<?page no="0"?> Marinus Wiedner Grammatical Gender in Old Occitan A Variationist Corpus-based Approach <?page no="1"?> Grammatical Gender in Old Occitan <?page no="2"?> Studia philologica Monacensia Edunt Andreas Dufter et Bernhard Teuber Volumen 27 · 2026 Comité scientifique - Advisory Board - Wissenschaftlicher Beirat Lina Bolzoni (Scuola Normale Superiore di Pisa) Anthony Cascardi (University of California at Berkeley) Pedro Cátedra (Universidad de Salamanca) Victoria Cirlot (Universitat Pompeu Fabra, Barcelona) Marie-Luce Démonet (Université François Rabelais, CESR, Tours) Carlos Garatea Grau (Pontificia Universidad Católica del Perú, Lima) Barbara Kuhn (Katholische Universität Eichstätt-Ingolstadt) Frank Lestringant (Université Paris-Sorbonne) María Jesús Mancho Duque (Universidad de Salamanca) Wolfgang Matzat (Eberhard-Karls-Universität Tübingen) Paulo de Sousa Aguiar de Medeiros (University of Warwick) Wolfram Nitsch (Universität zu Köln) Uli Reich (Freie Universität Berlin) Maria Selig (Universität Regensburg) Elisabeth Stark (Universität Zürich) Collegium consultorum <?page no="3"?> Marinus Wiedner Grammatical Gender in Old Occitan A Variationist Corpus-based Approach <?page no="4"?> Die vorliegende Arbeit wurde von der Philologischen Fakultät der Universität Freiburg als Dissertation angenommen. DOI: https: / / doi.org/ 10.24053/ 9783381165124 © 2026 · Marinus Wiedner https: / / orcid.org/ 0009-0005-9237-3334 Das Werk ist eine Open Access-Publikation. Es wird unter der Creative Commons Namensnennung - Weitergabe unter gleichen Bedingungen | CC BY-SA 4.0 (https: / / crea tivecommons.org/ licenses/ bysa/ 4.0/ ) veröffentlicht, welche die Nutzung, Vervielfältigung, Bearbeitung, Verbreitung und Wiedergabe in jeglichem Medium und Format erlaubt, solange Sie die ursprünglichen Autor: innen und die Quelle ordentlich nennen, einen Link zur Creative Commons-Lizenz anfügen und angeben, ob Änderungen vorgenommen wurden. Die in diesem Werk enthaltenen Bilder und sonstiges Drittmaterial unterliegen ebenfalls der genannten Creative Commons Lizenz, sofern sich aus der am Material vermerkten Legende nichts anderes ergibt. In diesen Fällen ist für die oben genannten Weiterverwendungen des Materials die Einwilligung der jeweiligen Rechteinhaber: innen einzuholen. Alle Informationen in diesem Buch wurden mit großer Sorgfalt erstellt. Fehler können dennoch nicht völlig ausgeschlossen werden. Weder Verlag noch Autor: innen oder Herausgeber: innen übernehmen deshalb eine Gewährleistung für die Korrektheit des Inhaltes und haften nicht für fehlerhafte Angaben und deren Folgen. Diese Publikation enthält gegebenenfalls Links zu externen Inhalten Dritter, auf die weder Verlag noch Autor: innen oder Herausgeber: innen Einfluss haben. Für die Inhalte der verlinkten Seiten sind stets die jeweiligen Anbieter: innen oder Betreiber: innen der Seiten verantwortlich. Narr Francke Attempto Verlag GmbH + Co. KG Dischingerweg 5 · D-72070 Tübingen Internet: www.narr.de eMail: info@narr.de Druck: Elanders Waiblingen GmbH ISSN 2365-3094 ISBN 978-3-381-16511-7 (Print) ISBN 978-3-381-16512-4 (ePDF) ISBN 978-3-381-16513-1 (ePub) Umschlagabbildung: aus COMETA (2025) Handschrift BmT, Cote 2883, f o . 125 r o ., c. 1, l. 13-16 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. <?page no="5"?> Table of Contents Acknowledgements 9 Abbreviations 11 1 Introduction 13 2 Definitions 19 2.1 Gender ...................................................................................... 19 2.2 Gender Doublets ........................................................................ 22 3 State of the Art 27 3.1 Gender in the Languages of the World ........................................ 27 3.1.1 Semantic Assignment Rules ............................................ 28 3.1.2 Formal Assignment Rules ............................................... 30 3.1.2.1 Morphological Assignment Rules ..................... 31 3.1.2.2 Phonological Assignment Rules ....................... 32 3.2 Gender in Latin ......................................................................... 33 3.3 Gender in the Romance Languages ............................................. 36 3.3.1 Romanian ...................................................................... 36 3.3.2 Italo-Romance................................................................ 38 3.3.2.1 Two-Gender Systems....................................... 38 3.3.2.2 Three-Gender Systems ..................................... 40 3.3.2.3 Four-Gender Systems ...................................... 41 3.3.2.4 Five-Gender Systems ....................................... 42 3.3.3 Ibero-Romance............................................................... 43 3.3.3.1 Old Ibero-Romance ......................................... 43 3.3.3.2 The Special Case of Asturian ............................ 44 3.3.4 Gallo-Romance .............................................................. 45 3.3.4.1 French ............................................................ 47 3.3.4.2 Franco-Provençal ............................................ 48 3.3.4.3 Occitan ........................................................... 49 3.4 Intermediate Summary............................................................... 51 3.5 Factors Influencing the Development of Gender Systems ............. 53 3.5.1 Sound Change in combination with morphological change 53 3.5.2 Frequency...................................................................... 54 <?page no="6"?> 6 Table of Contents 3.5.3 Analogy due to Semantic Relations ................................. 54 3.5.4 Analogy to Rhyming Words............................................ 55 3.5.5 Contact Languages ......................................................... 55 3.5.6 Normalisation and Standardisation ................................. 56 3.5.7 Changes in Society ......................................................... 57 4 The Old Occitan Nominal System—an Overview 59 4.1 Nouns and Adjectives ................................................................ 60 4.2 Pronouns and Determiners ......................................................... 63 5 Research Questions 67 6 Constitution of the Corpus COMETA 71 6.1 Choice of Manuscripts ............................................................... 71 6.2 Historical Context of the Manuscripts ......................................... 75 6.2.1 The Roman de Flamenca .................................................. 75 6.2.2 The Chanson de la Croisade contre les Albigeois ................. 77 6.2.3 The Vida de Sant Honorat ................................................ 84 6.2.4 The Vida de Sant Enimia ................................................. 88 6.2.5 The Vida de Santa Doucelina............................................ 90 6.2.6 The Roman de Philomena ................................................ 91 6.2.7 The Leys d’amors ............................................................ 94 6.2.8 Français 1049 ................................................................. 98 6.2.9 Harley 7403 ................................................................... 101 6.2.10 Nouvelle Acquisition Française 11151 .............................. 103 6.2.11 Nouvelle Acquisition Française 11180 .............................. 105 6.2.12 Remaining Manuscripts .................................................. 105 6.2.12.1 Harley 3041 .................................................... 106 6.2.12.2 Français 13504................................................. 106 6.2.12.3 Latin 901......................................................... 107 6.3 Implications for the Analysis ...................................................... 107 6.4 Intermediate Summary............................................................... 109 7 Technical Creation of the Corpus COMETA 111 7.1 Semi-Automatic Transcription using Transkribus ........................ 111 7.2 Part of Speech-Tagging using LLMs ............................................ 114 8 Gender Doublets in COMETA (2025) 119 8.1 Methodology ............................................................................. 120 8.2 Inherited Nouns from Latin ........................................................ 122 8.2.1 Second Declension Class ................................................ 123 <?page no="7"?> Table of Contents 7 8.2.2 Third Declension Class ................................................... 126 8.2.2.1 aigal ............................................................... 128 8.2.2.2 art .................................................................. 129 8.2.2.3 canal .............................................................. 133 8.2.2.4 color ............................................................... 134 8.2.2.5 crin ................................................................ 137 8.2.2.6 den ................................................................. 138 8.2.2.7 fam................................................................. 141 8.2.2.8 fornatz............................................................ 143 8.2.2.9 glan/ aglan....................................................... 144 8.2.2.10 labor............................................................... 146 8.2.2.11 lach ................................................................ 149 8.2.2.12 mar ................................................................ 150 8.2.2.13 marme ............................................................ 153 8.2.2.14 mel ................................................................. 155 8.2.2.15 orde................................................................ 157 8.2.2.16 ost .................................................................. 160 8.2.2.17 penche............................................................ 163 8.2.2.18 pols ................................................................ 164 8.2.2.19 ret .................................................................. 168 8.2.2.20 sal .................................................................. 169 8.2.2.21 salut ............................................................... 171 8.2.2.22 sanc................................................................ 176 8.2.2.23 semen ............................................................. 179 8.2.2.24 serp/ serpen ..................................................... 180 8.2.2.25 set .................................................................. 184 8.2.2.26 trau ................................................................ 187 8.2.3 Fourth Declension Class ................................................. 190 8.2.4 Fifth Declension Class .................................................... 194 8.2.5 Nouns Derived from Other Latin Word Classes ................ 197 8.2.5.1 semal .............................................................. 198 8.2.5.2 sera ................................................................ 199 8.3 Greek Loanwords ...................................................................... 201 8.3.1 evangelista .................................................................... 202 8.3.2 fantasma........................................................................ 204 8.3.3 papa .............................................................................. 206 8.3.4 patriarca ........................................................................ 209 8.3.5 profeta........................................................................... 212 8.3.6 salmista ......................................................................... 219 8.3.7 senebe ........................................................................... 221 8.4 Other Loanwords ....................................................................... 222 <?page no="8"?> 8 Table of Contents 9 Two Comparative Case Studies 225 9.1 Three Manuscripts of the Vida de Sant Honorat........................... 225 9.1.1 Gender Distribution Among the Manuscripts .................. 226 9.1.2 Detailed Gender Analysis ............................................... 228 9.2 The Chanson de la Croisade contre les Albigeois ......................... 237 10 Analysis and Discussion 243 10.1 Extralinguistic Factors ............................................................... 243 10.1.1 Diatopic Variation .......................................................... 243 10.1.2 Diachronic Variation ...................................................... 245 10.1.3 Interscribal Variation...................................................... 245 10.1.4 Intrascribal Variation...................................................... 247 10.2 Linguistic Factors ...................................................................... 248 10.2.1 Morphological Variation ................................................. 249 10.2.2 Semantic Variation ......................................................... 250 10.2.3 Morphosemantic Variation ............................................. 251 11 Conclusion 253 Bibliography 259 Language Index 299 List of Figures 303 List of Tables 305 <?page no="9"?> Acknowledgements This work would not have been possible without the help of many people. First and foremost, I want to thank my first supervisor Daniela Marzo. Thank you for your support throughout the process of writing this thesis and for encouraging me to pursue Old Occitan morphology and morphosyntax. Thank you also for your financial support in the form of funding from your chair for my stays at the archives in Paris and London; it would not have been possible to write this thesis without it. Also thank you for supporting me and letting me go to so many conferences, where I had the possibility to present my work and to discuss it with linguists from all around the world. I also want to thank my second supervisor, Maria Selig. You supported and guided me and my doctorate from the very beginning and were always available to discuss every aspect of it. Whether it was pressing issues or just some small side notes, you were always open to discuss everything. I want to express my gratitude to all my colleagues in the Kolloquium Romanistische Linguistik in Freiburg. Furthermore, I want to thank the participants in the different conferences, Oberseminare, Colloquia etc. who gave me ideas regarding my thesis and who thoroughly discussed my research with me. Thank you to all collegues involved in these discussions for supporting and challenging my project. Many thanks to Andreas Dufter and Bernhard Teuber for including my book in the Orbis Romanicus series and for the helpful suggestions and comments on my research. Furthermore, I want to thank Kathrin Heyng from the Narr Verlag for the very smooth communication and support in the publishing process of this book. I would especially like to thank Nicolas Gignac for proofreading the revised version of my thesis before publication, Simona Oberto for helping with the translation of the Old Occitan texts into English, and Fiona Gehring for coming up with the name of my corpus, COMETA. Thank you all for your support and for the time and energy you dedicated in supporting me. All mistakes remain my own. Moreover, I want to thank my colleagues from the ALMA-project in Munich as well as from the DOM, i.e., Matthias Schöffel, Paula Ruppert, Monika Tausend, Renate Peters, and Tobias Schmid, who helped me during this thesis by scanning some of the books available in the library of the DOM, by supporting the annotation process, and by helping with some tricky translations. <?page no="10"?