, raise, exalt iactare, sub-levare, ad-tollere, efferre, e-levare ess-in(de)- √gabiexceed, surpass, go beyond ex-cedere, ex-cidere imm-imm-√gabiavoid, shun, evade vitare, e-vitare, declinare, ef-fugere, de-serere to-air-uss-√gabiraise up, produce, cause pro-ferre, e-mergere, promere, de-promere to-ro-√gabicommit, transgress ad-mittere, com-mittere ess-com-√gariproclaim, give notice e-dicere, in-dicere for-com-√garicommand, order im-perare, prae-cipere, mandare in(de)-to-√garicall on, invoke in-vocare to-ad-√garideclare, tell pro-ferre to-air-in(de)-√garipromise polliceri, pro-mittere, de-signare imm-ad-√glādīconverse together ad-loquere fo-di-√gnībe sufficient suf-ficere air-ne-√guidpray orare <?page no="335"?> 7.2. Multiple preverbs in numbers 335 c omposite m eaning l atin counterpart ( s ) rem-uss-√guschoose beforehand, pre-elect prae-obtare com-ro-√iccmeet, encounter, join ac-cedere com-to-ad-ro-√iccconfer con-ferre fo-to-ad-ro-√icc(ī)substitute, supply sub-rogare for-com-√icchappen, occur, be made factum esse, posse imm-air-√iccbe appropriate to con-venire, com-petere in(de)-com-√iccstrike, cut, happen pulsare, con-tingere to-ad-ro-√icc(ī)cause, effect, bring about, bring ferre, af-ferre, con-ferre, de-ferre, in-ferre, per-ferre, prae-ferre to-in(de)-com-√iccstrike, hit a mark ac-cidere ad-di-en-√kwisregard, look at re-spicere dē-en-√kwislook at, behold, see intro-spicere, re-spicere, e-spicere, sentire eter-di-en-√kwisintrospect intro-spicere imm-ad-√kwislooks after, regard, examine, consider con-siderare rem-di-en-√kwisprovide for, force con-sulere, prae-videre to-in(de)-ad-√kwislook, gaze (at) per-spicere frith-ad-√kwislook forward to, expect, hope ex-spectare, prae-stolari, o-perire, sperare air-uss-√lēcīopen pandere, aperire, re-serare, patere com-air-√lēcīpermit, allow, let go di-mittere, ad-mittere, per-mittere to-ad-to-ad-√lēcīfondle, caress, pacify fovere, con-fovere, de-lenire <?page no="336"?> 336 7. Multiple preverbs in Old Irish c omposite m eaning l atin counterpart ( s ) to-fo-uss-√lēcīloosen, relax solvere, re-solvere ess-com-di-√logīdisrupt di-rumpere fo-ad-√logīlay low, prostrate, throw down con-sternare imm-fo-√longīcause, produce, make, affect ef-ficere, ex-inanire, nasci, ap-parire rem-fo-√longībear beforehand, anticipate anti-cipare, prae-venire ess-com-√ludepart, escape, set out, start pro-ficiscere com-fo-aith-√manibe commemorated com-memorare di-ro-√maniforget oblivisci for-aith-√manicall to mind, commemorate memini to-aith-√manicall to mind, commemorate commonere imm-ro-√medisin, transgress delinqui, peccare to-en-uss-√mligpromulgate, proclaim, announce elicet, promulgare di-fo-√nigwash, wash away luere, ab-luere, lavare to-imm-ad-√nocommand, enjoin mittere com-to-√orgstrike, pound con-terere, con-trire di-ess-√orgsmite, slay con-fundere frith-to-fo-√orgstrike ob-tondere to-fo-√orgcrush, grin, beat, pound at-terere, de-terere, pro-terere, triturare, pulsare to-imm-√orgpress, compress, chastise castigare, arctare, coartare, angere <?page no="337"?> 7.2. Multiple preverbs in numbers 337 c omposite m eaning l atin counterpart ( s ) for-com-√owkeep, preserve, retain servare, manere com-to-eter-√rethcomprise, comprehend com-prehendere fo-to-imm-di-√rethsubminister, fumigate suf-fire for-di-√rethpass through, go over in-lustrare, lustrare for-di-uss-√rethremain, remain over super-esse to-air-in(de)-√rethrun about, roam over per-errare to-eter-√rethcomprehend, grasp, overtake in-cludere, com-prehendere aith-ess-√rigrepeat, reiterate, change, emend emendare di-ess-√rig abandon, forsake nudare, de-serere ess-ess-√rigrise again, arise re-surgere to-di-uss-√rigbe enticed, be excited e-licere frith-to-fo-√rindāmark off, trace (a limit) ob-signare ad-to-fo-√rindā-* prick again re-pungere to-fo-√rindāexpress, signify significare, distinguere com-di-√sagask, seek, demand quaerere íarm-fo-√sagseek after, inquire about quaerere, per-quirere, in-quirere imm-to-√scarāstruggle together luctare di-uss-√scart(ā)remove, put aside ex-ponere com-uss-√scochīmove, change, removes, shake per-turbare, com-movere, com-mutare <?page no="338"?> 338 7. Multiple preverbs in Old Irish c omposite m eaning l atin counterpart ( s ) di-ro-uss-√scochīstand forth, distinguish oneself ante-cedere, ante-stare, e-minere, prae-esse, prae-stare rem-di-ro-uss-√scochībe pre-distinguished ante-ferre in(de)-com-√sechsignify beforehand in-dicare, in-tellegere, significare to-di-uss-√sechawake, arouse movere air-ni-√sedexpect, await ex-spectare fo-com-√selātake away, remove sub-ducere, ex-agere to-fo-com-√selā-* bring, drag over, draw over tractare to-ad-√selbīassign, ascribe ad-scribere, ad-signare, ad-dicere, in-dicere fo-to-ess-√sempour down, pour out suf-fundere to-ess-√semshed, pour out re-fundere, dis-pergere, con-cremare, fundere, ef-fundere to-uss-√sembring, bring forth (offspring) creare, germinare, parere, con-ditari, aedere to-air-√sistāstand, stay, remain ad-stare, stare frith-to-air-√sistāmake a stand against, oppose ad-stare, ob-sistere ad-ro-√slīdeserve, be entitled to merere, de-merere, pro-merere, bene-merere ad-com-√snīmake for, strive, seek ex-petere, petere ess-ro-√snīescape e-vadere fo-com-√snī steal, snatch away sub-ripere imm-frith-√snīcontend, dispute, disagree, gainsay in-fitari <?page no="339"?> 7.2. Multiple preverbs in numbers 339 c omposite m eaning l atin counterpart ( s ) com-to-√sowturn, convert, change con-vertere to-in(de)-√sowtranslate, turn back, return trans-ferre frith-to-in(de)-√swizdblow against ex-sufflare to-in(de)-√swizdblow, breathe, blow on, breathe on flare en-com-√tāget, obtain, procure petrare, mereri, im-petrare, con-sequi, ad-ipisci to-ess-√tābe absent, be lacking, be wanting ab-esse, de-esse com-en-√tēggo with, accompany, agree with con-stare frith-to-√tēgcome against, be at variance with contra-venire, ad-versare, ob-ire, op-ponere in(de)-uss-√tēgenter into in-ruere, in-ire, of-fendere, in-currere fo-in(de)-√tōsīunderstand sub-audire Tab. 47: The Latin counterparts of Old Irish composites <?page no="340"?> 340 7. Multiple preverbs in Old Irish Out of 158 Old Irish composites with a Latin counterpart, 77 contain one or more elements that are calqued from Latin. 21 As shown by the bold in Table 47, the EPs are frequently calques from Latin: thus, out of 77 composites, 62 contain an EP ~ EP calque. By contrast, there are only 10 cases in which the interior preverbs are copied from Latin. Rather, the reduced composite resulting from the deletion of the EP usually corresponds to a Latin simplex verb (42 / 62 composites), as exemplified in (19): (19) OI r. imm·tabair (imm-to-√ber-) ~ Lat. circum-dare ‘carries around’ (Ml.41c2). OI r. do·beir (to-√ber-) ~ Lat. dare ‘give’ (e. g. Ml.25b12; Sg.163a3, 209b26). In Ml.41c2, the influence of the Latin circum-dare even overrides the tendency of toto occur as EP (cf. Sections 7.2.3 and 7.6). The clearest Latin calques, underlined in bold in Table 47, follow: OIr. air- ~ Lat. pro- ‘with’; OI r. aith- ~ Lat. re- ‘with’; OI r. com- ~ Lat. con- ‘with’; OI r. eter- ~ Lat. inter- ‘between, among’; OI r. fo- ~ Lat. sub- ‘under’; OI r. frith- ~ Lat. ob-, contra-, ad- ‘against’; OI r. imm- ~ Lat. circum-, ambi- ‘about’; OI r. rem- ~ Lat. ante-, prae- ‘before, pre-’; OI r. sechmo- ~ Lat. praeter- ‘with’ . Less regular correspondences to Latin can also be identified, such as OI r. ad- ~ Lat. ad- ; OI r. di- ~ Lat. ex- , de- , ab- (a single Old Irish Source-preverb corresponds to three different Latin Source-preverbs; cf. also below OI r. ess- ) ; OI r. for- ~ Lat. super- ; OI r. ess- ~ Lat. ex- , de- ; OI r. in- (de)- ~ Lat. in- , ob- ; and OI r. to- ~ Lat. ad- , per- , pro- (cf. also Rossiter 2004: 21-97) . Interestingly, explicit evidence for calquing from Latin can occasionally be found in the glosses (this is similar to what Friedman & Joseph forthc. 2019: ch. 3 call “agentivity in contact-induced phenomena”). In (20), the meaning of a multiple preverb composite, used to translate the Latin corresponding composite intro-spicio , is obscure and elucidated by means of a reduced composite and an adverb equivalent to the EP : specifically, etar·décai ‘introspect’ is glossed by do·ecai ‘look at’ and ind=inmedonach ‘internally’. (20) .i. etir décai i. doécai i. e. inter-look_at. prs .3 sg i. e. look_at. prs .3 sg ind=inmedonach art . dat =internal. dat ‘i. e. he intro-spects, i. e. he looks internally’ (Ml.61a8) 21 The terms “calque” and “calquing” are used in a broad sense: cases in which there is a correspondence between form and meaning (“calques” proper) and cases in which only form correponds (“borrowing via copying”, “adoption”, “mirroring”; cf. Friedman & Joseph forthc. 2019) are both included in the count. <?page no="341"?> 7.2. Multiple preverbs in numbers 341 Calquing from Latin is by no mean mechanistic. To begin with, there are exceptions to the general principle that only Old Irish EP s are calqued from Latin. In some composites, a full sequence of Old Irish preverbs corresponds to a single Latin preverb. A case in point is do·róscai (di-ro-uss-scochī-) (literally) ‘fromforth-up-move’ → ‘stand forth, distinguish oneself ’, in which the combination di-ro-usstranslates the Latin preverbs ante-, e(x)-, or prae- (Latin equivalents for this composite are Lat. ante-cedere, ante-stare, e-minere, prae-esse, and prae-stare ) (for the examples containing preverb iteration, cf. Section 7.2.3). In other glosses, an Old Irish preverb corresponds to a Latin adverb, as shown in (21). Interestingly, the correspondence between the EP immand mutuo is also paralleled by the direct equivalence of the Old Irish IP ac-(ad-) with Latin ad- : 22 (21) Old Irish EP imm- ‘about, mutually’ ~ Latin mutuo ‘mutually’ OI r. immu-s-ac-aldat (imm-ad-√glādī-) p -3 pl . acc p -converse. prs .3 pl Lat. mutuo sé ad-locuntur mutually refl . acc to-speak. prs .3 pl ‘They converse together.’ (Ml.131c19) With another composite, calquing from Latin affects the verbal base but not the usage of preverbs, which is divergent (cf. KPV 487): (22) OI r. do·inmlig (to-en-uss-√mlig-) (literally) ‘to-in(to)-up- √ milk’ Lat. pro-mulgare ‘forth-milk’ PIE *h 2 melĝ- ‘milk’ ( LIV 2 279) ‘promulgate, proclaim, announce’ In (22), both Old Irish √mligand Lat. √mulggo back to the same PIE verbal root and are employed to make up a non-compositional composite with the meaning of ‘promulgate, proclaim, announce’. The preverbs attached to the verbal bases, however, are different in Old Irish and in Latin. Moreover, even the EP s that have been classified as clear calques from Latin do not always have an obvious Latin counterpart. For example, in the composites con·éitet (com-en-√tēg-) ‘go with, accompany, agree with’ and imm·tascra (imm-to-√scarā-) ‘struggle together’, the order of preverbs is not the usual one (in particular, it does not conform to McCone’s hierarchy of preverb ordering, 22 The Old Irish and Latin verbs in fact show a further similarity: OIr. ad·gládathar ‘converse together’ is usually a deponent verb (though, in (21) above, it takes active inflection), as is its Latin equivalent ad-loquor ‘converse’. <?page no="342"?> 342 7. Multiple preverbs in Old Irish on which see Section 7.6). Thus, one might suspect that the EP s immand comare subsequent additions, given the fact that these Old Irish preverbs frequently correspond to Latin circum-/ ambiand con- . This is, however, not the case: there are no Latin counterparts for con·éitet ‘agree with’ in the Priscian Glosses (Sg.197b17, 203a22), in which this composite is employed to elucidate, and not to translate, the main Latin text. Likewise, the Latin equivalent for imm·tascra ‘struggle together’, luctare ‘wrestle, struggle’, contains no preverb. 7.3. The form of composites 7.3.1. The allomorphy of Old Irish preverbs The form of Old Irish preverbs undergoes considerable variation, depending on whether they occur before, under, or after the accent. As pointed out by Anderson (2016: 210 ff.), the phonological content of preverbs occurring in pretonic position (i. e. before the so-called “juncture”) generally seem to be impoverished compared to the same preverbs occurring under the accent. As a result, a number of preverbs merge together when they occur in pretonic position. For example, the preverbs ad- ‘to, toward’, aith- ‘re-, ex-’, in(d)- ‘in(to)’, ess- ‘out of ’’, and uss- ‘up, off’ all can become atin prenuclear position before infixed pronouns ( GOI 495), as shown in Table 48, Table 49, and Table 50, which collect all the attested allomorphs for the preverbs ad- ‘to, toward’ , aith- ‘re-, ex-’, in(de)- ‘in, in(to)’ , es(s)- ‘out of, off’ , and us(s)- ‘up, off’ . Cit. Prenuclear Nuclear Gloss Position Orthography Phonology Notes _Ø, _φ ·ad- ·Øaðad ad·, at- _R ·á- ·ØaØ- ‘to, towards’ _C ·a- ·Øa g aith ad·, at- _all ·a(i)th- ·Øa ~ l - ‘re-, ex-’ Examples adad·rími (Wb.14d2) ni·áirmi (Wb.13d17) aithad·geúin (Wb.12c13) ni-n·aithgeuin (Ml.52x00) Tab. 48: Allomorphy of Old Irish preverbs: adand aith- (from Anderson 2016: 220) <?page no="343"?> 7.3. The form of composites 343 Cit. Prenuclear Nuclear Gloss Position Orthography Phonology Notes _t, _c ·é- ·ØʹaØʹ n in in·, at- _D, _ Ø ·i- ·Øʹə n - ‘in, into’ _R, _s ·e- ·Øʹa g - _T ·in- ·Øʹənʹ l ind in·, at- _φ, _s ·ind-, ·int- ·Øʹənʹdʹ l - ‘in, into’ _ Ø, _R ·ind- ·Øʹənʹdʹ l - Examples in(de)in·túaisi (Wb.13a11) ellachtae (Ml.84a6) Tab. 49: Allomorphy of Old Irish preverbs: inand ind- (from Anderson 2016: 218) 23 Cit. Prenuclear Nuclear Gloss Position Orthography Phonology Notes _ Ø ·es- ·Øʹases as·, at- _R ·é- ·ØʹaØʹ- ‘out of, off’ _C ·e- ·Øʹa g - _? ·as- ·Øasrare _ Ø ·os-, ·us- ·Ø g əs g us as·, at- _R ·ó-, ·úa-, ú- ·Ø g əØ g - ‘off’ _C ·o-, ·u-, ·Ø g ə g - Examples essas·beir (Wb.4d23) ni·epir (Wb. 25d4) us- (no relevant examples) con·úala Tab. 50: Allomorphy of Old Irish preverbs: es(s)and us(s)- (Anderson 2016: 221) As explained by McCone (1997: 4-5), pretonic position prevents a number of sandhi phenomena usually occurring at the boundaries between two preverbs, or between a preverb and a verb base. For example, in pretonic position, (a) vowels are not elided (e. g. to-air-√can- ‘foretell’ becomes -taircheta : prf .3 pl in prototonic forms, but remains du·aircet : pret .3 sg in deuterotonic ones); (b) consonant clusters are not assimilated or eliminated (e. g. ad-ro-√slī- ‘deserve, be entitled 23 According to Thurneysen (GOI 518 ff.), inand indbelong to a single lemma (contra VGK 451 ff.). <?page no="344"?> 344 7. Multiple preverbs in Old Irish to’ becomes -áirilli in prototonic form, but remains ad·roilli in deuterotonic form). By contrast, in pretonic position, preverbs undergo a number of changes that generally characterize Old Irish proclitics (McCone 1997: 5): for example, (a) the voicing of a dental in contact with a proclitic vowel (e. g. to-ess-√sem- ‘shed, pour out’ gives deuterotonic do·eissim , but prototonic -tessim ); (b) the tendency of retracting the articulation of proclitic vowels (e. g. u > o , cf. com-di-√sag- ‘ask, seek, demand’ gives deuterotonic con·dieig and prototonic -cuindig ). The switch to the prototonic forms also causes a number of changes that are triggered by the loss of stress undergone by the IP or by the verb base. Such changes include: (a) the shortening of long vowels in unstressed syllables (e. g. in(d)-√fēd- ‘tell, relate, make known’ gives deuterotonic in·fét, but prototonic -indet ), (b) a divergent syncope due to the adding of an extra preverb and the consequent resyllabification (e. g. com-uss-√anā- ‘cease, stop, desist, remain, end in’ gives deuterotonic con·os*na , but prototonic -cum*sana, in which the asterisk indicates the position of the syncope). Syncope in turn activates a number of further sandhi effects: for example, when it produces consonant clusters, the quality of those consonants undergoes assimilation. The sandhi effects, such as those described above, produce a wide range of allomorphs for Old Irish preverbs. Such allomorphs are thoroughly described in GOI (495 ff.) and investigated by Anderson (2016: 210 ff.), who provides tables similar to Table 48, Table 49, and Table 50 for the full catalog of Old Irish preverbs. 24 Importantly, these complex sandhi phenomena can obscure the morphological segmentation of composites, as anticipated in my comments concerning Table 44. For example, the composite in·árban ‘drive out, expel’ is analyzed differently by different authors, as shown in (23): (23) The segmentation of the composite in·árban ‘drive out, expel’ in(de)-air-uss-√ben- ‘in-before-up-cut’ ( VKG II 463; Anderson 2016) in(de)-ad-ro-uss-√ben- ‘in-to-forth-up-cut’ ( KPV 227; Milan and Priscian Glosses databases) 24 A number of apparent changes between deuterotonic and prototonic forms are merely orthographical (McCone 1997: 8; cf. also McCone’s useful Appendix on Old Irish spelling rules: McCone 1997: 267 ff.). <?page no="345"?> 7.3. The form of composites 345 7.3.2. Augment and preverbs As in other ancient Indo-European languages, Old Irish lexical preverbs interact with other pieces of preverbal morphology. Differently from the other languages, however, in which the augment goes back to an anaphoric / deictic particle (cf. e. g. LIPP II 179, and references therein), the Old Irish so-called “augments” or “temporal preverbs” ( ro- , rarely adand com- , and very rarely other preverbs; cf. Section 7.1.2.3) used to function only as lexical preverbs at a preceding stage, and as such used to be subject to the same positional constraints as lexical preverbs (McCone 2006: 204 ff.). The most widespread among Old Irish augments is the preverb ro-, which can occur either in a fixed or movable position. The so-called fixed rois usually placed directly before the verbal root ( GOI 339), as shown in (24). The augments less frequent than ro- , i. e. adand com- , also occur in this position ( GOI 344ff.), as exemplified in (25). The augments occupy this position especially with strong verbs. (24) ma-du-gneu inna-hui [huili] if-of-do. sbjv . prs .1 sg art . acc . pl indf . acc . pl remi-æ-r-burt fore-out_ofaug -say. prf .1 sg ‘If I would do everything that I have said previously.’ (Ml.23c24) (25) [du]-da-im-chom-arr di-a-chomallad to-3 pl . acc -aboutaug -press. sbjv . prs .1 sg to-3 sg . gen -fulfilling. dat tri-fochaidi 7 ingraimmen through-trial. acc . pl and persecution. acc . pl ‘… who would constrain them to its fulfillment through tribulations and persecutions.’ (Ml.77a12) With weak verbs, instead, the augment rois movable, in that it changes its positioning within the composite so as to invariably occur after the preverb or particle in pretonic position ( GOI 440), as shown in (26). In (26)a-b, the same composite occurs, that is, con·osna ( com-uss-√anā- ) ‘cease, stop, desist, remain, end in’; in (26)a, rois placed after the EP com- ‘with’, whereas in (26)b it is positioned more externally than the EP , after the negative particle ní . (26) a. The augment roafter the EP com- <?page no="346"?> 346 7. Multiple preverbs in Old Irish co-ro-s-an [conrosan] withaug -up-stop. prf .3 sg ‘That is stopped.’ (Ml.113c5) b. The augment roafter the negation and before the EP comní rú-chum-s-an-us-sa neg aug p p -stopprf .1 sg em .1 sg ‘I did not stop’ (Ml.94b14) The position of fixed ro- , which is undoubtedly the older of the two patterns (GOI 340; Lewis & Pedersen 1961[1937]: 252), can be explained as follows. A first motivation for its placement is connected with the interaction between the preverbal origin of these augments and the process of accretion (cf. Sections 7.2.3, 7.4.3, and 7.6.1) that characterizes the growth of multiple preverb composites in Old Irish. In addition, McCone (2006: 207) suggests that a relevant role might have been played by “the possibility of attaching a preverb such as telic roto certain others such as di- ‘from’ or ess- ‘out’ in order to bring out the completed nature of the action implied.” McCone backs up this suggestion based on the evidence that the sequences *eks-roand *di-roare known both in Old Irish nominal and verbal formations. Thus, they appear to make up stable pairs of preverbs (i. e. double preverbs) that could be added at once to a simplex verb. McCone (2006) also argues that even the original fixed rocould be displaced from the position immediately preceding the verbal base under certain conditions, specifically in the presence of a restricted number of IP s, including -ne- ‘down’, and -uss- ‘up, off’, as exemplified in (27): (27) a. The IP -ne- ‘down’ in ar·neät ‘expect, await’ ci-d ar-rú-n-eid even-be. sbjv . prs .3 sg p aug -down-expect. prf .3 sg ‘he even expected’ (Ml.68a6) b. The IP -uss- ‘up, off’ in con·oscaigi ‘move, change’ com-ro-[o]s-caigis-siu p aug -up-move. prf .2 sg em .2 sg ‘you have moved’ (Ml.21d7) The relative positioning of the grammaticalized augment roand the IP s -ne- ‘down’ and -uss- ‘up, off’ supports the hypothesis that -neand -ussshow an <?page no="347"?> 7.4. The semantics of multiple preverbs 347 outstanding lexical intergration with the verbal bases onto which they attach. In the same vein, Lewis & Pedersen (1961[1937]: 252) observe that, in compounds such as those in (27), the last (i. e. the innermost) preverb “had become inseparable from the verb.” 7.4. The semantics of multiple preverbs 7.4.1. Preverbs with spatial, abstract, and actional meanings As discussed in Sections 7.1.2.3 and 7.3.2, Old Irish possesses a number of fully grammaticalized preverbs, which either have perfectivizing function (i. e. ro-, ad-, com- ), or work as hosts under certain morphosyntactic conditions (i. e. no- ). Apart from these grammatical functions, Old Irish preverbs modify the verbs onto which they attach at a lexical level. This is also true for the preverbs that developed into perfective markers, as shown in (28)-(30). 25 For example, the paradigmaticized perfective preverb roonly rarely brings about lexical contributions to the verb; however, lexical contributions are far more frequently produced by the non-paradigmaticized perfective markers adand com- (cf. further the spatial meanings associated to essin examples (31) and (32)) . (28) Lexical ro-: do·róscai (di-ro-uss-√scōchī-) ‘stand forth’ (Lat. prae-stare ) duróscai stand_forth: prs .3 sg ‘It stands forth.’ (Ml.113a9) (29) Lexical ad-: imm·accaldathar (imm-ad-√glādī-) ‘address each other’ immu-s-ac-aldat (imm-ad-√glādī-) p -3 pl . acc p .converse. prs .3 pl ‘They converse together.’ (Ml.131c19 = (21)) 25 This is not the case for no-, which is never used as a lexical preverb. Its origins explain its behavior. The preverb nodoes not belong to the so-called category of Indo-European advs prevs adps (in Cuzzolin et al.’s 2006 terms). Instead, it goes back to the PIE temporal adverb *nú ‘now’, which shows reflexes also functioning as sentence connectors, notably MW neu and Hitt. nu (GOI 348; Lewis & Pedersen 1961[1937]: 259) . <?page no="348"?> 348 7. Multiple preverbs in Old Irish (30) Lexical usage of com- : com-en-√tēg- ‘go with, agree with’ (Lat. con-stare ) dús imbed [do]=duaid ɔeitsitis fa whether be. sbjv . pst .3 sg to= D . dat go_with. sbjv . pst .3 pl or di=a=naimtib to= poss .3 sg =enemy. dat . pl ‘Whether they should favor David or his enemies.’ (Ml.87c4) In (28), the IP romeans ‘forth’ within the composite di-ro-uss-√scōchī- (literally) ‘of-forth-up-move’, which develops the non-spatial meaning of standing forth, being pre-eminent. The entire combination of the Old Irish preverbs di-ro-usscorresponds to Latin prae- ‘pre-, fore-’ (cf. Section 7.2.4). In (29), the IP ad- ‘to, toward’ introduces the Addressee of a communication verb (the root √glādīis exclusively employed with adoccurring as IP , most likely due to the semantic solidarity between the preverb and the verbal base). In (30), the EP comretains its basic meaning of togetherness, whereas the whole composite develops the non-compositional meaning of ‘agreeing with’ from the basic meaning of ‘going with’. Preverbs can modify the verb, adding different types of meanings: (a) spatial; (b) abstract, or (c) actional. A number of preverbs such as ess- ‘out of ’ and fo- ‘under’ both retain their spatial meanings (31) and gain new but still spatial semantics (32). (31) Basic meanings of ess- ‘out of ’ and fo- ‘under’ a. is-áilgen do·neprinn be. prs .3 sg -mild. nom flow. prs .3 sg ‘Gently it flows’ (Sg.145a4) b. lase fu=tabair when under-place. prs .3 sg ‘When he places under …’ (Ml.40d2) (32) Non-basic spatial usage of ess- ‘out of ’ and fo- ‘under’ a. ní=æscomlai neg =go.away. prs .3 sg ‘It does not go away’ (Sg.3a6) <?page no="349"?> 7.4. The semantics of multiple preverbs 349 b. ní=ɔfitetar a-rrig foragabsat neg =know. pret .3 pl poss .3 pl -king. acc leave_behind. prf .3 pl di=a=nǽs of= poss .3 pl =track. dat ‘They do not know their king whom they had left behind them.’ (Ml.95a12) In (31), the composites do·eprainn (to-ess-√brenn-) ‘flow, trickle, gush’ and fo·tabair (fo-to-√ber-) ‘place under’ contain the IP essand the EP fo- , retaining their original spatial semantics. In (32), instead, the same preverbs develop the new spatial meanings of ‘away from’ and ‘behind’ in the composites as·comlai (ess-com-√lu-) ‘depart, escape, set out, start’, and fo·acaib (fo-ad-√gabi-) ‘leave behind’. The preverb essloses its elative component in favor of a more generic ablative value; the preverb focomes to mean ‘behind’ based on the following frequent analogical equation: ‘ above ’ : ‘ beneath ’ = ‘ before ’ : ‘ behind ’ (cf. Luraghi 2003: 226 on the cognate AG preposition hupó ‘under’; see also Section 5.4). Note further that essshows spatial semantics both as an EP and as an IP . This is a first suggestion that neither lexical nor actional meanings are associated to a specific positioning with respect to the verbal base. Old Irish preverbs also gain various types of abstract but still lexical meanings more or less directly connected with their basic meaning. For example, both the preverbs airand aithcan mean ‘again’, as shown by examples (33): (33) Meaning ‘again’ expressed by air- ‘before, for’ and aith- ‘re-, ex-’ a. ar·utaing (air-uss-√ding-) ‘build up, re-store, re-fresh’ arutaing refresh. prs .3 sg ‘Which refreshes.’ (Ml.64c20; Lat. re-ficere ) b. ad·éirrig (aith-ess-√rig-) ‘re-peat, re-iterate, change, emend’ is-airi aderrig-som be. prs .3 sg -for.3 sg . n repeat. prs .3 sg em .3 sg . m / n for=sa=nimchomarc fo=di on= art . acc . n =question. acc under=two. acc . du . f ‘It is therefore that he repeats the question twice.’ (Ml.46a21; Lat. re-petitio ) How do these preverbs come to mean ‘again’? The basic meaning of airis ‘before’: going back before an event can carry the implication of repeating such <?page no="350"?> 350 7. Multiple preverbs in Old Irish an event. The preverb aith- , instead, allegedly goes back to PIE *áto, áti ‘behind, again, away’ ( LIPP II 94 ff.): in parallel to what has been outlined for air- , going behind an event can bear the implication of going before it , and thus repeating such an event. These examples show that different preverbs can express quasi-equivalent meanings: thus, the occurrence of one preverb or another in a certain composite is an idiosyncratic lexical fact. Other preverbs with abstract meanings are exemplified in (34) and (35): (34) immethascrat struggle_together. prs .3 pl ‘who wrestle’ (Ml.118c11; Lat. luctantium : ptcp . prs . gen . pl ) (35) fothonsnát [fochosnat] steal. prs .3 pl ‘which steal’ (Ml.43a8; Lat. sub-ripiunt : prs .3 pl ) In (34), in the composite imm·tascra (imm-to-√scarā-) ‘struggle together’, the EP imm- ‘with’ (Comitative) develops a reciprocal meaning: if two humans are fighting together, they are likely to be fighting against one another (cf. also example (5)). In (35), fo- ‘under’ means ‘secretly’ on account of the following semantic shift: under > beneath > behind > invisibly > secretly . In addition, multiple preverbs of Old Irish can bring about various kinds of actional modifications: (a) ingressive, e. g. con·osna (com-uss-√anā-) ‘cease, stop, desist, remain, end in’ (36); (b) resultative, e. g. ar·foím (air-fo-√em-) ‘accept, receive, assume, take’ (37); (c) intensive, e. g. ad·cuimben (aith-com-√ben-) ‘cut, strike, wound, lacerate’ (38); (d) telic, e. g. etar·diben (eter-di-√ben-) ‘destroy’ (39). (36) Ingressive comin=ru-chumsan ptc . int = aug -cease. prs .3 sg ‘Has it ceased? ’ (Ml.32d26; Lat. con-quiescere ‘to become quiet’) (37) Resultative airacht a-frescastae .i. ni frisaiccai but art . acc . n -hoped. acc . n i. e. indf . acc . n hope. prs .2 sg 7 arafoimi iarum and receive. prs .2 sg after.3 sg . dat . n ‘But the thing hoped, i. e. something that you hope for and that you receive afterward.’ (Ml.68a8; cf. Lat. sus-cipere ‘take up’) <?page no="351"?> 7.4. The semantics of multiple preverbs 351 (38) Intensive aithand comhuilliu adcumnet indatae chlaidib more_greatly wound. prs .3 pl than sword. nom . pl ‘(It is) more greatly that they wound than swords.’ (Ml.77a1) (39) Telic eterand dico=etar=dam-dibitis-se in_order_that= p =1 sg . acc -destroy. sbjv . pst .3 pl em .1 sg ‘In order that they might destroy me.’ (Ml.54d14; cf. Lat. inter-ficere ) Neither lexical nor actional meanings are associated with a specific position with respect to the verbal base. An IP can have lexical or actional meanings: in (31), for example, the IP esshas a lexical spatial meaning; in (38)-(39), the IP s comand dibring about actional meanings, which are then reinforced by the addition of a further preverb, in some cases such as (39), based on a calque from Latin. In parallel, the above examples also show that EP s can carry about lexical (cf. (31)-(35)) and actional semantic contributions as well (cf. (36)-(37)). It is very difficult to find Old Irish composites in which two or more preverbs retain clearly detectable spatial meanings. Far more frequently, two (or more) preverbs with a similar spatial semantics attach onto the same verbal base. Otherwise, the meaning of (one of the) preverb(s) show(s) semantic solidarity with that of the modified verbal stem. In other words, Old Irish composites frequently show some kind of semantic redundancy (Section 7.4.3). This situation is arguably due to the high degree of lexicalization shown by Old Irish preverbs and to the process of accretion underlying multiple preverb composites, whereby a further preverb is attached to an already lexicalized composite (cf. Sections 7.2.3 and 7.6). However, a consistent group of composites containing two spatial specifications can be identified. In these formations, an interior toadds a deictic specification to a motion verb, which is then specified by a further spatial preverb, added more externally than topossibly resulting from a calque from Latin (cf. also (31)): (40) fris·taít (frith-to-√tēg-) ‘come against, be at variance with, oppose’ cf. Lat. contra-venire, ad-versare, ob-ire, op-ponere > do·tét ‘come’, that is, (literally) ‘go back’ > téit ‘go’ <?page no="352"?> 352 7. Multiple preverbs in Old Irish In (40), the addition of toto a verb of going results in a verb of coming. According to Stifter (2014: 238-239), this reversative meaning of tois connected with its ‘back’ meaning, shown by several Old Irish composites such as do·intai ‘turn back, return’, do·rá ‘row back’, and do·reith ‘run back’. By contrast, composites containing two actional or two lexical, but non-spatial, preverbs can be detected. For composites containing two actional preverbs, see examples (38)-(39). For a good example of a composite containing two non-spatial lexical preverbs, see (41): (41) Two non-spatial lexical preverbs: remi·epir (rem-ess-√ber-) ‘say beforehand, say previously’ (Lat. prae-dicere ) > as·beir (ess-√ber-) ‘say to, speak’ > beirid ‘carry, bring’ As emerges from the process of recomposition displayed in (41), the EP remis a later addition, probably modelled on Latin prae-, to an already non-compositional composite, that is, as·beir (ess-√ber-) ‘say to, speak’. The composite as·beir in turn contains the preverb ess- ‘out of ’. This preverb arguably describes the metaphorical motion performed by words or utterances out of speakers’ mouths / bodies, which are conceptualized as containers (on the Container-metaphor, cf. Sections 2.1.2 and 3.1.1). 26 7.4.2. Same preverbs, different meanings As is implied in the discussion of Section 7.4.1, Old Irish preverbs are polysemous elements. Some of them, specifically ro-, ad-, and com-, instantiate a twofold path of development: on the one hand, they underwent grammaticalization into bounder perfectives; on the other hand, they gained new lexical meanings, occasionally making up non-compositional composites. In Section 7.4.1, I also touched upon a number of new meanings gained by a number of Old Irish preverbs, that is, air- ‘before, for’ , aith- ‘re- , ex - ’, di- ‘of, from’ , ess- ‘out of ’ , eter- ‘between, among’, fo- ‘under’, and imm- ‘about, mutually’. It is impossible to discuss here all the semantic shifts summarized in Table 53, although each of them in principle deserves a separate treatment. Here, I illustrate the general development of Old Irish preverbs by examining fo- ‘under’, the most polysemous preverb in the Milan and in the Priscian Glosses (cf. Table 53). In Section 7.4.1, we saw that focan retain its basic meaning of ‘under’ (31) and develop the new spatial meaning of ‘behind’ (32); the linkage between these two meanings was also discussed. In addition, it was shown that the lack of 26 Cf. the Italian and English idioms tirar fuori le parole di bocca ‘get the words out of one’s mouth’ > ‘force one to speak’, or Engl. get it out! ‘tell this! ’. <?page no="353"?> 7.4. The semantics of multiple preverbs 353 visibility constitutes a clue about the semantic shift to ‘secretly’, shown in (35). Other meanings of foare more directly connected with ‘under’. The meaning of ‘support’, exemplified in (42), is one such: (42) The composite fodéinti (fo-di-√gnī-) ‘be sufficient’ (Lat. suf-ficere ) du-n-chlaind bed fodeinti toart . dat -child. dat be. sbjv . pst .3 sg be_sufficient. ger ‘For the child which should be sufficient’ (Ml.107a10) As remarked in e DIL (ie / 22 590), fodéinti is an artificial formation, in which the EP fois calqued from Latin sub- ( sufin (42)) and -déni (i. e. the prototonic form of do·gní ) corresponds to Latin faciō ‘do, make’ . The preverb fomeans ‘at a lower level’ in a couple of Old Irish composites: in fo·éitsi (fo-in(de)-√tōsī-) , it functions as ‘under-’ in ‘under-stand’ or ‘under-lying’, in which ‘under-’ indicates an implied, or subsumed, meaning: (43) foéitsider hi-suidiu deus understand. prs .3 sg . pass inana .3 sg . dat . n D. nom ‘ Deus is understood here.’ (Ml.34d5) The preverb fohas a similar value in fo·acain (fo-ad-√can-) that means ‘accompany in song’, that is, ‘play music in the background’ (cf. Sg.167a2). The value ‘at a lower level’ is also the key to drawing a link with the meaning of foin con·fodlai (com-fo-√dālī-) ‘sub-divide and share jointly (what has been previously divided)’. Another meaning derives from the lack of visibility occasionally implied by fo- , specifically ‘in place of ’, which is instantiated in fo·tairci (fo-to-ad-ro-√icc-) ‘sub-stitute, supply’. In this composite, only the preverb foretains a detectable meaning, arguably as it is a later addition calqued from Latin sub-rogare . The composite foindarbaide (fo-in(de)-ad-ro-uss-√ben-) ‘be relegated, be sub-jected’ instantiates a further group of metaphors typically associated with preverbs that originally mean ‘under, down’ or ‘on, up’: having control or force is up , lacking control or force is down (i. e. under ). Accordingly, fois also connected with the notions of impact, collision, and attack, as in fo·fúasna (fo-uss- √anā-) ‘perturb, disturb’ (cf. con·osna (com-uss-√anā-) ‘cease, stop, desist, remain, end in’, which lacks these notions, as it also lacks fo- ): impacts, collisions, and attacks imply lack of control on the part of the struck entity. When the notions of impact, collision, and attack are already implied in the verbal base, the preverb becomes obsolete in its spatial usage, and thus comes to indicate <?page no="354"?> 354 7. Multiple preverbs in Old Irish completion via the so-called Vey-Schooneveld effect (cf. in particular Chapter 6): e. g. do·fuiben (to-fo-√ben-) ‘cut down, cut out, destroy’ vs. benaid ‘cut, beat’. As preverbs are polysemous morphemes, their combinations can also be expected to express a range of different meanings. However, polysemy is not easy to observe for preverb combinations. First, most combinations (95 out of 115) are instantiated in only one or two composites (cf. Section 7.2.3). This suggests that multiple composition is (or used to be) a productive process, whereby preverbs, carrying their basic or non-basic meanings, successively combine with verbal bases as single units (but see Section 7.3.2, and McCone 2006: 2 on two possible fixed multiple preverb combinations). Second, the meaning of Old Irish MP s and IPs is frequently difficult to detect, as multiple preverb composites arguably originate in a step-by-step recomposition or accretion of an already lexicalized formation. Thus, only the semantic contribution of the EP tends to remain recognizable. 7.4.3. Different degrees of compositionality Identifing the semantic contribution brought about by each element of Old Irish composites is by no means trivial for various reasons. To begin with, Old Irish preverbs are very advanced in their lexicalization processes. Thus, frequently, only the EP , i. e. the last preverb being added to the composite, retains a clearly detectable meaning (cf. Section 7.4.2, and the discussion on the range of new meanings gained by fo- ‘under’). Moreover, multiple preverb composites can show different meanings in the two different collections of glosses (cf. Section 7.2.1), as well as in different contexts. These polysemous composites can also exhibit varying degrees of compositionality, as shown in (44) and (45) (Latin equivalents are indicated only if relevant): (44) Partially compositional vs. non-compositional meaning a. remi·décai (rem-di-en-√kwis-) ‘fore-see’ (Lat. prae-videre ) vs. ‘provide for’ (cf. Latin consulere )’ b. do·róscai (di-ro-uss-√scochī-) ‘stand forth’ (Lat. prae-stare ) vs. ‘distinguish oneself (Lat. eminere )’ c. con·oscaigi (com-uss-√scochī-) ‘(re)move’ vs. ‘change, upset’ d. con·toí (com-to-√sow-) ‘turn’ vs. ‘convert, change’ e. do·intai (to-in(de)-√sow-) ‘turn back, return’ vs. ‘translate’ f. sechmo·ella (sechmo-in(de)-√ell-) ‘pass by, pass’ (Lat. praeter-ire ) vs. ‘neglect’ (Lat. omittere ) (45) Compositional vs. non-compositional fris·taít (frith-to-√tēg-) ‘come against’ (Lat. contra-venire ) <?page no="355"?> 7.4. The semantics of multiple preverbs 355 vs. ‘is at variance with, oppose’ (Lat. op-ponere ) In addition, though a certain composite may result in a non-compositional formation, it might be the case that the semantic contribution of its parts is still traceable (certainly by the linguist, and possibly by the speaker as well). Cases in point are the following: ad·opair (ad-uss-√ber-) ‘sacrifice, offer up’ ((literally) ‘to-up-offer’); fo·acain (fo-ad-√can-) ‘sing to, accompany in song’ ((literally) ‘under-to-sing’); ar·foichlea (air-fo-√cēllā-) ‘look after, take care of, attend to’ ((literally) ‘before-under-go_around’); 27 con·imchloí (com-imm-√clow-) ‘change’ ((literally) ‘with-around-turn’); imm·fogni (imm-fo-√gnī-) ‘be construed with (gramm.)’ ((literally) ‘around-under-do’, ‘serve around’); fris·accai (frith-ad-√kwis-) ‘look forward to, expect, hope’ ((literally) ‘against-to-look’); do·romnathar (di-ro-√mani-) ‘forget’ ((literally) ‘away from-completely-think’); con·tetarrat (com-to-eter-√reth-) ‘comprise, comprehend’ ((literally) ‘with-to-between-run’). Last, the semantic contribution of certain preverbs, though still detectable, is redundant. Either the meanings of the preverbs overlap with one another (46), or the meaning of one of the preverbs, usually but not exclusively the IP (cf. (47)g), shows semantic solidarity with the verbal stem onto which it attaches (47). (46) Composites containing preverbs with overlapping meanings a. ad·tairbir (ad-to-air-√ber-) ‘bring back, deliver again’ (Goal+Goal) b. do·adbair (to-ad-uss-√ber-) ‘display, show, bring forward, offer’ (Goal+Goal) c. do·essuirg (di-ess-√org-) ‘smite, slay’ (Source+Source) d. do·inchain (to-in(de)-√can-) ‘chant, utter’ (Addressee+Addressee) e. in·togair (in(de)-to-√gari-) ‘call on, invoke’ (Addressee+Addressee) (47) Composites containing a redundant preverb a. do·eipen (to-ess-√ben-) ‘excise, cut (out of, off)’ (Sourcep +cut) b. ar·díben (air-di-√ben-) ‘cut off, slay, destroy’ (Sourcep +cut) c. etar·diben (eter-di-√ben-) ‘destroy’ (Sourcep +cut) d. imm·díben (imm-di-√ben-) ‘excise, circumcise’ (Sourcep +cut) e. do·eprainn (to-ess-√brenn-) ‘flow, trickle, gush’ (Sourcep +spring) f. fo·teissim (fo-to-ess-√sem-) pour down, pour out’ (Sourcep +pour) g. do·fonaig (di-fo-√nig-) ‘wash, wash away’ (Sourcep +wash) h. con·dieig (com-di-√sag-) ‘ask, seek, demand’ (Sourcep +ask) i. do·foscart(a) (di-uss-√scart(ā)-) ‘remove, put aside’ (Sourcep +up+remove) j. do·imchella (to-imm-√cēllā-) ‘surround, encompass’(around+go_around) k. do·aidlea (to-ad-√ell-) ‘come to, approach, visit, touch’ (Goalp +go) 27 Cf. the AG noun amphí-polos ‘maiden’, (literally) ‘the one who moves around (the master)’ for a similar semantic development (DELG 877). <?page no="356"?> 356 7. Multiple preverbs in Old Irish l. imm·accaldathar (imm-ad-√glādī-) ‘converse together’ (Addressee+talk) m. remi-uicsed (rem-uss-√gus-) ‘choose beforehand, pre-elect’ (up+choose) do·rochoíni (di-ro-√coin-) ‘despair of ’(intensive+lament) n. con·fodlai (com-fo-√dālī-) ‘share jointly, divide, apportion’ (sub+divide) The composite at (47)i is particularly noteworthy: the meanings of both the EP and the IP can be regarded as redundant. Both di- ‘of, from’ (Sourcep ) and uss- ‘up, off’ (up+Sourcep ) are subsumed by the semantics of the root √scart(ā)- ‘remove’: the event of removing implies a TR moving upward and away from a LM. I classify as ‘partially compositional’ all the composites outlined above, as they defy a clear-cut semantic categorization. Partially compositional composites can belong to different sub-categories: (i) composites in which only the EP retains a detectable meaning; (ii) polysemous composites; (iii) lexicalized composites still analyzable by the linguist (and possibly by the speaker); (iv) composites with redundant preverbs. Partially compositional composites, displayed in Table 51, constitute the majority of Old Irish composites (110 out of 178): c omposite s egmentation m eaning do·futhraccair di-fo-tre-√accdesire, wish imm·comairc imm-com-√arcquestion, ask, inquire of ar·díbdai air-di-√bādīsubmerge, drown, sink, wreck do·aithbig to-aith-√begdissolve, break up airdbidi air-di-√benbe destroyed, be cut off do·eipen to-ess-√benexcise, cut (out of, off) ad·cuimben aith-com-√bencut, strike, wound, lacerate ar·díben air-di-√bencut off, slay, destroy etar·diben eter-di-√bendestroy imm·díben imm-di-√benexcise, circumcise do·fuiben to-fo-√bencut, cut down, cut out, destroy do·immdiben to-imm-di-√bencut away, shorten ad·tairbir ad-to-air-√berbring back, deliver again ad·opair ad-uss-√bersacrifice, offer up remi·epir rem-ess-√bersay beforehand, say previously do·adbair to-ad-uss-√berdisplay, show, bring forward, offer <?page no="357"?> 7.4. The semantics of multiple preverbs 357 c omposite s egmentation m eaning do·eprainn to-ess-√brennflow, trickle, gush do·airchain to-air-√canprophesy, foretell do·inchain to-in(de)-√canchant, utter fo·acain fo-ad-√cansing to, accompany in song imm·timchella imm-to-imm-√cēllāsurround do·imchella to-imm-√cēllāsurround, encompass do·fuinchid to-fo-in(de)-√ciddescend remi·escaid rem-ess-√cidstretch forth as·rochoíli ess-ro-√coildefine, determine do·rochoíni di-ro-√coindespair of do·accradi to-ad-√crādexasperate, provoke do·aithchren to-aith-√crinabuy back, redeem ar·condla air-com-√dālīshare con·fodlai com-fo-√dālīshare jointly, divide, apportion do·aidlea to-ad-√ellcome to, approach, visit, touch ad·comla ad-com-√elljoint, unite do·ella di-en-√ellturn aside, deviate, bend, decline fo·accomla fo-ad-com-√ellsubjoin fo·indlea fo-in(de)-√ellwander, rove sechmo·ella sechmo-in(de)-√ellpass by, pass, neglect ar·foím air-fo-√emaccept, receive, assume, take as·toasci ess-to-√fāscīexpress as·indet ess-in(de)-√fēddeclare, relate, tell, set forth fo·tuidchet fo-to-di-com-√fēdsubdue remi·aisndet rem-ess-in(de)-√fēdtell beforehand, predict do·adbat to-ad-√fēdshow, manifest, set forth do·diat to-di-√fēdlead down, lead, bring con·foíra com-fo-√ferāprovide remi·foírea rem-fo-√ferāprovide previously <?page no="358"?> 358 7. Multiple preverbs in Old Irish c omposite s egmentation m eaning con·ocaib com-uss-√gabilift up, raise, exalt do·furgaib to-air-uss-√gabiraise up, produce, cause do·focaib to-uss-√gabirise imm·freccair imm-frith-√garicorrespond, answer, encounter in·togair in(de)-to-√garicall on, invoke imm·accaldathar imm-ad-√glādīconverse together do·eclainn to-ess-√glennpick out, select fo·tairci fo-to-ad-ro-√iccsubstitute, supply ad·déici ad-di-en-√kwisregard, look at do·écai dē-en-√kwislook at, behold, see etar·décai eter-di-en-√kwisintrospect fris·accai frith-ad-√kwislook forward to, expect, hope imm·accai imm-ad-√kwislook after, regard, examine, consider remi·décai rem-di-en-√kwisprovide for, force do·incai to-in(de)-ad-√kwislook, gaze (at) do·farcai to-for-ad-√kwislook down on, guard, fence around con·airléici com-air-√lēcīpermit, allow, let go do·fúasailci to-fo-uss-√lēcīloosen, relax fo·álgi fo-ad-√logīlay low, prostrate, throw down imm·folngai imm-fo-√longīcause, produce, make, affect remi·folngi rem-fo-√longībear beforehand, anticipate as·comlai ess-com-√ludepart, escape, set out, start comforaithmiti com-for-aith-√manibe commemorated do·romnathar di-ro-√maniforget for·aithminedar for-aith-√manicall to mind, commemorate, remember do·aithmenadar to-aith-√manicall to mind, commemorate, remember do·fonaig di-fo-√nigwash, wash away con·túairc com-to-√orgstrike, pound do·essuirg di-ess-√orgsmite, slay <?page no="359"?> 7.4. The semantics of multiple preverbs 359 c omposite s egmentation m eaning fris·túairc frith-to-fo-√orgthump, blunt do·fúairc to-fo-√orgcrush, grin, beat, pound do·immoirc to-imm-√orgpress, compress, chastise for·deret for-di-√rethpass through, go over do·airndret to-air-in(de)-√rethrun about, roam over do·íarmórat to-íarm-fo-√rethfollow, come after, go after ad·éirrig aith-ess-√rigrepeat, reiterate, change, emend do·érig di-ess-√rig abandon, forsake as·éirig ess-ess-√rigrise again, arise con·dieig com-di-√sagask, seek, demand íarmi·foich íarm-fo-√sagseek after, inquire about imm·tascra imm-to-√scarāstruggle together do·foscart(a) di-uss-√scart(ā)remove, put aside con·oscaigi com-uss-√scochīmove, change, remove, shake, upset do·róscai di-ro-uss-√scochīstand forth, distinguish oneself remderscaigthi rem-di-ro-uss-√scochībe pre-distinguished, di·fíuschi di-uss-√secharouse, excite, call into do·díuschi to-di-uss-√sechawake, arouse ar·neät air-ni-√sedexpect, await fo·coislea fo-com-√selātake away, remove do·fochsla to-fo-com-√selābring, drag over, draw over do·aissilbi to-ad-√selbīassign, ascribe fo·teissim fo-to-ess-√sempour down, pour out do·eissim to-ess-√semshed, pour out do·airissedar to-air-√sistāstand, stay, remain fris·tairissedar frith-to-air-√sistāmake a stand against, oppose as·roinni ess-ro-√snīescape fo·cosnai fo-com-√snī steal, snatch away con·toí com-to-√sowturn, convert, change do·intai to-in(de)-√sowturn back, return, translate <?page no="360"?> 360 7. Multiple preverbs in Old Irish c omposite s egmentation m eaning fris·tinfet frith-to-in(de)-√swizdblow against do·infet to-in(de)-√swizdblow, breathe, blow on, breath on do·esta to-ess-√tābe absent, lacking, wanting con·éitet com-en-√tēggo with, accompany, agree with fris·taít frith-to-√tēgcome against, be at variance with, oppose in·otat in(de)-uss-√tēgenter into Tab. 51: Old Irish partially compositional composites As discussed above in this Section, a number of partially compositional composites show semantic redundancy of some kind (cf. (46)-(47)). However, partial compositionality does not always originate from redundancy: in a number of composites, redundancy is the result of a lexicalization process which occurred at a preceding stage, after which a new preverb was added according to the process of accretion. A case in point follows in (48): Musterdatei NFA_Basis_A.dot 340 fo·cosnai fo-com-√snī steal, snatch away con·toí com-to-√sowturn, convert, change do·intai to-in(de)-√sowturn back, return, translate fris·tinfet frith-to-in(de)-√swizdblow against do·infet to-in(de)-√swizdblow, breathe, blow on, breath on do·esta to-ess-√tābe absent, lacking, wanting con·éitet com-en-√tēggo with, accompany, agree with fris·taít frith-to-√tēgcome against, be at variance with, oppose in·otat in(de)-uss-√tēgenter into Tab. 51: Old Irish partially compositional composites As discussed above in this Section, a number of partially compositional composites show semantic redundancy of some kind (cf. (46)-(47)). However, partial compositionality does not always originate from redundancy: in a number of composites, redundancy is the result of a lexicalization process which occurred at a preceding stage, after which a new preverb was added according to the process of accretion. A case in point follows in (48): (48) remi·epir [rem-[ess-√ber-]] ‘[say [beforehand/ previously]]’ > as·beir [ess-√ber-] ‘say, tell’, (literally) ‘out_of-bring’ > beirid [√ber-] ‘bring’ (= (41)) The composites that I regard as non-compositional are displayed in Table 52: these constitute the second largest group of Old Irish composites (64 out of 178). Together, partially and non-compositional composites cover almost the totality of the data (174 out of 178 composites): this per se reveals that the lexicalization process undergone by preverbs is very advanced in Old Irish. Composite Segmentation Meaning con·osna com-uss-√anācease, stop, desist, remain, end in fo·fúasna fo-uss-√anāperturb, disturb do·indnaig to-in(de)-√aneggive, bestow, grant, hand over foindarbaide fo-in(de)-ad-ro-uss-√benbe relegated, be subjected in·árban in(de)-ad-ro-uss-√bendrive out, expel do·forban to-for-√bencome, arrive, happen to, reach do·airbir to-air-√berbend, bend down, incline, lower do·opair to-uss-√bertake away, deprive, defraud accretion The composites that I regard as non-compositional are displayed in Table 52: these constitute the second largest group of Old Irish composites (64 out of 178). Together, partially and non-compositional composites cover almost the totality of the data (174 out of 178 composites): this per se reveals that the lexicalization process undergone by preverbs is very advanced in Old Irish. c omposite s egmentation m eaning con·osna com-uss-√anācease, stop, desist, remain, end in fo·fúasna fo-uss-√anāperturb, disturb do·indnaig to-in(de)-√aneggive, bestow, grant, hand over foindarbaide fo-in(de)-ad-ro-uss- √benbe relegated, be subjected in·árban in(de)-ad-ro-uss-√bendrive out, expel do·forban to-for-√bencome, arrive, happen to, reach <?page no="361"?> 7.4. The semantics of multiple preverbs 361 c omposite s egmentation m eaning do·airbir to-air-√berbend, bend down, incline, lower do·opair to-uss-√bertake away, deprive, defraud ar·foichlea air-fo-√cēllālook after, take care of, attend to con·érchloí com-air-√clowstir up, disturb, drive away, agitate con·imchloí com-imm-√clowchange do·aithchuiredar to-aith-√corireturn ad·cuimtig ad-com-uss-√dingbuild to, build up ar·utaing air-uss-√dingbuild up, restore, refresh con·utaing com-uss-√dingbuild, construct, build up, embellish do·inóla to-in(de)-uss-√ellgather, collect, assemble ar·coat air-com-√fēdprevent, injure ar·ingaib air-in(de)-√gabiavoid, flee from as·ingaib ess-ind(e)- √gabiexceed, surpass, go beyond fo·acaib fo-ad-√gabileave imm·imgaib imm-imm-√gabiavoid, shun, evade do·rogaib to-ro-√gabicommit, transgress do·fúarascaib to-for-ess-√gabiexpress, characterize as·congair ess-com-√gariproclaim, give notice for·congair for-com-√garicommand, order for·díngair for-di-en-√garisignify, express do·accair to-ad-√garideclare, tell do·airngir to-air-in(de)-√garipromise do·oggell dē-uss-√gellpurchase fodéinti fo-di-√gnībe sufficient imm·fogni imm-fo-√gnībe construed with (grammar) ar·neget air-ne-√guidpray remi·uicsed rem-uss-√guschoose beforehand, pre-elect con·táirci com-to-ad-ro-√iccconfer for·cumaing for-com-√icchappen, occur, be made, be brought about <?page no="362"?> 362 7. Multiple preverbs in Old Irish c omposite s egmentation m eaning imm·airicc imm-air-√iccbe appropriate to ad·cumaing in(de)-com-√iccstrike, cut, happen do·áirci to-ad-ro-√icccause, effect, induce, bring about do·ecmaing to-in(de)-com-√iccstrike, hit a mark ar·osailci air-uss-√lēcībe opened do·atailci to-ad-to-ad-√lēcīfondle, caress, pacify as·cuindligi ess-com-di-√logīdisrupt imm·ruimdethar imm-ro-√medisin, transgress do·inmlig to-en-uss-√mligpromulgate, proclaim, announce do·immna to-imm-ad-√nocommand, enjoin con·erairg com-air-√orggo astray for·comai for-com-√owkeep, preserve, retain con·tetarrat com-to-eter-√rethcomprise, comprehend fo·timmthirid fo-to-imm-di-√rethsubminister, fumigate for·díurat for-di-uss-√rethremain, remain over do·etarrat to-eter-√rethcomprehend, grasp, overtake du·dúrig to-di-uss-√rigbe enticed, excited ad·tóirndea ad-to-fo-√rindāprick again fris·tóirndea frith-to-fo-√rindāmark off, trace (a limit) do·foirndea to-fo-√rindāexpress, signify, denote in·coisig in(de)-com-√sechsignify beforehand fo·uisim fo-uss-√sembe stored, be put away do·fuissim to-uss-√sembring, bring forth con·tairissedar com-to-air-√sistāremain constant, consist ad·roilli ad-ro-√slīdeserve, be entitled to ad·cosnai ad-com-√snīmake for, strive, seek imm·fresnai imm-frith-√snīcontend, disagree, gainsay ad·cota en-com-√tāget, obtain, procure fo·éitsi fo-in(de)-√tōsīunderstand Tab. 52: Old Irish non-compositional composites <?page no="363"?> 7.4. The semantics of multiple preverbs 363 Only four composites out of 178 can be regarded as fully compositional: (a) fo·tabair (fo-to-√ber-) ‘place under’; (b) fris·tabair (frith-to-√ber-) ‘set against, oppose’; (c) imm·tabair (imm-to-√ber-) ‘carry round, surround’; (d) con·ricc (com-ro-√icc-) ‘meet, encounter, join’. Under example (40), I already discuss composites (a)-(c): they contain to- ‘to, toward’ as an IP , which provides a deictic orientation to the verb base, and a further spatial specification as an EP . In regard to composite (d), the root √iccbelongs to a motion verb, the IP roemphasizes the Path of motion, and the EP comexpresses togetherness (Comitative). 7.4.4. Summarizing the meanings of preverbs in multiple preverb combinations Table 53 summarizes the different meanings of Old Irish multiple preverbs. Each meaning is exemplified at least by one composite; if the same meaning is expressed by preverbs occurring both in exterior and interior position, a relevant example for each position is provided. Several semantic shifts displayed in Table 53 are discussed in Sections 7.4.1, 7.4.2, and 7.4.3. p reverb m eaning e xample adto (Goal) to (Recipient) to (Addressee) to (Stimulus) to-ad-√ell- ‘come to’ ad-uss-√ber- ‘offer up, sacrifice’ imm-ad-√glādī- ‘address each other’ frith-ad-√kwis- ‘look forward to’ airbefore (Goal) before (Time) again (Time) back, away (Goal) for (Beneficiary) for (Purpose) resultative intensive to-air-in(de)-√reth- ‘run about (back & forth)’ to-air-fo-√can- ‘foretell’ air-uss-√ding- ‘build up, restore, refresh’ air-in(de)-√gabi- ‘avoid’ air-fo-√cēllā- ‘take care of ’ air-ne-√guid- ‘pray’ air-fo-√em- ‘accept, receive’ air-di-√ben- ‘cut off, destroy’ aithagain (Time) back (Goal) intensive aith-ess-√rig- ‘repeat, reiterate’ to-aith-√cori- ‘return’ aith-com-√ben- ‘wound’ <?page no="364"?> 364 7. Multiple preverbs in Old Irish p reverb m eaning e xample comwith (Comitative) togetherness inchoative, ingressive completion com-en-√tēg- ‘go with’ com-to-air-√sistā- ‘remain constant’ com-uss-√anā- ‘cease, stop’ com-to-√sow- ‘convert, change’ diaway from (Source) aside (Goal) out of (a group) (Source) from (Origin) because of (Cause) completely di-fo-√nig- ‘wash away’ di-en-√ell- ‘turn aside’ di-ro-uss-√schochī- ‘stand forth, distinguish oneself ’ di-uss-√gell- ‘purchase’ di-ro-√coin- ‘despair of ’ eter-di-√ben- ‘destroy’ essout of (Source) away from (Source) off (Source) beyond (cf. Lat. ex- ) metaphorical ‘out of ’ (an event ; cf. Lat. di- ) metaphorical ‘out of ’ ( body / container ) out of (a group) away from (absent) to-ess-√brenn- ‘spring out of, gush’ ess-com-√lu- ‘depart, escape’ to-ess-√ben- ‘cut off’ ess-in(de)-√gabi- ‘exceed, surpass’ ess-com-di-√logī- ‘interrupt’ rem-ess-√ber- ‘tell before’ to-ess-√glenn- ‘pick out’ to-ess-√tā- ‘be absent’ eterbetween, inside (cf. Lat. intro- ) (Goal) among, completely completely (cf. Lat. inter- ) eter-di-en-√kwis- ‘introspect’ to-eter-√reth- ‘encompass, comprehend’ eter-di-√ben- ‘destroy’ founder (Goal) under (as a support) (Location) under- (cf. Engl. under-stand ) under (at a lower level) sub- (cf. Engl. sub-divide ) behind (Location) secretly in the place of (Substitute) lacking control impact, collision, attack completion fo-to-√ber- ‘bring under’ fo-di-√gnī- ‘be sufficient’ fo-in(de)-√tōsī- ‘understand’ fo-ad-√can- ‘accompany in song’ com-fo-√dāli- ‘subdivide and share jointly’ fo-ad-√gabi- ‘leave behind’ fo-com-√snī- ‘steal’ fo-to-ad-ro-√icc- ‘sub-stitute’ fo-in(de)-ad-ro-uss-√ben- ‘be sub-jected’ fo-uss-√anā- ‘perturb, disturb’ to-fo-√ben- ‘cut down’ <?page no="365"?