> 10 Acknowledgements This thesis would not have been possible without the support of my family. First of all, I want to thank my brother Anian and my aunt Jenny, who proofread the whole thesis and gave me very helpful suggestions to make it more idiomatic; thank you for the enormous amount of work and time you dedicated to my thesis. Furthermore, I want to thank my brother Florian for his support in technical aspects. Thank you for always reacting so quickly and for always finding a solution for my sometimes rather peculiar problems. Finally, I want to thank my whole family: my parents Barbara and Bernhard, as well as my brothers Anian, Florian, and Sebastian. You were always there for me and understood when I did not respond for days because I was too focused on an article or my thesis. Thank you for your endless support and understanding! <?page no="11"?> Abbreviations acc accusative. ADJ adjective. ADP adposition. ADV adverb. AUX auxiliary verb. BL British Library. BmC Bibliothèque municipale de Carcassonne. BmT Bibliothèque municipale de Toulouse. BnF Bibliothèque nationale de France. c column. CCONJ coordinating conjunction. CER character error rate. DET determiner. F feminine. f o folio. gen gender. GT Ground Truth. HTR Handwritten Text Recognition. l line. LLM large language model. M masculine. NA not applicable. NAF Nouvelle Acquisition Française. nom nominative. num number. obl oblique. PART participle. pl plural. PoS Part of Speech. PRON pronoun. PROPN proper noun. PUNCT punctuation. r o recto. SCONJ subjunction. sg singular. SYM symbol. UD Universal Dependencies. v o verso. <?page no="13"?> 1 Introduction La vita è bello. Ist doch egal, wie das heißt. Du weißt schon was ich mein’ (Refrain of the song “Expresso & Tschianti” by Josh. 2021) The chorus of the song “Expresso & Tschianti” by Josh, loosely translated as ‘Life is beautiful—doesn’t matter how it’s said—you know what I mean either way’, is a parody of German speakers’ mispronunciations of Italian. However, the song goes beyond mere mispronunciation. In the first part of the refrain, the noun vita is preceded by the expected feminine article, but the predicative adjective shows masculine gender. This mismatch between prenominal and postnominal agreement is unusual. The subsequent German text of the refrain suggests that the meaning of the sentence can be understood no matter how it is said, and therefore whether or not the gender agreement is consistent. Such gender mismatches are striking, at least for L1 speakers of a given language, but the semantics of the sentence would most likely still be understood. The notion of gender has different meanings based on context. In this book, the focus is on grammatical or linguistic gender, which is treated (more or less) independently of social gender and biological gender. As Section 2.1 shows, biological and grammatical gender are sometimes connected, with biological gender constituting the semantic core of grammatical gender in some languages (for a more detailed discussion, see Aikhenvald 2016). However, even in languages where biological gender constitutes the semantic core of gender assignment, grammatical gender sometimes does not match biological gender. For example, the German noun Mädchen ‘girl’ designates a young female but triggers neuter agreement because it is a diminutive form, and diminutives are neuter in German. The question of whether biological gender can serve as a semantic basis for the gender system is only relevant in some languages. Other languages do not have a semantic core based on biological sex, but rather on animate versus inanimate. This book’s main focus is understanding the factors that led to variation in gender within the Romance languages, with Old Occitan serving as the object language. Therefore, existing studies on gender in Old Occitan will be supplemented and gaps in the research will be filled. This study differs from previous ones in its socio-historical approach, which considers the texts and scribes <?page no="14"?> 14 Introduction (where possible) and seeks a more sophisticated, multi-factorial explanation of the existing variation in medieval languages. Until now, all studies on gender in Old Occitan have merely noted the existence of gender variation; no detailed analysis of diatopic, micro-diachronic, or text-genre-specific variation has been conducted. By addressing the temporal and spatial distribution of gender variation, this study incorporates sociolinguistic and dialectological perspectives. Additionally, it examines the influence of scribes’ individual attitudes and competencies on gender assignment by systematically contrasting different manuscripts of the same text or parts of a text that can be traced back to different scribes or authors. Due to the methodological approach applied in this book, an additional outcome of this study concerns the influence of modern editions on gender marking; this too is entirely novel. It is well known that editions tend to normalise morphological markings, so the high degree of graphic variation in medieval vernacular manuscripts is not reflected in many editions. The reason for these editorial interventions can be found in their origin in historiographic contexts, which do not fulfil the requirements of current linguistic research (see Selig 2005: 256-257), particularly where the accuracy of transcription is of relevance, i.e., for morphological or morphosyntactic analyses. To address this issue, this research is based on the original manuscripts rather than modern editions of the texts. This provides a more accurate picture of gender variation in the Middle Ages. The following example illustrates the problem with many editions for this type of research. Example (1) shows an extract from the Roman de Flamenca, displaying my own transcription (Example 1a) alongside the transcriptions (Example 1b) found in different editions of this text. (1) a. Quant es dins iun li pantecosta ‘When in June is Pentecost’ 1 (BmC, Cote 34 13 th century: folio (f o ). 45 recto (r o ), line (l) 7) b. Quant es dins iun la pantecosta (Meyer 1901: 96; Scrivner et al. 2013: vers 2578; Zufferey/ Fasseur 2014: 290) Even without knowing the gender value of the articles li and la in Example (1), it is clear that the editors changed the vowel quality, which may also have 1 All translations are my own unless stated otherwise. Many thanks to Simona Oberto and Fiona Gehring for their help in proofreading the translations. <?page no="15"?> Introduction 15 changed the gender of the respective noun. This example is certainly not an isolated case, as similar examples can be found in different editions of Old Occitan manuscripts. Therefore, it was decided to (re)transcribe a certain number of manuscripts semi-automatically in order to create a new corpus. This timeconsuming process limits the amount of data that could be used for this research, thereby constraining the representativeness of the findings for the Old Occitan linguistic area. However, it ensures the validity of the findings in the context of the genesis of these manuscripts, while avoiding findings of gender variation (or the lack thereof) due to errors introduced by the editors. The book’s research areas thus demonstrate that it pursues a broadly grammatical aim on the one hand: to determine how variation in gender assignment can be situated within the overall set of factors that shape nominal morphology and agreement phenomena. On the other hand, the innovative preparation of the corpus opens up a perspective that situates gender marking within a historical context of norms. The main research question for this book is therefore “Can the study of gender assignment in medieval Occitan be used to draw conclusions about the reasons, extent, and dynamics of varying gender assignment? ”. To answer this question and provide further clarification, the book is structured as follows: Chapter 2 introduces the main concepts pertinent to this research, starting with a definition of grammatical gender (Section 2.1), followed by some remarks on gender doublets (Section 2.2). Chapter 3 proposes a typological survey of gender systems with an expanded focus that embeds each language-specific gender system within the broader set of morphological and phonetic structures. This approach remains open to multiple, and sometimes competing, embeddings of gender assignment. This chapter serves as an introduction to gender marking in the Romance language family (and other languages of the world) and as a perspective on this phenomenon relating it to other levels of structure within the nominal system. This chapter is structured as follows: Section 3.1 introduces the major types of gender systems found across the world’s languages, as well as the semantic and formal principles that govern gender assignment. Section 3.2 then outlines the Latin point of departure, after which the remainder of the chapter surveys the diversity of gender systems found in the Romance languages of Europe. This overview begins with the case of Romanian, highlighting the existence of what is known as Genus Alternans (Section 3.3.1). Section 3.3.2 then presents some Italo-Romance varieties with different numbers of genders, ranging from two to five. Section 3.3.3 focuses on older stages of Ibero-Romance and the unique agreement system in contemporary Asturian within the Romance languages. The final part of the presentation of gender systems in the Romance languages of Europe is dedicated to Gallo-Romance varieties (see Section 3.3.4). After a summary of the information <?page no="16"?> 16 Introduction presented on gender in Romance languages in Section 3.4, Chapter 3 concludes with an overview of the factors known to influence the development of gender from Latin to the Romance languages, and changes in gender systems in general (Section 3.5). The final introductory chapter (Chapter 4) provides a brief overview of the Old Occitan nominal system. It introduces the basic properties of gender, number, and case in Old Occitan, and describes the inflectional paradigms of nouns, adjectives, participles, pronouns, and articles. As Old Occitan is a non-standardised medieval vernacular, this account is necessarily approximate and cannot capture the full range of variation attested in the manuscripts. The main body of the book begins with a contextualisation and presentation of the research questions that led to this work (see Chapter 5). These questions are based on the definitions and the state of the art presented in Chapters 2 to 4. These issues give rise to the following research areas: 1. The temporal and spatial distribution of gender doublets in Old Occitan, especially in the Languedoc and Provence linguistic areas. 2. The development of the Latin neuter in Old Occitan. 3. The influence of scribes and their personal profiles. Chapters 6 and 7 describe the texts included in the COMETA (2025) 2 corpus as well as the methodology employed for the transcription of the manuscripts. The COMETA (2025) corpus is the main source of data for this analysis and was created specifically for this project. To this end, 17 manuscripts were transcribed semi-automatically. The manuscripts included are mainly religious texts, written between 1250 and 1400 in the Languedocian or the Provençal linguistic areas. The approach described provides a solid empirical foundation for the study of gender marking. To contextualise the findings as clearly as possible, information on the origin (place and time), the author, the scribe, and the context of the texts will be presented. This information will be given in as much detail as possible for each manuscript in Chapter 6. Alongside this detailed discussion of the manuscripts’ contexts, the COMETA (2025) corpus will be divided into two sub-corpora according to the region in which the respective manuscript was written. The resulting sub-corpora comprise approximately 410,000 tokens for the Languedocian linguistic area and approximately 405,000 tokens for the Provençal linguistic area. These two corpora are almost evenly balanced in terms of the number of tokens, making it possible to compare the data according to region. 2 I would like to express my sincerest gratitude to Fiona Gehring for suggesting this name. <?page no="17"?> Introduction 17 Chapters 8 and 9 present the corpus analysis from two complementary perspectives. Chapter 8 focuses on gender doublets in Old Occitan, drawing on a list of nouns identified as gender-ambiguous in the Old Occitan lexicographic literature, most notably in the DOMél (2025). The aim of this initial analysis is to contextualise and supplement existing research by providing new data taken directly from the manuscripts. The distribution of gender for these gender doublets is shown across the two linguistic areas assessed in this book. This also enriches existing studies by adding an analysis of diatopic variation. Of the 61 gender doublets indexed in the DOMél (2025), 39 are present in the COMETA (2025) corpus and are therefore included in the analysis. Chapter 9 consists of two case studies which aim to shed light on the potential influence of scribes on gender variation in these texts. Three manuscripts of the Vida de Sant Honorat will therefore be compared with respect to gender assignment within each manuscript. While these manuscripts are not direct copies of one another, they still share the same narrative and contain some parallel passages that are well aligned across all manuscripts. Differences in gender assignment in these comparable passages will be interpreted as gender variation based on the scribe. A second case study aims to corroborate the findings of the previous case study and is based on the manuscript of the Chanson de la Croisade contre les Albigeois, a text written by a single scribe but composed by two different authors. In contrast to the first case study, any gender variation observed between the two parts of the manuscript is interpreted as reflecting the authors’ original usage rather than scribal intervention. Chapter 10 summarises the findings of the two corpus studies, examining them from two complementary perspectives: extralinguistic variation (i.e., diatopic and diachronic variation, variation within texts written by a single scribe, and differences due to different scribes) and linguistic factors (i.e., morphologically and semantically conditioned variation). The final part of this chapter discusses possible combinations of these explanations for gender variation. Finally, Chapter 11 summarises the findings and contextualises the results within the Romance languages. The concluding section outlines possible directions for future research on gender in Old Occitan, as well as in the Romance languages. <?page no="19"?