> 7.4. The semantics of multiple preverbs 365 p reverb m eaning e xample forover (Location; cf. Lat. super- ) over, beyond (Goal) over (protection) (Beneficiary) over (Area) having control for (Purpose) for-di-uss-√reth- ‘remain over’ for-di-√reth- ‘go over’ to-for-ad-√kwis- ‘look down on, guard’ com-for-aith-√mani- ‘commemorate’ for-com-√gari- ‘order, command’ for-com-√ow- ‘keep, retain for’ frithagainst (Goal) expectation, hope (Stimulus) frith-to-√ber- ‘set against’ frith-ad-√kwis- ‘hope’ íarmafter (Time) to-íarm-fo-√reth- ‘follow’ immaround (Goal) metaphorically ‘around’ all around (cf. Lat. circum- ) around-thoroughly reciprocal imm-to-√ber- ‘carry around’ imm-fo-√gnī- ‘be construed with’ (grammar) imm-di-√ben- ‘cut off around’ imm-ad-√kwis- ‘examine’ imm-to-√scarā- ‘struggle one another’ in(de)in(to) (Goal) in(to) (Addressee) in(de)-uss-√tēg- ‘enter into’ to-in(de)-√can- ‘chant to’ -nedown (Path) aith-ni-√sed- ‘await’ remforth (Path) before (Time) rem-ess-√cid- ‘stretch forth’ rem-ess-√ber- ‘say beforehand’ roforward (Path) forth (Path) intensive (‘loudly’) com-ro-√icc- ‘meet’ di-ro-uss-√schochī- ‘stand forth, distinguish oneself ’ di-ro-√coin- ‘despairs of ’ sechmobeyond (also metaphorically) sechmo-in(de)-√ell- ‘bypass, neglect’ toto, toward (Goal) to (Addressee) to-reversative (Recipient) reversative completion to-di-√fēd- ‘lead, bring’ to-in(de)-√can- ‘chant to’ to-ad-√selbī- ‘assign’ frith-to-√tēg- ‘come against’ to-ess-√ben- ‘cut off’ <?page no="366"?> 366 7. Multiple preverbs in Old Irish p reverb m eaning e xample -ussupward (Path) up (removing an obstacle) up, out of (a group) (Source) upon (Goal) up (excitement) to-uss-√gabi- ‘raise to’ aith-uss-√lēcī- ‘open’ rem-uss-√tēg- ‘pre-elect’ in(de)-uss-√tēg- ‘come upon’ to-di-uss-√rig- ‘excite’ Tab. 53: The meanings of Old Irish multiple preverbs 7.5. The syntax of multiple preverb composites As discussed in Sections 7.1.2 and 7.4, Old Irish preverbs are much more advanced in their lexicalization process than their Vedic and Homeric Greek counterparts. In addition, a number of Old Irish preverbs, specifically ro- , ad-, com- (and more marginally others), also show a parallel development: specifically, they underwent full grammaticalization into aspectual markers, as well as functioning as lexical modifiers in word-formation. Due to the advanced lexicalization and grammaticalization of Old Irish preverbs, their meanings and behavior diverge considerably from those of the etymologically related prepositions (cf. Dickey 2012, who labels the Slavic preverbs that behave the same way as the just mentioned Old Irish preverbs as “orphan prefixes”; cf. Section 6.1.1). In turn, Old Irish prepositions are also quite grammaticalized, and Old Irish prepositionless cases retain only a very limited ability to express semantic roles (cf. Section 7.1.2.2). These issues relating to preverbs occurring in multiple preverb composites are addressed in the following Section. 7.5.1. No alternative constructions to multiple preverb composites By rearranging the elements that constitute a multiple preverb composite in a different order, one obtains a construction that is still grammatical, but carries a different meaning. Put another way, the construction with multiple preverbs cannot freely alternate with a construction displaying a single preverb composite and a PP . This is shown in (49)-(50) with fris·tabair (frith-to-√ber-) ‘set against’, as opposed to do·beir ‘give’ taking the PP fri + acc (example (49) is also stylistically marked, as it contains a so-called figura etymologica: fritabar ‘opposition’ is the verbal noun associated with fris·tabair ). <?page no="367"?> 7.5. The syntax of multiple preverb composites 367 (49) do-nd fritobairt maill fromart . dat . f opposition( f ). dat slow. dat . f fri-ta-taibret na-dorche against-3 pl . acc set . prs .3 pl art . nom . pl . n -dark. nom . pl do=n-ṡoilsi to= art . dat . f -light( f ). dat ‘From the slow opposition with which the darknesses oppose themselves to the light …’ (Sg.183b3) (50) con-ducthar nomen fri-s. until-give. sbjv . prs . sg . pass name. nom againstacc .3 sg . m / n ‘(The possession is indefinite …) until a name is put beside it.’ (Sg.200b13 = (11)) Similarly as in (49)-(50), it is possible for another composite containing the same sequence of preverbs, frith-to-, to produce alternative constructions: both are grammatical, but they express two different meanings: the multiple preverb composite fris·taít (frith-to-√tēg-) retains the more compositional meaning of ‘come against, oppose’ (51), whereas the reduced composite do·tét ‘come’ in combination with fri + acc develops the lexicalixed meaning of ‘go with, apply to, be construed with (grammar)’ ((52); cf. also Ml.53a8, Sg.158a3, 158a4, 159a3, 213a10). (51) .i. co-nna bí ní frestai i. e. so_thatconj . neg ex . prs .3 sg indf . nom oppose. sbjv .3 sg á mes ón 3 sg . gen . m / n evaluation. acc dem . n ‘I. e. so that there is nothing which opposes its evaluation.’ (Ml.31d6) (52) .i. ní=taet chomsuidigud fri-u i. e. neg =come. prs .3 sg composition. nom against-3 pl . acc in nominativo nisi in paucis in nominative. abl if_not in few. abl . pl ‘I. e. except in a few instances, there is no composition with them in the nominative.’ (Sg.197a4) <?page no="368"?> 368 7. Multiple preverbs in Old Irish Another interesting case is the composite do·intai (to-in(de)-√sow-), which can retain a partially compositional meaning ‘turn back, return’ (Ml.54d3), as well as develop the non-compositional one ‘translate’ (example (53) below, Ml.3a13, 37a10, 89d6, 3a7, Sg.26b4, 20b10; cf. (44)). If one rearranges the elements of the composite into a different construction such as that in (54) with soïd and the PP in+ dat (with the ip employed as a preposition), one also obtains the new meaning of ‘transform x into y’. 28 Notably, it makes sense to compare the multiple preverb composite do·intai with the simplex verb soïd because the reduced composite *in(de)-√sowis not attested: as discussed in Section 7.2.3, this is connected with the fact that composites containing toand lacking it are frequently near-equivalents. (53) cia=[a]s-id-ru-bart-sa nad= although= p -3 sg . n aug -say. prf .1 sg em .1 sg conj . neg = tintae-siu a=llatin do=gregaib· translate. sbjv . prs .2 sg em .2 sg from= L . dat to= G. dat . pl ‘Although I have said you should not translate out of Latin for Greeks.’ (Ml.3a15) (54) r-a-soisit-si ón i-ngoi 7 aug -3 sg . n -turn. prf .2 pl em .2 pl dem . n in-falsehood. acc and an-fír un-true. acc . n ‘(After I had chosen you in passing righteous judgment upon all,) you have turned it into falsehood and untruth.’ (Ml.103c15) Thus, the lexicalization undergone by preverbs leads them to follow divergent paths from those of the corresponding prepositions. For example, as·ingaib ‘go beyond, surpass’ is a non-compositional composite, made up by ess- ‘out of ’+ ind(e)- ‘into’ + √gabi- ‘grasp’ (cf. in·gaib (ind(e)-√gabi-) ‘reproach, reprove’; gaibid ‘grasp, reach, go’). The EP ess- , if used as a preposition ( a ), takes the dative case and indicates Source. The PP that means ‘beyond, across, over’ is instead sech(mo)/ tar + acc (e DIL .ie/ 36 742; e DIL .ie / 40 049; Vendryes 1923: 147). 28 On the etymological and semantic linkage between the preposition do ‘to’ and the preverb to- ‘to, toward’, see Section 7.2.3, Stifter (2014), and references therein. <?page no="369"?> 7.5. The syntax of multiple preverb composites 369 7.5.2. Preverb repetition Apart from the prepositionless accusative, which regularly expresses Goal with a number of very frequent motion verbs including téit ‘go’, prepositionless cases are only residually used to express semantic roles in Old Irish (cf. Section 7.1.2.2). In parallel, Old Irish prepositions function as the heads of the phrases in which they occur (on the notion of head within PP s, cf. Section 3.1.3). Case alternation is meaningful only with the prepositions air ‘before, for’, fo ‘under’, for ‘on, over’, in(de) ‘in(to)’ and distinguishes Goal from Location. Given the situation just outlined, it is not suprising that preverbs are frequently repeated outside the preverbal context as prepositions, as in (55)-(56) below. (55) doroschither dano diib stand_forth. prs .3 sg . pass also of. dat .3 pl ‘He is also distinguished from them.’ (Ml.107b5; Lat. prae-ponere ) (56) fris-tait fris-[s]om against-come. prs .3 sg against.3 sg . acc em .3 sg ‘who opposes him’ (Ml.23c11; Lat. ad-versarii : gen ) In (55), the EP of do·róscai (di-ro-uss-√scōchī-) ‘stand forth, distinguish oneself from’ also occurs as a conjugated preposition, i. e. diib (cf. further Ml.119d3; the same composite can also take sech(mo) + acc , as in Ml.84b1). Example (56) contains the composite fris·táit (frith-to-√tēg) ‘come against’; its EP frithis also repeated as a conjugated preposition in fris (cf. also Ml.17c5, 140b6). 29 As clearly shown by (56), the EP can be repeated outside the preverbal context, even though its meaning is not bleached. Given the advanced grammaticalization of prepositions, preverbs are arguably repeated as prepositions for two reasons: first, they contribute to avoiding semantic ambiguity, as happens for example in Homeric Greek (Chapter 5; Zanchi 2017b). Second, prepositional phrases are the preferred means by which Old Irish expresses semantic roles. 29 The composite fris·táit can also be employed as a lexicalized transitive verb with the direct object (cf. example (51)). However, the prepositionless accusative is possibly an innovation rather than a residual usage. <?page no="370"?> 370 7. Multiple preverbs in Old Irish 7.5.3. Preverbs as modifiers of verb argument structure As pointed out by Vendryes (1923: 241), Old Irish preverbs mainly carry out the function of lexically modifying the meaning of the verbal bases onto which they attach: so, for example, ad·cí (ad-√ kwis-) means ‘see’ (the root √kwisis only attested in composition; cf. Table 45), whereas fris·accai (frith-ad-√kwis-) means ‘look forward to, expect, hope’, after the addition of frith- ‘against’. Both the single preverb and the multiple preverb composites are transitive. Similarly, scuchaid, scuichid ‘go, move’ is a motion / caused motion verb that takes the accusative of the moved TR and various PP s indicating the Goal of motion ( LM ). The derived composite con·oscaigi (com-uss-√scochī-) develops a different meaning, ‘move, change, remove, shake, upset’ (cf. example (44)), but is constructed in a similar way, in that it is accompanied by the direct object in the accusative, and different PP s expressing Goal (including i+ dat , imm + acc , and tre+ acc ). However, it is not infrequent that Old Irish preverbs bring about semantic changes that have the side-effect of transitivizing peripheral arguments. In other words, Old Irish preverbs can function as applicatives (on applicatives from a typological perspective, cf. Austin 1997; Shibatani 2000; and Peterson 2007). The first few relevant examples contain the intransitive motion verb √cēllā- ‘go around’. When this root is compounded with preverbs, it results in transitive composites such as imm·timchella (imm-to-imm-√cēllā-) ‘surround’ and do·imchella (to-imm-√cēllā-) ‘surround, encompass’. 30 The root √rethis also intransitive and indicates a manner of motion verb: rethid ‘run’ takes various PP s expressing the Goal of motion such as sech(mo) + acc (Ml.120b2 = (57)) and in(de) + acc (Ml.138d6). If modified by various preverbs, it turns into the transitive composite do·íarmórat (to-íarm-fo-√reth-) ‘follow, come / go after’ (58). (57) air-ní-derb lin-nai etarcnae forneg -certain. nom . n beside.1 pl . acc em .1 pl knowledge. nom inna-lloc sech a retham art . gen . pl -place. gen . pl beyond rel . ptc run. prs .1 pl ‘For the knowledge of places past which we run is not certain to us.’(Ml.120b2) 30 The composite imm·cella (imm-√cēllā-) ‘surround’ is also attested in the Milan Glosses (Ml.67d12), but is infrequent, and used without an overt second argument. <?page no="371"?> 7.5. The syntax of multiple preverb composites 371 (58) co-ndermanammar-ni so_that-forget. sbjv . prs .1 pl em .1 pl inna-imned-sin im-biam art . acc . pl -tribulation. acc . pl dem inex . prs .1 pl i-sind-laithiu tri-chumsanad inna inart . dat -day. dat through-resting. acc art . gen aidche do-d=iarm-o-rat night. gen p -3 sg . acc . n = p p -follow. prs .3 sg ‘So that we may forget those troubles in which we are in the day through the repose of the night that follows it.’ (Ml.21c3) In (57), the simplex verb retham takes the PP sech(a) expressing the Goal of motion, whereas the composite do-d-iarmorat contains the third person singular infixed pronoun -dfunctioning as a direct object. The posture verb √sed- ‘sit’ behaves the same: the simplex verb saidid ‘sit’ is intransitive and takes different PP s indicating Location such as for + dat (59). After the addition of the preverbs air-ne- , the meaning of the verb is changed from sitting to waiting for (ar·neät (aith-ni-√sed-) ‘before-down-sit’ → ‘expect, await’), and accordingly the composite becomes transitive (60). (59) is do saidi-siu for=hirubinaib be. prs .3 sg to.3 sg . n . dat sit. prs .2 sg em .2 sg over= C . dat . pl ‘It is for this that you sit on the Cherubim.’ (Ml.101c6-7) (60) ar-ro-t=ne-ithius beforeaug -2 sg =down-sit. pst .1 sg sa-du-m-fortacht a-dǽ em .1 sg -to-1 sg . gen -helping. dat ptc - G . voc ‘I awaited you to help me, o God.’ (Ml.46b20) In (59), the PP for=hirubinaib plays the role of Location in combination with the simplex verb form saidi, whereas in (60) the composite arro-t=neithius, the infixed pronoun -t= expresses the direct object (on the relative position of the agument -roand the lexical preverbs in the perfect of ar·neät, see the discussion under example (27)). <?page no="372"?> 372 7. Multiple preverbs in Old Irish 7.6. Preverb ordering A number of Old Irish preverbs only rarely occur in multiple preverb composites. In particular, preverbs that are instantiated in less than 10 combinations are the following: aith- ‘re-, ex-’, eter- ‘between, among’, for- ‘on, over’, frith- ‘against’, íarm- ‘after’, ne- ‘down’, rem- ‘before, pre-’, sechmo- ‘past, beyond’, and tre- ‘through’. For this reason, it is difficult to provide conclusive quantitative data on preverb ordering. These data are nevertheless displayed in Table 54, which must be viewed with this caveat in mind. Table 54 reports the positioning of Old Irish preverbs and their frequencies. The sum of frequencies of all positions (exterior, medial, and interior) gives the total number of preverb combinations containing a certain preverb. Medial position is regarded so as to include McCone’s (1997, 2006) second, third, and fourth positions (cf. Table 55 and Table 56), that is, all positions other than the exterior and the interior ones in combinations that contain more than two preverbs. Second, third, and fourth positions can be treated together, as not many composites include more than three preverbs: thus, having more than one MP is relativey rare (cf. Section 7.2.3). In Table 54, only the reconstructed form of preverbs is reported (cf. Section 7.3.1 on the changes undergone by preverbs occurring in different positions). This choice is based on McCone’s (1997, 2006): reporting the same form for preverbs as McCone facilitates the comparison between his hierarchy (Table 55) and the data of this study (Table 56). p reverb i nterior m edial e xterior ad 8 (35 %) 8 (35 %) 7 (30 %) air 6 (40 %) 3 (20 %) 6 (40 %) aith 3 (60 %) 0 2 (40 %) com 12 (46 %) 2 (8 %) 12 (46 %) di 9 (38 %) 8 (33 %) 7 (29 %) ess 7 (50 %) 1 (7 %) 6 (43 %) eter 2 (67 %) 0 1 (33 %) fo 10 (40 %) 2 (8 %) 13 (52 %) for 1 (13 %) 2 (26 %) 5 (61 %) frith 1 (17 %) 0 5 (83 %) íarm 0 1 (50 %) 1 (50 %) imm 4 (23 %) 3 (18 %) 10 (59 %) <?page no="373"?> 7.6. Preverb ordering 373 p reverb i nterior m edial e xterior in(de) 15 (66 %) 4 (17 %) 4 (17 %) ne 1 (100 %) 0 0 remi 0 0 6 (100 %) ro 10 (71 %) 4 (29 %) 0 sechmo 0 0 1 (100 %) to 6 (12 %) 14 (29 %) 29 (59 %) tre 1 (100 %) 0 0 uss 19 (100 %) 0 0 Tab. 54: The positioning of Old Irish preverbs and their frequencies As Table 54 shows, not all preverbs can occur in all positions: íarm- ‘after’, rem- ‘before, pre-’, and sechmo- ‘past, beyond’ are never interior, whereas ne- ‘down’ , ro- (etymologically) ‘forth’ , tre- ‘through’ , and uss- ‘up, off’ are never exterior. Other preverbs have clear, though not absolute, positional inclinations: imm- ‘about, mutually’, eter- ‘between, among’, for- ‘on, over’, frith- ‘against’, and to- ‘to, toward’ tend to occur in the exterior position, whereas in(de)- ‘in(to)’ preferably selects the interior position. The positional tendencies of other preverbs are not clear: ad- ‘to, toward’ , air- ‘before, for’ , aith- ‘re-, ex-’, com- ‘with’ , di- ‘of, from’ , ess- ‘out of ’ , and fo- ‘under’. 7.6.1. McCone’s hierarchy of preverb ordering In his monograph on the early Irish verb, McCone (1997) tries to draw conclusions about Old Irish preverb ordering in primary composition, resulting in the orderings displayed in Table 55. The hierarchy of Table 55 represents the relative positional properties of preverbs to one another and not the position of preverbs with respect to the verbal base. Specifically, McCone assigns each preverb a positional slot from 1 to 5, in which 1 represents the outermost and 5 the innermost position relative to the verbal base. Importantly, preverbs grouped within the same slot are not necessarily equivalent in regard to their positioning: simply, their relative ranking cannot be determined due to contradictory or insufficient evidence (McCone 1997: 94). According to this ranking, there are preverbs that tend to select the exterior position (1-2), specifically, to- ‘to, toward’, for- ‘on, over’, frith- ‘against’, eter- ‘between, among’, and imm- ‘about, mutually’; by contrast, the interior position (4-5) is the favorite one for com- ‘with’, ro- (etymologically) ‘forth’, uss- ‘up, <?page no="374"?> 374 7. Multiple preverbs in Old Irish off’, and ne- ‘down’. Medial position (3) is selected by ad- ‘to, toward’, aith- ‘re-, ex-’, air- ‘before, for’, di- ‘of, from’, ess- ‘out of ’, fo- ‘under’, and in(de)- ‘in(to)’. These data are roughly consistent with mine, except for the positioning of com-, which is preferably interior according to McCone, but does not seem to favor either placement in the composites from the Milan and Priscian Glosses , which constitute the corpus of my study . 1 2 3 4 5 6 to ‘to, toward’ for ‘on, over’ ad ‘to, toward’ com ‘with’ uss ‘up, off’ verbal base frith ‘against’ aith ‘re-, ex-’ ro ‘forth’ ne ‘down’ eter ‘between, among’ air ‘before, for’ imm ‘about, mutually’ di ‘of, from’ ess ‘out of ’ fo ‘under’ in(de) ‘in(to)’ Tab. 55: Relative ordering of Old Irish preverbs in primary composition In order to explain the numerous exceptions to the hierarchy just outlined, McCone takes into account deviations due to calques from Latin (McCone 1997: 94, 2006: 178-179; Sections 7.2.4 and 7.6.3). In addition, he distinguishes two layers of composition, which he calls “primary” and “secondary composition” . According to McCone (1997: 95), “primary composition presumably reflects an older stage at which the hierarchy in [Table 55] was still operative and [simultaneous] compounding with up to three or four preverbs was still an active process, whereas secondary composition belongs to a later stage in which one of a restricted range of preverbs could be prefixed where appropriate to inherited compounds now perceived as relatively inflexible units.” More explicitly, McCone (1997: 94) writes that “the term secondary composition refers here to the prefixing of a preverb to a pre-existent compound treated as a single unit without regard to the above sequence.” However, the assumption of a clear-cut differentiation between two layers of composition has been challenged by Rossiter (2004) to the point that even McCone (2006: 180 ff., 187 f.) changed his mind. Rossiter (2004) shows that the majority of Old Irish composites can be reduced by successively removing the <?page no="375"?> 7.6. Preverb ordering 375 outermost preverb. The possibility of reduction mirrors the fact that multiple composition was probably an incremental process throughout the prehistory of Old Irish, which McCone accordingly names “recomposition” or “accretion”, as represented below (cf. further Sections 7.1.1, 7.2.3, and examples therein): Musterdatei NFA_Basis_A.dot 353 (61) as·congair ‘proclaim, give notice’ (ess-com-√gari-) for·congair ‘command, order’ (for-com-√gari-) > con·gair ‘cry, shout (out, loudly)’ (com-√gari-) > gairid ‘call’ (√gari-) 7.6.2. Preverb ordering in the Milan and Priscian Glosses: against McCone’s generalizations Table 56 shows preverb combinations attested in composites occurring in the Milan and Priscian Glosses arranged as in McCone’s hierarchy (cf. Table 55). The rightmost column of Table 56 shows whether a given combination is consistent with the mentioned hierarchy or not: the sign + stands for a positive response, the sign for a negative one, whereas nd means that the given combination contains a preverb that McCone does not take into consideration. EP MP( S ) IP M C C ONE (1987) 1 2 3 4 5 ad - - com + ad - com uss + ad - di in(de) + ad - - uss + ad - - ro + ad - to air ad - to fo air - - com + air - - di + air - - fo + air - - in(de) + air - - ne + air - - uss + aith - - com + aith - - ess + com - - air com - - di com - - fo com - - imm accretion reduction 7.6.2. Preverb ordering in the Milan and Priscian Glosses: beyond McCone’s generalizations Table 56 shows preverb combinations attested in composites occurring in the Milan and Priscian Glosses arranged as in McCone’s hierarchy (cf. Table 55). The rightmost column of Table 56 shows whether a given combination is consistent with the mentioned hierarchy or not: the sign + stands for a positive response, the sign - for a negative one, whereas nd means that the given combination contains a preverb that McCone does not take into consideration. ep mp( s ) ip m c c one (1987) 1 2 3 4 5 ad - - com + ad - com uss + ad - di in(de) + ad - - uss + ad - - ro + ad - to air ad - to fo air - - com + air - - di + air - - fo + air - - in(de) + <?page no="376"?> 376 7. Multiple preverbs in Old Irish ep mp( s ) ip m c c one (1987) 1 2 3 4 5 air - - ne + air - - uss + aith - - com + aith - - ess + com - - air com - - di com - - fo com - - imm com - - in(de) com - - uss + com - - ro + com - - to com to ad ro com - to air com - to eter com - for aith di - - ess + di - - fo + di - fo tre nd di - - in(de) + di - - uss + di - - ro + di - ro uss + ess - - com + ess - com di ess - - ess + ess - - in(de) + ess - - ro + <?page no="377"?> 7.6. Preverb ordering 377 ep mp( s ) ip m c c one (1987) 1 2 3 4 5 ess - - to eter - di in(de) + eter - - di + fo - - ad + fo - ad com + fo - - com + fo - - di + fo - - in(de) + fo in(de) ad ro uss + fo - - uss + fo - - to fo to ad ro fo to di com fo - to ess fo - imm di for - - aith + for - - com + for - - di + for - di in(de) + for - di uss + frith - - ad + frith - - to frith - to air frith - to fo frith - to in(de) íarm - - fo imm - - ad + imm - - air + <?page no="378"?> 378 7. Multiple preverbs in Old Irish ep mp( s ) ip m c c one (1987) 1 2 3 4 5 imm - - com + imm - - di + imm - - fo + imm - - frith + imm - - imm + imm - - ro + imm - - to imm - to imm in(de) ad ro uss + in(de) - - com + in(de) - - uss + in(de) - - to rem - di in(de) nd rem di ro uss nd rem - - ess nd rem - ess in(de) nd rem - - fo nd rem - - uss nd sech - - in(de) nd to - - ad + to - ad uss + to - ad ro + to ad to ad to - - air + to - air ro + to - air in(de) + to - air uss + to - - aith + <?page no="379"?> 7.6. Preverb ordering 379 ep mp( s ) ip m c c one (1987) 1 2 3 4 5 to - - di + to - di uss + to - - ess + to - - eter + to - - fo + to - fo com + to - fo in(de) + to - fo uss + to - - for + to - for ad + to - íarm fo nd to - - imm + to - imm ad + to - imm di + to - - in(de) + to - in(de) ad + to - in(de) com + to - in(de) uss + to - - uss + to - - ro + Tab. 56: Old Irish preverb combinations arranged as in McCone (1997) As shown in Table 56, out of 115 combinations, 27 are not consistent with Mc- Cone’s hierarchy. This means that 69 out of 178 composites do not conform to McCone’s preverb ordering outlined in Table 55 (McCone 1997, 2006). The combinations / composites can be inconsistent in the following respects: (a) the preverb to- ‘to, toward’ is medial / interior and not exterior (e. g. ad-to-fo-; comto-eter-; fo-to-ad-ro-; frith-to-fo-; imm-to- ); (b) the preverb com- ‘with’ is exterior rather than interior (e. g. com-fo- ); (c) the preverb diis more interior than comin the combination ess-com-di- (cf. also Section 7.2.3 on preverb combinations). <?page no="380"?> 380 7. Multiple preverbs in Old Irish Deviations (a) and (b) can frequently, though not absolutely, be explained as calques from Latin, as shown in (62). (62) a. Type (a) deviations OI r. fo-to-√berplace under Lat. sub-ducere OI r. imm-to-√bercarry round, surround Lat. circum-dare OI r.com-to-eter-√rethcomprise, comprehend Lat. com-prehendere 31 OI r. frith-to-fo-√rindāmark off, trace (a limit) Lat. ob-signare 32 b. Type (b) deviations OI r. com-fo-√ferāprovide Lat. com-parare Type (a) and (b) deviations that cannot be explained by means of Latin influence can be accounted for in different ways. For example, ad-toinstantiates type (a) deviation, with tointerior rather than exterior. The combination of two near-synonymic preverbs such as ad- ‘to, toward’ and to- ‘to, toward’ (two Goal-preverbs) results in an iterative meaning. Accordingly, this combination consistently translates Latin re- : e. g. ad-to-air-√ber- ‘bring back, delivers again’, Lat. re-digere; ad-to-fo-√rindā- ‘prick again’, Lat. re-pungere . 33 The composite con·dieig (com-di-√sag-) ‘ask, seek, demand’ instead contains a type (b) deviation. In this formation, the exterior comcan be considered a later addition to a preexisting *di·saigid, which can be assumed on the basis of the Old Irish attested forms i ndegaid ‘(in quest of,) after’, and the verbal noun saigid ‘seeking, aiming at’ and its compounds (McCone 1995: 157, 2006: 180). In addition, as shown in Section 7.4.3, discussing example (47), the Source meaning of the IP diis consistent with the semantics of the verbal base √sag- ‘ask’: this semantic solidarity also contributes to explaining the interior position of di-, the early reanalysis of *di·saigid as a single lexical unit, and the subsequent addition of a further preverb, i. e. com- ‘with’. The composite con·imchloí (com-imm- √clow-) ‘change’ can be similarly accounted for. Specifically, the EP comappears to be a later addition to an existing imm·cloí ‘change’, which came early to be perceived as a single lexical unit by virtue of the semantic solidarity between its two components: imm- ‘around, about’ and √clow- ‘turn’. In general, it seems 31 Note that the reduced composite lacking com- , to-eter-√reth- , is nearly equivalent, as shown by the fact that it also translates Latin com-prehendere (OIr. to-eter-√reth- ‘comprehend, grasp, overtake’, Lat. in-cludere , com-prehendere ; cf. Table 47). 32 Cf. the reduced composite lacking the EP : to-fo-√rindā- ‘express, signify’, Lat. significare , distinguere . 33 Cf. fn. 32. <?page no="381"?> 7.6. Preverb ordering 381 that comhas long remained available for composition and recomposition, possibly also because it was particularly prone to being equated to Latin con-, given their formal and semantic similarity (McCone 2006: 178-179). Type (c) deviations, that is, the interior positioning of diin the combination ess-com-di- , instantiated by the composite as·cuindligi (ess-com-di-√logī-) ‘disrupt’, can also be explained by virtue of frequent recomposition with com- . In ess-com-di-√logī-, the EP ess- ‘out of ’ (Source-preverb) is possibly a calque from Latin di-rumpere, in which diis also a Source-preverb. Once essis removed, the MP comcan be then equated to Latin conexpressing completion: di-√logī- (literally) ‘put away’ → ‘forgive’ acquires a telic nuance after the addition of com- (‘completion+put away’ → ‘disrupt’). 