> 2 Definitions This chapter provides a brief overview of the notions of gender and gender doublets used in this book. As these definitions are mostly well known, only brief preliminary and theoretical definitions are provided here, with more concrete examples given in Chapter 3. This chapter provides an overview of the different types of gender systems found in the languages of the world, with a focus on Romance languages later on. 2.1 Gender The definition of gender that is most frequently encountered, and adopted in this study, is the following: “Genders are classes of nouns reflected in the behavior of associated words” (Hockett 1962: 231). Languages with grammatical gender have at least two genders. In Indo-European languages, the gender assigned to an animate noun often coincides with the biological gender of its referent. However, the gender assignment of other nouns to one of the genders is often arbitrary from a semantic point of view (see Corbett 1991: 1; Schafroth 2015). Nevertheless, the semantic core plays a role in the attributes assigned to objects. For instance, Konishi (1993) demonstrated in a study of German and Spanish that the same object, which was feminine in Spanish and masculine in German (e.g., Spanish manzana and German Apfel, both ‘apple’), was assigned stereotypically feminine or masculine attributes, respectively (see also McConnell-Ginet 2014). In general, gender is assumed to play an important role in cognitive language processing and language acquisition. For example, when two consecutive nouns belong to the same gender, the second noun is activated more quickly than when the nouns belong to different genders (see, for example, Vigliocco et al. 2005; Okan/ S. M. Müller/ García-Retamero 2010; Cubelli et al. 2011; Boutonnet/ Athanasopoulos/ Thierry 2012; Pavlidou/ Alvanoudi 2014). While gender facilitates processing by contributing to faster comprehension, it makes acquiring a native-like command of a given language that has grammatical gender more difficult for non-native speakers (see J. Wolf 2024). Even L1 speakers of a given language—in this case French—do not acquire gender with 100% accuracy (see Ayoun 2018: 138). Gender is acquired quite early in first language acquisition, first based on formal cues and only later on semantic cues (see Lust 2006: 250- 251). <?page no="20"?> 20 2 Definitions According to the World Atlas of Language Structures (WALS) (see Corbett 2013b), biological sex constitutes the semantic core of gender assignment in 75% of the languages with gender systems. Overall, nearly half of the languages included show gender distinctions, primarily in the form of binary or tripartite gender systems. However, some languages show four or more distinct gender values (see Corbett 2013a). Many Indo-European languages show the grammatical feature of gender, often distinguishing between two genders (e.g., masculine and feminine in French) or three genders (e.g., masculine, feminine, and neuter in German). Others show minimal gender differentiation (e.g., English distinguishes gender only for thirdperson singular pronouns) (see Corbett 1991: 1-2). Some languages have overt gender, whereby the gender of a noun can be determined solely from its morphophonological form (see Corbett 1991: 62-63). However, gender is not always overtly marked on nouns by morphological or phonological means. In many languages, gender is covert and is revealed only by other words that show agreement according to the gender of the agreement controller. As the name suggests, these agreement controllers trigger agreement with other words, the agreement targets (see Corbett 2006a: 4). Agreement can be defined as follows (for an overview of the definitions of agreement, see i.e., Corbett (2003)): a rule that says, basically, that the presence of the feature A (out of a limited number of features) on the lexical unit X (the controller) requires the presence of A on the lexical unit Y (the target), within a syntactic domain D. The unity X and Y can, in principle, belong to any lexical class that the language in question distinguishes (e.g. nouns, verbs, adjectives and articles) (Matasović 2004: 13). Agreement controllers are mainly nouns and pronouns, but there are also nonlexical controllers, such as clauses and predicates (see, for example, Acquaviva 2020: 70). Agreement targets also differ from one language to another. For instance, in English, gender agreement only applies to certain pronouns (as mentioned before; see also Corbett 1991: 12), whereas other languages show agreement across a broader range of targets (for a typology of possible targets, see, for example, Matasović 2017). In Romance languages, these agreement targets include articles, adjectives, pronouns, and participles (see Loporcaro 2018: 8). The following Example (2) illustrates gender agreement for the French noun prophète ‘prophet’. <?page no="21"?> 2.1 Gender 21 (2) a. prophète b. le prophète c. la prophète d. les prophètes saints / vivants e. les prophètes saintes / vivantes f. les prophètes, ils sont là g. les prophètes, elles sont là In the first example here (Example 2a), the gender of the noun is not visible, as French is a language without overt gender marking. This noun could be either feminine or masculine, most likely depending on whether it refers to a woman or a man. Therefore, additional information is required to determine the gender of the noun. In the subsequent examples (2b) to (2g), gender can be determined due to the agreement targets. In Examples (2d) to (2g), the plural form is used to assure that gender is always indicated by just one agreement target because French plural determiners do not show gender agreement. In (2b) and (2c), the gender is indicated by the definite articles; in (2d) and (2e), by the qualificative adjectives or adjectival participles; and in (2f) and (2g), by the respective personal pronouns. Thus, we can determine that the noun is masculine in (2b), (2d), and (2f), and feminine in (2c), (2e), and (2g), while the gender remains undetermined in (2a). Hockett (1962: 231) supplements his definition as follows: “To qualify as a gender system, the classification must be exhaustive and must not involve extensive intersection: that is, every noun must belong to one of the classes, and very few can belong to more than one”. He confirms that only a small number of nouns should belong to more than one gender. This definition applies once gender is treated as a grammatical category produced and subsequently stabilised as a consequence of grammaticalisation. In these gender systems, the original semantic motivation—often visible in non-grammaticalised systems that reveal their ontological foundation transparently—no longer plays an important role anymore. It is then nothing more than a purely grammatical distribution. Gender assignment in fully grammaticalised gender systems still follows certain rules based on semantic, morphological, phonological, or combined criteria. These different systems of gender assignment will be addressed in Section 3.1. The next section (Section 2.2) will focus on the rare cases of nouns that trigger agreement with not one, but two (or more) different genders. <?page no="22"?> 22 2 Definitions 2.2 Gender Doublets Canonically, each noun should trigger agreement with exactly one gender (see Corbett/ Fedden 2016: 495). Nevertheless, some nouns do not conform to this principle. Such nouns are known as gender doublets, generally defined as two variants of a single substantival word form differentiated by their agreement with two different genders (see Spiewok 1975: 127-128). However, this is a very broad definition—too broad for the purposes of this study—because a closer examination reveals that the phenomenon may be rooted in a variety of distinct underlying conditions. From a synchronic perspective, several types of gender doublet can be distinguished, most of which can be explained on the basis of etymology or semantics (i.e., homonyms and polysemes). Only one type of gender doublet is relevant to the present study. To situate this type within the broader framework, I will briefly outline the overall category and its subtypes. Homonyms are two phonologically and morphologically identical signs that are regarded as distinct lexical items due to their different etymologies and lack of shared semantic properties (see Corbett 1991: 67; Matthews 1991: 44-49). An example of gender variation due to homonymy in French can be found in Examples (3c) and (3d). An example of homonymy in Old Occitan would be the nouns cor ‘heart’ and cor ‘body’, the former derived from Latin coR ‘heart’, the latter from Latin coRpus ‘body’. Polysemes, on the other hand, are formally identical word forms with different meanings that are semantically related to each other (see Marzo 2013b: 59). Examples of polysemes can be found in Examples (3a) and (3b) as well as in Examples (3e) and (3f). In modern French, a significant number of gender doublets with semantic differentiation in combination with a different gender are still found. From a synchronic point of view, these are gender doublets, but from a diachronic point of view, they are homonyms or polysemes. This clearly demonstrates that gender doublets fall into several categories that differ with regard to their underlying conditions. The Académie française recognises about 300 of these doublets with semantic differentiation, not including epicene nouns. Epicene nouns (see Corbett 1991: 68) or common nouns (see Ogneva 2020: 283) are nouns that denote animates of both genders with one form but are differentiated through agreement with agreement targets of a different gender. An example is the French noun élève ‘pupil’, which has the same form for both male and female pupils. The Examples in (3) show some of these gender doublets in French, taken from Darcos (2021). (3) a. un aide ‘assistant’ b. une aide ‘action of helping’ c. un vase ‘vase’ d. une vase ‘dirt, mud’ <?page no="23"?> 2.2 Gender Doublets 23 e. un œuvre ‘global result’ f. une œuvre ‘(art) work, piece’ This type of variation is not considered to be ‘variation’ in the strict sense, but rather a lexicalised gender distinction that is therefore not included in the following study. Another factor contributing to gender doublets with semantic differentiation in some languages is evaluative marking; that is, the assignment of a “value which is different from that of the standard or default” (Grandi/ Körtvélyessy 2015: 13), such as diminutives or augmentatives. These evaluative markings can alter the gender of a noun in some languages, including Romance languages (see Corbett 2014: 123; Corbett/ Fedden 2016: 526). This is a case of suffixation, in which an evaluative derivational suffix leads to a semantic change. Therefore, it is a case of word formation rather than gender variation. Similar cases include Italian mela ‘apple’ and Italian melo ‘apple tree’, where the change in semantics from fruit to fructiferous tree is achieved through suffixation that is also accompanied by a change in gender. Therefore, from a synchronic point of view, these nouns can be considered gender doublets with semantic differences, just like the homonyms discussed above. Another example of a change in semantics due to a change in gender is the Italian masculine fiasco ‘flask’, where the feminine pendant fiasca means ‘large flask’, and the Catalan masculine plat ‘plate’, where the feminine pendant plata means ‘large plate’ (see Grandi 2011: 12). These augmentatives always involve a change in gender from masculine to feminine (see Grandi 2011: 12) and are derived from the Latin neuter collective ending -a (and also Proto-Indoeuropean, see Schön 1971; for more on collectives in Romance languages, see Mihatsch 2021). Conversely, dedicated evaluative suffixes in Romance languages usually maintain the gender of the suffixed noun. For instance, in Italian, the feminine noun finestra ‘window’ has the diminutive finestrina ‘small window’, which are both feminine. A masculine diminutive form (finestrino) exists, but this is also associated with a change in semantics; specifically, it refers to the ‘side-window of a car’ (see Rainer 2016: 2721). Such variations resulting from suffixation processes are excluded from the following analysis. Another point to considere carefully concerns nouns designating animates. In these cases, the agreement difference of the noun usually expresses a difference in sex of the referred animate, e.g., Spanish el guitarrista ‘the male guitarist’ versus la guitarrista ‘the female guitarist’ (see Ogneva 2020: 283; see also the French epicene noun élève ‘pupil’ as mentioned previously). However, for these epicene nouns, there are sometimes discrepancies between biological sex and gender, as in the example given by Ogneva (2020: 290): “mi mama [sic! ] es un buen médico ‘my mum is a-M good-M doctor’, mi mama [sic! ] es una buena médico ‘my mum is a-F good-F doctor’ or mi mama [sic! ] es una buena médica <?page no="24"?> 24 2 Definitions ‘my mum is a-F good-F doctor’”. In this case, the variation is due to medico being a masculine noun that was used as a generic term for both male and female doctors. Semantically, the agreement would be feminine when referring to the mother, but grammatically, the agreement would require a masculine gender. Thus, it depends on whether someone wants to use gender-sensitive language, which is a topic of ongoing debate in many linguistic communities (see Zapf 2024: 18-19). In this case, therefore, the variation in gender can be explained by ideological convictions. As this type of variation only arose in the second half of the 20 th century (see Zapf 2024: 18), it is not expected to have played a role in gender variation in the Middle Ages. This is another matter for epicene nouns such as the Old Occitan profeta, which appeared in the manuscripts with both masculine (e.g., lo profeta) and feminine (e.g., la profeta) agreement targets (see Jensen 1973: 398; Wiedner 2026: 80-83; see also Section 8.3.5), regardless of the gender of the person being referred to (see Section 8.3.5 for examples where sex and gender do not coincide). In this case, gender variation can be observed despite the semantic criterion of biological sex, as is the case with medico as shown above. However, the variation appears to be independent of semantics, as a male prophet can be referred to with either masculine or feminine agreement. Having briefly introduced the categories that will not be considered further, I now turn to the type of gender doublet that constitutes the focus of the present analysis: gender doublets without semantic differences according to the gender assignment. This phenomenon can be illustrated with the French Example (4): this word has no fixed gender and the Académie Française dictionary specifies that it is a “nom masculin ou féminin invariable (on doit préférer le masculin)” (Académie Française 1980-2024: s.v. après-midi). (4) a. une après-midi ‘an afternoon’ b. un après-midi ‘an afternoon’ The focus of this book is on this particular type of gender doublet, exemplified in (4), where gender differentiation does not correspond to a semantic distinction and gender variation is not lexicalised. Nouns whose gender alternation is associated with a difference in meaning are considered instances of lexicalised gender variation and are therefore excluded from the subsequent analyses. However, this exclusion does not imply that all the nouns examined in this study lack semantically conditioned gender variation. For example, the Old Occitan noun salut exhibits gender variation only when it means ‘salutation’, whereas its gender remains fixed in its other meanings (see, for example, Sections 8.2.2.21 and 10.2.2). It is exactly this type of gender variation that cannot be explained simply on the basis of semantic or etymological differences and that is often <?page no="25"?> 2.2 Gender Doublets 25 marginalised in studies on gender, as it does not fit the broader system that is investigated in the following chapters. Thus, the definition of gender doublets employed in the present study is as follows: gender doublets are two variants of a single substantival word form that differ in their agreement with two different genders but do not show etymological or semantic differences. <?page no="27"?> 3 State of the Art Gender is a widespread noun categorisation system present in around 20% of the world’s languages. Allassonnière-Tang et al. (2021: 2) show this based on a study of 3,077 languages, 634 of which have the feature of gender. Due to the scope of this monograph, only a brief overview is possible, focusing on different types of gender assignment strategies. Following a preliminary introduction to gender in some languages of the world (Section 3.1), the focus turns on gender in Latin (Section 3.2), followed by an overview of gender in Romance languages (Section 3.3), which highlights some noteworthy cases. 3.1 Gender in the Languages of the World This section provides an overview of the scope of gender systems worldwide, presented according to the widely accepted typology of gender assignment systems (see, for example, Aikhenvald 2016; Aikhenvald/ Mihas 2019; Aikhenvald 2025; Corbett 1991, 2006b, 2008, 2012, 2014; Corbett/ Fraser 2010). It begins with semantic assignment rules (see Section 3.1.1) and then moves on to formal assignment rules (see Section 3.1.2), which are divided into morphological rules (see Section 3.1.2.1) and phonological ones (see Section 3.1.2.2). The information presented in the following sub-chapters is mainly based on works by Corbett (1991) and Aikhenvald (2025). Before going into the details, a short summary of the discussion of whether some of these languages have the feature of noun classes rather than gender is provided (for a more detailed discussion, see also Corbett 2007). The terms noun class and gender are often seen as synonymous in larger systems with more than four genders, while the notion of gender is also used for systems with only two or three distinctions (see Aikhenvald 2025: 25). Corbett/ Fraser (2010: 293) state that “the term ‘gender’ covers what in some traditions is called ‘noun class’”. The main difference between these two terms is that gender refers to a morphosyntactic category, i.e., something shown by agreement on other word classes, whereas noun classes refer to an inherent morphological or phonological category marked directly in nouns. As will be shown later, there is often a correlation between noun classes and their respective genders. However, this correlation is not absolute in all languages (see i.e., Section 3.1.2, Table 3.2). <?page no="28"?> 28 3 State of the Art The linguistic term of gender, used in medieval Latin school grammars, stems from the Latin genus ‘kind, sort’ (see Corbett 1991: 1). Haspelmath (2014: 492) argues about the use of terms from the Latin grammar tradition: “While it is certainly possible to describe all languages with the categories of medieval Latin school grammar […], such descriptions are usually seen as lacking insight and as distorting the true picture of the languages”. Aikhenvald (2012) clarifies the difference between genders and noun classes: When Europeans came to study African languages, they discovered larger genderlike systems with eight or more possibilities in languages like Swahili. These often did not include a masculine-feminine distinction. The term “noun class” came to be used for systems of this type; this term and “gender” are also often used interchangeably (Aikhenvald 2012: 78). Many linguists assert that there should only be a very small number of genders (see, for example, Dixon 2010: 11; Kilarski 2013: 156; Aikhenvald 2006: 463). For instance, Aikhenvald (2000: 18) specifies “I shall reserve the term gender for small systems of two to three distinctions” while Mel’čuk (2013: 739) includes systems with up to four distinctions. Corbett (1991: 1) merely states that “[a] language may have two or more such classes or genders” and Aikhenvald (2006: 463) agrees and specifies that “[t]here is a limited, countable number of classes”. As this book focuses on gender in Old Occitan—an Indo-European language with only two gender distinctions—and thus fulfilling even the most restrictive requirements on gender as presented above, the term gender is used as proposed by Aikhenvald (2012: 78) for the remainder of this study: “To avoid confusion, I use only the term “gender” here”. Using the term gender for all sorts of noun classes yields the following types of language: languages with two or three genders, such as many Romance, Germanic, and Slavic languages as well as languages from the Sepik region of New Guinea; and languages with four to six genders, such as many Australian and Caucasian languages. Some languages even have more than ten different genders, as seen particularly in Bantu languages in Africa (see Aikhenvald 2025: 25). In general, it can be said that gender is a morphosyntactic category, only shown through the agreement of other word classes (the agreement targets). Other categories, such as declension, may correlate with gender assignment in languages with morphological gender assignment. However, declension is a strict morphological category and must be considered separately from gender (see Section 3.2 and Table 3.2 on the correlation and distinction). 3.1.1 Semantic Assignment Rules The first type of language presented here is one with semantic assignment rules. There are different degrees to which languages belong to this class: languages <?page no="29"?> 3.1 Gender in the Languages of the World 29 with exclusively semantic assignment rules and languages with predominantly semantic ones. The prototypical and mostly used languages for this first case are Dravidian languages, which are primarily spoken in India. One of the major Dravidian languages is Tamil (see Corbett 1991: 8; Aikhenvald 2025: 29). Gender assignment in Tamil is divided into rational and non-rational categories. The rational criterion covers humans, gods and goddesses, as well as demons, and is divided into two sub-groups: masculine rational includes male humans and gods, while feminine rational includes goddesses and female human beings. Conversely, the non-rational gender includes all other nouns, including nonhuman animates. As this gender system is based only on these criteria, no further information is necessary other than whether the noun is rational, and if so, whether it is masculine or feminine (see Corbett 1991: 9; Aikhenvald 2025: 29). Similar systems are often found in North East Caucasian languages, which also divide nouns into male humans, female humans, and others (see Corbett 1991: 9; Aikhenvald 2025: 29). Some Dravidian languages have a twofold gender system, dividing nouns into masculine and non-masculine, the latter of which includes all nouns that do not designate a male (see Aikhenvald 2025: 29). An opposite system can be found in Diyari, which is spoken in Australia and in most Arawak languages spoken in South and Central America. It can also be found in Kaingang, which is spoken in Brazil. Here, gender assignment is based on the criterion of female humans versus all other nouns (see Aikhenvald 2025: 29-30). The gender assignment system in English is also based on semantic criteria: masculine agreement is triggered by male humans, and feminine agreement by female humans. Anything else is neuter, with exceptions only in the gender assigned to domestic animals, as well as to animals in children’s stories, which either corresponds to biological sex in cases of domestic animals or is conventionalised in tales. The reason for these exceptions is affective gender marking (see Corbett 1991: 12). Alongside these gender systems with strict semantic assignment rules, some languages belong to what Corbett (1991: 13) calls “Predominantly semantic systems”. An example of this type of language is Dyirbal, spoken in northeast Queensland, Australia. This language has four different genders: gender I contains male humans and non-human animates; gender II contains female humans, water, fire, and fighting; gender III contains non-flesh food and edible vegetables; and gender IV contains everything else (see Corbett 1991: 15-16; Aikhenvald 2025: 31-33). However, there are some exceptions to these rules. For example, birds belong to gender II even though they should be assigned to gender I. This exception can be explained through mythology: birds are culturally believed to carry the spirits of deceased human females, and therefore belong to gender II. Similarly, fishing line belongs to gender I due to its association with fish, <?page no="30"?> 30 3 State of the Art which are animate and therefore belong to that gender (see Corbett 1991: 15- 16). Therefore, the gender system in Dyirbal is a semantic-based system, but for some cases, it is important to understand the culture’s mythology or conceptual associations. It is also at this point that it becomes predominantly a semantic system, as it is not possible to determine merely from semantics whether a certain concept is strongly enough associated with another to change its gender. 3.1.2 Formal Assignment Rules The next two types of gender assignment rule are based on formal criteria, according to either morphology or phonology. In these cases, it is important to remember that “[n]o gender system is completely ‘mechanical’—that is, devoid of semantic motivation and based on principles other than meanings. There will always be a ‘core’ where meaning is crucial” (Aikhenvald 2025: 40). This means that while there are gender systems where assignment to a given gender is based solely on semantic cues, there are no gender systems where the assignment is based solely on formal cues—semantics always play a role (see Corbett 2013c). In languages that show a combination of semantic, morphological, or phonological assignment rules, the relative importance of these features is a matter of debate. Corbett/ Fraser (2010) propose that semantic rules play a more important role than formal ones, while Rice (2006) considers both types of rules to be equally important. Opposing these two theories, Basirat/ Allassonnière- Tang/ Berdicevskis (2021) conducted a computer simulation which calculated that there is a hierarchy of formal cues over semantic rules, as well as over the combination of formal and semantic rules. For this computer simulation, French, German, and Russian were compared. The model was trained to predict grammatical gender using character-level embeddings that capture only formal properties, context-based embeddings that capture only semantic information, or a combination of the two (see Basirat/ Allassonnière-Tang/ Berdicevskis 2021: 4). The results clearly state that formal cues are a much stronger predictor of gender assignment than semantic features and even slightly stronger than the combination of formal and semantic cues (see Basirat/ Allassonnière-Tang/ Berdicevskis 2021: 7-8). This also corresponds to the order of language acquisition in first language (L1) settings, where formal cues are acquired, followed by semantic cues (see Lust 2006: 250-251; see also Section 2.1). In a similar study on Old Occitan, which included only character-level information, Wiedner/ Schöffel (2026) could show similar results, attributing gender to a noun in Old Occitan with over 90% of accuracy based on the Latin etymon and gender, and the Old Occitan word form. Evaluating the attention machanism used shows that the model focuses on the ending, emphasising its major role in gender assignment in Romance languages. This underlines the importance of formal cues in gender as- <?page no="31"?> 3.1 Gender in the Languages of the World 31 signment, which is also supported by another computer simulation in Chatterjee et al. (2026). 3.1.2.1 Morphological Assignment Rules The first example of languages with morphological assignment rules is given with Bantu languages spoken in Africa. Here, Swahili is used as an example. This language also has a semantic core, with augmentatives belonging to one class (5/ 6), diminutives to another (7/ 8), and all remaining animates belong to gender 1/ 2 (see Table 3.1). However, apart from this semantic core, gender assignment is based on morphological cues. For example, all nouns starting with the prefix m-/ mibelong to gender 3/ 4, while those starting with ki-/ vibelong to gender 7/ 8 (see Corbett 1991: 47). Controller Typical morphological Target gender form gender form (prefixes on noun) (verbal agreements) 1/ 2 m-/ waa-/ wa- 3/ 4 m-/ miu-/ i- 5/ 6 ∅ ∼ ji-/ mali-/ ya- 7/ 8 ki-/ viki-/ vi- 9/ 10 N-/ Ni-/ zi- 11/ 10 u-/ Nu-/ zi- 15 kuku- Table 3.1: Morphological gender assignment in Swahili, taken from Corbett (1991: 47). Another language with mainly morphological gender assignment is Russian 1 . Apart from the semantic core based on biological sex, the assignment of a noun to a given gender is based on the declension class of the noun in question and is therefore highly predictable. Those nouns belonging to declension type I are masculine, those belonging to types II and III are feminine, and the remainder belong to the neuter gender (see Corbett 1991: 34-36). There are, however, some exceptions to the morphological rule: for instance, the noun врач ‘medical doctor’ belongs to the first declension class and is therefore masculine, but shows a mixture of masculine and feminine agreement when denoting a female doc- 1 Aikhenvald (2025: 41) lists Russian as a language with both phonological and morphological gender assignment rules. For example, nouns ending in -a typically belong to the feminine gender. However, as these nouns typically belong to the first declension (which I assume is the second according to Corbett (1991: 41)), they should also belong to the feminine gender based on morphology. Therefore, solely based on this data, it is not possible to state whether phonology is responsible for gender assignment, or whether phonology is the reason for a noun belonging to a certain declension class, which in fact determines the gender. <?page no="32"?> 32 3 State of the Art tor. Semantic assignment rules supersede morphological assignment, at least to some extent (see Corbett 1991: 38). Figure 3.1 shows the diagram of the complex assignment system in Russian and the interplay of semantics and morphology (the letter E is the point of entry for the noun). Figure 3.1: Gender assignment rules in Russian, taken from Corbett (1991: 41). Another language with semantic and morphological gender assignment rules is German, which has three genders (masculine, feminine and neuter). In German, gender agreement is only shown in the singular (see Aikhenvald 2025: 41), making it a convergent gender system, according to Corbett (1991: 155), with three genders in the singular, but no gender distinction in the plural. The semantic basis is the biological gender of animate nouns. Morphologically, suffixes provide information on the gender values, e.g., nouns ending in -ung are feminine (for instance Prüfung ‘exam’ or Erziehung ‘education’), while nouns ending in -chen are neuter. This includes diminutives, even when denoting a human with a biological gender, as previously mentioned in Chapter 1, Mädchen ‘girl’ is neuter in German (see Aikhenvald 2025: 41-42). In this case, morphology takes precedent over semantics, but in contemporary German, we often find instances of constructio ad sensum (see e.g., P. Koch/ Oesterreicher 2011: 82-84), where the semantic gender is adopted later on (sie ‘she’ in the case of the girl), contradicting the grammatical gender (which would be es ‘it’) (see Garbo/ Miestamo 2019: 24; Aikhenvald 2025: 42). 