7.6.3. An integrated account of preverb ordering Preverb ordering in Old Irish is thus motivated by different factors. As discussed in Sections 7.2.4 and 7.6.2, the Latin source text is one such factor. A number of EP s, especially but not exclusively those that do not conform to McCone’s hierarchy (1997), can be accounted for as calques from Latin: as the Latin preverb is attached to a simplex verb, Old Irish EP s appear to be later additions to an already lexicalized composite, i. e. to a morphological formation perceived as a single (i. e. simplex) lexical unit. A particularly clear example of this is the series of composites built on do·beir ‘give, place’ shown in (63) (cf. further McCone 2006: 178): Musterdatei NFA_Basis_A.dot 360 McCone’s hierarchy (1997), can be accounted for as calques from Latin: as the Latin preverb is attached to a simplex verb, Old Irish EPs appear to be later additions to an already lexicalized composite, i.e. to a morphological formation perceived as a single (i.e. simplex) lexical unit. A particularly clear example of this is the series of composites built on do·beir ‘give, place’ shown in (63) (cf. further McCone 2006: 178): (63) Multiple preverb composites built on do·beir ‘give, place’ fo·tabair ‘under-give’; Lat. sub-dare *b h erbeirid → do·beir fris·tabair ‘against-give’; Lat. ob-dare ‘bring’ ‘to-bring’ imm·tabair ‘around-give’; Lat. circum-dare Further motivations for preverb ordering are specific paths of development undergone by specific Old Irish preverbs. For example, apart from the exceptions described in Section 7.6.2, the preverb to- ‘to, toward’ is especially inclined to be exterior: in the Milan and Priscian Glosses , 30 out of a total of 116 combinations include toas EP. This positional behavior is arguably related to the etymology of to- : as shown by Stifter (2014), tois probably the result of the coalescence of two distinct Old Irish forms, specifically the preposition *to 1 - ‘back, re-’ (cf. ? Lyd. ta- < PIE *tó 1 ‘to’; LIPP II 772, Stifter 2014: 237, and references therein) and the clausal connector *to 2 - (cf. Hitt. ta- ‘then’ < PIE *tó 2 ; LIPP II 775, and references therein). The former explains some meanings expressed by to- (Section 7.4.1), whereas the latter accounts for its tendency to occur as EP and to retain this status (cf. Section 7.2.3). Old Irish is a VSO language (cf. Section 7.1); thus, if a preverb functions as EP, this basically implies its occurrence in clause initial position, which is also the typical placement for clausal connectors. Similarly, the exclusively exterior EPs íarm- ‘after’, rem- ‘before, pre-’, and sech(mo)- ‘beyond’ are secondary formations, made up by an originally spatial adverb or a pronominal stem and a suffix -mi, -mo, -mu, -ma . A development of PIE *epi possibly is the basis for Old Irish íar- (cf. GOI 516; LIPP II 294); the preverb remis probably connected with Latin prior, prius ‘former’, and thus with PIE *pr-ó (allative), *pr-í (locative), and *pr̥- ‘forward, forth’ (LIPP II 633 ff.). If this etymology is correct, the -mof remmight go back to the superlative form *prisamo- (cf. Lat. prīmus ‘first’). The preverb sech(mo)is related to Latin secus ‘beside, otherwise’. According to LIPP II (758), the Latin and Old Further motivations for preverb ordering are specific paths of development undergone by specific Old Irish preverbs. For example, apart from the exceptions described in Section 7.6.2, the preverb to- ‘to, toward’ is especially inclined to be exterior: in the Milan and Priscian Glosses , 30 out of a total of 116 combinations include toas EP . This positional behavior is arguably related to the etymology of to- : as shown by Stifter (2014), tois probably the result of the coalescence of two distinct Old Irish forms, specifically the preposition *to 1 - ‘back, re-’ (cf. ? Lyd. ta- < PIE *tó 1 ‘to’; LIPP II 772, Stifter 2014: 237, and references therein) and the clausal connector *to 2 - (cf. Hitt. ta- ‘then’ < PIE *tó 2 ; LIPP II 775, and references therein). The former explains some meanings expressed by to- (Sec- <?page no="382"?> 382 7. Multiple preverbs in Old Irish tion 7.4.1), whereas the latter accounts for its tendency to occur as EP and to retain this status (cf. Section 7.2.3). Old Irish is a VSO language (cf. Section 7.1); thus, if a preverb functions as EP , this basically implies its occurrence in clause initial position, which is also the typical placement for clausal connectors. Similarly, the exclusively exterior EP s íarm- ‘after’, rem- ‘before, pre-’, and sech(mo)- ‘beyond’ are secondary formations, made up by an originally spatial adverb or a pronominal stem and a suffix -mi, -mo, -mu, -ma . A development of PIE *epi possibly is the basis for Old Irish íar- (cf. GOI 516; LIPP II 294); the preverb remis probably connected with Latin prior, prius ‘former’, and thus with PIE *pr-ó (allative), *pr-í (locative), and *pr̥- ‘forward, forth’ ( LIPP II 633 ff.). If this etymology is correct, the -mof remmight go back to the superlative form *prisamo- (cf. Lat. prīmus ‘first’). The preverb sech(mo)is related to Latin secus ‘beside, otherwise’. According to LIPP II (758), the Latin and Old Irish formations both go back to the reflexive pronominal stem *su̯e- . Whatever their etymologies are, the later formation of these preverbs arguably contributes to explaining their exterior positioning and their low degree of semantic bleaching. To summarize, the exterior placement of some preverbs may have been motivated by Latin calques and the unique development of these preverbs. Why, though, do certain other preverbs favor interior placement? Arguably, preverb ordering is ruled by the following principle: the higher the semantic solidarity between a preverb and the verb, the closer that preverb occurs to the verb. I call this the “redundancy principle”. Such a tendency ruling preverb ordering recalls Bybee’s (1985) “order” and “relevance principle”, as is previously discussed (cf. Sections 4.6, 5.6 and 6.6). However, in the case of preverbs, the meaning of IP s is so close that it may come to overlap with that of the verbal bases: semantic closeness implies redundancy rather than relevance for preverbs. Therefore, the composite is prone to be reanalyzed as a single lexical unit due to the semantic overlap among its parts. Clear examples for this redundancy principle are offered by the non-reducible preverbs -ne- ‘down’ and -uss- ‘up, off’, as shown in (64). (64) a. do·foscart(a) (di-uss-√scart(ā)-) ‘remove, put aside’ (up+remove) b. remi·uicsed (rem-uss-√gus-) ‘choose beforehand, pre-elect’ (up+choose) c. ar·neät (aith-ni-√sed-) ‘expect, await’(down+sit) In addition, both -ne- ‘down’ and -uss- ‘up, off’ are etymologically Path preverbs: as Path is not a particularly relevant component for the encoding of spatial relations, it is infrequent that Path-preverbs are further specified by additional PP s. This arguably constitutes another reason for their internal placement and <?page no="383"?> 7.6. Preverb ordering 383 consequent reanalysis as part of the verbal base (the so-called “lower salience” principle). Interestingly, Path is also the semantic role expressed by ro- (etymologically) ‘forward, forth’ in its spatial and etymological meaning. Accordingly, rois strictly interior, and its Path meaning frequently overlaps with the meaning of the motion verbs onto which it attaches, as in examples (65): (65) a. con·táirci (com-to-ad-ro-√icc-) ‘confer’(Pathp +reach) b. fo·tairci (fo-to-ad-ro-√icc-) ‘substitute, supply’ (Pathp +reach) c. as·roinni (ess-ro-√snī-) ‘escape’ (Pathp +turn) The fact that, in Old Irish, preverbs are advanced in their lexicalization process complicates this analysis, as their precise semantic contributions are not always easy to discern (cf. Section 7.4). However, when Old Irish composites at least partially retain their semantic compositionality, the “redundancy principle” seems to apply. For example, a confirmation of this principle is found in the behavior of the preverbs that do not clearly select either the exterior or the interior position, such as fo- ‘under’: (66) a. Exterior fofo·acain (fo-ad-√can-) ‘accompany in song’ [under+[sing to]] fo·álgi (fo-ad-√logī-) ‘lay low, prostrate, throw down’ [under+[put to]] b. Interior focon·fodlai (com-fo-√dālī-) ‘share jointly, divide, apportion’ [together[sub+divide]] do·fuiben (to-fo-√ben-) ‘ cut, cut down, cut out, destroy’ [completion[sub+cut]] In (66)a, the EP fomodifies the remaining composite as a whole: in both composites, the semantics of ad- ‘to, toward’, which indicates both Addressee and Goal, is closer to that of the verbal base than to the semantics of fo-. By contrast, in (66)b, the IP fohas a meaning close to the verbal bases onto which it attaches, that is, √dālī- ‘divide’ and √ben- ‘cut’. The EP s then modify the remaining composite as a whole: comadds a meaning of togetherness, while toexpresses completion. <?page no="385"?> 8. Multiple preverbs in ancient Indo-European languages: differences, similarities and concluding remarks 8.1. Multiple preverbs: differences among Vedic, Homeric Greek, Old Church Slavic, and Old Irish 8.1.1. Multiple preverbs in numbers: a comparison Table 57 summarizes the quantitative data that I outlined in Sections 4.2, 5.2, 6.2, and 7.2. l anguage c omposites h apaxes o ccurrences v erbal r oots p reverb c ombinations Vedic 116 88 (76 %) 186 56 52 Homeric Greek 64 41 (64 %) 138 43 31 Old Church Slavic 23 5 (21 %) 363 10 15 Old Irish 178 61 (34 %) 1240 43 116 Tab. 57: Multiple preverbs in numbers: a general overview In Vedic, most composites (88 out of 116) are hapaxes; conversely, only one composite, ā́ ní √sad- ‘sit down on, cause to sit down, establish’, occurs in more than 10 R̥g-Vedic passages, and thus seems to make up a conventionalized unit. A deeper look at the data, however, reveals that only the IP -verb combination ní √sadis kept in the later language (on the semantic solidarity between ní ‘down’ and √sad- ‘sit’, cf. further Section 8.2.3 below). These data indicate a fluid system, in which the internal dependencies between the elements making up the composite are still loose. This apparent lack of conventionalization is backed up by the fact that only 45 out of 118 composites are still attested in post-R̥g-Vedic texts. In addition, most preverb combinations only modify one verb, with the notable exception of abhí prá , which is instantiated in no less than 14 composites. <?page no="386"?> The Homeric Greek situation is somewhat similar to that of Vedic. Once again, most composites (41 out of 64) occur only once in the poems. However, three of them, i. e. eis-ana-baínō ‘go upward to’, eis-aph-iknéomai ‘arrive at, come to’, and hup-ek-pheúgō ‘flee away secretly’, reach or surpass the frequency of 10 occurrences, which significantly contributes to increasing the number of Homeric passages with multiple preverbs. In post-Homeric prose, only 21 out of 64 composites are still used. Notably, eis-ana-baínō ‘go upward to’, one of the most frequent composites, does not belong with this group, which suggests that frequent attestation does not necessarily correspond to high conventionalization. In Homeric Greek, two preverb combinations are notably frequent: ex-apo-, containing two Source-preverbs, which as such are particularly prone to undergo grammaticalization (cf. Zanchi 2017b), and para-ex- , which make up a highly conventionalized double adverb and preposition in Homeric and post-Homeric Greek. Old Church Slavic quantitative data contrast strikingly with that profiled above for Vedic and Homeric Greek. Old Church Slavic shows a relatively narrow set of multiple preverb composites, which are, however, instantiated in a far higher number of occurrences. Old Church Slavic data only include five hapaxes, which either (a) at least partially retain their compositional meaning ( iz-obrětati ‘find out’, prěd ъ -po-lagati ‘distribute to’, s ъ -po-žiti ‘live for a while with’, s ъ -prě-byvati ‘remain together with’), or (b) contain the imperfectivizing suffix -ova- , which has no meaning clearly distinct from the far more frequent competing suffix -aj- (Section 6.3.3); thus, pairs such as pro-po-vědovati ‘announce, proclaim’ and pro-po-vědati were rival composites, and only one variant later survived. Notably, in hapaxes of the (a)-type, the EP has a direct Greek counterpart: ex- , para- , and sun- , respectively. All in all, Old Church Slavic conventionalizes or displays an array of already conventionalized multiple preverb composites which are still used in present-day Bulgarian with similar or less-compositional meanings (13 out of 23). Notably, in later Slavic languages, the system of multiple preverbs seems to expand; however, the latter constitutes a quite distinct phenomenon from what we have observed for Old Church Slavic. As discussed in Section 7.2, Old Irish displays a far more numerous set of multiple preverb composites and occurrences than the other Indo-European languages of the sample. A relatively low number of verbal roots are contained in multiple preverbs composites, whereas the number of preverb combinations is high. This abundance of combinations possibly results from the process of accretion or recomposition that lies behind these multiple preverb formations: many preverb combinations are identical to one another except for the EP. In addition, given the high number of occurrences with respect to lemmas, the high number of attestations (more than 30 composites in more than 10 passages), and 386 8. Multiple preverbs in ancient Indo-European languages <?page no="387"?> the relatively low number of hapaxes (61 out of 178 composites), the Old Irish system of multiple composition can be said to be quite conventionalized. Once again, these data confirm the Old Irish exceptional preference for this construction, already highlighted by Lewis & Pedersen (1961[1937]: 267) among others. 8.1.2. Different degrees of univerbation, lexicalization, and grammaticalization Table 58 provides an overview of the semantic analyses outlined in Sections 4.4, 5.4, 6.4, and 7.4. l anguage m otion / l oca tion r oots f ully c ompositional p artially c om positional n on composi tional Vedic 26 (46 %) 37 (32 %) 51 (44 %) 26 (24 %) Homeric Greek 26 (60 %) 23 (36 %) 34 (53 %) 5 (11 %) Old Church Slavic 3 (30 %) 3 (12 %) 10 (44 %) 10 (44 %) Old Irish 19 (43 %) 4 (2 %) 110 (62 %) 64 (36 %) Tab. 58: Semantic parameters: a general overview With the exception of Old Church Slavic, most composites in each language are classified as partially compositional. Arguably, this partially compositional categorization results from two different factors. On the one hand, preverbs are polysemous elements (cf. Sections 4.4.4, 5.4.4, 6.4.4, and 7.4.4), and as such, they can simultaneously retain their basic meanings and also develop lexicalized ones. On the other hand, these composites constitute the most heterogeneous group, including: (a) polysemous composites; (b) composites with redundant elements (semantic redundancy potentially holding between the two or more preverbs or between a preverb, most frequently the IP , and the verbal stem); (c) composites made up from elements whose semantic contributions are still traceable, certainly by linguists and possibly by speakers (cf. Sections 4.4.3, 5.4.3, 6.4.3, and 7.4.3). As for the rest of composites, in the early attested varieties, i. e. Vedic and Homeric Greek, the number of compositional composites is higher than that of non-compositional ones. In contrast, for the later varieties, Old Church Slavic and Old Irish, the opposite is true. Notably, in Old Irish, the language from my sample that extended the system of multiple preverbs far beyond the others (cf. Section 8.1.1 and Chapter 7), the number of fully compositional composites is exceptionally low (only 4 out of 178). 8.1. Multiple preverbs: differences among Vedic, Homeric Greek, OCS, and Old Irish 387 <?page no="388"?> The occurrence of a motion or location verb proper does not necessarily show per se the (low) degree of lexicalization for a number of reasons. First, the majority of non-motion and non-location roots can be assimilated to motion and location verbs, in that, in a non-spatial event, different types of metaphorical TR s can be located or directed toward many types of metaphorical LM s (cf. the discussions in Sections 4.2.2, 5.2.2, 6.2.2, and 7.2.2). Second, the composites containing a motion or location verb proper can develop non-compositional meanings. In this respect, Old Irish is particularly instructive: see for example ar·foichlea (air-fo-√cēllā-) ‘look after, take care of, attend to’ (simplex √cēllā- ‘go around’), sechmo·ella (sechmo-in(de)-√ell-) ‘neglect’ (simplex -√ell- ‘go, put in motion’), and do·róscai (di-ro-uss-√scochī-) ‘stand forth, distinguish oneself ’ (simplex √scochī- ‘move, start’). The conclusions suggested by the quantitative data and the semantic analyses outlined above are backed up by the syntactic behavior of multiple preverbs and by the argument structure that multiple preverb composites display in each language. These parameters are summarized in Table 59 (cf. Sections 4.5, 5.5, 6.5, and 7.5) and discussed briefly below, one by one. (a) (b) (c) (d) (e) l anguage t mesis a lternative c onstructions o ptionality r epetition t ransitivizing a bility Vedic + + + - - Homeric Greek + + + - +/ - Old Church Slavic - + - + + Old Irish +/ - - - + + Tab. 59: Syntactic parameters: a general overview (a)Tmesis. Both lexical and non-lexical tmeses (in Bertrand’s 2014 terms) are allowed both in Vedic and in Homeric Greek (cf. Sections 4.1 and 5.1). Both the EP and the IP - but more frequently the EP - can be displaced from the immediate preverbal position. Instead, apart from the allegedly archaic tmesis and Bergin’s Rule patterns, only the non-lexical tmesis of the EP is possible in Old Irish (cf. Section 7.1.2.1.3). Given that Old Irish is a VSO language, with V occurring in first position, and that the linguistic material that intervenes in between the EP and the remaining composite comprises P2 (Wackernagel’s) clitic particles or pronouns, it appears that Wackernagel’s Law somewhat supersedes the tenden- 388 8. Multiple preverbs in ancient Indo-European languages <?page no="389"?> cy toward univerbation (cf. Section 3.1.2). Thus, in Old Irish, the EP retains a proclitic status, whereas the MP s and IP develop into fully-fledged affixes. The few exceptions to this rule can be motivated as calques from the Latin main text (cf. Section 7.1.2.1.2). By contrast, the earliest written records of Slavic do not attest tmesis at all. (b) Alternative constructions (or preverbs’ movability). Construction alternation involving multiple preverb composites and single preverb composites + PP is widespread in Vedic and Homeric Greek (cf. Sections 4.5.1 and 5.5.1). By contrast, it is only residual in Slavic. Notably, construction alternation is allowed with two fully compositional composites, specifically prěd ь -po-lagati ‘distribute to’ and v ъ s-pri-im а ti / v ъ s-pri-jęti ‘receive in return’ (cf. Section 6.5.1). No semantically equivalent constructions are available in Old Irish (cf. Section 7.5.1). Notably, when alternation is possible, the difference between the construction with multiple preverbs and the construction with a single preverb and a PP lies in the explicit mention of one of the event’s participants. In particular, the construction with multiple preverbs is more likely to omit a participant. Crucially, preverbs’ capacity to refer back to and recover discourse active participants is arguably one of the reasons for their univerbation, grammaticalization, and lexicalization (cf. Section 8.2.3). (c)-(d) Optionality and repetition. Unsurprisingly, these two parameters operate very differently in Vedic and Homeric Greek on the one hand, and in Old Church Slavic and Old Irish on the other (see Table 59). Parameter (c), optionality, occurs in the language varieties in which preverbs retain much of their adverbial status and thus are clearly adjuncts (i. e. modifiers) to what can be regarded as a basic sentence (cf. Boley 2004: 52; Section 3.1.3). Accordingly, in Vedic and Homeric Greek, morphological cases at least partially preserve their original concrete meanings and functions. In contrast, in Old Church Slavic and Old Irish, preverbs are usually repeated outside the preverbal context: these repetitions show that prepositions are the preferred - though by no means exclusive - way of expressing semantic roles in these languages. Thus, preverbs are well differentiated from their cognate prepositions and clearly show verbal rather than nominal orientation (cf. the discussion of the opposite Vedic situation in Section 4.1.2.4) (e) Transitivizing ability. The results relating to this parameter also conform to expectations: the more advanced preverbs are in their grammaticalization and lexicalization paths, the more able they are to affect the argument structure of the verbal bases onto which they attach. Transitivizing potential is connected with preverbs’ grammaticalization into actional markers. As is discussed in Section 2.3.2, one among the factors that characterize prototypical transitivity is telicity, an actional trait that is frequently brought about by the addition of a 8.1. Multiple preverbs: differences among Vedic, Homeric Greek, OCS, and Old Irish 389 <?page no="390"?> 390 8. Multiple preverbs in ancient Indo-European languages preverb. However, transitivization can also result as a side-effect of the semantic changes brought about by preverbs’ lexicalization (cf. Sections 4.5.3, 5.5.3, 6.5.3, and 7.5.3). The formal aspects of composites, discussed in Sections 4.3, 5.3, 6.3, and 7.3, also fit the picture above: in Vedic and Homeric Greek, the sandhi effects occurring between the elements of the composite are not typically word-internal (Sections 4.3, 5.3, 6.3, and 7.3). For the Homeric Greek data, this analysis also finds confirmation in the metrical structure of the hexameter: in slightly fewer than half of the occurrences, a metrical pause can be assumed, restoring original word boundaries splitting multiple preverbs from verbal stems (Section 5.3.1). In Old Irish, by contrast, the forms of preverbs vary tremendously depending on their occurrence before, under, or after the accent (Section 7.3); this suggests a high degree of integration of the so-called nuclear (i. e. tonic and post-tonic) preverbs. In addition, as shown for Vedic and Homeric Greek, other pieces of preverbal morphology (the augment in particular) usually occur between the preverbs and the verbal stem. The few exceptions to this rule can be motivated by philological factors (cf. in particular Section 5.3.3). In Old Irish, the grammaticalized preverb rooccurs in the same position as the augments of Vedic and Homeric Greek with strong verbs, specifically between the IP and the verbal stem (so-called “fixed ro- ”) (Section 7.3.2). This pattern, the so-called fixed ro- , most likely represents the most ancient pattern. With weak verbs, instead, roalways occurs after the pretonic position, either between the EP and the remaining composite (in deuterotonic forms) or between the proclitic particle and the whole composite (in prototonic forms) (so-called movable ro-). The positions of Vedic and Homeric Greek augments, as well as that of fixed ro- , suggest a low level of internal dependency between preverbs and the verbal stems that they modify. In Old Irish, the development of the movable ropattern can be regarded as a piece of evidence in favor of the increasing integration of medial and IP s with the verbal stem. 8.2. Multiple preverbs: similarities among Vedic, Homeric Greek, Old Church Slavic, and Old Irish 8.2.1. Preverb ordering: the common reasons behind it The question of whether certain sequences of preverbs or preverb ordering in the daughter languages represent the conventionalization of practices already known in Proto-Indo-European still needs to be addressed. However, recon- <?page no="391"?> 8.2. Multiple preverbs: similarities among Vedic, Homeric Greek, OCS, and Old Irish 391 structing Proto-Indo-European preverb combinations is not an easy task for several reasons. To begin with, several preverbs occurring across Indo-European languages are actually cognates. However, though several preverbs stem from the same Proto-Indo-European root, they are not always straightforwardly comparable. For example, the following set of preverbs goes back to the same Proto-Indo-European adverb *pr̥- ‘forward, forth’: for the preverbs in the language sample of this work, see Vedic prá ‘forward, forth’, pári ‘around’, Homeric Greek pro- ‘forward, forth’, peri- ‘around’, para- ‘beside’, Old Church Slavic pro- ‘forward’, pri- ‘beside’, prě-(d ъ )- ‘beside, in front of ’, and Old Irish ro- (etymologically) ‘forward’, (? )rem- ‘pre-’ ( IEW 810-814; LIPP II 633-655). Though these preverbs are probably cognates, they actually reflect different Proto-Indo-European forms, with different case endings, different ablaut grades, or different derivational suffixes: specifically, * pró : all , *prei : dat , *per-i : loc , *per-ā̆ : ins , *pri-ām : adv (cf. also the examples below in (1)). Thus, for example, is Old Irish rem- , which goes back to *pri-ām, directly comparable with the other preverbs, which go back either to * pró or to *pri ? Following this analysis, should we relate Old Irish remto Vedic prá , whose most direct reflex is however Old Irish ro- ? In addition, no preverb combination is attested in all languages of the present sample, possibly due to the relatively low number of multiple preverb composites and combinations in Old Church Slavic. However, cognate combinations in two or even three languages are not infrequent, as is shown in (1)a-d: (1) Examples of cognate preverb combinations a. Ved. abhí prá PIE *ánb h i + *pr-ó OI r. imm-ro- *ánb h i + *pr-ó b. Hom.Gr. ex-apo- PIE *éĝ h s + *áp-o OCS iz-po- *éĝ h + *poc. Hom.Gr. para-ex- PIE *per-ā̆ + *éĝ h s OCS pri-iz- *pr-i + *éĝ h OI r. rem-ess- *pri-ām + *éĝ h s d. Hom.Gr. ek-pro- PIE *éĝ h s + *pr-ó OCS iz-pro- *éĝ h + *pr-ó OI r. ess-ro- *éĝ h s + *pr-ó (cf. IEW 53, 292-293, 810-814; LIPP II 71-74, 204-205, 633-655) <?page no="392"?> 392 8. Multiple preverbs in ancient Indo-European languages Moreover, even though common preverb combinations do occur in the sample, this does not necessarily imply that they contain preverbs that attached onto verbal stems at an early stage. Therefore, such combinations are also not always comparable. For example (1)c, the Homeric Greek combinations para-exis cognate with the Old Church Slavic and Old Irish combinations pri-izand rem-ess-. However, on the one hand, the Greek combination para-exalso makes up a lexicalized double adverb and a lexicalized preposition and thus its occurrence as a preverb combination may be accounted for in this way. On the other hand, in Old Church Slavic and Old Irish, the additions of the EP s priand remare strongly suspected to be later calques from the corresponding Greek paraand Latin ante- . Therefore, in this case, common ordering cannot be regarded as a matter of inheritance. As for the proposed inherited relative ordering posited by Papke (2010: 145) and reported in example (2), similar counter arguments can be made. (2) Vedic, Homeric Greek, and Old Irish relative ordering (adapted from Papke 2010: 145) Ved. abhí pári prá Hom.Gr. amphiperipro- OI r. immairro- PIE *ánb h i *péri *pr-ó ( LIPP II 35 f.) ( LIPP II 618 f.) ( LIPP II 636) Most importantly, the Old Irish EP immsimply cannot be directly compared with the other two EP s for several reasons. First, the Old Irish addition of immas an EP is frequently a calque from Latin. Furthermore, in regard to the relative ordering of Homeric Greek amphiand peri- , Papke overlooks the fact that these two preverbs occur elsewhere in the Homeric poems as a lexicalized double adverb. In short, Papke’s generalization does not account for crucial pieces of information contained in the concrete data. Another methohdological issue also arises with Papke’s analysis. Specifically, Papke (2010) focuses on a shared relative ordering of preverbs, rather than the ordering within common preverb combinations or common multiple preverb composites. This is problematic for the reconstruction itself: by definition, reconstruction is based on concrete linguistic items, which can be grouped into cognate sets. By comparing these cognate sets of concrete items, one infers a reconstructed form, which is an abstraction. However, Papke includes abstract items rather than actual forms in her cognate set. In fact, the relative ordering of <?page no="393"?