3.1.2.2 Phonological Assignment Rules Different types of phonological assignment are possible, based on segmental and supra-segmental features. An example of the latter is Qafar, an East Cushitic lan- <?page no="33"?> 3.2 Gender in Latin 33 guage spoken in northeastern Ethiopia and Djibouti. Once again, the semantic core of the gender system is based on biological gender: male humans and animals are masculine, and female humans and animals are feminine. In addition to this semantic basis, nouns ending in accented vowels are feminine, while all other nouns (those ending in a consonant or a non-accented vowel) are masculine (see Corbett 1991: 51; Aikhenvald 2025: 40). Another example for phonological-based assignment rules is Limilngan, an Australian language with four genders. The semantic core is as follows: Gender I includes humans, gender II animals, gender III plants, and gender IV is for everything else. However, phonology also plays a role in the assignment of nouns to genders II and III: nouns with an initial / l/ or / d/ are assigned to gender II, while those with an initial / m/ are mainly assigned to gender III (see Aikhenvald 2025: 40). In Hausa, a Chadic language spoken in Nigeria, the phonological assignment rule states that all nouns ending in -aa are feminine (see Corbett 1991: 52-43; Aikhenvald 2025: 40). 3.2 Gender in Latin Latin has three genders: masculine, feminine and neuter. Attribution to a certain gender was made according to two types of criteria: first, according to semantic criteria, and second, to morphological ones. Semantic criteria predominate over morphological criteria. They concern the animacy of nouns, dividing them into animate and inanimate categories. Animate nouns are attributed to either the masculine or the feminine gender according to their respective sex; in this case, therefore, there is an interplay between gender and sex. For instance, vir ‘man’ is a masculine noun, while uxor ‘wife’ is feminine (see Wang 2023: 71-72). Some inanimate nouns are also assigned gender according to semantic criteria: winds (e.g., aquilo ‘North-wind’), rivers (e.g., tiberis ‘Tiber’), mountains (e.g., Vesuvius ‘Vesuvius’) and the names of months (e.g., ianuarius ‘January’) are masculine, while islands (e.g., sardinia ‘Sardinia’), plants (e.g., rosa ‘rose’), especially trees (e.g., mālus ‘apple tree’), and names of countries (e.g., gallia ‘Gaul’) are usually feminine (see Pinkster 2015: 39). The rest of the inanimate nouns are assigned a gender based on their declension class: nouns from the first and fifth declensions are mainly feminine, while those from the second and fourth are mainly masculine (see Polinsky/ van Everbroeck 2003: 362). Exceptions to this rule form those nouns of the second declension class with the nominative ending -um, and nouns of the fourth declension class ending in -u, which are neuter. The third declension class is more complex with regard to gender assignment. The following list shows the tendencies for the gender assignment of third <?page no="34"?> 34 3 State of the Art declension nouns according to their endings (list following Panhuis 2015: 22; Rubenbauer/ Hofmann 1995: 40-42) • Mainly masculine are the nouns ending in - -or (dolor, doloris ‘pain’) - -os (flos, floris ‘flower’) - -er (aer, aeris ‘air’) • Mainly feminine are the nouns ending in - -ō (oratiō, orationis ‘orison’) - -is (navis, navis ‘ship, nave’ ; cassis, cassidis ‘helmet’) - parisyllables ending in -es (nubes, nubis ‘cloud’), imparisyllables in -as, -es and -us with the stem ending in -d/ -t (aetas, aetatis ‘age’; merces, mercedis ‘wages’; salus, salutis ‘sanity, health’) - consonants followed by -s (frons, frontis ‘front’; lex, legis ‘law’) • Mainly neuter are the nouns ending in - -e (mare, maris ‘sea’) - -l (animal, animalis ‘animal’) - -ur and -us with the inflectional suffix for the genitive ending in -ris (robur, roboris ‘strength’; corpus, corporis ‘body’) - -men (nomen, nominis ‘name’) Exceptions to the morphological rules are mainly found in the third declension. For instance, arbor ‘tree’ is feminine (contradicting the morphological rule but confirming the semantic rule, which outweighs the morphological rule); cor ‘heart’ is neuter; and sermō ‘discourse’ is masculine, etc. (see Rubenbauer/ Hofmann 1995: 40-42; Panhuis 2015: 22). There is also a lot of variation, particularly among nouns of the third declension class. Gender assignment does not seem to be as stable in the third declension as it is in the first and second. This is likely due to the more diverse endings and greater differences in declension. For example, the Latin noun mare was primarily neuter, but could also appear as masculine or feminine (see TLL 2025: s.v. “mare”; see also Section 8.2.2.12). It is therefore not surprising that gender variation in the transition from Latin to the Romance languages is mainly found in nouns deriving from the third declension class and is strengthened by the disappearance of the neuter (see de Dardel 1965: 50-65). <?page no="35"?> 3.2 Gender in Latin 35 The third declension class comprises the largest number of different nouns among the most frequent Latin nouns 2 , accounting for 52.6% of these nouns. The first declension class comprises 21.6%, the second 23.7%, the fourth 1.4%, and the fifth 0.7%. Of the 1,500 most frequent nouns, 38.1% are masculine, 40.8% are feminine, and 21.1% are neuter (see Polinsky/ van Everbroeck 2003: 362). The illustrations of the Latin gender system presented above clearly demonstrate the difference between grammatical gender (a morphosyntactic category) and declension class (a purely morphological category). There is a tendency suggesting a correlation between declension class and gender, but not an absolute correlation. This illustrates the difference between agreement controllers (substantives divided into the five declension classes) and agreement targets (e.g., the adjectives reflecting the division into the three genders). Table 3.2 provides an overview of the correlation between declension class and gender assignment in Latin. Masculine Feminine Neuter Animate yes yes no Inanimate yes yes yes Female no yes no Male yes no no 1DEC rare yes no 2DEC yes very rare yes 3DEC yes yes yes 4DEC rare yes rare 5DEC no (one exception) yes no Table 3.2: Correlation between gender and declension class in Latin, from Maiden (2010: 167). 2 All indications here are made for the 1,500 most frequent nouns according to Polinsky/ van Everbroeck (2003). These numbers are given for the most frequent nouns, as frequency plays a crucial role in the development of gender (see Section 3.5.2) <?page no="36"?> 36 3 State of the Art 3.3 Gender in the Romance Languages Some developments are shared by all Romance languages. For instance, the neuter has disappeared to a certain degree (see i.e., Romanian for a discussion of the preservation of a third gender). Mostly, masculine nouns have replaced the neuter singular nouns from the second declension that end in -um due to the omission of word final consonants as a result of sound change. Conversely, the plural forms ending in -a are often reanalysed as the feminine singular (see, for example, Meyer-Lübke 1883: 12-17; Diez 1882: 404; Chircu-Buftea 2011: 10), e.g., neuter sg. folium > Italian masc. sg. foglio ‘sheet’, while neuter pl. folia > Italian fem. sg. foglia ‘leaf ’. This led to a parallel binary gender system that is shared by all modern standard Romance languages with the exception of Romanian (see Loporcaro 2018: 33-34). Furthermore, differences in gender assignment are common from one Romance language to another, especially for nouns inherited from the third declension class in Latin, For example, the Latin third declension neuter noun maRe is masculine in Italian, Portuguese, and Romansh, feminine in French and Romanian, and agrees with both feminine and masculine in Catalan, Spanish, Provençal, and Sardinian (see Section 8.2.2.12; see also de Dardel 1965: 66-91). This is because third declension nouns that were formerly neuter have neither semantic, morphological, nor phonological cues that would facilitate attribution to a certain gender. This lack of clear morphological gender cues is maintained in the transition to the Romance languages, as they did not adopt the gender unambiguous endings of the first or second declension classes. Due to the continued absence of these cues for nouns inherited from the third declension class, their gender often varies between Romance languages and even within their respective varieties (see Maiden 2010: 169). Following the loss of the neuter gender, a binary gender distinction remains between feminine and masculine, with masculine being the default gender. In general, formerly feminine nouns from Latin remain feminine in Romance languages, while non-feminine ones remain non-feminine, also based on phonological cues (-a is mainly feminine, -o masculine) (see Maiden 2010: 168-169). The semantic core of the Latin gender system, whereby male animates belong to the masculine gender and female animates to the feminine, is continued in the Romance languages (see Loporcaro 2018: 52-56). 3.3.1 Romanian The question of whether Romanian has a twoor three-gender system is disputed, as Loporcaro (2018: 92) demonstrates. He lists a total of 17 different scientific works that argue for a three-gender system in Romanian, yet he also lists <?page no="37"?> 3.3 Gender in the Romance Languages 37 15 works arguing for a binary gender system. Table 3.3 gives an overview of the gender system in Romanian. Here, N represents the disputed gender, which is sometimes referred to as neuter, but more often as Genus Alternans ‘alternating gender’. The reason for the difference in interpretation of the Romanian gender system as binary or ternary originates in the definition of gender and the difference between controller and target genders. These two types of gender are defined by the division of nouns into “controller genders, the genders into which nouns are divided, from target genders, the genders which are marked on adjectives, verbs and so on” (Corbett 1991: 151; italics in the original). SINGULAR PLURAL Romanian M pantof-ul e bun pantofi-i sunt bun-i shoe(m)-def.m.sg is good[m.sg] shoe(m)-def.m.pl are good-m.pl N creion-ul e bun creioane-le sunt bun-e pencil(n)-def.m.sg is good[m.sg] pencil(n)-def.f.pl are good-f.pl F băutur-a e bun-ă băuturi-le sunt bun-e drink(f)-def.f.sg is good-f.sg drink(f)-def.f.pl are good-f.pl ‘the shoe/ pencil/ drink is good’ ‘the shoes/ pencils/ drinks are good’ Table 3.3: The gender system in Romanian, from Loporcaro (2018: 92). As Table 3.3 shows, Romanian has a series of nouns that trigger masculine agreement in the singular, but feminine agreement in the plural (see also Windisch 1973; Corbett 1991: 150-153; Maiden 2016; Loporcaro 2018: 92-109; Maiden et al. 2021: 53-73; Aikhenvald 2026: 73-75). It can therefore be argued that Romanian has two target genders (masculine and feminine), and three controller genders, including the Genus Alternans alongside masculine and feminine. This reconciles differing views on the number of genders in Romanian, suggesting that both interpretations of the Romanian gender system may be valid to some extent. Mostly former neuter nouns from the second declension class are assigned to the Genus Alternans, which is easily understandable from morphological and phonological perspectives: after the elision of word final consonants and the collapse of the declension system, these nouns morphologically resemble masculine nouns from the second declension class in the singular and those from the first declension class in the plural. This led to the adoption of feminine agreement in the plural (see Maiden et al. 2021: 59-61). It is beyond the scope of this monograph to discuss whether analysis of Genus Alternans as a third gender is correct (see, for example, Loporcaro 2017, 2018, 2022) or misleading (see e.g., Maiden 2016; Maiden et al. 2021). The important point here is that the Romanian nominal system has retained the Genus Alter- <?page no="38"?> 38 3 State of the Art nans, which is not present in any other contemporary standard Romance language, but, as Loporcaro (2018) and Wang (2023, 2025) argue, this is a common phenomenon in the transition from Latin to Romance languages. However, it only existed in very early stages of most varieties, so that it is quite likely that it was no longer existent when the texts in the COMETA (2025) corpus were written. 3.3.2 Italo-Romance Many different types of gender assignment can be found in Italo-Romance varieties. The following examples are based on the number of genders in their linguistic system, in order to show that there is more variation in Italo-Romance compared to standard Italian. 3.3.2.1 Two-Gender Systems Standard Italian has a parallel binary gender system. This means that Italian nouns have two types of controller gender, which trigger either masculine or feminine gender agreement. Conversely, it also has a binary target gender system, whereby the agreement targets express either masculine or feminine agreement. This agreement is found in both the singular and the plural forms, making it a four-cell/ four-form paradigm concerning the agreement targets, not the agreement controllers (see Corbett 1991: 155; Loporcaro 2018: 34), as presented in Table 3.4. Gender Suffix (sg) Suffix (pl) Example (sg) Example (pl) Masculine -o -i il libro ‘the book’ i libri -e -i il cane ‘the dog’ i cani -a -i il ciclista ‘the (male) biker’ i ciclisti Feminine -a -e la casa ‘the house’ le case -e -i la luce ‘the light’ le luci Table 3.4: Simplified gender system of Modern Italian. As Table 3.4 shows, the gender system in Italian is not an overt gender system, given that, for example, cane ‘dog’, which ends in -e, is masculine singular, while luce ‘light’, which also ends in -e, is feminine singular, and case ‘houses’, which ends in -e on the other hand is feminine plural. The most ambiguous vowels seem to be -i and e, which are relevant to both numbers and genders. Information on <?page no="39"?