> 8.2. Multiple preverbs: similarities among Vedic, Homeric Greek, OCS, and Old Irish 393 Vedic, Homeric Greek, and Old Irish in (2) in itself represents an abstract pattern, inferred from an array of concrete composites; therefore, it cannot be employed for linguistic reconstruction. In addition, Papke (2010: 154) addresses lexicalization as a pivotal factor in the linear transmission of preverb ordering. However, once again, a relative ordering such as that in (2) is itself an abstract pattern and as such cannot undergo lexicalization and consequently be inherited. Thus, the common relative ordering, which does occur in the sample (cf. Sections 4.6, 5.6, 6.6, and 7.6), calls for explanations other than genealogical inheritance. At a general level, one such explanation is the “redundancy principle” introduced in this work: the more the semantics of a preverb shows semantic solidarity with that of the verb onto which it attaches, the stronger its tendency to occur in the immediate vicinity of that verb. This principle somewhat resembles Bybee’s (1985) well-known ‘order and relevance principle’ of affix ordering, except that in the case of preverbs, semantic solidarity with a verb makes a particular preverb redundant, rather than relevant. Notably, semantic redundancy is arguably also one of the factors triggering reanalysis and consequent developments of preverbs (cf. Section 8.2.3). A further motivation in preverb ordering is the tendency typical of certain preverbs, but not others, to receive further semantic specifications in the form of verbal dependents. Specifically, preverbs that tend not to get further specification preferably select the interior position and thus are prone to be interpreted as verb-oriented; in contrast, preverbs that are frequently specified by further verbal dependents preferably occur in the exterior position. As is well-known from linguistic typology (cf. Ikegami 1987; Ungerer & Schmidt 1996; Verspoor, Dirven & Radden 1999; various papers in Luraghi et al. 2017 and references therein), the overt expression of the Goal-participant is far more frequent than the overt expression of the Sourceand of the Path-participants. Consequently, Sourceand Pathpreverbs, which do not frequently receive a further specification, tend to occupy the interior position. This principle for preverb ordering may be the “lower salience” principle, in that Source and Path do no usually constitute salient pieces of information for the event conceptualization. At a more detailed level, particular paths of development of specific preverbs can also ground their positional preferences. For example, the etymology of the Old Irish tofully accounts for its usual exterior position (cf. Section 7.6.3). In Ancient Greek, the occurrence of certain preverb combinations outside the immediate preverbal contexts, with adverbial or prepositional function, clearly explains their relative ordering (cf. Section 5.2.3). Moreover, as mentioned previously, the interplay with the Greek or Latin sourceor main texts must always be kept in mind for the cases of Old Church Slavic and Old Irish composites (cf. Sections 6.2.4 and 7.2.4). <?page no="394"?> 394 8. Multiple preverbs in ancient Indo-European languages Importantly, after the reanalysis of preverbs as actional markers had taken place (cf. Section 8.2.3), new orders arguably became possible. Thus, for example, once the telic meanings frequently associated with Greek Source preverbs (Zanchi 2017b) are established, such telic preverbs, most notably ex- , often occur in exterior position, as they modify the temporal structure of the whole (lexicalized) composite onto which they attach (cf. e. g. the post-Homeric composites ex-epí-stamai ‘know thoroughly’, ex-epi-sphragízomai ‘be stamped deep on’, ex-uper-optáō ‘bake or dry extremely’) (cf. also the Homeric composites mentioned in Section 5.6.3). A similar process might constitute the basis for the development of the modern Slavic system of multiple preverbs (usually called “prefixes”), whereby the exterior (usually called “external”) preverb always has predictable, quantizing, and actional meanings. Such meanings apply to the rest of the composite, which is modified as a single unit. In Old Church Slavic, we initially observed a large quantity of data that provide evidence for this development (cf. Section 6.3). In particular, see vъz-n е -n а -viděti ‘come to hate’ and vъs-po-męnǫti ‘start remembering, remind’, as opposed to n е -n а -viděti ‘hate’ and po-męnǫti ‘remember’. Further cases in point are iz-ob-rěsti/ iz-ob-rětati ‘find out’ and pri-ob-rěsti ‘acquire’, containing telic izand resultative pri- . Needless to say, this development was still at its inception in Old Church Slavic, as can be easily assessed, for example, from the fact that po- , which is almost exclusively exterior in modern Slavic, only occurs internally in Old Church Slavic. 8.2.2. Common process of formation of multiple preverb composites Apart from cases with double adv sadp sprev s (cf. Sections 5.2.3 and 5.4.2; Section 7.2.3), what lies behind the formation of multiple preverb composites is a process of accretion or recomposition (in Rossiter’s 2004 and McCone’s 2006 terms), which is exemplified in (3) from Old Church Slavic and (4) from Old Irish: Musterdatei NFA_Basis_A.dot 373 the exterior (usually called “external”) preverb always has predictable, quantizing, and actional meanings. Such meanings apply to the rest of the composite, which is modified as a single unit. In Old Church Slavic, we initially observed a large quantity of data that provide evidence for this development (cf. Section 6.3). In particular, see vъz-nе-nа-viděti ‘come to hate’ and vъs-pomęnǫti ‘start remembering, remind’, as opposed to nе-nа-viděti ‘hate’ and pomęnǫti ‘remember’. Further cases in point are iz-ob-rěsti/ iz-ob-rětati ‘find out’ and pri-ob-rěsti ‘acquire’, containing telic izand resultative pri- . Needless to say, this development was still at its inception in Old Church Slavic, as can be easily assessed, for example, from the fact that po- , which is almost exclusively exterior in modern Slavic, only occurs internally in Old Church Slavic. 8.2.2. Common process of formation of multiple preverb composites Apart from cases with double ADV s- ADP s- PREV s (cf. Sections 5.2.3 and 5.4.2; Section 7.2.3), what lies behind the formation of multiple preverb composites is a process of accretion or recomposition (in Rossiter’s 2004 and McCone’s 2006 terms), which is exemplified in (3) from Old Church Slavic and (4) from Old Irish: iz-po-věděti ‘confess, explain’ (3) věděti → po-věděti pro-po-věděti ‘proclaim, predict’ ‘know’ ‘tell’ za-po-věděti ‘order’ fo·tabair ‘under-give’ (4) beirid → do·beir fris·tabair ‘against-give’ ‘bring’ ‘to-bring’ imm·tabair ‘around-give’ Recomposition can be better observed for the latest varieties of the sample, i.e. Old Church Slavic and Old Irish (cf. the many old Irish examples mentioned throughout Chapter 7). Arguably, the reason for this is two-fold: (a) the more advanced lexicalization of composites made them readily available for further composition, i.e. re-composition (cf. Papke 2010: 155); (b) Old Church Slavic and Old Irish composites can be compared with their Greek and Latin counterparts. From this comparison, it becomes apparent that most frequently the <?page no="395"?> 8.2. Multiple preverbs: similarities among Vedic, Homeric Greek, OCS, and Old Irish 395 Recomposition can be better observed for the latest varieties of the sample, i. e. Old Church Slavic and Old Irish (cf. the many old Irish examples mentioned throughout Chapter 7). Arguably, the reason for this is two-fold: (a) the more advanced lexicalization of composites made them readily available for further composition, i. e. re-composition (cf. Papke 2010: 155); (b) Old Church Slavic and Old Irish composites can be compared with their Greek and Latin counterparts. From this comparison, it becomes apparent that most frequently the exterior preverb seems to be a later addition and frequently a calque. The reduced composite frequently corresponds to a simplex lexical unit in the sourceor main language (cf. Sections 6.2.4 and 7.2.4). This suggests that the innermost part of the Old Church Slavic and Old Irish verbs was actually a composite, but a lexicalized one, i. e. a composite perceived as a single lexical unit. Such process of recomposition is much harder to assess for Vedic and Homeric Greek. These early language varieties attest to a very early system of verbal composition, in which it is difficult to determine whether multiple preverb composites represented actual compound units (cf. Sections 4.1.2 and 5.1.2 on the unclear categorial status of Vedic and Homeric preverbs). Nonetheless, relevant examples can be found, as reported in (5) for Vedic (on which, cf. also examples in (73)b and the couple áti prá √vr̥- ‘outstrip, surpass’ and ví prá √vr̥- ‘lie down beside’) and in (6) for Homeric Greek (cf., on Vedic, Papke 2010: 155, who also argues in favor of a process of formation that I would call recomposition or accretion): 8.2.3. Grammaticalization and lexicalization: the common reason behind two distinct developments While it is generally acknowledged that preverbs originally functioned as freestanding adverbs with spatial semantics in Proto-Indo-European, and that in later languages they underwent grammaticalization into actional (and eventually aspectual) markers or lexicalization into fully-fledge compounds, the exact reasons for these developments are far from clear. In general, preverbs’ devel- <?page no="396"?> 396 8. Multiple preverbs in ancient Indo-European languages opments can be regarded as reanalyses, triggered by the semantic solidarity that holds between preverbs and the verbal stem onto which they attach. Such semantic solidarity can initiate two divergent developments: on the one hand, the preverb can be reanalyzed as part of the verbal stem, which leads to its lexicalization (cf. the discussed and revealing examples of Ved. ā́ ní √sad- ‘sit down’, Section 4.3.1; Hom.Gr. pro-kath-ízō ‘perch forth’, Section 5.4.3; and of OIr. ar·neät (aith-ni-√sed-) ‘expect, await’, Sections 7.3.2 and 7.6.3). On the other hand, the preverb can start to be perceived as redundant. Consequently, speakers reanalyze its meaning and start interpreting it as a marker of actionality (so-called Vey-Schooneveld effect, cf. Chapter 6). In this way, preverbs are reassigned a salient piece of information regarding the internal temporal structure of events. Later on, from functioning as actional markers, preverbs can further develop into aspectual markers. A separate issue is why these small uninflected morphemes with originally locative semantics develop into specifically actional markers and not into other types of grammatical morphemes. One answer comes from semantic broadening through metaphor: preverbs, while bringing about spatial meanings, are able to add inherent endpoints to spatial events. This ability was later extended to non-spatial events as well (cf. among others Shull 2003; Wiemer and Seržant 2017). As indicated earlier (cf. e. g. Section 7.1.2.3), however, this explanation works only for preverbs that etymologically express the Goal of motion. For preverbs with other original semantics, conceptual analogy comes into play, specifically in the form of the following cognitive metaphor: events are locations . Thus, for the widespread development of Source-preverbs into telic markers, the explanation now becomes straightforward: departing from an event (i. e. a location ) can imply that the said event is completed. Different conceptual analogies are required to account for the developments of preverbs with different original semantics. For example, preverbs with Comitative semantics develop into telic markers, as completion can be thought of as togetherness (cf. Lat. con- , Ved. sám , and OI r. com- ). Path-preverbs, such as Old Church Slavic po- , can come to be used as distributive markers, as covering a Path can subsume covering all intermediate steps that make up the Path itself (cf. Figure 4). A final example is the Old Church Slavic preverb v ъ z-, which originally means ‘upward’ and later on acquires ingressive meanings. This semantic shift involves a cluster of metaphors, specifically: (a) more is up , less is down ; (b) events can be thought of as piles ; (c) going upward along a pile is going toward the culmination of an event (cf. Figure 5). A further factor possibly contributed to strengthening the motivations for preverbs’ grammaticalization: specifically, preverbs’ ability to refer back to dis- <?page no="397"?> 8.2. Multiple preverbs: similarities among Vedic, Homeric Greek, OCS, and Old Irish 397 course active (i. e. topical) participants (cf. Sections 3.1.1, 4.5.4, and 5.5.5). The link that holds between topicality and telicity was correctly pointed out by Viti (2008a, 2008b): topical participants are conceptualized as entire in space and as complete in time. In this specific sense, thus, the development of preverbs can be regarded as a discourse-oriented grammaticalization (for a possible typological parallels, cf. the ‘relative preverbs’ of Severn Ojibwe, Section 3.3.2). 8.2.4. Common semantic developments As the semantic analyses outlined in Sections 4.4, 5.4, 6.4, and 7.4 contribute to showing, cognate preverbs can instantiate similar semantic shifts. In addition, preverbs that are not etymologically related, but have close original semantics, can develop similar meanings, due to the cognitive basis of metaphor. I exemplify the principle outlined above by means of one preverb, specifically Proto-Indo-European *sup ‘down’, *sup-  ‘downward’ ( LIPP II 746) and semantically similar preverbs. This preverb has reflexes in Vedic úpa ‘near’, Homeric Greek hupo- ‘under’, and Old Irish fo- ‘under’. As a touchstone, I choose Old Irish fo-, as it displays the widest range of semantic shifts. These shifts are summarized in (7) below and fully discussed in Sections 7.4.1 and 7.4.2: (7) Meanings of Old Irish fo- (etymologically) ‘under’ m eaning e xample ‘under’ (Goal) fo-to-√ber- ‘bring under’ ‘under’ (as a support) (Location) fo-de-√gnī- ‘be sufficient’ ‘under- ’ (cf. Engl. under-stand) fo-in-√tōsī- ‘understand’ ‘under’ (at a lower level) fo-ad-√can- ‘accompany in song’ ‘sub-’ (cf. Engl. sub-division) com-fo-√dāli- ‘divide and share jointly’ ‘behind’ (Location) fo-ad-√gabi- ‘leave behind’ ‘secretly’ fo-con-√snī- ‘steal’ ‘in the place of ’ (Substitute) fo-to-ad-ro-√icc- ‘sub-stitute’ ‘lacking control’ fo-inde-ad-ro-uss-√ben- ‘be sub-jected’ ‘impact, collision, attack’ fo-uss-√anā- ‘perturb, disturb’ ‘completion’ to-fo-√bina- ‘cut down’ <?page no="398"?> 398 8. Multiple preverbs in ancient Indo-European languages A number of developments in (7) are also attested in the other ancient Indo-European languages of the sample, even in the relatively small corpus of multiple preverb composites. To being with, (7)c the reflex of *sup also occurs as an EP in the Vedic composite that means ‘understand’, úpa prá √vid-. The widespread semantic shift from ‘down, under, below’ to ‘behind’ (7)f is also shared by Homeric Greek hup-ek-pro-théō ‘run forth from behind, outstrip’. The metaphor under is secretly (7)g is instantiated in both Homeric Greek hup-ek-pro-pheúgō ‘flee away secretly from’ and Old Irish fo-con-√snī- ‘steal’. The Vedic composite ápa ní √lī- ‘hide oneself, disappear completely’ is also a related case, though it contains as an IP the preverb ní (etymologically) ‘down’, which is not etymologically related, but rather semantically similar, to Greek hupoand Old Irish fo- (< PIE *ní ‘below’, cf. OI r. -ne- ; LIPP II 559; cf. Section 4.4.2). The connection between under and lack of control (7)i is evidenced in all three languages: specifically, in Vedic úpa ā́ √car- ‘be of service’, Homeric Greek hup-ek-sṓizō ‘save (by drawing) away from the control of ’, and fo-inde-ad-ro-uss-√ben- ‘be subjected’. The Vedic composite ní ā́ √kr̥- ‘hold back’ shows that the preverb ní (etymologically) ‘down’ (cf. above) also instantiates this semantic shift. Finally, (7)j the meaning of ‘impact, collision, attack’ connects Old Irish fowith Old Church Slavic za- ( OCS za-po-věděti ‘forbid, order’, OI r. fo-uss-√anā- ‘perturb, disturb’). The etymological locative meaning of zais ‘behind’ rather than ‘below, under’ (< PIE *ĝ h óh 1 ‘behind’, cf. LIPP II 277), but this connection is not surprising: as mentioned above, the meanings ‘below’ and ‘behind’ are frequently paired with one another (cf. also Luraghi 2003: 226). By contrast, no preverb combinations show common semantic shifts, probably because the meanings that preverbs bring to multiple preverb composites appear to emerge from their one-by-one addition during the recomposition or accretion process, rather than from the interplay among the preverbs. An exception to this principle is represented by the iterative and intensive meanings that result from preverb iteration (cf. Hom.Gr. pro-pro-kulíndomai ‘keep rolling in front of ’; OI r. ess-ess-√rig- ‘rise again’, imm-imm-√gabi- ‘go around around’ → ‘avoid’). However, both the iterative and the intensive meanings are simply iconic developments of reduplication, and as such are cross-linguistically common within world’s languages (cf. e. g. Moravcsik 1978: 317 “the most outstanding single concept that reduplicative constructions recurrently express in various languages is the concept of increased quantity”, which is in turn related to both iteration and intensification; Kajitani 2005; Fischer 2001a). <?page no="399"?> 8.2. Multiple preverbs: similarities among Vedic, Homeric Greek, OCS, and Old Irish 399 8.3. Brief concluding remarks This work describes and analyzes multiple preverb composites in a sample of ancient Indo-European languages, including Vedic ( R̥g-Veda ), Homeric Greek ( Iliad, Odyssey ), Old Church Slavic ( Codices Marianus, Zographensis, Suprasliensis ), and Old Irish ( Milan and Priscian Glosses ). After an introduction describing the aims of this work and the sample texts, this work opens with a theoretical chapter devoted to the theoretical tools necessary to study preverbs (Chapter 2), and with a general and typological overview of preverbs (Chapter 3). The book provided thereafter quantitative data as for the number of multiple preverb composites, multiple preverb combinations, and verbal roots modified by multiple preverbs (cf. Sections 4.2, 5.2, 6.2, and 7.2). Moreover, it thoroughly carried out philological, formal, semantic, and syntactic analyses on multiple preverb composites. The results summarized above (Section 8.1.2) provide two similar scenarios for Vedic (Chapter 4) and Homeric Greek (Chapter 5), in which multiple preverbs still retain much of their original function and syntactic behavior. By contrast, the grammaticalization and lexicalization paths are far more advanced in Old Church Slavic (Chapter 6) and in Old Irish (Chapter 7). This work also highlights a number of similarities among the developments undergone by Vedic, Homeric, Old Church Slavic, and Old Irish multiple preverbs (cf. Section 8.2). In particular, it is concluded that a process of recomposition or accretion most likely accounts for the formation of multiple preverb composites in all four languages; in addition, preverb ordering can be similarly explained by an analysis integrating different kinds of explanatory factors: specifically, (a) semantic solidarity holding between preverbs and verbs; (b) preverbs’ tendency to be specified by further event participants; (c) specific etymologies of specific preverbs; (d) calques from other languages. It was also pointed out that cognate or semantically similar preverbs are likely to undergo similar semantic shifts. Crucially, by analyzing a relatively small array of multiple preverb composites and by integrating this analysis with the findings of other linguists on different languages, this book contributes to illuminating the reasons for grammaticalization and lexicalization of Indo-European preverbs. These developments are understood as two distinct re-analyses, both triggered by the same pivotal factor, namely, the semantic solidarity that came to make preverbs’ semantic contributions be perceived as redundant. In consequence, preverbs were re-assigned salient pieces of information as markers of actionality (grammaticalization), or were re-interpreted as part of the verbal stem (lexicalization). <?page no="401"?> Web Resources 401 Web Resources A dictionary of the Old-Irish glosses in the Milan Codex Ambrosianus C 301 . 〈http: / / www.univie.ac.at/ indogermanistik/ milan_glosses〉 Electronic Dictionary of the Irish Language (e DIL ) 〈http: / / www.dil.ie/ 〉 Monier Williams Sanskrit-English Dictionary 〈http: / / www.sanskrit-lexicon.uni-koeln. de/ monier/ 〉 Perseus Digital Library 〈http: / / www.perseus.tufts.edu/ hopper/ 〉 RigVeda: a Metrically Restored Text 〈https: / / liberalarts.utexas.edu/ lrc/ rigveda/ index.php〉 The Codex Suprasliensis Project 〈http: / / csup.ilit.bas.bg/ node/ 1〉 The online database of the Old Irish Priscian glosses. 〈http: / / www.univie.ac.at/ indogermanistik/ priscian/ 〉 The Online Liddell-Scott-Jones Greek-English Lexicon ( LSJ ) 〈http: / / stephanus.tlg.uci.edu/ lsj/ #eid=1&context=lsj〉 The Thesaurus Linguae Graecae ( TLG ) 〈http: / / stephanus.tlg.uci.edu/ 〉 The Tromsø Old Russian and OCS Treebank ( TOROT ) 〈https: / / nestor.uit.no/ 〉 <?page no="403"?> References 403 References Ackerman, Farrell & Webelhuth, Gerth. 1998. A Theory of Complex Predicates . Stanford: Center for the Study of Language and Information ( CSLI ). Adger, David. 2000. First Position and the Syntax / Prosody Interface: Old Irish Preverbs. In WCCFL 19 Proceedings , Roger Billerey & Brook D. Lillehaugen (eds), 1-14. Somerville ( MA ): Cascadilla Press. Agrell, Sigurd. 1908. Aspektänderung und Aktionsartbildung beim polnischen Zeitworte: in Beitrag zum Studium der indogermanischen Präverbia und ihrer Bedeutungsfunktionen . Lund: Lunds Universitets årsskrift. Ahlqvist, Anders. 1980. On word order in Irish. In Papers from the 4th international conference on historical linguistics, Elizabeth C. Traugott, Rebecca Labrum & Susan Shepherd (eds), 107-113. Amsterdam: Benjamins. Aitzetmüller, Rudolf. 1991. Altbulgarische Grammatik. Als Einführung in die Slavische Sprachwissenschaft. 2., verbesserte und erweiterte Auflage . Freiburg: Weiher. Allen, Sidney W. 1973. Accent and Rhythm. Prosodic Features of Greek and Latin: a Study in Theory and Reconstruction . Cambridge: Cambridge University Press. Alonso Déniz, Alcorac. 2014a. Contraction. In Encyclopedia of Ancient Greek Language and Linguistics , Georgios K. Giannakis et al. (eds), 383-384. Leiden: Brill. Alonso Déniz, Alcorac. 2014b. Elision. In Encyclopedia of Ancient Greek Language and Linguistics , Georgios K. Giannakis et al. (eds), 545-546. Leiden: Brill. Amse-De Jong, Tine H. 1974. The meaning of the finite verb forms in the Old Church Slavonic Codex Suprasliensis: A synchronic study . The Hague: Mouton. Andersen, Paul K. 1979. Word Order Typology and Prepositions in Old Indic. In Studies in Diachronic, Synchronic, and Typological Linguistics: Festschrift for Oswald Szemérenyi on the Occasion of his 65th Birthday [Current Issues in Linguistic Theory 11] , Bela Brogyanyi (ed), 23-34. Amsterdam: Benjamins. Anderson, Cormac. 2016. Consonant colour and vocalism in the history of Irish. Adam Mickiewicz University of Poznań (Poland), Ph.D. thesis. 〈https: / / repozytorium.amu. edu.pl/ bitstream/ 10593/ 14780/ 1/ Thesis%2020160615%20final.pdf〉 Andrews, Avery D. 1985. The major functions of the noun phrase. In Language typology and syntactic description. Vol. 1: Clause structure , Timothy Shopen (ed), 62-154. Cambridge: Cambridge University Press. Anthony, David W. 2007. The Horse The Wheel And Language. How Bronze-Age Riders From the Eurasian Steppes Shaped The Modern World . Princeton: Princeton University Press. Anttila, Raimo. 1989[1972]. Historical and Comparative Linguistics , 2nd revised edn. Amsterdam: Benjamins. Arkadiev, Petr M. 2015. Areal’naja tipologija prefiksal’nogo perfektiva (na materiale jazykov Evropy i Kavkaza) [Areal Typology of the prefixed perfective (based on ma- <?page no="404"?> 404 References terials from the languages of Europe and the Caucasus)]. Moskva: Jazyki slavjanskoj kul’tury. Arlotto, Anthony. 1972. Introduction to Historical Linguistics . Boston: Houghton. Austin, Peter. 1997. Causatives and applicatives in Australian aboriginal languages. In The dative and related phenomena , Kazuto Matsumura & Tooru Hayasi (eds), 165-225. Tokyo: Hitsuji Shobo. Auwera, Johan van Der. 2002. More thoughts on degrammaticalization. In New Reflections on Grammaticalization [Typological Studies in Language 49] , Ilse Wischer & Gabriele Diewald (eds), 19-29. Amsterdam: Benjamins. Avilova, N. S. 1959. O kategorii vida v sovremennom russkom literaturnom jazyke [On the Category of Aspect in Contemporary Russian Literary Language]. Russkij jazyk v nacional’noj škole 4: 21-26. Avilova, N. S. 1976. Vid Glagola i Semantika Glagol'nogo Slova [Verbal Aspect and the Semantics of the Verb]. Moscow. Babkho-Malaya, Olga. 1999. Zero Morphology: a Study on Aspect, Argument Structure and Case . New Brunswick ( NJ ): Rutgers. Bader, Françoise. 1997. Prépositions dans les langues indo-européennes anciennes. Faits de langues 9: 49-60. Baldi, Philip. 1979. Typology and the Indo-European prepositions. Indogermanische Forschungen 84: 49-61. Bauer, Bernhard & Schumacher, Stefan. 2014. The online database of the Old Irish Priscian glosses . Institut für Sprachwissenschaft, Universität Wien (Austria). Beekes, Robert S. P. 1995. Comparative Indo-European Linguistics, an Introduction. Amsterdam: Benjamins. Beekes, Robert S. P. 2010. Etymological Dictionary of Greek. Leiden: Brill. Belaj, Branimir. 2008. Jezik, prostor i konceptualizacija. Shematična značenja hrvatskih glagolskih prefiksa [Language, space, and conceptualization. Schematic meanings of the Croatian verbal prefixes]. Osijek: Sveučilište Josipa Jurja Strossmayera, Filozofski fakultet. Benveniste, Émile. 1968. Mutations of Linguistic Categories. In Directions for Historical Linguistics: a Symposium , Winfred P. Lehmann & Yakov Malkiel (eds), 83-94. Austin: University of Texas Press. Bergin, Osborn. 1938. On the syntax of the verb in Old Irish. Ériu 12: 197-214. Bermel, Neil. 1997. Context and the Lexicon in the Development of Russian Aspect . Berkely: University of California Press. Bertinetto, Pier Marco. 1986. Tempo, aspetto e azione nel verbo italiano: il sistema dell’indicativo . Firenze: Accademia della Crusca. Bertinetto, Pier Marco. 2001. On a frequent misunderstanding in the temporal-aspectual domain. The ‘Perfective = Telic Confusion’. In Semantic Interfaces. Reference, Anaphora and Aspect, Carlo Cecchetto et al. (eds), 177-210. Stanford: CSLI Publications. Bertinetto, Pier Marco & Civardi, Eugenio. 2015. The semantics of Degree Verbs and the telicity issue. Borealis 4(1): 55-77. <?page no="405"?> References 405 Bertrand, Nicolas. 2014. On tmesis, word order and noun incorporation in Homeric Greek. In The Greek verb. Morphology, Syntax, and Semantics. Proceedings of the 8th International Meeting on Greek Linguistics (Agrigento, October 1-3, 2009), Annamaria Bartolotta (ed), 11-30 . Louvain: Peeters. Boley, Jaqueline. 2004. Tmesis in Proto-Indo-European Syntax . Innsbruck: Innsbrucker Beiträge zur Sprachwissenschaft. Boling, Bruce D. 1972. Some Problems of the Phonology and Morphology of the Old Irish Verb. Ériu 23: 73-102. Bolinger, Dwight. 1971. The phrasal verb in English . Cambridge: Harvard University Press. Booij, Geert E. 1990. The boundary between morphology and syntax: Separable complex verbs in Dutch. In Yearbook of Morphology 1990, Geert E. Booij & Jaap van Marle (eds), 45-63. Dordrecht: Foris. Booij, Geert E. 2002a. The Morphology of Dutch . Oxford: Oxford University Press. Booij, Geert E. 2002b. Separable complex verbs in Dutch, a case of periphrastic word formation. In Structural Aspects of Semantically Complex Verbs , Nicole Dehé & Anja Wanner (eds), 21-42. Berlin: Peter Lang. Booij, Geert E. & van Kemenade, Ans. 2003. Preverbs: An introduction. In Yearbook of Morphology 2003 , Geert E. Booij & Jaap van Marle (eds), 1-12. Dordrecht: Kluwer. Booij, Geert & van Marle, Jaap (eds). 2003. Yearbook of Morphology 2003. Dordrecht: Kluwer. Bopp, Franz. 1816. Über das Conjugationssystem der Sanskritsprache in Vergleichung mit jenem der griechischen, lateinischen, persischen und germanischen Sprache . Frankfurt am Main: Andreäische. Borchio, Luisa. 2018. Preverbi e participi passati di radici intransitive in vedico. University of Pavia (Italy), MA thesis. Börjars, Kersti & Vincent, Nigel. 2011. Grammaticalization and directionality. In The Oxford Handbook of Grammaticalization , Heiko Narrog & Bernd Heine (eds), 163-176. Oxford: Oxford University Press. Borodič, Vera V. 1953. K voprosu o formirovanii soveršennogo i nesoveršennogo vida v slavjanskich jazykach [On the question of the formation of perfective and imperfective aspect in the Slavic languages]. Voprosy jazykoznanija 1953(2): 68-86. Bortone, Pietro. 2010. Greek Prepositions: From Antiquity to the Present. Oxford: Oxford University Press. Botne, Robert. 2003. To die across languages: toward a typology of achievement verbs. Linguistic Typology 7: 233-278. Breu, Walter. 1992. Zur Rolle der Präfigierung bei der Entstehung von Aspektsystemen. In Linguistique et slavistique. Melanges offerts à Paul Garde , Marguerite Guiraud-Weber & Charles Zaremba (eds), 119-135. Paris: Presses universitaires de Provence. Brinton, Laurel J. & Traugott, Elizabeth C. 2005. Lexicalization and Language Change [Research Surveys in Linguistics]. Cambridge: Cambridge University Press. Brugmann, Claudia. 1988. The story of Over : Polysemy, semantics and the structure of the lexicon. New York: Garland. <?page no="406"?> 406 References Brugmann, Karl. 1885. Griechische Grammatik . Munich: Beck. Brugmann, Karl. 1890. Griechische und lateinische Sprachwissenschaft. Munich: Beck. Brugmann, Karl & Thumb, Albert. 1913. Griechische Grammatik . 4 Aufl. Munich: Beck. Brunel, Jean. 1939. L’aspect verbal et l’emploi des préverbes en grec, particulièrement en attique . Paris: Klincksieck. Budassi, Marco. 2017. The development of Insular Celtic double system of inflection. Paper presented at the Second European Symposium in Celtic Studies, Bangor (Wales), July 31 st -August 8 th 2017. Budassi, Marco & Roma, Elisa. 2018. On the origin of the absolute vs. conjunct opposition in Insular Celtic. Indogermanische Forschungen 123(1): 293-338. Bybee, Joan L. 1985. Morphology: A Study on the Relation Between Meaning and Form. Amsterdam: Benjamins. Bybee, Joan L. 1988. Semantic Substance vs. Contrast in the Development of Grammatical Meaning. Berkeley Linguistics Society: Proceedings of the Annual Meeting 14: 247-264. Bybee, Joan L. 2008. Grammaticization: implications for a theory of language. In Crosslinguistic Approaches to the Psychology of Language: Research in the Tradition of Dan Isaac Slobin, Jianshing Guo et al. (eds), 345-356. London: Taylor & Francis. Bybee, Joan L. 2011. Usage-based theory and grammaticalization. In The Oxford Handbook of Grammaticalization, Heiko Narrog & Bernd Heine (eds), 69-78. Oxford: Oxford University Press. Bybee, Joan L. & Dahl, Östen. 1989. The creation of tense and aspect systems in the languages of the world. Studies in Language 13(1): 5-103. Bybee, Joan L. & Hopper, Paul J. (eds). 2001. Frequency and the Emergence of Linguistic Structures [Typological Studies in Language 45]. Amsterdam: Benjamins. Bybee, Joan, Perkins, Revere & Pagliuca, William. 1994. The Evolution of Grammar: Tense, Aspect and Modality in the Languages of the World . Chicago: University of Chicago Press. Campbell, Lyle. 2001. What’s wrong with grammaticalization? In Grammaticalization: A Critical Assessment [Special Issue of Language Sciences 23(2-3)], Lyle Campbell (ed), 113-161. London: Pergamon. Capaldo, Mario. 1984. Limiti e novità della nuova edizione del Suprasliensis. Europa Orientalis : 219-235. Capell, Arthur. 1979. Classification of verbs in Australian languages. In Australian Linguistic Studies [Pacific Linguistics C-54], Stephen A. Wurm (ed), 229-322. Canberra: Research School of Pacific and Asian Studies. Carnie, Andrew & Guilfoyle, Eithne (eds). 2000. The Syntax of the verb initial languages . New York: Oxford University Press. Carnie, Andrew, Dooley, Sheila A. & Harley, Heidi (eds). 2005. Verb first: On the syntax of verb-initial languages . Amsterdam: Benjamins. Carr, Thérèse M. 2000. Wunambal: A language of the North-West Kimberley Region, Western Australia . University of New England (Armidale), MA thesis. <?page no="407"?> References 407 Casaretto, Antje. 2010a. Syntax und Wortarten der Lokalpartikeln des R̥gveda VI : práti . International Journal of Diachronic Linguistics and Linguistic Reconstruction 7: 1-53. Casaretto, Antje. 2010b. Syntax und Wortarten der Lokalpartikeln des R̥gveda VIII : abhí. Historiche Sprachforschung 123: 97-156. Casaretto, Antje. 2011a. Syntax und Wortarten der Lokalpartikeln des R̥gveda IX : ánu. Münchener Studien zur Sprachwissenschaft 65: 7-64. Casaretto, Antje. 2011b. Syntax und Wortarten der Lokalpartikeln des R̥gveda XVI : áti und tirás . International Journal of Diachronic Linguistics and Linguistic Reconstruction 8: 117-160. Casaretto, Antje. 2011c. Syntax und Wortarten der Lokalpartikeln des R̥gveda XII : ví . Historische Sprachforschung 124: 134-177. Casaretto, Antje. 2011d. Lokalpartikel im R̥gveda XIV (Folge 14): ácchā . Zeitschrift für Indologie und Sudasienstudien 28: 1-27. Casaretto, Antje. 2012. Syntax und Wortarten der Lokalpartikeln des R̥gveda XXI : purás, purástāt, purā́. Münchener Studien zur Sprachwissenschaft 66(1): 11-53. Casaretto, Antje. 2013. Syntax und Wortarten der Lokalpartikeln des R̥gveda XIX : prá . Indogermanische Forschungen 117: 15-74. Casaretto, Antje & Schneider, Caroline. 2015. Vedic local particles at the syntax-semantics interface. In Language Change at the Syntax-Semantics Interface [Trends in Linguistics. Studies and Monographs 278], Chiara Gianollo, Agnes Jäger & Doris Penka (eds), 223-260. Berlin: de Gruyter. Čertkova, Marina Yu. 1996. Grammatičeskaja kategorija vida v sovremennom russkom jazyke [The grammatical category of aspect in contemporary Russian]. Moscow: Izdatel ’ stvo Moskovskogo Universiteta. Chantraine, Pierre. 1942. Le rôle et la valeur de en dans la composition. Revue de Philologie 68: 115-125. Chantraine, Pierre. 1945. Morphologie historique du Grec. Paris: Klincksieck. Chantraine, Pierre. 1953. Grammaire homérique . Tome 2: Syntaxe . Paris: Klincksieck. Chantraine, Pierre. 1968. Dictionnaire étymologique de la langue grecque. Histoire des mots. Paris: Klincksieck. Charitonidis, Chariton. 2013. The Role of Pragmatics in Verbal Derivation: The Case of apo- , ek(s)- , and kse- Verbs in Modern Greek. Bucharest Working Papers in Linguistics 15(1): 79-112. Citraro, Cinzia. 2014. Compound verbs’ meaning in Homer: The case for ἀνά/ ἐπί/ (ὑπέρ). In The Greek Verb. Morphology, Syntax, Semantics. Proceedings of the 8th International Meeting on Greek Linguistics, Annamaria Bartolotta (ed), 53-70. Louvain: Peeters. Ciucci, Luca. 2014. Tracce di contatto tra la famiglia zamuco (ayoreo, chamacoco) e altre lingue del Chaco: prime prospezioni. Quaderni del laboratorio di linguistica SNS 13: 2-52. Claudi, Ulrike & Heine, Bernd. 1986. On the metaphorical base of grammar. Studies in Language 10: 297-335. <?page no="408"?> 408 References Coleman, Robert. 1991. Latin prepositional syntax in Indo-European perspective . In New Studies in Latin Linguistics: Proceedings of the 4th International Colloquium on Latin Linguistics, Cambridge, April 1987 [Studies in Language Companion Series 21], Robert Coleman (ed), 323-338. Amsterdam: Benjamins. Collinge, Neville. 1985. The Laws of Indo‐European . Amsterdam: Benjamins. Comrie, Bernard. 1976. Aspect . Cambridge: Cambridge Univesity Press. Conforti, Costanza. 2014. La tmesi gotica. University of Pavia (Italy), BA thesis. Cook, Anthony. 1986. Participle sentences in Wakiman. In Complex Sentence Constructions in Australian Languages , Peter Austin (ed), 69-95. Amsterdam: Benjamins. Cotticelli-Kurras, Paola. 2015. Periphrastic constructions, phasal verbs, and Aktionsart in Hittite. In Verbis 1: 43-61. Cowgill, Warren. 1975. The Origins of the Insular Celtic Conjunct and Absolute Verbal Endings. In Flexion und Wortbildung [Akten der V. Fachtagung der Indogermanischen Gesellschaft], Helmut Rix (ed), 40-70. Wiesbaden: Reichelt. Craig, Colette H. 1991. Ways to go in Rama: a case study in polygrammaticalization. In Approaches to Grammaticalization: Vol. 2. Types of grammatical markers [Typological Studies in Language 19(2)] , Elizabeth C. Traugott & Bernd Heine (eds), 455-492. Amsterdam: Benjamins. Craig, Colette H. 1993. Jakaltek directionals: their meaning and discourse function. Languages of the World 7(2): 23-36. Craig, Colette H. & Hale, Kenneth. 1988. Relational preverbs in some languages of the Americas: typological and historical perspectives. Language 64: 312-344. Croft, William. 1993. The role of domains in the interpretation of metaphors and metonymies. Cognitive Linguistics 4(4): 335-370. Croft, William. 2001. Radical Construction Grammar: Syntactic Theory in Typological Perspective . Oxford: Oxford University Press. Croft, William. 2006. Typology. In The Handbook of Linguistics [Blackwell Handbooks in Linguistics], Mark Aronoff & Janie Rees-Miller (eds), 337-368. Oxford: Blackwell. Croft, William. 2012. Verbs: Aspect and Causal Structure . Oxford: Oxford University Press. Croft, William. 2017. Force-dynamic image schemas: between verb and argument structure construction [Slides of a lecture held at University of Pavia, October 17 th 2017]. Curtius, Georg. 1846. Die Bildung der Tempora und Modi im Griechischen und Lateinischen sprachvergleichend dargestellt . Berlin: Besser. Cuzzolin, Pierluigi. 1995. A proposito di sub vos placo e della grammaticalizzazione delle adposizioni. Archivio Glottologico Italiano 80: 122-142. Cuzzolin, Pierluigi & Haverling, Gerd. 2010. Syntax, sociolinguistics, and literary genres. In New Perspectives on Historical Latin Syntax , Vol. 1 [Trends in Linguistics Studies and Monographs 180(1)], Pierluigi Cuzzolin & Philip Baldi (eds), 19-64. Berlin: de Gruyter. Cuzzolin, Pierluigi, Putzu, Ignazio & Ramat, Paolo. 2006. The Indo-European Adverb in Diachronic and Typological Perspective. Indogermanische Forschungen 111: 1-38. <?page no="409"?> References 409 Dahl, Eystein. 2010. Time, Tense and Aspect in Early Vedic Grammar. Exploring Inflectional Semantics in the Rigveda. Leiden: Brill. Dahl, Östen. 1985. Tense and aspect systems . Oxford: Basil Blackwell. Dalle Ceste, Carlo. 2014. Functions of the Old Irish preverb ro- (< *pro) in the Milan Glosses. University of Pavia (Italy), MA thesis. Danesi, Serena. 2009. Uso e significato della preverbazione in vedico: la testimonianza del R̥gveda. University of Pisa (Italy), Ph.D. thesis. Danesi, Serena. 2013. Particle-verb Constructions in Vedic: The Case of ápa . Studi e Saggi Linguistici 51: 57-100. Daniel, Michael & Rind-Pawlowski, Monika. 2014. A Survey of Khinalug preverbs. Paper presented at East Caucasian preverbs and the compounding: derivation - inflection continuum, held in connection with SWL 6 , Pavia (Italy), September 8 th -10 th 2014. 〈https: / / www.academia.edu/ 8964489/ A_Survey_of_Khinalug_preverbs〉 De Angelis, Alessandro. 2004. Forme di “tmesi” nel greco omerico, la legge di Wackernagel e un caso di rianalisi sintattica. In Dialetti dialettalismi, generi letterari e funzioni sociali. Atti del V Colloquio Internazionale di Linguistica Greca (Milano, 12-13 settembre 2002), Giovanna Rocca (ed), 179-214. Alessandria: Edizioni dell’Orso. De Angelis, Alessandro & Gasbarra, Valentina. 2010. Tra morfologia e sintassi: mic. e-pi-ko-wo / o-pi-ko-wo e le forme omeriche del tipo ἐπήρετμος. In La morfologia del Greco tra tipologia e diacronia, Ignazio Putzu, Giulio Paulis, Gianfranco Nieddu & Pierluigi Cuzzolin (eds), 150-167. Milano: Franco Angeli. De Vaan, Michiel. 2008. Etymological Dictionary of Latin and the other Italic Languages. Leiden: Brill. Dedio, Stefan & Widmer, Paul. 2017. S, A, and P argument demotion with preverbal imm-(a- N ) in Old and Middle Irish. Études Celtiques 43: 187-206. Dehé, Nicole & Wanner, Anja (eds). 2001. Structural Aspects of Semantically Complex Verbs . Berlin: Peter Lang. Dehé, Nicole, Jackendoff, Ray, McIntyre, Andrew & Urban, Silke (eds). 2002. Verb Particle Explorations . Berlin: de Gruyter. DeLancey, Scott. 2001. Lectures on Functional Syntax . LSA Summer Institute, UC Santa Barbara. Delbrück, Berthold. 1878. Die altindische Wortfolge aus dem Śatapathabrāhmaṇa dargestellt. [Syntaktische Forschungen 3] . Halle: Buchhandlung des Weisenhauses. Delbrück, Berthold. 1888. Altindische Syntax [Syntaktische Forschungen 5]. Halle: Buchhandlung des Weisenhauses. Delbrück, Berthold. 1897. Vergleichende Syntax der indogermanischen Sprachen , vol. 2 [= Grundiss, vol. 4]. Strassburg: Trübner. Dickey, Eleanor. 2006. The Use of Latin sis as a Focus-marking clitic particle. Oxford University Working Papers in Linguistics, Philology & Phonetics 11: 21-26. Dickey, Stephen M. 2007. A prototype account of the development of delimitative po in Russian. In Cognitive Paths into the Slavic Domain, Dagmar Divjak & Agata Kochańska (eds), 329-374. Berlin: de Gruyter. <?page no="410"?> 410 References Dickey, Stephen M. 2011. The varying role of poin the grammaticalization of Slavic aspectual systems: Sequences of events, delimitatives, and German language contact. Journal of Slavic Linguistics 19(2): 175-230. Dickey, Stephen M. 2012. Orphan prefixes and the grammaticalization of aspect in South Slavic. Jezikoslovlje 13(1): 71-105. Dickey, Stephen M. & Janda, Laura A. 2015. Slavic aspectual prefixes and numeral classifiers: Two kinds of lexico-grammatical unitizers. Lingua 158: 57-84. Dieterich, Karl. 1909. Die präpositionalen präfixe in der griechischen Sprachentwicklung mit besonderer Berücksichtigung des Mittel- und Neugriechischen . Strassburg: Trübner. Dik, Helma. 1995. Word Order in Ancient Greek . Leiden: Brill. Dik, Helma. 2007. Word Order in Greek Tragic Dialogue . Oxford: Oxford University Press. Dillon, Myles. 1943. On the structure of the Celtic verb. Language 19(3): 252-255. Dillon, Myles. 1962. History of the preverb to. Éigse 10: 120-126. Dixon, R. M. W. 1979. Ergativity. Language 55(1): 59-138. Doherty, Cathal. 1999. Tmesis and verb second in Early Irish syntax. Proceedings of the Berkeley Linguistic Society 25: 98-108. Doherty, Cathal. 2000. Tmesis and Verb Second in Early Irish Syntax. Berkeley Linguistics Society: Proceedings of the Annual Meeting 25: 98-108. 〈http: / / journals.linguisticsociety.org/ proceedings/ index.php/ BLS/ article/ viewFile/ 1206/ 988〉 Dostál, Antonin. 1954. Studie o vidovém systému v staroslověnštině [Studies on the aspectual system in old Slovak] . Prague: Státní pedagogické nakladatelství. Dover, Kenneth J. 1960. Greek Word Order . Cambridge: Cambridge University Press. Dowty, David. 1979. Word meaning and Montague Grammar . Dordrecht: Reidel. Drinka, Brigette. 2011. The Sacral Stamp of Greek: Periphrastic Constructions in New Testament Translations of Latin, Gothic, and Old Church Slavonic. Oslo Studies in language 3(3): 41-73. Duhoux, Yves. 1998. Autour de la tmèse grecque. Situation dialectale à l’époque mycénienne; datation de l’épopée. In Quaestiones Homericae. Acta Colloquii Namuricensis, Lambert Isebaert & René Lebrun (eds), 71-80. Louvain: Peeters. Dunkel, George E. 1976. Repetition and Deletion of Preverbs and Verbs in Early Indic and Greek . University of Pennsylvania (Philadelphia, PA ), Ph.D. thesis. Dunkel, George E. 1979. Preverb repetition. Münchener Studien zur Sprachwissenschaft 38: 41-82. Dunkel, George E. 1981a. Āmreḍita and iteration of preverbs in Vedic and Hittite. Zeitschrift für vergleichende Sprachforschung 95(2): 214-226. Dunkel, George E. 1981b. Further traces of preverbal Āmreḍita in Greek and Latin. Zeitschrift für vergleichende Sprachforschung 95(2): 227-231. Dunkel, George E. 1982a. Autour de la ā́ R̥g-Védique (A Lexical Study). Indo-Iranian Journal 24(2): 89-102. Dunkel, George E. 1982b. σύν, ξύν. Glotta 60(1): 55-61. <?page no="411"?> References 411 Dunkel, George E. 2014. Lexikon der indogermanischen Partikeln und Pronominalstämme. Band 2: Lexicon . Heidelberg: Winter. Eckhoff, Hanne M. & Haug, Dag T. T. 2015. Aspect and prefixation in Old Chruch Slavonic. Diachronica 32(2): 186-230. Eckhoff, Hanne M. & Berdicevskis, Aleksandrs. 2015. Linguistics vs. digital editions: The Tromsø Old Russian and OCS Treebank. Scripta & e-Scripta 14-15: 9-25. Ernout, Alfred & Thomas, Francois. 1964[1951]. Syntaxe Latine . Paris: Klincksieck. Eska, Joseph. 1994. Rethinking the Evolution of the Celtic Constituent Configuration. Münchener Studien zur Sprachwissenschaft 55: 7-39 Eska, Joseph. 2007. Bergin’s Rule. Syntactic diachrony and discourse strategy. Diachronica 24(2): 253-278. Eska, Joseph. 2012. Absolute and Conjunct, Cowgill and Apocope. In The Indo-European Verb. Proceedings of the Conference of the Society for Indo-European Studies, Los Angeles 13-15 September 2010 , H. Craig Melchert (ed), 51-59. Wiesbaden: Reichert. Fil’, Julija V. 2011. O glagol’noj poliprefiksacii v staroslavjanskom jazyke [On verbal multiple prefixation in Old Church Slavic]. Vestnik Tomskogo gosudarstvennogo universiteta 352: 37-41. Filip, Hana. 1999. Aspect, Eventuality Types, and Noun Phrase Semantics . New York: Garland. Filip, Hana. 2003. Prefixes and the Delimitation of Events. Journal of Slavic Linguistics 11(1): 55-101. Filip, Hana. 2012. Lexical Aspect. In The Oxford Handbook of Tense and Aspect , Robert I. Binnick (ed), 721-751. Oxford: Oxford University Press. Fillmore, Charles J. 1968. The Case for Case. In Universals in Linguistic Theory, Emmon W. Bach & Robert T. Harms (eds), 1-88. New York: Holt, Rinehart & Winston. Fillmore, Charles J. 1982. Frame semantics. Linguistics in the morning calm , The Linguistic Society of Korea (ed), 111-137. Seoul: Hanshin. Fintel, Kai von. 1995. The formal semantics of grammaticalization. North East Linguistics Society 25: 175-190. Fischer, Olga. 2008. On analogy as the motivation for grammaticalization. Studies in Language 32(2): 336-382. Fischer, Olga. 2011a. Cognitive iconic grounding of reduplication in language. In Semblance and signification, Pascal Michelucci, Olga Fischer & Christina Ljungberg (eds), 55-81. Amsterdam: Benjamins. Fischer, Olga. 2011b. Grammaticalization as analogically driven change? In The Oxford Handbook of Grammaticalization, Heiko Narrog & Bernd Heine (eds), 31-42. Oxford: Oxford University Press. Fischer, Olga, Rosenbach, Anette & Stein, Dieter (eds). 2000. Pathways of Change: Grammaticalization in English . Amsterdam: Benjamins. Fischer, Olga, Norde, Muriel & Perridon, Harry. 2004. Introduction. In Up and Down the Cline - the Nature of Grammaticalization [Typological Studies in Language 59] , Olga Fischer, Muriel Norde & Harry Perridon (eds), 73-96. Amsterdam: Benjamins. Forsyth, James. 1970. A Grammar of Aspect. Cambridge: Cambridge University Press. <?page no="412"?> 412 References Forsyth, James. 1972. The Nature and Development of the Aspectual Opposition in the Russian Verb. The Slavonic and East European Review 50(121): 493-506. Fortson, Benjamim W. IV . 2004. Indo-European Language and Culture: An Introduction . Oxford: Blackwell. Fränkel, Hermann. 1923. Homerische Wörter. In Antidōron: Festschrift Jakob Wackernagel zur Vollendung des 70. Lebensjahres, 274-282. Göttingen: Vandenhoeck & Ruprecht. Fraenkel, Eduard. 1964[1933]. Kolon und Satz, II : Beobachtungen zur Gliederung des antiken Satzes. In Kleine Beiträge zur klassischen Philologie , vol. 1, Eduard Fraenkel (ed), 93-130. Roma: Edizioni di Storia e Letteratura. [Originally published in Nachrichten der Göttinger Gesellschaft der Wissenschaften , Phil.-hist. Klasse: 319-354.] Friedman, Victor A. & Joseph, Brian D. Forthc 2019. The Balkan Languages . Cambridge: Cambridge University Press. Friedrich, Paul. 1974. On aspect theory and Homeric aspect. International Journal of American Linguistics 40: 1-44. Friedrich, Paul. 1976. The devil’s case: PIE as type II . In Linguistic studies offered to Joseph Greenberg on the occasion of his 60th birthday , vol. 3, Alphonse Juilland (ed), 463-480. Saratoga ( CA ): Anma Libri. Friedrich, Paul. 1987. The Proto-Indo-European Adpreps (Spatio-temporal Auxiliaries). In Festschrift for Henry Hoenigswald: on the occasion of his 70th birthday, George Cardona & Norman H. Zide (eds), 131-142. Tübingen: Narr. Frigione, Chiara. 2015. Materiali per un lessico verbale del paleoslavo di impostazione morfosintattica . Università per Stranieri di Siena (Italy)/ Universität zu Köln (Germany), Ph.D. thesis. Frisk, Hjalmar. 1960. Griechisches Etymologisches Woerterbuch . Heidelberg: Winter. Fritz, Matthias. 1997. Die semantischen und syntaktischen Relationen von Lokalpartikeln mit drei Kasus bei Homer . Berlin: Fachbereich Altertumswissenschaften. Gabelentz, Georg von der. 1901[1891]. Die Sprachwissenschaft, ihre Aufgaben, Methoden und bisherigen Ergebnisse , 2nd edn. Leipzig: Weigel. García Ramón, José L. 2002. Zu Verbalcharakter, morphologischer Aktionsart und Aspekt in der Indogermanischen Rekonstruktion. In Indogermanische Syntax. Fragen und Perspektiven, Heinrich Hettrich (ed), 105-136. Wiesbaden: Reichert. Garde, Paul. 2004. La préposition-préverbe, marque de fabrique de l’indo-européen . Paper presented at University of Provence, accessed on May 27 th 2010. 〈http: / / sites.univprovence.fr/ wclaix/ cal04-05.htm〉 Geiss, Heinz. 1957. Bemerkungen zu ἅλλομαι and πάλλομαι bei Homer. Münchener Studien zur Sprachwissenschaft 11(1): 62-66. Gelderen, Elly Van. 2004. Grammaticalization as Economy. Amsterdam: Benjamins. Gelderen, Elly Van. 2011. Grammaticalization and generative grammar: a difficult liaison. In The Oxford Handbook of Grammaticalization, Heiko Narrog & Bernd Heine (eds), 43-55. Oxford: Oxford University Press. Geldner, Karl F. 1951-1957. Der Rig-Veda: aus dem Sanskrit ins Deutsche Übersetzt . Cambridge ( MA ): Harvard Oriental Studies. <?page no="413"?> References 413 Giannakis, Giorgios K. 1997. Studies in the Syntax and Semantics of the Reduplicated Presents of Homeric Greek and Indo-European . Innsbruck: Innsbrucker Beiträge zur Sprachwissenschaft. Gil, David. 2011. Para-linguistic usages of clicks. In The world atlas of language structures online , chapter 142, Matthew S. Dryer & Martin Haspelmath (eds), Munich: Max Planck Digital Library. 〈http: / / wals.info/ chapter/ 142〉 Givón, Talmy. 1971. Historical syntax and synchronic morphology. Chicago Linguistics Society Proceedings 7: 394-415. Givón, Talmy. 1979. On Understanding Grammar . New York: Academic Press. Givón, Talmy. 1984. Syntax. A Functional-Typological Introduction. Amsterdam: Benjamins. Goldstein, David M. 2010. Wackernagel’s Law in Fifth-Century Greek . University of California (Berkeley, CA ), Ph.D. thesis. 〈http: / / digitalassets.lib.berkeley.edu/ etd/ ucb/ text/ Goldstein_berkeley_0028E_10542.pdf〉 Gonda, Jan. 1962. The Aspectual Function of the Rig-Vedic Present and Aorist . The Hague: Mouton. Gonda, Jan. 1968. The Sanskrit particle api . Lingua 21: 183-196. Grassi, Cesare. 1966. Problemi di sintassi Latina: consecutio temporum e aspetto nel verbo latino . Florence: La Nuova Italia. Grassmann, Hermann. 1936[1873]. Worterbuch zum Rig-Veda . Leipzig: Brockhaus. Greč, Nikolaj I. 1827. Praktičeskaja russkaja grammatika [Practical Russian Grammar]. Saint-Petersburg: Verlagsort. Greene, David. 1977. Archaic Irish. In Indogermanisch und Keltisch, Karl H. Schmidt (ed), 11-33. Wiesbaden: Reichert. Griffith, Aaron & Stifter, David. 2007-2013. A dictionary of the Old-Irish glosses in the Milan Codex Ambrosianus C 301 . Institut für Sprachwissenschaft, Universität Wien. Grinevald, Colette. 2003. The expression of static location in a typological perspective. In Space in Languages: Linguistic Systems and Cognitive Categories [Typological Studies in Language 66], Maya Hickmann & Stéphane Robert (eds), 29-58. Amsterdam: Benjamins. Haas, Florian. 2007. The development of English each other : grammaticalization, lexicalization, or both? English Language and Linguistics 1(1): 31-50. Hagège, Claude. 2010. Adpositions . Oxford: Oxford University Press. Hale, Kenneth. 1983. Walpiri and the grammar of non-configurational languages. Natural Language & Linguistic Theory 1(1): 5-47. Hamp, Eric P. 1982. Review of Studies in descriptive and historical linguistics. Festschrift for Winfred P. Lehmann ed. by Paul J. Hopper. Amsterdam: Benjamins, 1977. Zeitschrift für celtische Philologie 39: 343-346. Harðarson Jón Axel. 1993. Studien zum urindogermanischen Wurzelaorist und dessen Vertretung im Indoiranischen und Griechischen. Innsbruck: Innsbrucker Beiträge zur Sprachwissenschaft. Harris, Alice. 2003. Preverbs and their origins in Georgian and Udi. In Yearbook of Morphology 2003 , Geert E. Booij & Jaap van Marle (eds), 61-78. Dordrecht: Kluwer. <?page no="414"?> 414 References Hartmann, Daniel. 1994. Particles. In The Encyclopedia of Language and Linguistics, vol. 1, R. E. Asher & J. M. Y Simpson (eds), 2953-2958. Oxford: Pergamon. Haspelmath, Martin. 1997. From space to time . München: Lincom. Haspelmath, Martin. 1999. Why is grammaticalization irreversible? Linguistics 37: 1034-68. Haspelmath, Martin. 2001. Word classes/ parts of speech. In International Encyclopedia of Social and Behavioral Sciences, Paul B. Baltes & Neil J. Smelser (eds), 16 538-16 545. Amsterdam: Pergamon. Haspelmath, Martin. 2004. On directionality in language change with particular reference to grammaticalization. In Up and Down the Cline - the Nature of Grammaticalization [Typological Studies in Language 59] , Olga Fischer, Muriel Norde & Harry Perridon (eds), 73-96. Amsterdam: Benjamins. Haspelmath, Martin. 2006. Terminology of case. In The Oxford Handbook of Case , Andrej L. Malchukov & Andrew Spencer (eds), 505-517. Oxford: Oxford University Press. Haspelmath, Martin. 2007a. Coordination. In Language typology and syntactic description, vol. 2: Complex constructions 2nd edn. , Timothy Shopen (ed), 1-51. Cambridge: Cambridge University Press. Haspelmath, Martin. 2007b. Pre-established categories don’t exist: consequences for language description and typology. Linguistic Typology 11(1): 119-132. Haspelmath, Martin. 2011. The gradual coalescence into ‘words’ in grammaticalization. In The Oxford Handbook of Grammaticalization, Heiko Narrog & Bernd Heine (eds), 342-355. Oxford: Oxford University Press. Haug, Dag T. T. 2009. Does Homeric Greek have prepositions? Or local adverbs? (And what’s the difference anyway? ). In Grammatical Change in Indo-European Languages: Papers presented at the workshop on Indo-European Linguistics at the XVIII th International Conference on Historical Linguistics, Montreal, 2007 [Current Issues in Linguistic Theory 305], Vit Bubenik & John Hewson (eds), 103-120. Amsterdam: Benjamins. Haug, Dag T. T. & Jøhndal, Marius L. 2008. Creating a Parallel Treebank of the Old Indo-European Bible Translations. In Proceedings of the Second Workshop on Language Technology for Cultural Heritage Data (LaTe CH 2008), Caroline Sporleder & Kiril Ribarov (eds), 27-34. 〈http: / / www.hf.uio.no/ ifikk/ english/ research/ projects/ proiel/ Activities/ proiel/ publications/ marrakech.pdf〉 Hauge, Kjetil Rå. 2002. At the Boundaries of the Balkan Sprachbund - Pragmatic and Paralinguistic Isomorphisms in the Balkans and Beyond. Mediterranean Language Review 14: 21-40. Haverling, Gerd. 2003. On prefixes and actionality in Classical and Late Latin. Acta Linguistica Hungarica 50(1): 113-135. Haverling, Gerd. 2008. On the development of actionality, tense, and viewpoint from Early to Late Latin. In Interdependence of Diachronic and Synchronic Analyses , Folke Josephson & Ingmar Söhrman (eds), 73-104. Amsterdam: Benjamins. <?page no="415"?> References 415 Haverling, Gerd. 2010. Actionality, Tense, and Viewpoint. In Perspectives on Historical Latin Syntax (chap. 4, vol. II ), Philip Baldi & Pierluigi Cuzzolin (eds), 277-524. Berlin: de Gruyter. Heinämäki, Orvokki. 1984. Aspect in Finnish. In Aspect Bound. A voyage into the realm of Germanic, Slavonic and Finno-Ugrian aspectology, Casper de Groot & Hannu Tommola (eds), 153-178. Dordrecht: Foris. Heine, Bernd. 1997. Grammaticalization theory and its relevance for African linguistics. In African Linguistics at the Crossroads, Robert K. Herbert (ed), 1-15. Cologne: Köppe. Heine, Bernd. 2003. On degrammaticalization. In Historical Linguistics 2001 , Barry Blake & Kate Burridge (eds), 163-179. Amsterdam: Benjamins. Heine, Bernd, Claudi, Ulrike & Hünnemeyer, Friederike. 1991. Grammaticalization: A Conceptual Framework . Chicago: University of Chicago Press. Heine, Bernd & Kuteva, Tania. 2005. Language Contact and Grammatical Change . Cambridge: Cambridge University Press. Heine, Bernd & Kuteva, Tania. 2006. The Changing Languages of Europe . Oxford: Oxford University Press. Heine, Bernd & Kuteva, Tania. 2007. The Genesis of Grammar . Oxford: Oxford University Press. Heine, Bernd & Reh, Mechthild. 1982. Patterns of Grammaticalization in African Languages . Institut für Sprachwissenschaft, Universität zu Köln. Hettrich, Heinrich. 1991. Syntax und Wortarten der Lokalpartikeln des R̥gveda I: ádhi . Münchener Studien zur Sprachwissenschaft 52: 27-76. Hettrich, Heinrich. 1993. Syntax und Wortarten der Lokalpartikeln des R̥gveda II : antár . Münchener Studien zur Sprachwissenschaft 54: 147-176. Hettrich, Heinrich. 2002. Syntax und Wortarten der Lokalpartikeln des R̥gveda III : pári . In Novalis Indogermanica. Festschrift für Günter Neumann zum 80. Geburtstag, Matthias Fritz & Susanne Zeilfelder (eds), 215-242. Graz: Leykam. Hettrich, Heinrich. 2007. Materialien zu einer Kasussyntax des R̥gveda . Würzburg: Universität Würzburg, Institut für Altertumswissenschaften / Lehrstuhl für Vergleichende Sprachwissenschaft. 