> 3.3 Gender in the Romance Languages 39 number being included 3 in the final vowel of the word in Italian (see Ayoun/ Maranzana 2020: 309). Neither gender nor number is completely overt in Italian or any other Romance language (see Loporcaro 2018: 39). There are nevertheless morphological cues, such as the inflectional class. In Italian, nouns belonging to the second inflectional class 4 , such as those ending with -a / -e are feminine (see Thornton 2003: 3, 2009: 18; Loporcaro 2018: 57). In addition to the aforementioned nouns, some masculine nouns distinguish between countable and collective meanings in the plural, which is also reflected in their agreement with different genders. For instance, the masculine nouns il muro ‘the wall’, il ginocchio ‘the knee’ or l’osso ‘the bone’ behave like masculine nouns in the plural when used as countable nouns: i muri ‘the walls’, i ginocchi ‘the knees’ or gli ossi ‘the bones’. On the other hand, they behave like feminine nouns when they have a collective meaning: le mura ‘walls’, le ginocchia ‘knees’ or le ossa ‘bones’. This collective feminine plural agreement can be traced back to the Latin neuter plural ending -a of the second declension class neuters and is a remnant of the Latin neuter. Such nouns with alternating gender are rare in Modern Italian, but were more common in Old Italian (see Section 3.3.2.2). The syntactically unmarked gender value in the major Romance languages is masculine, which is used by default. This results in the following simplified gender resolution in modern standard Romance languages: if everything involved is feminine, the agreement is also feminine; however, if at least one conjunct is masculine, the agreement is masculine. The place of the default gender marker occupied by the neuter gender in Latin is now filled by the masculine gender in Romance languages (see Loporcaro 2018: 38-39). Loporcaro (2018: 35) shows this with the following example: “lo sapevo (che saresti venuto) (It.) ‘I knew (you would come)’”. The pronoun lo is lost in translation; a more literal, nonidiomatic translation would be ‘I knew it (, that you would come)’. The referent of this pronoun lo ‘it’ is not further specified, which is why it appears in the default gender. Further interesting questions remain open, such as the gender assignment for loanwords (see e.g., Thornton 2001) or the gender assignment in deverbal conversion in Italian (see e.g., Thornton 2004; Marzo 2013a). For a proposition to resolve this issue, see Marzo/ Wiedner (2025). 3 It should nevertheless be noted that gender and number are two distinct features in Romance languages that must be considered separately. Both a feminine and a masculine noun can appear as a singular or a plural noun, while a noun in the singular can be either masculine or feminine (see Fedden/ Corbett 2017: 9). This applies regardless of the differentiation between sigmatic plural formation, as in Spanish, or vocalic plural formation, as in Italian (see Lampitelli 2014). 4 The terminology used here is taken from Thornton (2001: 485) and D’Achille/ Thornton (2003). <?page no="40"?> 40 3 State of the Art 3.3.2.2 Three-Gender Systems Taking a closer look at other Italo-Romance varieties reveals that many of them do not have a binary gender system. Old Italian is a good place to start, as it has a tripartite controller gender system with a binary target gender system (see, for example, Benincà/ Penello 2010; Faraoni 2013; Faraoni/ Gardani/ Loporcaro 2013; Loporcaro/ Faraoni/ Gardani 2014). This gender system in the earlier stages of Italian resembles the gender system in Romanian, which was discussed in Section 3.3.1. The gender system in Old Italian, as it developed from Latin, is shown in Figure 3.2. -us -um -a -i -a -ae SG PL a. Latin M N F I II III > -o -a -i -e SG PL b. Old Italian M F I III II Figure 3.2: The gender system in Old Italian, from Loporcaro/ Faraoni/ Gardani (2014: 7). These findings for Old Italian contradict previous assumptions that the neuter had already disappeared in Vulgar Latin (see Schön 1971: 4; Rovai 2012), as they clearly demonstrate a third gender in Old Italian after the presumed disappearance of the ternary gender distinction. This is also supported by recent findings concerning the Genus Alternans in Old Ibero-Romance (see Wang 2023, 2025: see also Section 3.3.3.1). Even in contemporary standard Italian, some remnants of this third, alternating gender are preserved, as seen in the well-known example of il braccio / le braccia ‘the arm(s)’ demonstrates (see Faraoni 2013: 29; Faraoni/ Gardani/ Loporcaro 2013: 172; Loporcaro/ Faraoni/ Gardani 2014: 5; Paciaroni/ Nolè/ Loporcaro 2013: 89). However, unlike in Modern Italian, this class of nouns was productive in Old Italian, attracting other nouns, such as neologisms formed by conversion. For example, the noun grido ‘cry, scream’ with the plural form grida, from the verb gridare ‘to cry, to scream’, was integrated into the alternating gender (see Gardani 2013: 387-394). Similar systems can be found in contemporary Italo-Romance varieties, especially in central and southern areas, e.g., in the Verbicarese variety, the variety of Verbicaro spoken in northwestern Calabria (see Loporcaro/ Silvestri 2016). Remnants of a similar three-gender system can also be found in other Italo-Romance varieties, such as in northern Corsican (see Faraoni/ Loporcaro 2014). <?page no="41"?> 3.3 Gender in the Romance Languages 41 3.3.2.3 Four-Gender Systems Some Italo-Romance varieties have an even more complex gender system with four genders. There are two types of four-gender systems in Italo-Romance varieties. The first type has four different controller genders but remains with two target genders. An additional alternating gender triggers feminine agreement in the singular and masculine agreement in the plural, as proposed for Old Romanesco (see Formentin/ Loporcaro 2012), as well as in Modern Bocchiglierese (see Cappellaro 2015). This is exemplified in Figure 3.3, where lo castiello with its plural le castella ‘the castle(s)’ belong to gender III, while la torre and its plural li torri ‘the tower(s)’ belong to gender IV (see Loporcaro 2015: 112). -o (lo) -a (la) M. N. F. -i (li) -e (le) I II III IV Figure 3.3: Gender system for the dialects of San Giovanni in Fiore and Bocchigliero = Old Romanesco/ Old Salentino, from Loporcaro (2015: 112). The other, more common type of four-gender system found in some central Italo-Romance varieties features Genus Alternans, with masculine agreement in the singular and feminine agreement in the plural. However, it also includes a fourth gender that only applies to the class of mass nouns (e.g., poison, salt) and therefore only exists in the singular. This fourth gender is often referred to as neutro di materia, whose status as a gender in its own right is sometimes overlooked, with it beeing seen either as a remnant of neuter (see Maiden 2010: 170), or as part of the non-feminine gender—a kind of sub-gender of masculine (see Haase 2000: 226-227). However, as this class of nouns has distinct agreement targets, it should be considered a gender in its own right 5 . Varieties that show this mass neuter therefore have four controller genders and three target genders (see Loporcaro/ Paciaroni 2011; Paciaroni/ Nolè/ Loporcaro 2013). This gender system can be found, for instance, in Agnonese (see Loporcaro/ Pedrazzoli 2016), in Maceratese (see Paciaroni 2017), in Ripano (see Paciaroni/ Loporcaro 2018a,b), or in Molfettese (see Loporcaro/ Breimaier/ Manzari 2021). It is also documented in older stages, such as in Old Neapolitan (and is still preserved in its modern variety, see Loporcaro/ Paciaroni 2011: 423; Ledgeway 2009). For a more detailed overview, see Loporcaro/ Paciaroni (2011) and Paciaroni/ Nolè/ Loporcaro (2013) and especially Chapter 4 in Loporcaro (2018). 5 For further discussion, see Loporcaro/ Pedrazzoli (2016: 83-84) and Loporcaro (2018: 145-149). <?page no="42"?> 42 3 State of the Art Figure 3.4 illustrates the gender system in Agnonese, which is an example of a four-gender system in Italy (see also Loporcaro/ Paciaroni 2011: 418-129; Loporcaro/ Pedrazzoli 2016: 82). Agnonese SINGULAR PLURAL N lə s E æ “ lə vuo “ nə ∅ ‘good salt’ M ru d E ndə vuo “ nə rə die “ ndə vuo “ nə ‘the good tooth/ teeth’ A ru ləndzuo “ rə vuo “ nə lə ləndzeu “ r 5 veu “ nə ‘the good bed sheet/ -s’ F la recc 5 veu “ n 5 lə reccə veu “ nə ‘the good ear/ -s’ Figure 3.4: Four-gender system in Agnonese, from Loporcaro (2018: 140). 3.3.2.4 Five-Gender Systems It is argued that some Italo-Romance varieties have five genders, combining both the four-gender systems presented in Section 3.3.2.3 and the respective alternating gender including a distinct form for mass nouns. This gender system is very rare and uncertain, as suggested by the title of the section in Loporcaro (2018: 277): “Romance dialects with five gender values? ”. Poggio S. Romualdo SINGULAR PLURAL N o ka : ču ∅ ‘DEF.N cheese’ M u mun E llu i mun E lli ‘the kid/ -s’ A 1 u pulsu e pulsa ‘the wrist/ -s’ F a vakka e vakke ‘the cow/ -s’ A 2 a pje : gu i pje : gu / pέ : gora ‘the sheep/ PL’ Figure 3.5: Five-gender system in Poggio S. Romualdo, from Loporcaro (2018: 280). Figure 3.5 shows this possible five-gender system in Poggio S. Romualdo, a variety spoken near Fabriano. A similar system can be found in the Amaseno variety <?page no="43"?> 3.3 Gender in the Romance Languages 43 (see Loporcaro 2018: 280-282). The gender labelled A2, the second alternating gender, only comprises a very small number of nouns and is no longer productive, so it can be considered an inquorate gender. However, it was probably productive for a significant amount of time in the past (see Loporcaro 2018: 280). These varieties therefore have three target genders, but five different controller genders. 3.3.3 Ibero-Romance The gender system in the major Ibero-Romance languages is quite similar to the gender system in Italian presented in Section 3.3.2.1, so it will not be presented in more detail here. For more information on gender in modern Catalan, see e.g., Messalles (2020), for gender in modern Spanish e.g., Azofra Sierra (2019) and Diebowski (2021), and for gender in modern Portuguese, see e.g., Mota (2016). Here, merely a short overview of new findings in older stages of Ibero-Romance varieties is given in Section 3.3.3.1. Furthermore, a special case of gender in Asturian is demonstrated in Section 3.3.3.2. 3.3.3.1 Old Ibero-Romance In two recent studies, Wang (2023, 2025) could show the presence of a stable Genus Alternans in early Old Ibero-Romance varieties. This confirms Loporcaro’s hypothesis of an intermediate stage in the transition from Latin to Romance characterised by the presence of such a gender (see Chapter 7 in Loporcaro 2018). Wang (2023: 245) identified 69 nouns belonging to the alternating gender (versus 70 in Wang (2025: 15)), with masculine agreement in the singular and feminine in the plural in Vulgar Latin and early Proto-Ibero-Romance. For these nouns, a feminised plural form ending in -as, which triggers feminine agreement, was noted early on. However, a corresponding feminine singular was only documented from the 11 th century onwards for the vast majority of these nouns (see Wang 2025: 17). Wang (2025: 20) classifies the nouns into eight semantic fields, e.g., buildings and constructions with nouns like Latin edificium ‘building’ or cimiterium ‘graveyard’. Another category comprises natural concrete entities including nouns such as Latin animal ‘animal’ or lignum ‘firewood’ (see Wang 2025: 20). Regarding its disappearance, he concluded that semantic individuation versus collectiveness plays a crucial role. Early-levelled neuter nouns were all associated with clearly individual objects, whereas those that retained Genus Alternans typically denoted collectives. Some collective nouns were eventually levelled from the singular when their components were usually interpreted as discrete individuals, despite occasional collective use, thus triggering masculine agreement. In con- <?page no="44"?> 44 3 State of the Art trast, nouns that were levelled from the plural and became feminine in Medieval Spanish consistently referred to closely connected collective entities. Pluralia tantum that became either plural feminines or uncountable feminine singulars denoted granular objects without clear individuation, leading to their reanalysis in medieval usage (see Wang 2023: 245-246). 3.3.3.2 The Special Case of Asturian The gender system in Asturian is of special interest, as it includes the difference between mass and count nouns directly in the agreement, at least in postnominal agreement. As Ordóñez (2006, 2007a) shows, this phenomenon is not only found in Asturia and Cantabria, but also in Castilla in the region between Cordillera Cantábrica and Toledo. In these varieties, prenominal agreement exhibits a binary contrast between masculine and feminine, whereas postnominal agreement features a tripartite system that distinguishes between masculine and feminine count nouns, masculine and feminine mass nouns, and neuter (see Loporcaro 2018: 161-162). This is commonly referred to as neutro de materia (see e.g., Ordóñez 2007b). The mass noun agreement is already found in late medieval texts, mainly in the thirdperson pronoun system (see Seibane 2010). The contrast between prenominal and other agreement is illustrated in Figure 3.6. Central Asturian (Lena) gender countability det N Adj a. F count la casa fri-a ‘DEF.F.SG cold house’ b. mass tsiche fri-o ‘DEF.F.SG cold milk’ c. M el café ‘DEF.M.SG cold coffee’ d. count pie fri-u ‘DEF.M.SG cold foot’ Figure 3.6: Preand post-nominal agreement in Asturian, taken from Loporcaro (2018: 160). The prenominal agreement, which distinguishes between masculine and feminine, works the same way as in most other Romance languages. It is based on a semantic core, but it is not solely a semantic system, as inanimates are also included in both genders. Regarding gender assignment for inanimates, formal criteria come into play, which is very similar to e.g., Spanish and Catalan (see Loporcaro 2018: 163). <?page no="45"?> 3.3 Gender in the Romance Languages 45 The more difficult and disputed issue is the three-way contrast in the other domains as well as the presence of a dedicated suffix for these agreement occasions (see Loporcaro 2018: 163). There are doublets in which agreement in the elsewhere domain changes the meaning from referring to a single unit of something to referring to a noncount or mass noun: “el papel blancu ‘the white piece of paper’ vs. papel blanco ‘white paper (noncount)’, una fueya seca ‘a dry leaf ’, fueya seco ‘dry folliage’” (Hualde 2020: 12). Different analyses have been proposed for this third agreement, ranging from a different gender value (e.g., masculine, feminine, neuter) over an additional number value (e.g., singular, mass, plural), to a semantic feature (e.g., count versus noncount) (for a summary, see Loporcaro 2018: 163-164). Building upon previous studies and proposals for analysing this agreement, Loporcaro (2018: 172-192) puts forward a new way of interpreting the agreement system of Asturian, arguing for two concurrent gender systems. As he convincingly shows, concurrent gender systems within one and the same language are uncommon in Romance languages, but are well known in other language families (see also Fedden/ Corbett 2017). The proposed gender system is depicted in Figure 3.7. System 1: Two Values M el pie/ café def.m.sg foot(m,mc)/ coffee(m,n) F l-a casa/ tsiche def-f.sg house(f,fc)/ milk(f,n) Domain: [ __ N ] NP System 2: ThRee Values M Count pie fri-u foot(m,mc) cold-mc Neuter tsiche/ café fri-o milk(f,n)/ coffee(m,n) cold-n F Count casa fri-a house(f,fc) cold-fc Domain: elsewhere Figure 3.7: Summary of the two concurrent gender systems of Central Asturian, as proposed in Loporcaro (2018: 172). 3.3.4 Gallo-Romance This section begins with an overview of the gender system in French (see Section 3.3.4.1), followed by information on gender in Franco-Provençal (see Section 3.3.4.2), before concluding with an overview of the state of the art on gender in Occitan (see Section 3.3.4.3). Given the fact that Gallo-Romance languages and varieties all have a binary gender system, the focus here is not on how many gen- <?page no="46"?