〈http: / / www.vergl-sprachwissenschaft.phil1. uni-wuerzburg.de/ fileadmin/ 04080400/ Materialien.pdf〉 Hettrich, Heinrich. 2012. Präpositionalausdrücke bei Homer. In Homer, gedeutet durch ein großes Lexikon [Akten des Hamburger Kolloquiums vom 6.-8. Oktober 2010 zum Abschluss des Lexikons des Frühgriechischen Epos], Michael Meier-Brügger (ed), 45-61. Berlin: de Gruyter. Hettrich, Heinrich, Casaretto, Antje & Schneider, Caroline. 2010. Syntax and Wortarten der Lokalpartikeln des R̥gVeda. IV : I. Allgemeines (H. Hettrich); II . úpa (A. Casaretto); III . áva (C. Schneider). Münchener Studien zur Sprachwissenschaft 64: 17-130. Hewitt, Steve. 2009. The question of a Hamito-Semitic substratum in Insular Celtic. Language and Linguistics Compass 3: 972-995. Hewson, John & Bubenik, Vit. 2006. From Case to Adposition. The development of configurational syntax in Indo-European languages. Amsterdam: Benjamins . <?page no="416"?> 416 References Himmelmann, Nikolaus P. 2004. Lexicalization and grammaticization: Opposite or orthogonal? In What Makes Grammaticalization - A Look from its Fringes and its Components [Trends in Linguistics: Studies and Monographs 158], Walter Bisang, Nikolaus P. Himmelmann & Björn Wiemer (eds), 19-40. Berlin: de Gruyter. Hinrichs, Erhard W. 1985. A Compositional Semantics for Aktionsarten and NP Reference in English . Ohio State University (Columbus, OH ), Ph.D. thesis. Hoddinott, William G. & Kofod, Frances M. 1976. Topic E: Simple and compound verbs: conjugation by auxiliaries in Australian verbal systems. Djamindjungan. In Grammatical Categories in Australian Languages , R. M. W. Dixon (ed), 698-704. Canberra: Australian Institute of Aboriginal Studies. Hofman, Rijcklof. 1993. The Linguistic Preoccupations of the Glossators of the St Gall Priscian. Historiographia Linguistica 20(1): 111-126. Hopper, Paul J. 1987. Emergent grammar. Berkeley Linguistics Society: Proceedings of the Annual Meeting 13: 139-157. Hopper, Paul J. & Thompson, Sandra. 1980. Transitivity in grammar and discourse. Language 56: 251-299. Hopper, Paul J. & Traugott, Elisabeth C. 1993. Grammaticalization . Cambridge: Cambridge University Press. Hopper, Paul J. & Traugott, Elisabeth C. 2003. Grammaticalization , 2nd edn. Cambridge: Cambridge University Press. Horrocks, Geoffrey C. 1981. Space and Time in Homer: Prepositional and Adverbial Particles in the Greek Epic . New York: Arno Press. Horrocks, Geoffrey C. 1997. Homer’s dialect. In A new companion to Homer , Ian Morris & Barry B. Powell (eds), 193-217. Leiden: Brill. Horrocks, Geoffrey C. 2010. Greek. A History of the Language and its Speakers . Oxford: Blackwell. Humbert, Jean. 1960. Syntaxe grecque. Paris: Klincksieck. Humboldt, Wilhelm Von. 1822. Über die Entstehung der grammatischen Formen und ihren Einfluss auf die Ideenentwicklung . Abhandlungen der Akademie der Wissenschaften zu Berlin. [Reprinted 1972, Darmstadt: Wissenschaftliche Buchgesellschaft.] Iacobini, Claudio, Corona, Luisa, De Pasquale, Noemi & Buoniconto, Alfonsina. 2017. How should a “classical” Satellite-Framed Language behave? Path encoding asymmetries in Ancient Greek and Latin. In Space in Diachrony, Silvia Luraghi, Tatiana Nikitina & Chiara Zanchi (eds), 95-118. Amsterdam: Benjamins. Iacobini, Claudio & Masini, Francesca. 2006. The emergence of verb-particle constructions in Italian: locative and actional meanings. Morphology 16: 155-188. Ikegami, Yoshihiko. 1987. ‘Source’ and ‘Goal’: A case of linguistic dissymmetry. In Concept of Case , René Dirven & Günter Radden (eds), 122-146. Tübingen: Gunter Narr. Imbert, Caroline. 2006. Ancient languages in a functional-typological perspective: The coding strategy(-ies) of Path in Homeric Greek. Paper presented at the Linguistics Department Colloquium , Eugene (OR), May 25 th 2006. <?page no="417"?> References 417 Imbert, Caroline. 2007. Typology meets ancient languages. Grammaticalization and Path coding in Homeric Greek. Paper presented at the Deuxième Colloque International de l'Association Française de Linguistique Cognitive ( AFL iCo), Lille (France), May 10 th -12 th 2007. Imbert, Caroline. 2008. Dynamique des systemes et motivations fonctionnelles dans lʼencodage de la trajectoir: Description typologique du grec homérique et du vieil-anglais . Université Lumière Lyon 2 (France), Ph.D. thesis. Imbert, Caroline .2009. Affix-order constraints in Path coding: About satellites, adpositions and the gradience of categories. Paper presented at the Association for Linguistic Typology - 8th Biennial Meeting ( ALT 8) , Berkeley ( CA ), July 23 th -26 th 2009. Imbert, Caroline. 2010. Multiple preverbation in Homeric Greek: A typological insight. Cogni-Textes [En ligne], Vol. 4 . 〈http: / / cognitextes.revues.org/ 387〉 Imbert, Caroline & Craig, Grinevald. 2004. Twenty years of relational preverbs: a grammaticalization account.Paper presented at the conference New Reflections on Grammaticalization 4 ( NRG 4) , Leuven (Belgium), July 16 th -19 th 2008. Inglese, Guglielmo. 2016. Verbal roots and lexical aspect in Hittite: a cognitive linguistics approach. Paper presented at the 15 th Fachtagung der Indogermanischen Gesellshaft , Wien (Austria), September 13 th -16 th 2016. Inglese, Guglielmo. 2017. A synchronic and diachronic typology of Hittite reciprocal constructions. Studies in Language 41(4): 956-1006. Inglese, Guglielmo & Zanchi, Chiara. Forthc. Reciprocal constructions in Homeric Greek: a typological and corpus-based approach. To appear on Folia Linguistica Historica. Irigoin, Jean. 1954. Note sur l’anastrophe des prépositions en grec. Glotta 33: 90-100. Isaac, Graham R. 1993. Issues in the reconstruction and analysis of insular Celtic syntax and phonology. Ériu 44: 1-32. Isaac, Graham R. 2000. The Most Recent Model of the Development of Absolute and Conjunct Flexion. Ériu 51: 63-68. Istratkova, Vyara. 2004. On multiple prefixation in Bulgarian. Nordlyd 32(2): 301-321. Ivanov, Vjačeslav V. 1973. Tipologija razvitija slavjanskich i indoevropejskich predlogov i poslelogov [The typological development of Slavic and Indo-European prepositions and postpositions]. In Strukturno-tipologičeskie issledovanija v oblasti grammatiki slavjanskich jazykov [Structural-typological Investigations of the Grammar of the Slavic Languages], 51-60 . Moscow: Nauka. Jackendoff, Ray. 1972. Semantic Interpretation in Generative Grammar . Cambridge ( MA ): MIT Press. Jackendoff, Ray. 1990. Semantic Structures . Cambridge ( MA ): MIT Press. Jacobsohn, Hermann. 1926. Wackernagel, Vorlesungen über Syntax. Gnomon 2: 369-395. Jamison, Stephanie W. & Brereton, Joel P. 2014. The Rigveda. The earliest poetry of India (3 vols.). Oxford: Oxford University Press. <?page no="418"?> 418 References Janda, Richard. 2001. Beyond “pathways” and “unidirectionality”: on the discontinuity of language transmission and the counterability of grammaticalization. Language Sciences 23: 265-340. Janda, Laura A. 2007. Aspectual clusters of Russian verbs. Studies in Language 3: 607-648. Janda, Laura A. 2012. Russkie pristavki kak Sistema glagol’nych klassifikatorov [Russian prefixes as a system of verbal classificators]. Voprosy jazykoznanija 6: 3-47. Janda, Laura A. & Šarić, Ljiljana. 2009. Cognitive Linguistics in Exploration of Slavic Languages: A Bibliography of Cognitive Linguistic Analyses of Slavic Data . 〈https: / / www.academia.edu/ 7479592/ Slavic_Cognitive_Linguistics_A_Bibliography_with_ Laura_Janda_〉 Janda, Laura A., Endersen, Anna, Kuznetsova, Julia, Lyashevskaya, Olga, Makarova, Anastasia, Nesset, Tore, Sokolova S. 2013. Why Russian Aspectual Prefixes Aren’t Empty. Prefixes as Verb Classifiers . Bloomington ( IN ): Slavica. Janda, Laura A. & Olga Lyashevskaya. 2013. Semantic Profiles on Five Russian Prefixes: po-, s-, za-, na-, pro-. Journal of Slavic Linguistics 21(2): 211-258. Janda, Richard D. & Joseph, Brian D. 1991. Meta-Templates and the Underlying (Dis-) Unity of Reduplication in Sanskrit. In ESCOL ‘91. Proceedings of the Eighth Eastern States Conference on Linguistics , Germán F. Westphal, Benjamin Ao & Hee-Rahk Chae (eds), 160-173 . Columbus: The Ohio State University Department of Linguistics. Joseph, Brian D. 2000. On the Development of Modern Greek óxi ‘no’. In Naturally! Linguistic Studies in Honour of Wolfgang Ulrich Dressler presented on the occasion of his 60th birthday, Chris Schaner-Wolles, John Rennison & Friedrich Neubarth (eds), 207-214. Torino: Rosenberg & Sellier. Joseph, Brian D. 2001. Is there such a thing as “grammaticalization”? In Grammaticalization: A Critical Assessment [special issue of Language Sciences 23(2-3)], Lyle Campbell (ed), 163-86. London: Pergamon. Joseph, Brian D. 2003. Morphologization from syntax. In The Handbook of Historical Linguistics, Brian D. Joseph & Richard D. Janda (eds), 472-492. Oxford: Blackwell. Joseph, Brian D. 2004. Rescuing traditional (historical) linguistics from grammaticalization theory. In Up and Down the Cline - the Nature of Grammaticalization [Typological Studies in Language 59] , Olga Fischer, Muriel Norde & Harry Perridon (eds), 45-71. Amsterdam: Benjamins. Joseph, Brian D. 2006. How accommodating of change is grammaticalization? The case of “lateral shifts”. Logos and Language 6(2): 1-7. Joseph, Brian D. 2011. Grammaticalization: A General Critique. In The Oxford Handbook of Grammaticalization, Heiko Narrog & Bernd Heine (eds), 193-205. Oxford: Oxford University Press. Joseph, Brian D. 2014. What counts as (an instance of) Grammaticalization? In Refining Grammaticalization [special issue of Folia Linguistica 48(2)], Ferdinand von Mengden & Horst Simon (eds), 1-23. Berlin: de Gruyter. <?page no="419"?> References 419 Joseph, Brian D. 2017. Modern Greek Confronting its Past: Archaism and Innovation in Language. Paper presented at the 29th Annual Thomas Leontis Lecture, Columbus ( OH ) March 30 th 2017. Joseph, Brian D. & Janda, Richard D. (eds). 2003. The Handbook of Historical Linguistics . Oxford: Blackwell. Kahane, Αhuvia. 1994. The Interpretation of Order: a Study in the Poetics of Homeric Repetition . Oxford: Oxford University Press. Kajitani, Motomi. 2005. Semantic Properties of Reduplication among the World’s Languages. LSO Working Papers in Linguistics 5: 93-106. Kamphuis, Jaap. 2016. Verbal aspect in Old Church Slavonic . University of Leiden (Netherlands), Ph.D. thesis. Kemenade, Ans van & Los, Bettelou. 2003. Particles and prefixes in Dutch and English. In Yearbook of Morphology 2003, Geert E. Booij & Jaap van Marle (eds), 79-119. Dordrecht: Kluwer. Keller, Rudi. 1990. Sprachwandel: Von der unsichtbaren Hand in der Sprache . Tübingen: Francke. Kiparsky, Paul. 2011. Grammaticalization as optimization. In Grammatical Change: Origins, Nature, Outcomes , Diane Jonas, John Whitman & Andrew Garrett (eds), 15-51. Oxford: Oxford University Press. Kittilä, Seppo, Västi, Katja & Ylikoski, Jussi (eds). 2011. Case, Animacy, and Semantic Roles [Typological Studies in Language 99]. Amsterdam: Benjamins. Klein, Jared S. 1987. The Two Senses of the Term anaphora and their Functional Unity: Evidence from the RigVeda. In Festschrift for Henry Hoenigswald: on the occasion of his 70th birthday, George Cardona & Norman H. Zide (eds), 193-199. Tübingen: Narr. Klein, Jared S. 2003. Āmreḍitas and Related Constellations in the Rigveda. Journal of the American Oriental Society 123(4): 773-802. Klein, Jared S. 2007. On the Nature and Function of Preverb Repetition in the Rigveda. Studien zur Indologie und Iranistik 24: 91-103. Klein, Jared S. 2008. Adverbial Repetition in the Rigveda. In Indologica. Elizarenkova Memorial Volume , vol. 1, Leonid Kulikov & Maxim Rusanov (eds), 167-182. Moscow: Russian State University for the Humanities. Klein, Jared S. 2012. Categories and Types of Interstanzaic Repetition in the Rigveda. Historische Sprachforschung 125: 191-210. Klikovac, Duška. 2004. Metafore u mišljenju i jeziku [Metaphors in cognition and language]. Belgrade: Biblioteka XX vek. Klikovac, Duška. 2006. Semantika predloga: studija iz kognitivne lingvistike [The semantics of prepositions: studies from Cognitive Linguistics]. Belgrade: Filološki fakultet. Knittel, Marie-Laurence. 2015. Preverbs, aspect and nominalization in Hungarian. Recherches linguistiques de Vincennes 43(1): 47-76. Koch, John T. 1987. Prosody and the Old Celtic verbal complex. Ériu 38: 143-176. Koffka, Kurt. 1935. Principles of Gestalt Psychology. London: Lund Humphries. <?page no="420"?> 420 References Kofod, Frances M. 1996. Kija Learner’s Grammar . Halls Creek ( WA ): Kimberley Language Resource Centre. Köhler, Wolfgang. 1929. Gestalt psychology . New York: Liveright. Kortlandt, Frederik. 1979. The Old Irish absolute and conjunct endings and questions of relative chronology. Ériu 30: 35-53. Kortlandt, Frederik. 1994. Absolute and Conjunct Again. Münchener Studien zur Sprachwissenschaft 55: 61-68. Kozhanov, Kirill. 2016. Verbal prefixation and argument structure in Lithuanian. In Argument Realization in Baltic [Valency, Argument Realization and Grammatical Relations in Baltic 3], Axel Holvoet & Nicole Nau (eds), 363-402. Amsterdam: Benjamins. Krause, Wolfgang. 1953. Handbuch des Gotischen. München: Beck. Krisch, Thomas. 1984 . Konstruktionsmuster und Bedeutungswandel indogermanischer Verben. Anwendungsversuche von Valenztheorie und Kasusgrammatik auf Diachronie und Rekonstruktion [Europäische Hochschulschriften : Reihe 21, Linguistik; Bd. 35]. Frankfurt am Main: Lang. Krisch, Thomas. 2006. RIVELEX . Rigveda-Lexikon. Unter Mitarbeit von Christina Katsikadeli, Stefan Niederreiter, Thomas Kaltenbacher. Band 1: Wörter beginnend mit "a“ . Graz: Leykam. Krisch, Thomas. 2012. RIVELEX . Rigveda-Lexikon. Unter Mitarbeit von Christina Katsikadeli, Stefan Niederreiter, Konstantinos Sampanis, Sabine Ziegler. Band 2: Wörter beginnend mit anderen Vokalen als „a“. Graz: Leykam. Kulikov, Leonid. 2007. Reciprocal constructions in Vedic. In Reciprocal Constructions [Typological Studies in Language 71], Vladimir P. Nedjalkov (ed), 709-738. Amsterdam: Benjamins. Kulikov, Leonid. 2012. Vedic Preverbs as Markers of Valency-Changing Derivations. Studies in Language 34(4): 721-746. Kulikov, Leonid. 2017. Indo-Aryan. In The Indo-European Languages, Mate Kapović (ed), 214-262. London: Routledge. Kuryłowicz, Jerzy. 1929. La genèse d’aspects verbaux slaves. Prace filologiczne 14: 644-657. Kuryłowicz, Jerzy. 1964. The Inflectional Categories of Indo-European . Heidelberg: Winter. Kuryłowicz, Jerzy. 1975[1965]. The evolution of grammatical categories. Esquisses linguistiques 2: 38-54. [ Diogenes 1965: 55-71.] Kurz, Josef et al. 1966-1997. Lexicon linguae palaeoslovenicae.Slovnik jazyka staroslověnskeho (52 volumes). Praha: Nakl. Československé akademie věd. Lakoff, George. 1977. Linguistic Gestalts. Chicago Linguistic Society Proceedings 13: 236-287. Lakoff, George. 1987. Women, Fire and Dangerous Things. Chicago: Chicago University Press. Lakoff, George & Johnson, Mark. 1980. Metaphors we live by . Chicago: Chicago University Press. <?page no="421"?> References 421 Langacker, Ronald W. 1987. Foundations of Cognitive Grammar. Stanford: Stanford University Press. Langacker, Ronald W. 2008. Cognitive Grammar: A Basic Introduction . New York: Oxford University Press. Laum, Bernhard. 1928. Das alexandrinische Akzentuationssystem . Paderborn: Schöning. Lazarevska-Stančevska, Jovanka. 2004. Metaforizacija na over , under i out kaj kompleksite zborovi vo angliskiot jazik i nivnite ekvivalenti vo makedonskiot jazik [Metaphorization of over, under and out as complex words in English and their equivalents in Macedonian]. UKIM (Skopje, Macedonia), Ph.D. thesis. Łazorczyk, Agnieszka Agata. 2010. Decomposing Slavic aspect: The role of aspectual morphology in Polish and other Slavic languages . University of Southern California (Los Angeles, CA ), Ph.D. thesis. Le Blanc, Nicholas L. 2010. The polysemy of an “empty” prefix: A corpus-based cognitive semantic analysis of the Russian verbal prefix po-. University of North Carolina at Chapel Hill, Ph.D. thesis. Lehmann, Christian. 1989. Grammatikalisierung und Lexikalisierung. Zeitschrift für Phonetik, Sprachwissenschaft und Kommunikationsforschung 42: 11-19. Lehmann, Christian. 1995[1982]. Thoughts on Grammaticalization . Munich: Lincom Europa [revision of Thoughts on Grammaticalization: A Programmatic Sketch , Arbeiten des Kölner Universalienprojekts 49]. Lehmann, Christian. 2002. New reflections on grammaticalization and lexicalization. In New Reflections on Grammaticalization [Typological Studies in Language 49] , Ilse Wischer & Gabriele Diewald (eds), 1-18. Amsterdam: Benjamins. Lehmann, Christian. 2004. Theory and method in grammaticalization. Zeitschrift für Germanistische Linguistik 32(2): 152-187. [Reprinted in Grammatikalisierung, Gabriele Diewald (ed), 152-187. Berlin: de Gruyter.] Lehmann, Christian. 2015. Thoughts on Grammaticalization , 3rd edn. [Classics in Linguistics]. Berlin: Language Science Press. Leroy, Maurice. 1958. L’aspect verbal en grec ancien: esquisse de problème. Revue Belge de Philologie et d’Histoire 36: 128-138. Leskien, August. 1922[1971]. Handbuch der altbulgarischen (Altkirchenslavischen) Sprache: Grammatik - Texte - Glossar . Heidelberg: Winter. Levin, Beth. 1993. English Verb Classes and Alternations. Chicago: The University of Chicago Press. Levinson, Stephen C. 1996. Frames of reference and Molyneux's question: Cross-linguistic evidence. In Language and space, Paul Bloom, Mary A. Peterson, Lynn Nadel & Merrill F. Garrett (eds), 109-169. Cambridge ( MA ): MIT Press. Lewis, Henry & Pedersen, Holger. 1961. A Concise Comparative Celtic Grammar. Göttingen: Vandenhoeck & Ruprecht. Lightfoot, Douglas. 2011. Grammaticalization and lexicalization. In The Oxford Handbook of Grammaticalization, Heiko Narrog & Bernd Heine (eds), 438-449. Oxford: Oxford University Press. <?page no="422"?> 422 References Lindsay, Wallace M. 1913. Sexti Pompei Festi De verborum significatu quae supersunt cum Pauli epitome . Leipzig: Teubner. Lord, Albert B. 1960. The singer of tales. Cambridge ( MA ): Harvard University Press. Lowe, John J. 2015. Participles in Rigvedic Sanskrit. Oxford: Oxford University Press. Lubotsky, Alexander M. 1994. RV . ávidhat. In Früh-, Mittel-, Spätindogermanisch. Akten der IX . Fachtagung der Indogermanischen Gesellschaft vom 5. bis 9. Oktober 1992 in Zürich, George Dunkel et al. (eds), 201-206. Wiesbaden: Reichert. Lüdeling, Anke. 2001. On Particle Verbs and Similar Constructions in German . Stanford: CSLI . Lunt, Horace G. 1965. Old Church Slavonic Grammar. The Hague: Mouton. Lunt, Horace. G. 1977. Limitations of Old Church Slavonic in Representing Greek. In The early versions of the New Testament: Their Origin, Transmission, and Limitations , Bruce M. Metzger (ed), 431-442. Oxford: Clarendon Press. Lupaş, Liana 1972. Phonologie du grec attique [= Janua Linguarum, ser. pract. 164]. The Hague: Mouton. Luraghi, Silvia. 2001. The development of local particles and adverbs in Anatolian as a grammaticalization process. Diachronica 28(1): 31-58. Luraghi, Silvia. 2003. On the Meaning of Prepositions and Cases: The Expression of Semantic Roles in Ancient Greek [Studies in Language Companion Series 67]. Amsterdam: Benjamins. Luraghi, Silvia. 2006. Greek prepositions: patterns of polysemization and semantic bleaching. In Word Classes and Related Topics , Emilio Crespo, Jesús de la Villa & Antonio R. Revuelta Puigdollers (eds), 487-499. Louvain: Peeters. Luraghi, Silvia. 2009. The internal structure of adpositional phrases and the notion of government. In Form and Function in Language Research: Papers in Honor of Christian Lehmann , Johannes Helmbrecht, Yoko Nishina, Zong-Min Shin, Stavros Skopeteas & Elisabeth Verhoeven (eds), 231-246. Berlin: de Gruyter. Luraghi, Silvia. 2010. The rise (and possible downfall) of configurationality. In Continuum Companion to Historical Linguistics , Silvia Luraghi & Vit Bubenik (eds), 212-229. London: Continuum. Luraghi, Silvia. 2012. The spatial meaning of diá with the accusative in Homeric Greek. Mnemosyne 65(3): 357-386. Luraghi, Silvia. Forthc. a. External possessor constructions in Indo-European. In Reconstructing Syntax , Jóhanna Barðdal, Spike Gildea & Eugenio R. Lujan Martines (eds). Leiden: Brill. Luraghi, Silvia. Forthc. b. Configurational language. In WSK 13 Linguistic Typology. Berlin: De Gruyter. Luraghi, Silvia. Forthc. c. Configurationality. In WSK 13 Linguistic Typology . Berlin: De Gruyter. Luraghi, Silvia. Forthc. d. Coordination reduction. In WSK 13 Linguistic Typology . Berlin: De Gruyter. Luraghi, Silvia & Narrog, Heiko (eds). 2014. Perspectives on Semantic Roles [Typological Studies in Language 106] . Amsterdam: Benjamins. <?page no="423"?> References 423 Luraghi, Silvia & Zanchi, Chiara. 2018. Double accusative constructions and ditransitives in Ancient Greek. In Ditransitives, Agnes Korn, Andrej Malchukov & Carina Jahani (eds). Wiesbaden: Reichert. Lyons, John. 1977. Semantics . vol. 2. Cambridge: Cambridge University Press. Mac Coisdealbha, Pádraig. 1998. The syntax of the sentence in Old Irish: Selected studies from a descriptive, historical and comparative point of view [ed. by Graham R. Isaac]. Tübingen: Max Niemeyer. Macdonell, Arthur A. 1910. Vedic Grammar . Delhi: Bhartiya Publishing House. Macdonell, Arthur A. 1916. A Vedic Grammar for students. Oxford: Clarendon Press. Macdonell, Arthur A. 1917. A Vedic Reader for Students . Oxford: Oxford University Press. MacRobert, C. M. 1986. Foreign, Naturalized and Native Syntax in Old Church Slavonic. Transactions of the Philological Society 84(1): 142-166. Maisak, Timur. 2014. Preverbs in Agul: an elaborate system of locative prefixation. Paper presented at East Caucasian preverbs and the compounding: derivation - inflection continuum, held in connection with SWL 6 , Pavia (Italy), September 8 th -10 th 2014. 〈https: / / www.academia.edu/ 18376455/ Preverbs_in_Agul_an_elaborate_system_of_ locative_prefixation_2014_〉 Malchukov, Andrej & Spencer, Andrew (eds). 2009. The Oxford Handbook of Case . Oxford: Oxford University Press. Mallory, James P. & Adams, Douglas Q. 1997. Encyclopedia of Indo-European Culture . London: Fitzroy Dearborn Publishers. Marcialis, Nicoletta. 2007. Introduzione alla lingua paleoslava . Firenze: Firenze University Press. Martinelli, Maria C. 2001. Da Fränkel a Kahane. Considerazioni sulla divisione in cola dell’esametro omerico. Gaia: revue interdisciplinaire sur la Grèce Archaïque 5: 119-129. Masini, Francesca. 2009. Phrasal lexemes, compounds and phrases: a constructionist perspective. Word Structure 2(2): 254-271. Maslov, Jurij S. 1959. Voprosy proischoždenija glagol'nogo vida na IV . meždunarodnom s"ezde slavistov [Questions about the origin of the verbal aspect in the 4 th International Congress of slavists]. Voprosy jazykoznanija 2: 151-157. Matasović, Ranko. 2012a. The substratum in Insular Celtic. Journal of Language Relationship 8: 153-159. Matasović, Ranko. 2012b. The substratum in Insular Celtic (a response to Tatyana Mikhailova). Journal of Language Relationship 8: 165-168. Mayrhofer, Manfred. 1986-2001. Etymologisches Wörterbuch des Altindoarischen . Heidelberg: Indogermanische Bibliothek. McCone, Kim. 1979. Pretonic preverbs and the absolute verbal endings in Old Irish. Ériu 30: 1-34. McCone, Kim. 1982. Further to absolute and conjunct. Ériu 33: 1-29. McCone, Kim. 1985a. The origin of the Old Irish absolute and conjunct verbal inflection. In Grammatische Kategorien. Funktion und Geschichte [Akten der VII Fachta- <?page no="424"?> 424 References gung der Indogermanischen Gesellschaft], Bernfried Schlerath (ed), 261-270. Berlin: Wiesbaden. McCone, Kim. 1985b. The Würzburg and Milan Glosses: Our Earliest Sources of ‘Middle Irish’. Ériu 36: 85-106. McCone, Kim. 1995. Old Irish con·dieig “asks, seeks”, Verbal Noun cuin(d)gid : a Problem of Syncope and Verbal Composition. Éigse 28: 156-159. McCone, Kim. 1997. The Early Irish Verb. Maynooth: Department of Old Irish, National University of Ireland, Maynooth. McCone, Kim. 2006. The Origins and Development of the Insular Celtic Verbal Complex. Maynooth: Department of Old Irish, National University of Ireland, Maynooth. McGregor, William B. 2002. Verb Classification in Australian Languages . Berlin: de Gruyter. McIntyre, Andrew. 2000. German Double Particles as Preverbs. Morphology and Conceptual Semantics [Studien zur deutschen Grammatik 61]. Tübingen: Stauffenburg. McIntyre, Andrew. 2001a. Argument blockages induced by verb particles in English and German: Event modification and secondary predication. In Structural Aspects of Semantically Complex Verbs, Nicole Dehé & Anja Wanner (eds), 131-164. Berlin: Peter Lang. McIntyre, Andrew. 2001b. Idiosyncracy in particle verbs. In Structural Aspects of Semantically Complex Verbs, Nicole Dehé & Anja Wanner (eds), 95-118. Berlin: Peter Lang. McIntyre, Andrew. 2003. Preverbs, argument linking and verb semantics: Germanic prefixes and particles. In Yearbook of Morphology 2003, Geert E. Booij & Jaap van Marle (eds), 1-12. Dordrecht: Kluwer. Meillet, Antoine. 1903. Introduction à l’étude comparative des langues indoeuropéennes . Paris: Hachette. Meillet, Antoine. 1912. L’évolution des formes grammaticales. Scientia (Rivista di scienza) 12(6): 384-400. [Repr. in Meillet 1958: 130-148.] Meillet, Antoine. 1922. Introduction a l’étude comparative des langues classiques . Paris: Champion. Meillet, Antoine. 1924. Le slave commun. Paris: Champion. Meillet, Antoine. 1948. Aperc̨u d’histoire de la langue grecque , 6th edn. Paris: Hachette. Meillet, Antoine. 1958[1912]. Linguistique historique et linguistique générale . Paris: Champion. Meillet, Antoine & Vendryes, Joseph. 1924. Grammaire comparée des langues classiques . Paris: Champion. Melchert, Craig H. 2009. Local Adverbs in Hittite: Synchrony and Diachrony. Language and Linguistics Compass 3(2): 607-620. Merlan, Francesca C. 1994. A Grammar of Wardaman. A language of the Northern Territory of Australia. Berlin: de Gruyter. Metslang, Helle. 2001. Lexical evidence for the parallel development of the Latvian and Livonian verb particles. In The Circum-Baltic Languages. Typology and Contact. <?page no="425"?> References 425 Vol. 2. Grammar and Typology [Studies in Language Companion Series 55] , Östen Dahl & Maria Koptjevskaja-Tamm (eds), 442-479. Amsterdam: Benjamins. Mikailova, Tatyana. 2012. Once again on the pre-Celtic substratum in the British Islands (a response to Ratko Matasović). Journal of Language Relationship 8: 160-164. Milićević, Nataša. 2004. The lexical and superlexical verbal prefix izand its role in the stacking of prefixes. Nordlyd 32(2): 279-300. Mithun, Marianne. 2011. Grammaticalization and explanation. In The Oxford Handbook of Grammaticalization, Heiko Narrog & Bernd Heine (eds), 177-192. Oxford: Oxford University Press. Monro, David B. & Allen, Thomas W. 1920. Homeri Opera . Oxford: Clarendon Press. Moore, Norman. 1882. A Concise Irish Grammar [Trans. of Windisch 1879]. Cambridge: Cambridge University Press. Moravcsick, Edith A. 1978. Reduplicative Constructions. In Universals of Human Language, vol. 3, Greenberg Joseph H. (ed), 297-334. Stanford: Stanford University Press. Morris, William. 1981. The American Heritage Dictionary of the English Language . Boston: Houghton Mifflin. Morris-Jones, John. 1899. Pre-Aryan syntax in Insular Celtic. In The Welsh People , John Rhys & David Brynmor-Jones (eds), 617-641. Oxford: Oxford University Press. Napoli, Maria. 2006. Aspect and Actionality in Homeric Greek. A contrastive analysis . Milano: Franco Angeli. Narrog, Heiko. 2014. The grammaticalization chain of case functions - extension and reanalysis of case marking vs. universals of grammaticalization. In Perspectives on Semantic Roles, Silvia Luraghi & Heiko Narrog (eds), 71-100. Amsterdam: Benjamins. Narrog, Heiko & Heine, Bernd. 2011. Introduction. In The Oxford Handbook of Grammaticalization, Heiko Narrog & Bernd Heine (eds), 1-18. Oxford: Oxford University Press. Nash, David. 1982. Warlpiri verb roots and preverbs. In Papers in Warlpiri Grammar, in Memory of Lothar Jagst . [Work Papers of SIL - AAB , Series A, Vol. 6], Stephen Swartz (ed), 165-216. Berrimah ( NT ): Summer Institute of Linguistics. Nash, David. 1986. Topics in Warlpiri Grammar. New York: Garland. Neri, Sergio. 2007. Cadere e abbattere in indoeuropeo. Sull’etimologia del tedesco fallen , latino aboleo e greco ἀπόλλυμι. Innsbruck: Innsbrucker Beiträge zur Sprachwissenschaft. Neri, Sergio. 2013. Zum urindogermanischen Wort für ‘Hand’. In Multi Nominis Grammaticus. Studies in Classical and Indo-European Linguistics in Honor of Alan J. Nussbaum on the Occasion of his Sixty-Fifth Birthday , Adam I. Cooper, Jeremy Rau & Michael Weiss (eds), 185-205. Ann Arbor: Beech Stave Press. Nesset, Tore, Janda, Laura A. & Endresen, Anna. 2011. Two ways to get out: Radial category profiling and the Russian prefixes vyand iz- . Zeitschrift für Slavistik 56: 377-402. <?page no="426"?