> 46 3 State of the Art ders there are or on differences in gender assignment depending on the place of agreement. This section focuses therefore on the type of gender assignment and on gender variation within the respective binary system (or in comparison to French as the major standard language). Figure 3.8: Variation in gender of 165 nouns compared to French with the data of the ALF, taken from Brun-Trigaud/ Sauzet/ Allassonnière-Tang (2023: 18). The map in Figure 3.8 shows the variation in gender across Gallo-Romance varieties, based on a calculation of gender variation for 165 nouns in the data of the Atlas Linguistique de la France (ALF). This variation is based on standard French (see Brun-Trigaud/ Sauzet/ Allassonnière-Tang 2023: 18), which certainly explains the higher percentage of variation in the Occitan varieties compared <?page no="47"?> 3.3 Gender in the Romance Languages 47 to the varieties d’oïl in the north. Nevertheless, it shows that there is substantial variation in gender assignment between the Languedocian and Provençal linguistic areas. Brun-Trigaud/ Sauzet/ Allassonnière-Tang (2023) identify the main factors of gender variation as semantics (mainly concerning nouns in the semantic fields of food or daily life), etymology (the most significant variation is seen in former neuter nouns inherited from the third declension class) as well as phonology (word-final vowels that are maintained in Occitan and in Franco-Provençal, but not in French). Platz (1918) proposes an additional factor of variation in gender: the superposition. This means that a word may change gender when it conceptually replaces another noun and adopts the gender of the former noun. This theory, confirmed by Brun-Trigaud/ Sauzet/ Allassonnière-Tang (2023: 17-18), can be exemplified with the noun noce ‘wedding’, which is predominantly feminine in Gallo-Romania but sometimes masculine in the north, liekely due to the superposition of the other noun for wedding, mariage, which is a masculine noun. 3.3.4.1 French In contrast to the other major Romance languages, which have a binary gender distinction, French has a tendency towards a binary convergent system. In French, most agreement targets preserved distinct gender markings in the singular, while their forms merged in the plural due to the effects of sound change (see Loporcaro 2018: 41), as shown in Figure 3.9. Nevertheless, some agreement targets also preserve a gender distinction in the plural, e.g., the masculine nouveaux [ nu.vo ] versus the feminine nouvelles [ nu.vEl ], both ‘new’. There is no difference in sound between the singular and plural forms, but there is a distinction between the masculine and feminine forms. a. def = DO clitic SG PL M lə le(z) F la b. ‘this’ SG PL M sə se(z) F s E t c. ‘my’ SG PL M mõ me(z) F ma Figure 3.9: Syncretism in the plural formation of French, from Loporcaro (2018: 41). Due to sound change, word-final vowels and consonants are less significant for gender assignment in French than in other Romance languages. Even the word final -a was weakened to an [ @ ] and eventually disappeared completely (though there are regional differences, see for example D. Müller/ Wiedner 2022), making gender differentiation more difficult based solely on phonological criteria (see Corbett 1991: 58; Brun-Trigaud/ Sauzet/ Allassonnière-Tang 2023: 14). This <?page no="48"?> 48 3 State of the Art is also reflected in the high number of agreement errors for learners of French as a foreign language (see Lambelet 2012; J. Wolf 2024). Corbett (1991: 57-61) argues for a very complex assignment system based on phonological cues as presented in Table 3.5. He argues against the idea of gender assignment being totally arbitrary in French (apart from the semantic core of sexuated animates). As it is not an overt gender system, most of the endings in the table are not one hundred percent unambiguous for one gender or the other, but there are nevertheless strong tendencies. This interpretation of the French gender system as one based on formal criteria was first proposed by Armbruster (1888), who saw formal criteria as one of the main reasons for changes in gender. Nouns ending in a consonant Nouns ending in a vowel Final consonant Number of nouns % MASC Final vowel Number of nouns % MASC / ʒ/ 1453 94.2 / œ̃/ 17 100 / m/ 1406 91.9 / ɑ̃ / 1963 99.3 / f/ 301 89.0 / ɛ̃ / 938 99.0 / r/ 5175 76.8 / ø/ 189 97.4 / g/ 235 73.2 / o/ 865 97.2 / k/ 833 66.6 / E / 625 90.2 / b/ 129 65.1 / u/ 171 87.7 / l/ 1581 58.4 / a/ 970 82.6 / t/ 2269 51.2 / y/ 201 71.6 / p/ 214 48.6 / e/ 2791 50.1 / ï / 69 39.0 / ɔ̃ / 2665 29.8 / s/ 1380 38.5 / i/ 2336 24.6 / d/ 714 38.1 / ʃ/ 290 34.0 / j/ 352 32.4 / v/ 143 31.5 / n/ 1135 31.5 / z/ 612 10.0 Table 3.5: Phonological gender assignment in French, from Corbett (1991: 59). 3.3.4.2 Franco-Provençal Information and studies on gender in Franco-Provençal are scarce, particularly with regard to older language stages. As Carles/ Glessgen (2019) show, research on medieval Franco-Provençal lacks systematic scientific study: “Il n’existe aucune synthèse, aucune vue d’ensemble qui établirait les trajectoires de l’écrit francoprovençal au Moyen Âge et qui décrirait la physionomie de sa scripta.” (Carles/ Glessgen 2019: 74). It is therefore not possible to give any information on the gender system of Old-Franco-Provençal. This would be an important topic for future research. <?page no="49"?> 3.3 Gender in the Romance Languages 49 In modern varieties of this language, Brun-Trigaud/ Sauzet (2023) show the gender variation for 74 lexical units taken from the ALF, which was collected in the late 19 th century. They demonstrate that almost all of these nouns vary in gender within the Franco-Provençal linguistic area, but the maps are also interesting for the other Gallo-Romance varieties, since they include all ALF data points. Generally, there is significant variation in gender, both compared to French and Occitan, and within different Franco-Provençal varieties. In his Franco-Provençal grammar, Stich (1998: 88-91) gives some further information on the gender system. He begins by stating that gender assignment in Franco-Provençal basically works in a similar way to French. He also lists some nouns where the gender differs between the two languages, for example the Franco-Provençal feminine noun la sâl ‘the salt’ compared to the masculine French noun le sel. Furthermore, he mentions three nouns that he calls épicène (see Stich 1998: 88), which are in fact gender doublets with a diatopic distribution. For example, mensonge ‘lie’ appears with both masculine and feminine agreement depending on the variety (see Stich 1998: 88). In the part of the grammar dedicated to the nominal system and the glossary at the end of the book combined, Stich (1998: 88; 379-409) lists a total of 6 gender doublets, namely afâre ‘matter’, blèche ‘Band-Aid’, cllédar ‘lattice gate’, corvâ ‘drudgery’, lace ‘lace, shoestring’, and the aforementioned mensonge ‘lie’. In addition, for the noun sarpent ‘serpent, snake’, he indicates mostly a feminine use (see Stich 1998: 404), which suggests that there is also a masculine use for this noun, as it is only mostly and not exclusively feminine. The nouns listed here as gender doublets in Franco-Provençal or where a difference in gender to French is indicated are often nouns that also show gender variation from one Romance language to another (see e.g., French feminine l’affaire versus Italian masculine l’affare; Italian masculine il sale versus Spanish feminin la sal to name but a few examples). Apart from this small amount of information on gender variation in Franco- Provençal, there are no further studies on this subject to my knowledge. There are some dictionary entries, which will not be discussed here, as they are included in the analysis of gender doublets in Section 8.1. 3.3.4.3 Occitan Loporcaro (2018: 203-207) dedicates only five pages to grammatical gender in Old Gallo-Romance, with a focus on Old French and presents little data on Old Occitan. He begins this short section with the following statement: Ever since their earliest documentation, Old French and Old Occitan seem to have been more advanced than Old Italian or Romanian along the path which led to estab- <?page no="50"?> 50 3 State of the Art lishing a binary (controller) gender system, with full reassignment of all previously neuter nouns to the remaining two genders (Loporcaro 2018: 203). He argues for the preservation of a dedicated neutral agreement target form, a specific default form, only for pronouns and adjectives derived directly from the former Latin neuter. These neuter forms started to disappear early in Old Gallo-Romance, leaving the masculine forms as the default (see Loporcaro 2018: 204). Apart from that, nouns generally kept the gender inherited from Latin, reassigning neuter forms to either the masculine or feminine gender, as described for the Romance languages in Section 3.3 (see Donaldson/ Sibille 2024: 202). Chambon (2003: 362) argues for an additional gender in Old Occitan for three nouns, namely carra ‘cartload’, paira which translates approximately as ‘one day of work by a pair of oxen owed as payment’, and semoia ‘half a bushel (a measure of capacity)’ (for the translations, see DOMél 2025: s.v. “carra”; “paira2”; “semoia”). These three nouns occur with the numerals doa and tria, which, according to Chambon (2003: 362), do not appear in other contexts. He argues that: Il est donc préférable d’analyser / ’-a/ dans / do+a/ et / tri+a/ comme la marque (qu’on notera {’-a2}) d’un troisième genre ou d’un second nombre. […] Il vaut donc mieux, nous semble-t-il, rendre compte de front et de l’emploi au pluriel seulement et de l’indéterminabilité du genre, et dire que / c’arra/ , / p’ajra/ , / sem’oja/ sont à la fois des pluriel II et des substantifs du troisième genre (Chambon 2003: 362). This proposed third gender only involves three nouns, which only agree with the cardinal numerals for two and three, and no other agreement targets. Therefore, it is questionable to attribute these nouns to a third gender on their own. If it is a gender in its own right, it is more accurately described as an inquorate gender: Inquorate gender values are agreement classes which comprise a small number of nouns, and whose agreements can be readily specified as an unusual combination of forms available for agreement with nouns with the normal gender values (Corbett 2012: 84). Apart from the very low number of nouns included in the gender, the prerequisite for a gender to be inquorate is the appearance of agreement targets involved in other contexts. This is the case at least for doa, which also appears in the COMETA (2025) corpus, as Example (5) below demonstrates. Therefore, it is unjustified to analyse these nouns as belonging to a third gender, let alone an inquorate gender, since the combination of agreement targets is not unusual. <?page no="51"?> 3.4 Intermediate Summary 51 (5) ambe doa milia cavalliers. ‘with two thousand knights’ (BnF, Français 13509 14 th century: f o 22 verso (v o ), column (c) 1, l 25) Apart from the information presented here, there is little research on grammatical gender in Old Occitan, with only some lists of nouns that have an unclear or double gender value. As these lists form the basis of the analysis presented in Chapter 8, they will not be presented here. The same applies to Wiedner (2026, forthcoming), which explore nine examples of gender doublets in Old Occitan based on data from COMETA (2025). This article is included in the analysis of gender doublets later in Chapter 8. A special case in contemporary Languedocian is the three nouns amor, art, and onor, which are instances of Genus Alternans in Languedocian, as they act as masculine nouns in the singular and as feminine nouns in the plural (see Alibèrt 1976: 46). In addition to these three nouns, Alibèrt (1976: 46) gives a list of 49 further gender doublets in Languedocian. About half of these were also listed as gender doublets in Old Occitan and are therefore part of the analysis in Chapter 8. Most of the nouns listed in Alibèrt (1976: 46) either designate abstract concepts (e.g., carèma ‘Lent’ or freg ‘cold’) or are mass nouns (e.g., poison ‘poison’ or sal ‘salt’). Bach (2024: 320) highlights the fluctuation between masculine and feminine gender especially for the nouns designating blood, honey, and milK, all being mass nouns. This variability in gender assignment exists also in other Romance varieties and languages, see the discussion on mass nouns and a dedicated mass neuter in Sections 3.3.2.2 and 3.3.3.2. 3.4 Intermediate Summary In general, it has been shown that the gender system evolved from a complex three-gender system to a simpler two-gender system during the transition from Latin to Romance languages. This transition is influenced by morphological and phonological developments, as well as semantic relations. The loss of the neuter gender has different implications for the development of gender in the Romance languages, with different stages being seen in contemporary Romance languages. The loss of neuter has two main consequences: the first concerning the gender system as a whole, and the second concerning the gender of specific nouns. <?page no="52"?> 52 3 State of the Art Regarding the former, it has been shown that different gender systems have evolved from the Latin starting point, with up to five distinctions on a controller gender level. On the target gender level, most varieties have either two or a maximum of three target genders. This is the case in Romanian, where there are three controller genders, but only two target genders, the third controller gender being the Genus Alternans. In some respect, this type of gender is the expected outcome of the loss of the neuter gender, given that the neuter endings merged with the masculine endings in the singular and the feminine endings in the plural. A type of alternating gender is therefore postulated for older stages of the Romance languages (see Loporcaro 2018: 223). Regarding the latter, the second consequence of the loss of neuter is that neuter nouns had to be reassigned to one of the remaining genders. In languages and varieties that preserve the alternating gender, most neuters belong to that gender. However, other varieties that lost the alternating gender and developed a binary gender contrast had to assign a new gender to all former neuter nouns. This leads to a considerable number of nouns that trigger a different gender in various Romance varieties, and occasionally within one and the same variety. In general, the nominal system in Latin is more complex than in the standard Romance languages. This is partly because Latin still has a complex declension system with five different declension classes and six cases, as well as the three genders. The transition to Romance languages displays a general simplification of the nominal system, with a reduction in the number of declension classes and the loss of the case system. However, this simplification process can lead to a more complex intermediate state, which is still evident in some Italo-Romance varieties, with up to five different genders. It is interesting to note that gender-variable nouns in different Romance language varieties are often nouns that vary from one language to another. Most of these nouns are derived from the Latin third declension class (especially, but not limited to, former neuters from this declension class), which was already susceptible to gender variation in Latin (see Section 3.2). For instance, nouns inherited from the Latin maRe show strong variation within Romance languages and their varieties. Furthermore, mass nouns and collectives also seem susceptible to variation in Romance languages; for example, consider the aforementioned nouns describing the concepts of salt or blood. As shown in Section 3.3.4, Gallo-Romance varieties all have a binary gender system. Therefore, the interesting question regarding these varieties is the gender assignment for former neuters and how this varies both within and among varieties. This book therefore focuses on gender assignment rather than gender systems as a whole, the latter being a more interesting topic regarding, for example, Italo-Romance varieties. Due to the scope and goals of this book it was not possible to provide detailed information on every Romance variety and its <?page no="53"?