> 426 References Nevis, Joel A. & Joseph, Brian D. 1992. Wackernagel affixes: evidence from Balto-Slavic. In Yearbook of Morphology 1992, Geert E. Booij & Jaap van Marle. Dordrecht: Kluwer. Newmeyer, Frederick J. 2001. Deconstructing grammaticalization. Grammaticalization: A Critical Assessment [Special Issue of Language Sciences 23(2 - 3)], Lyle Campbell (ed), 187-230. London: Pergamon. Nooten, Barend A. van & Holland, Gary B. (eds). 1994. Rig Veda: a metrically restored text with an introduction and notes . Cambridge: Harvard University Press. Norde, Muriel. 2009. Degrammaticalization . Oxford: Oxford University Press. Norde, Muriel. 2010. Lehmann’s parameters revisited. In Grammaticalization and Language Change: Origins, Criteria, and Outcomes, Tine Breban, Liselotte Brems, Kristin Davidse & Tanja Mortelmans (eds), 73-110. Amsterdam: Benjamins. Nuti, Andrea. 2010. Having something that you don’t own: apud possessive constructions in Latin and a comparison with locative possessive sentences in Irish. In Studies in Classical Linguistics in Honor of Philip Baldi, B. Richard Page & Aaron D. Rubin (eds), 57-74. Leiden: Brill. O’Brien, Michael A. 1938. Varia II . Ériu 12: 236-244. O’Brien, Michael A. (ed). 1962. Corpus genealogiarum Hiberniae i . Dublin: Dublin Institute for Advanced Studies. Papanastassiou, George. 2011. The preverb ἀπό in Ancient Greek. In Proceedings of the 9th International Conference on Greek Linguistics , Katerina Chatzopoulou, Alexandra Ioannidou & Suwon Yoon (eds), 97-11. Columbus: Ohio State University Department of Linguistics. Parry, Adam (ed). 1971. The making of Homeric verse: the collected papers of Milman Parry . Oxford: Oxford University Press. Papke, Julia K. 2010. Classical Sanskrit preverb ordering: a diachronic study . Ohio State University (Columbus, OH ), Ph.D. thesis. 〈https: / / etd.ohiolink.edu/ 〉 Patri, Sylvain. 2007. Remarque sur le positionnement syntaxique du déterminant adverbial (préverbe) en indo-européen. Indogermanische Forschungen 11(2): 26-38. Pedersen, Holger. 1913. Vergleichende Grammatik der keltischen Sprachen. 2. Band. Bedeutungslehre (Wortlehre) . Göttingen: Vandenhoeck & Ruprecht. Pentland, David H. 2005. Preverbs and particles in Algonquian. Papers of the Algonquian Conference 36: 323-338. Pernée, Lucien. 1983. L’aspect en grec ancient: problèmes d’analyse. Les Études Classiques 51: 297-302. Peterson, David A. 2007. Applicative Constructions . Oxford: Oxford University Press. Petit, Daniel. 2017. Reconstructing Proto-Indo-European Syntax. Course held in connection with the Fourth Pavia International Summer School for Indo-European Linguistics (Pavia, September 4 th -9 th 2017). Pfeiffer, Rudolf. 1968. History of classical scholarship from the beginnings to the end of the Hellenistic age . Oxford: Clarendon. Pinault, Georges-Jean. 1995. Le Problème du Préverbe en Indo-Européen. In Les Préverbes dans les Langues d' Europe: Introduction à l‘Étude de la Préverbation , André Rousseau (ed), 35-59. Lille: Sepentrion. <?page no="427"?> References 427 Plungian, Vladimir A. 2011. Vvedenie v grammatičeskuiu semantiku: grammatičeskie značenija i grammatičeskie sistemy jazykov mira [Introduction to grammatical semantics: Grammatical meanings and grammatical systems of the world’s languages]. Moscow: Russian State University for the Humanities. Pokorny, Julius. 1914. A concise Old Irish grammar and reader . Dublin: Niemeyer. Pokorny, Julius. 1949. Zum nicht-indogermanischen Substrat im Inselkeltischen. Die Sprache 1: 235-245. Pokorny, Julius. 1959. Indogermanisches etymologisches Wörterbuch. Bern: Francke. Poldauf, Ivan. 1954. Spojování s předponami při tvoření dokonavých sloves v češtině. Slovo a slovesnost 15(2): 49-65. Pompei, Anna. 2010. De l’expression de l’espace à l’expression du temps (et de l’aspect) en latin: le cas des préverbes. De lingua latina 3: 1-20. Pompeo, Flavia. 2002. Dall’avverbio localistico alla preposizione in Omero [Biblioteca di Ricerche linguistiche e filologiche 53]. Roma: Il Calamo. Probert, Philomen. 2003. A New Short Guide to the Accentuation of Ancient Greek . London: Bristol Classical Press. Puhvel, Jaan. 1984. Hittite Etymological Dictionary. Volume 1: Words beginning with A . Berlin: de Gruyter. Puhvel, Jaan. 2011 . Hittite Etymological Dictionary . Volume 8: Words beginning with PA . Berlin: de Gruyter. Ramat, Paolo. 1992. Thoughts on degrammaticalization. Linguistics 30: 549-560. Ramchand, Gillian. 2004. Time and the event: The semantics of Russian prefixes. Nordlyd 32(2): 323-361. Rappaport, Malka & Levin, Beth. 1985. A Case Study in Lexical Analysis: The Locative Alternation . Cambridge ( MA ): MIT Press. Reid, Nicholas J. 1990. Ngan’gityemerri. A language of the Daly River region, Northern Territory of Australia . Australian National University (Canberra, Australia), Ph.D. thesis. Reinöhl, Uta. 2016 . Grammaticalization and the Rise of Configurationality in Indo-Aryan . Oxford: Oxford University Press. Reinöhl, Uta & Casaretto, Antje. 2018. When grammaticalization does not occur—Prosody-syntax mismatches in Indo-Aryan. Diachronica 35(2): 238-276. Renou, Louis. 1933. La séparation du préverbe et du verbe en védique. Bulletin de la Société de Linguistique de Paris 34: 49-96. Renou, Louis. 1935. Grammaire sanscrite . Paris: Maisonneuve. Renou, Louis. 1952. Grammaire de la langue védique . Lyon: IAC . Renou, Louis. 1956. Préposition et préverbe dans le R̥gveda. Etudes védiques et paninéennes II : 113-127. Revuelta Puigdollers, Antonio R. 2014. Postpositives. In Encyclopedia of Ancient Greek Language and Linguistics , Georgios K. Giannakis et al. (eds). Leiden: Brill. Revuelta Puigdollers, Antonio R. 2014. Some verbs prefixed by periin Ancient Greek. In The Greek verb. Morphology, Syntax, and Semantics. Proceedings of the 8th Inter- <?page no="428"?> 428 References national Meeting on Greek Linguistics (Agrigento, October 1-3, 2009), Annamaria Bartolotta (ed), 291-310 . Louvain: Peeters. Rice, Keren. 2000. Morpheme order and semantic scope: Word formation in the Athapaskan verb . Cambridge: Cambridge University Press. Rissanen, Matti, Kytö, Merja & Heikkonen, Kirsi (eds). 1997. Grammaticalization at Work: Studies of Long‐Term Developments in English . Berlin: de Gruyter. Rix, Helmut et al. 2001. Lexicon der Indogermanischen Verben. Die Wurzeln und ihre Primärstammbildungen. Wiesbaden: Reichert. Romagno, Domenica. 2004. Ancora su preverbazione e sistemi verbali. Il caso dei preverbi greci. Archivio Glottologico Italiano 89(2): 165-180. Romanova, Eugenia. 2004. Superlexical vs. lexical prefixes. Nordlyd 32(2): 255-278. Roost, Juliana. 2013. Stages of Language Development in the Old Irish Glosses on Priscian . University College Dublin (Ireland), BA thesis. Rosch, Eleanor H. 1973. Natural categories. Cognitive Psychology 4: 328-350. Rosch, Eleanor H. 1975. Cognitive representations of semantic categories. Journal of Experimental Psychology: General 104: 192-233. Rosch, Eleanor H. 1978. Principles of categorization. In Cognition and categorization, Eleanor H. Rosch & Barbara B. Lloyd (eds), 27-48. Hillsdale ( NJ ): Erlbaum. Rosenbach, Anette. 2004. The English sgenitive. A case of degrammaticalization? In Up and Down the Cline - the Nature of Grammaticalization [Typological Studies in Language 59] , Olga Fischer, Muriel Norde & Harry Perridon (eds), 73-96. Amsterdam: Benjamins. Rossiter, Trudy. 2004. Verbal Composition in Old Irish with Special Reference to Multi-preverb compounds. National University of Ireland (Maynooth, Ireland), Ph.D. thesis. Rothstein, Susan. 2004. Structuring Events: A Study in the Semantics of Lexical Aspect. Malden ( MA ): Blackwell. Rousseau, André (ed). 1995. Les préverbes dans les langues d’Europe. Introduction à l’étude de la préverbation . Lille: Presses Universitaires du Septentrion. Roussel, Louis. 1958. L’aspect en grec attique . Montpellier: Presses Universitaires de France. Ruipérez, Martín S. 1954. Estructura del sistema de aspectos y tiempos del verbo griego antiguo. Análisis Functional Sincrónico. Salamanca: Consejo Superioe de Investigaciones Científicas. Russell, Paul. 1988. The Celtic Preverb ussand Related Matters. Ériu 39: 95-126. Ruvoletto, Luisa. 2016. I prefissi verbali nella Povest’ vremennych let. Per un’analisi del processo di formazione dell’aspetto verbale in russo. Firenze: Firenze University Press. Růžička, Rudolf. 1957. Der Verbalaspekt in der altrussischen Nestorchronik [Veröffentlichungen des Instituts für Slawistik 14]. Berlin: Akademie-Verlag. Sasse, Hans-Jürgen. 2002. Recent activity in the theory of aspect: Accomplishments, achievements, or just non-progressive state? Linguistic Typology 6(2): 199-217. Sathe, Jayashree. 1993. The position of anu in some Vedic Mantras. Bulletin of the Deccan College Research Institute 53: 339-342 <?page no="429"?> References 429 Sausa, Eleonora. 2015. Argument structure constructions in Homeric Greek. A study on bivalent verbs . University of Pavia (Italy), Ph.D. thesis. Saussure, Ferdinand de. 1922. Recueil des publications scientifiques de F. de Saussure , Charles Bally & Léopold Gautier (eds). Lausanne and Geneva: Payot. Schain, Rachel. 2011. Exploring Aktionsart in Corpora: A Case Study of Koine Greek Erchomai and Eiserchomai. Journal of Greek Linguistics 11: 221-248. Schrijver, Peter. 1994. The Celtic Adverbs for ‘Against’ and ‘With’ and the Early Apocope of *-i. Ériu 45 151-189. Schrijver, Peter. 1997. Studies in the history of Celtic pronouns and particles . Maynooth: Maynooth Studies in Celtic Linguistics. Schneider, Carolin. 2009. Syntax und Wortarten der Lokalpartikeln des R̥gveda V: ní . Historische Sprachforschung 122: 118-169. Schneider, Carolin. 2010a. Syntax und Wortarten der Lokalpartikeln des R̥gveda VII : níṣ . International Journal of Diachronic Linguistics and Linguistic Reconstruction 7: 149-193. Schneider, Carolin. 2010b. Lokalpartikel im R̥gveda (Folge 13): ápa . Studien zur Indologie und Iranistik 27: 1-35. Schneider, Carolin. 2011. Syntax und Wortarten der Lokalpartikeln des R̥gveda X: úd . Münchener Studien zur Sprachwissenschaft 65: 185-241. Schneider, Carolin. 2012. Local Particles in the R̥gveda. Part XVII : párā . Münchener Studien zur Sprachwissenschaft 66: 221-246. Schneider, Carolin. 2013a. Local Particles in the R̥gveda. Part XVIII : parás . International Journal of Diachronic Linguistics and Linguistic Reconstruction 10: 55-90. Schneider, Carolin. 2013b. Syntax und Wortarten der Lokalpartikeln des Ṛgveda XI : sám . Historische Sprachforschung 126: 142-206. Schneider, Carolin. 2013c. Syntax und Wortarten der Lokalpartikeln des R̥gveda XX : ápi . Indogermanische Forschungen 118: 23-54. Schooneveld, Cornelis H. van. 1958. The so-called préverbes vides and neutralization. In Dutch Contributions to the Fourth International Congress of Slavicists : 159-161 . The Hague: Mouton. Schourup, Lawrence. 1999. Discourse markers. Lingua 107: 227-265. Schultze-Berndt, Eva. 2000. Simple and Complex Verbs in Jaminjung. A study of Event Categorization in an Australian Language. Katholieke Universiteit Nijmegen (Netherlands), Ph.D. thesis. Schultze-Berndt, Eva. 2001. Ideophone-like characteristics of uninflecting predicates in Jaminjung (Australia). In Ideophones, F. K. Erhard Voeltz & Christa Kilian-Hatz (eds), 355-373. Amsterdam: Benjamins. Schultze-Berndt, Eva. 2003. Preverbs as an open class in Northern Australian languages: Synchronic and diachronic considerations. In Yearbook of Morphology 2003 , Geert E. Booij & Jaap van Marle (eds), 145-177. Dordrecht: Kluwer. Schumacher, Stefan. 1999. Randbemerkungen zu absolut und konjunkt: Mittelkymrisch hanfot. In Studia Celtica et Indogermanica. Festschrift für Wolfgang Meid zum 70. Geburtstag , Peter Anreiter & Erzsébet Jerem (eds), 453-464. Budapest: Archaeolingua. <?page no="430"?> 430 References Schumacher, Stefan. 2004. Die keltischen Primärverben: Ein vergleichendes, etymologisches und morphologisches Lexikon . Innsbruck: Innsbrucker Beiträge zur Sprachwissenschaft. Schwyzer, Eduard. 1950. Griechische Grammatik . Band 2: Syntax . München: Beck. Seiter, William J. 1979. Instrumental Advancement in Niuean. Linguistic Inquiry 10: 595-621. Shibatani, Masayoshi. 2000. Issues in transitivity and voice: Japanese perspective. Bullettin Faculty of Letters, University of Kobe 27: 523-586. Shull, Sarah. 2003. The experience of space: the privileged role of spatial prefixation in Czech and Russian. München: Sagner. Sims-Williams, Patrick. 1984. The double system of verbal inflexion in Old Irish. Transactions of the Philological Society 82: 138-201. Simpson, Jane. 1991. Warlpiri Morpho-Syntax. A Lexicalist Approach . Dordrecht: Kluwer. Sgarioto, Laura. 1999. I preverbi dell’ungherese. Aspetti sintattici e semantici. University of Pavia (Italy), BA thesis. Slavin, Tanya. 2006a. Some semantic consequences of the structural position of preverbs in Severn Ojibwe. Papers of the Algonquian Conference 37: 293-310. Slavin, Tanya. 2006b. Some issues in the ordering of preverbs in Severn Ojibwe. In Proceedings of the 2006 Annual Conference of the Canadian Linguistic Association, Claire Gurski & Milica Radisic (eds). 〈http: / / westernlinguistics.ca/ Publications/ CLA2006/ Slavin.pdf.〉 Sommer, Ferdinand. 1926. Ein eigenartiger Fall von Tmesis bei Homer. In Festschrift für Universitäts-prof. Hofrat Dr. Paul Kretschmer. Beiträge zur griechischen und lateinischen Sprachforschung , 257-261. Wien: Deutscher Verlag für Jugend und Volk. Stifter, David. 2014. The history of the Old Irish preverb to- . In Linguistic and Philological Studies in Early Irish, Elisa Roma & David Stifter (eds), 203-246. Lewiston: The Edwin Mellen Press. Stokes, Whitley (ed). 1899-1900. The Bodleian Amra Choluimb Chille . Revue celtique 20: 30-55,132-183, 248-287, 400-437, 21: 133-136. Stokes, Whitley & Strachan, John (eds). 1901-1903. Thesaurus Palaeohibernicus: a collection of Old-Irish glosses, scholia, prose, and verse, 3 vols . Cambridge: Cambridge University Press. Strachan, John. 1901. Some Notes on the Irish Glosses of Würzburg and St. Gall. Zeitschrift für celtische Philologie 3(1): 55-60. Strachan, John. 1903. On the Language of the St. Gall Glosses. Zeitschrift für celtische Philologie 4(1): 470-492. Streitberg, Wilhelm A. 1900. Urgermanische Grammatik . Heidelberg: Winter. Strömberg, Reinhold. 1946. Greek prefix studies. On the use of adjective particles . Göteborg: Elanders Boktryckeri Aktiebolag. Strunk, Klaus. 1994. Relative chronology and Indo-European verb-system. Journal of Indo-European Studies 22: 417-434. <?page no="431"?> References 431 Sturm, Julia C. Patterson. 2014. Idiomatization of Preverb+Verb Compound in the R̥gVeda . University of Georgia (Athens, GA ), MA thesis. Svenonius, Peter. 2004a. Slavic Prefixes and Morphology: An Introduction to the Nordlyd volume. Nordlyd 32(2): 177-205. Svenonius, Peter. 2004b. Slavic prefixes inside and outside VP . Nordlyd 32(2): 205-253. Svorou, Soteria. 1994. The Grammar of Space [Typological Studies in Language 25]. Amsterdam: Benjamins. Šarić, Ljiljana. 2008. Spatial Concepts in Slavic: A Cognitive Linguistic Study of Prepositions and Cases . Wiesbaden: Harrassowitz. Šarić, Ljiljana. 2010. Cognitive semantics and analysis of prefixes. Zbornik radova Njegoševi dani 2: 341-350. Šarić, Ljiljana. 2012. Introduction: A cognitive linguistic view of South Slavic prepositions and prefixes. Jezikoslovlje 13(1): 5-17. Švedova, Natalija J. et al. (eds). Russkaja grammatika [Russian grammar], vol.1. Moscow: Nauka. Talmy, Leonard. 1983. How language structures space. In Spatial orientation: Theory, research and application, Herbert L. Pick & Linda P. Acredolo (eds), 225-282. New York: Plenum Press. Talmy, Leonard. 2000. Toward a Cognitive Semantics. Vol. 1: Concept Structuring Systems. Cambridge ( MA ): MIT Press. Tatevosov, Sergej. 2002. The parameter of actionality. Linguistic Typology 6(3): 317-401. Tatevosov, Sergej. 2008. Intermediate prefixes in Russian. In Formal approaches to Slavic linguistics 2007 , Andrei Antonenko, Christina Y. Bethin & John F. Baylin (eds), 423-442. Ann Arbor: University of Michigan Press. Taylor, John R. 1989b. Linguistic Categorization: Prototypes in Linguistic Theory . Oxford: Clarendon Press. Tchizmarova, Ivelina, K. 2005. Verbal Prefixes in Bulgarian and Their Correspondences in American English: A Cognitive Linguistic Analysis . Ball State University (Muncie, IN ), Ph.D. thesis. Tchizmarova, Ivelina, K. 2006. A Cognitive Linguistic Analysis of the Bulgarian Verbal Prefix Pre- ‘across, through, over’. Glossos 7: 1-49. 〈http: / / seelrc.org/ glossos/ issues/ 7〉 Tenny, Carol. 1994. Aspectual Roles and the Syntax-Semantics Interface . Dordrecht: Kluwer Academic Publishers. Teselkin, Avenir S. & Alieva, Natalija F. 1960. Indonezijskij jazyk [Indonesian language]. Moscow: Nauka. Tesnière, Lucien. 1959. Éléments de syntaxe structural . Paris: Klincksieck. Thomason, Olga. 2006. Prepositional Systems in Biblical Greek, Gothic, Classical Armenian, and Old Church Slavic . The University of Georgia (Athens, GA ), Ph.D. thesis. Thumb, Albert. 1905. Handbuch des Sanskrit: eine Einführung in das sprachwissenschaftliche Studium des Altindischen , 2 vol. Heidelberg: Winter. Thurneysen, Rudolf. 1946. A Grammar of Old Irish. Dublin: Dublin Institute for Advanced Studies. <?page no="432"?> 432 References Tichonov, Aleksandr N. 1964. Čistovidovye pristavki v sisteme russkogo vidovogo formoobrazovanija [Purely aspectual prefixes in the system of the Russian aspectual formation]. Voprosy jazykoznanija 1: 42-52. Tichonov, Aleksandr N. 1998. Vidovye korrelacii v sovremennom russkom jazyke [Aspectual correlations in contemporary Russian]. In Tipologija vida: Problemy, poiski, rešenija [Typology of aspect: Problems, searches, solutions], Marina Ju. Čertkova (ed), 466-477. Moscow: Jazyki russkoj kul’tury. Tischendorf, Constantin von. 1869-1872. Novum Testamentum Graece. Editio octava critica maior . Leipzig: Gieseche & Devrient. Traugott, Elizabeth C. 1988. Pragmatic strengthening and grammaticalization. In Proceedings of the 14th Annual Meeting of the Berkeley Linguistic Society , Shelley Axmaker, Annie Jaisser & Helen Singmaster (eds), 406-416. Traugott, Elizabeth C. 2001. Legitimate counterexamples to unidirectionality . Paper presented at University of Freiburg. 〈www.stanford.edu/ ~traugott/ papers/ Freiburg. Unidirect.pdf〉 Traugott, Elizabeth C. 2010. Grammaticalization. In The Continuum Companion to Historical Linguistics , Silvia Luraghi & Vit Bubenik (eds), 271-285. London: Continuum. Traugott, Elizabeth C. & Heine, Bernd (eds). 1991a. Approaches to Grammaticalization , vol. 1. Amsterdam: Benjamins. Traugott, Elizabeth C. & Heine, Bernd (eds). 1991b. Approaches to Grammaticalization , vol. 2. Amsterdam: Benjamins. Traugott, Elizabeth C. & Trousdale, Graeme (eds). 2010. Gradience, Gradualness, and Grammaticalization [Typological Studies in Language 90]. Amsterdam: Benjamins. Trousdale, Graeme. 2008a. A constructional approach to lexicalization processes in the history of English: evidence from possessive constructions. Word Structure 1: 156-177. Trousdale, Graeme. 2008b. Constructions in grammaticalization and lexicalization: evidence from the history of a composite predicate construction in English. In Constructional Approaches to English Grammar, Graeme Trousdale & Nikolas Gisborne (eds), 33-67. Berlin: de Gruyter. Tyler, Andrea & Evans, Vyvyan. 2003. The semantics of English prepositions: spatial scenes, embodied meaning, and cognition . Cambridge: Cambridge University Press. Ungerer, Freidrich & Schmid, Hans-Jörg. 1996. An Introduction to Cognitive Linguistics . London: Longman. Vaillant, André. 1939. L’Aspect verbal du slave commun: sa morphologisation. Revue des études slaves 19: 289-314. Vaillant, André. 1946. La dépréverbation. Revue des études slaves 22: 5-45. Vaillant, André. 1966. La Grammaire comparée des langues slaves . Tome 3: Le Verbe . Paris: Klincksieck. Vaillant, André. 1977. La Grammaire comparée des langues slaves . Tome 5: La syntaxe . Paris: Klincksieck. Valentine, Randolph J. 2001. Nishnaabemwin Reference Grammar. Toronto: University of Toronto Press. <?page no="433"?> References 433 Vasmer, Max. 1953-1958. Russisches etymologisches Wörterbuch. 3 Bände. Heidelberg: Winter. Vendler, Zeno. 1957. Verbs and times. Philosophical Review 66: 143-160. Vendryes, Joseph. 1904. Traité d’accentuation grecque . Paris: Klincksieck. Vendryes, Joseph. 1912. La place du verbe en celtique. Mémoires de la Société de Linguistique de Paris 17: 337-351. Vendryes, Joseph. 1923. Grammaire du Viel-Irlandais. Paris: E. Guilmoto. Vendryes, Joseph & Lambert, Pierre-Yves. 1959-1996. Lexique etymologique de l'irlandais ancien (7 vols). Dublin: Dublin Institute for Advanced Studies. Verkuyl, Henk J. 1972. On the compositional nature of the aspect . Dordrecht: Reidel. Verkuyl, Henk J. 1993. A Theory of Aspectuality: The Interaction between. Temporal and Atemporal Structure . Cambridge: Cambridge University Press. Verkuyl, Henk J. 2005. Aspectual composition: surveying the ingredients. In Perspectives on Aspect, Henk J. Verkuy, Henriette de Swart & Angeliek van Hout (eds), 19-40. Dordrecht: Springer. Verspoor, Marjolijn, Dirven, René & Radden, Günter. 1998. Putting concepts together: Syntax. In Cognitive Exploration of Language and Linguistics [Cognitive Linguistics in Practice 1], René Dirven & Marjolijn Verspoor (eds), 79-105. Amsterdam: Benjamins. Vey, Mark. 1952. Les préverbes «vides» en tchèque moderne. Revue des études slaves 29(1): 82-107. Vincent, Nigel. 1999. The evolution of c-structure: prepositions and PP s from Indo-European to Romance. Linguistics 37: 1111-1154. Viti, Carlotta. 2008a. Coding spatial relations in Homeric Greek: preverbs vs. prepositions. Historische Sprachforschung 121: 114-161. Viti, Carlotta. 2008b. From space words to transitive markers: the case of Ancient Greek en ‘in’. Transactions of the Philological Society 106: 375-413. Wackernagel, Jacob. 1892. Über ein Gesetz der indo-germanischen Wortstellung. Indogermanische Forschungen 1: 333-436. Wackernagel, Jacob. 1924. Vorlesungen über Syntax. Zweite Reihe. Basel: Emil Birkhäuser. Wagner, Heinrich. 1959. Das Verbum in den Sprachen der britischen Inseln . Tübingen: Niemeyer. Wagner, Heinrich. 1964. Nordeuropäische Lautgeographie. Zeitschrift für celtische Philologie 29: 225-298. Wagner, Heinrich. 1967. Zur unregelmäßigen Wortstellung in der altirischen Alliterationsdichtung. In Beiträge zur Indogermanistik und Keltologie, Julius Pokorny zum 80. Geburtstag gewidmet, Wolfgang Meid (ed), 289-314. Innsbruck: Institut für Sprachwissenschaft der Universität. Wälchli, Bernhard. 2001. Lexical evidence for the parallel development of the Latvian and Livonian verb particles. In The Circum-Baltic Languages. Typology and Contact. Vol. 2. Grammar and Typology [Studies in Language Companion Series 55] , Östen Dahl & Maria Koptjevskaja-Tamm (eds), 413-441. Amsterdam: Benjamins. <?page no="434"?> 434 References Walsh, Michael. 1996. Vouns and nerbs: a category squish in Murrinh-Patha (Northern Australian). In Studies in Kimberley Languages in Honour of Howard Coate , William B. McGregor (ed), 227-252. München: Lincom Europa. Watkins, Calvert. 1963. Preliminaries to a historical and comparative analysis of the syntax of the Old Irish verb. Celtica 6: 1-49. Watkins, Calvert. 1964. Preliminaries to the Reconstruction of Indo-European Sentence Structure. Proceedings of the Ninth International Congress of Linguists : 1035-1045. Watkins, Calvert. 1976. Toward Proto-Indo-European syntax: Problems and pseudo-problems. In Proceedings of the Chicago Linguistics Society: Papers from the parasession on diachronic syntax, Sanford B. Steever, Carol A. Walker & Salikoko S. Mufwene (eds), 305-326 . Chicago: Chicago Linguistics Society. Watkins, Calvert. 2000. The American Heritage Dictionary of Indo-European Roots . 2nd edn. Boston: Houghton Mifflin. Weir, E. M. Helen. 1986. Footprints of Yesterday’s Syntax: Diachronic Development of Certain Verb Prefixes in an OSV Language (Nadëb). Lingua 68: 291-316. West, Martin L. (ed). 1998. Homeri Ilias. Volumen I. Rhapsodiae I- XII . [Bibliotheca scriptorum Graecorum et Romanorum Teubneriana]. Stuttgart: Teubner. West, Martin L. (ed). 2000. Homeri Ilias. Volumen II . Rhapsodiae XIII - XXIV . Indices . [Bibliotheca scriptorum Graecorum et Romanorum Teubneriana]. Stuttgart: Teubner. Whitney, William D. 1955[1879]. A Sanskrit grammar, including both the classical language, and the older dialects, of Veda and Brahmana. Cambridge ( MA ): Harvard University Press. Wiemer, Björn & Seržant, Ilja A. 2017. Diachrony and typology of Slavic aspect: What does morphology tell us? In Unity and diversity in grammaticalization scenarios , Walter Bisang & Andrej Malchukov (eds), 239-307. Berlin: Language Science Press. Wijk, Nicolaas van. 1929. Sur l’origine des aspects du verbe slave. Revue des études slaves 9(3): 237-259. Willis, David. 2007. Syntactic lexicalization as a new type of degrammaticalization. Linguistics 45(2): 271-310. Wilson, Stephen. 1999. Coverbs and Complex Predicates in Wagiman . Stanford ( CA ): CSLI Publications. Windisch, Ernst. 1879. Kurzgefasste Irische Grammatik . Leipzig: Brockhaus. Wischer, Ilse. 2000. Grammaticalization versus lexicalization: “methinks” there is some confusion. In Pathways of Change: Grammaticalization in English, Olga Fischer, Anette Rosenbach & Dieter Stein (eds), 355-370. Amsterdam: Benjamins. Witzel, Michael. 1995. Early Sanskritization: Origin and Development of the Kuru state. Electronic Journal of Vedic Studies 1(4): 1-26. Wüllner, Franz. 1831. Über Ursprung und Urbedeutung der sprachlichen Formen. Münster: Theissingsche Buchhandlung. Zaimov, Jordan & Capaldo, Mario (eds). 1982. Suprasŭlski ili Retkov sbornik: v dva to ma . Sofia: Izdatel’stvo na Bŭlgarskata Akademija na Naukite. Zaliznjak, Anna A. & Šmelev, Aleksej D. 2000. Vvedenie v russkuju aspektologiju [Introduction to Russian Aspectology]. Moscow: Jazyki russkoj kul’tury. <?page no="435"?> References 435 Zanchi, Chiara. 2014. Multiple Preverbation in Homeric Greek . University of Pavia (Italy), MA thesis. Zanchi, Chiara. 2016. La semantica della preposizione ὑπέρ nel greco omerico. Emerita, Revista de Lingüística y Filología Clásica 84(1): 1-30. Zanchi, Chiara. 2017a. Metro e confini di parola: il caso dei preverbi multipli in Omero. In Ancient Greek Linguistics: new approaches, insights, perspectives, Felicia Logozzo & Paolo Poccetti (eds), 77-100 . Berlin: de Gruyter. Zanchi, Chiara. 2017b. New evidence for the Source-Goal Asymmetry: Ancient-Greek preverbs. In Space in Diachrony, Silvia Luraghi, Tatiana Nikitina & Chiara Zanchi (eds), 147-178. Amsterdam: Benjamins. Zanchi, Chiara & Naccarato, Chiara. 2016. Multiple prefixation in Old Church Slavonic and Old Russian. In Tipologija morfosintaksičeskich parametrov. Materialy meždunarodnoj konferencii “Tipologija morfosintaksičeskich parametrov 2016 ” , Maria B. Konošenko, Ekaterina A. Ljutikova & Anton V. Zimmerling (eds), 359-390. Moskva: MPGU . Zeller, Jochen. 2001. Particle Verbs and Local Domains . Amsterdam: Benjamins. Ziffer, Giorgio. 2005. Lo slavo ecclesiastico antico: questioni vecchie e nuove. Incontri linguistici 28: 118-125. <?page no="436"?> Sprachvergleich Studien zur synchronen und diachronen Sprachwissenschaft Band 2 The book investigates multiple preverbs (PVs) in some ancient IE languages (Vedic, Homeric Greek, Old Church Slavic, and Old Irish). After an introduction, it opens with the theoretical framework and a typologically-oriented overview of PVs. It then gives quantitative data about multiple PV composites and carries out philological, formal, semantic, and syntactic analyses on them. The comparison among these languages suggests that a process of accumulation lies behind multiple PV composites. Also, PV ordering is explained by diƒerent factors: semantic solidarity between PVs and verbs PVs’ tendency to be speci†ed by event participants, PVs’ etymologies, in‡uence from other languages. The book also contributes to casting light on the reasons for PVs’ grammaticalization and lexicalization. These are two distinct reanalyses triggered by the same factor, i.e. the mentioned semantic solidarity, which makes PVs be felt as redundant. They are thus reassigned salient pieces of information as actional markers (grammaticalization) or reinterpreted as part of the verb (lexicalization). ISBN 978-3-8233-8274-4 Zanchi Multiple Preverbs Chiara Zanchi Multiple Preverbs in Ancient Indo-European Languages A comparative study on Vedic, Homeric Greek, Old Church Slavic, and Old Irish 18274_Umschlag.indd Alle Seiten 06.08.2019 11: 17: 27