> 3.5 Factors Influencing the Development of Gender Systems 53 gender system(s) 6 . A summary of the different types of gender system in the Romance languages in Europe is given in Figure 3.10. Figure 3.10: Romance gender systems in Europe, adapted from Loporcaro (2018: 60). 3.5 Factors Influencing the Development of Gender Systems The following sections aim to illustrate the main forces and factors influencing changes in gender systems. This is not intended to be understood as an exhaustive list. 3.5.1 Sound Change in combination with morphological change One of the main reasons for a change in gender systems is sound change in combination with morphological change. For example, with the omission of wordfinal consonants, the morphological distinction between neuter nouns from the second declension ending in -um and masculine nouns from that declension ending in -us was lost, as both now ended in -u (see Diez 1882: 419; Maiden 2010: 167-168). The same is true for formerly feminine nouns from the fourth declension class ending in -us that often became masculine after the collapse of the case system (see Meyer-Lübke 1894: 5-25). The morphological (and phonolog- 6 For a more detailed discussion and overview, the reader is referred to the seminal work by Loporcaro (2018), who presents a comprehensive overview of the gender systems in the Romance varieties in Europe. <?page no="54"?> 54 3 State of the Art ical) distinction between nouns from the second and fourth declension, which was previously visible in the genitive, for example, was no longer possible because those cases no longer existed. Following the collapse of the case system and the distinction between the two declension classes, most nouns from the fourth declension class became masculine nouns and were absorbed into the second declension class (see Diez 1882: 419), apart from nouns such as manus ‘hand’, which remained feminine or gender ambiguous (see Núñez-Méndez 2026: 96-97; see also Section 8.2.3). The same applies to formerly neuter plurals from the second Latin declension class, which ended in -a. This plural form was mainly reanalysed as the feminine singular for nouns that designate concepts associated with a plurality of referents. These nouns formally converged with feminine nouns of the first declension class while sometimes keeping a collective value (see Schön 1971: 12; Mihatsch 2021: 9; see also the Italian feminine plural collective noun, e.g., le mura). On the effects of morphophonological erosion on gender reduction and possibly even gender loss, see Garbo/ Miestamo (2019). 3.5.2 Frequency The frequency of a noun may also influence gender development in a language. In a computer simulation of gender development from Latin to Old French, Polinsky/ van Everbroeck (2003: 380) show a correlation between the frequency of a certain noun and the preservation of gender for other nouns: the more frequent a noun is, the more resistant it is to changes in gender. In addition to maintaining their own gender more stably, the most frequent nouns also contribute to the maintaining of gender for less frequent nouns. 3.5.3 Analogy due to Semantic Relations The next factor that may play an important role in the development of gender for individual nouns is analogical change due to semantic relations. The first example to mention is that of antonyms. In German, for example, the feminine die Nacht ‘the night’ becomes (at least in literary contexts) a masculine noun des Nachts ‘at night’ in the genitive case, in analogy to the masculine noun des Tages ‘by day’ (see Malkiel 1951: 508). A similar argument can be found for the masculine gender change of the French noun val ‘valley’ through its close semantic relationship with the masculine noun mont ‘mountain’, as discussed by Armbruster (1888: 5). A possible reason for the feminine gender of the former neuter (but already gender-variable) noun maRe in some Romance varieties might be the clear gender value of the semantically closely connected noun teRRa. The feminine noun teRRa attracts maRe to be feminine as well (see H. Sachs 1886: <?page no="55"?> 3.5 Factors Influencing the Development of Gender Systems 55 44; Meyer-Lübke 1894: 380; Malkiel 1951: 508). However, the validity of this assumption is unclear, as Malkiel (1951: 508) questions: Should the conclusion be drawn, from It. masc. mare, that ‘land’ and ‘sea’ have been less vividly associated in Italy than, say, in France? The hypothesis is open to no scientific test, but common sense rules it out, since a glance at the map shows the incomparably greater dependence of Italy on the sea-lanes than of any other Romance country, with the possible exception of Portugal and Catalonia, which, incidentally, exhibit a conflicting pattern of gender (Malkiel 1951: 508; italics in the original). A similar scenario of analogical change in gender assignment due to a semantic relation is proposed in the FEW (1922-2003: s.v. “marmor”) for the Latin third declension neuter noun maRmoR, which may have changed gender from neuter to feminine in relation to the feminine hyperonym petRa. This sort of analogical change from one gender to another is mainly documented for formerly neuter nouns which had to be reattributed to another gender during the loss of the neuter gender. In these cases, semantically related concepts with a stable gender value (both petRa and teRRa are feminine nouns ending in -a, showing a stable gender value) might facilitate the change in gender for formerly neuter third declension class nouns (see Malkiel 1951: 508-509). 3.5.4 Analogy to Rhyming Words Meyer-Lübke (1894: 426-427) proposes a similar factor in combination with phonological form, but without a semantic relation. He argues for the influence of rhyming words. He gives the example of the three masculine Latin nouns fons, mons, and pons influencing the formerly feminine noun fRons to become masculine in French. He suggests the same reason for the gender change of the Latin noun dens, which became feminine due to the influence of the feminine noun gens. Malkiel (1951: 508-509) agrees with this factor of gender assignment change and lists about twenty nouns that may have changed gender due to the influence of rhyming words. 3.5.5 Contact Languages Another factor that may have contributed to the development of gender systems in the transition from Latin to Romance languages is substrate languages (see Polinsky/ van Everbroeck 2003: 379-380). The language of the conquered population in a certain territory died out eventually, but left traces in the stratum, the language that survived (for a definition of substrate languages, see Krefeld 2003). In the case of Old French (and also Old Occitan, except Gascon), this substrate language is Gaulish, which has a binary gender contrast that distinguishes <?page no="56"?> 56 3 State of the Art between masculine and feminine. The Gaulish data significantly altered the outcome of the computer simulation by Polinsky/ van Everbroeck (see 2003: 379- 380), contributing to the realistic results observed in Old French. The Gaulish data accelerated the disappearance of the neuter gender in the computer simulation. The possibility of gender loss 7 in language-contact scenarios has been documented for other languages as well. For example, the Tamian dialect of Latvian lost its gender due to contact with Livonian, a Balto-Finnic language without gender (see Aikhenvald 2016: 87). While the results presented by Polinsky/ van Everbroeck (2003) impressively demonstrate the importance of substrate languages for gender development, it remains unclear, from their article, where they obtained their data on Gaulish. Consequently, both the quality and validity of the data remain uncertain, as the authors themselves acknowledge (see Polinsky/ van Everbroeck 2003: 366). This is compounded by the scarcity of Gaulish data (see Blom 2023). An additional factor related to contact languages is gender variation in loanwords. In Latin and Romance languages, Greek is an important source of loanwords such as Old Occitan profeta ‘prophet’ and papa ‘pope’ (see Section 8.3), or Spanish dogma ‘dogma’ and problema ‘problem’ (see Núñez-Méndez 2026: 96). Irrespective of their ending in <-a>, these nouns are all masculine, which is a frequent phenomenon in Spanish. For example, Harris (1991: 37) found around 600 instances of masculine Spanish nouns ending in -a. However, some of these nouns show gender variation and trigger both masculine and feminine agreement. Examples include the Greek loanwords dracma ‘drachm’ or reúma ‘rheumatism’ in Spanish. This gender variation is caused by the different gender endings in Greek and the Romance languages; in Greek, -a is a neuter ending, whereas in the Romance languages, this vowel mainly appears in feminine nouns. This phenomenon is not limited to loanwords from Greek, but also applies to loanwords from other languages in Spanish, such as the Finnish loanword sauna ‘sauna’ (see Núñez-Méndez 2026: 131). 3.5.6 Normalisation and Standardisation The presence of gender doublets in French can be explained, at least in part, by the difference between the gender of a noun as it has developed over time and is applied by the speakers on the one hand, and a gender introduced by normative and re-latinisation tendencies of those who wanted to restore the “original” gender of a noun. This was the case in France in the 16 th century. Such actions 7 There is also the opposite possibility, whereby a language develops a new gender (or gender system) as a consequence of language contact (see Aikhenvald 2016: 88-90; Garbo/ Miestamo 2019: 40-44). See Loporcaro/ Gardani/ Giudici (2021) for a case study on the complexification of the gender system in Istro-Romanian due to contact with the Croatian language. <?page no="57"?> 3.5 Factors Influencing the Development of Gender Systems 57 sometimes lead to discrepancies between the norm and actual usage. For example, the noun enzyme is used as a masculine noun, while the Académie française indicates a feminine gender due to its etymology (see Jeanmaire 2014: 100-101). A similar scenario can be seen with the gender for Covid in French during the Covid pandemic. This acronym for Coronavirus Disease was mainly used as a masculine noun in French (see Balnat 2020: 11). Contrary to this entrenched masculine use, on 7 May 2020 the Académie Française opted for a feminine use of covid due to the semantic basis of maladie ‘disease’, which is feminine in French (see Académie française 2020). This sort of normative and prescriptive language policies may also influence the gender assigned to certain nouns. 3.5.7 Changes in Society An additional factor that may influence the development of gender is correlated with changes in society. A current example of this can be seen in modern tendencies towards gender-inclusive languages 8 . The goal of including all genders in job titles, for example, influences gender assignment in languages, as Sarradilla Castaño (2023: 258-278) concludes in her article: La visión de la naturaleza en la Antigüedad se reflejaba en el género; como también lo hacía la lucha entre las tendencias culta y popular al introducirse nuevas voces; y como lo hace hoy en día la incorporación de la mujer a determinadas actividades antes reservadas solo a los hombres o la reivindicación de su visibilización dentro de los distintos colectivos. Como señalaba al principio, pues, lengua y sociedad a menudo van de la mano y, afortunadamente, el cambio sigue siendo posible (Sarradilla Castaño 2023: 278-279). 8 As gender-inclusive language is not the focus of this study, it is mentioned here mainly as a possible reason for language change. Only some indications for further literature on the subject are given below. For Spanish, see Scotto/ Perez (2020) and Zapf (2024). For genderinclusive language in French, see Armstrong/ Bauvois/ Beeching (2001), Becquer et al. (1999), and Schafroth (2015). For feminine job titles in French, Spanish and Italian, see Schafroth (2013). For an overview of historic developments in the Romance languages, see Heinemann/ Neumann-Holzschuh (2009). <?page no="59"?> 4 The Old Occitan Nominal System—an Overview This section will focus not only on gender, but also on case and number, as well as the different agreement targets. The aim is to provide an overview so that the examples given later can be interpreted based on this knowledge. The presentation of the different declension types that follows is a normalised approximation of the Old Occitan nominal system. It is important to note that Old Occitan is a non-standardised historical language that shows a considerable amount of variation at different levels, including the graphic level (e.g., thera are at least eight different ways to write the sound [ L ], see Meisenburg (1996: 370)), and some features are not applied consistently. This should be kept in mind when considering the data presented in Chapters 8 and 9. In addition to the notable variation within a single Old Occitan variety, there is also considerable diatopic variation which cannot be discussed in this brief overview. The main part of the literature on the Old Occitan nominal system focuses on early Old Occitan, whereas the data used in this study dates from a later period (1250-1400). For instance, Grafström (1968) provides an overview of the morphology of the earliest Old Languedocian charts, Reimann (1882) analyses the language up to 1300, and Loos (1884) only marginally uses texts originating from after the decline of the troubadour lyric after the Albigensian Crusade (1209-1229). The same applies to Grandgent (1905), Roncaglia (1965), Schultz- Gora (1973), Paden (2003), and Romieu/ Bianchi (2025), whose grammatical treatises are based on the troubadour lyrics. Grandgent (1905: 1), for example, states: “lacking words or forms must occasionally be sought in writings later than the 13th [century; MW]”. This is almost exactly the case in Skårup (1997: 5): “La langue écrite est l’ancien occitan (dit aussi ancien provençal) d’avant 1300 […] Ne sont cités qu’exceptionellement des textes postérieurs”. Therefore, the following presentation is based on works that also include data on Old Occitan after the mid-13 th century. These are 1 : Anglade (1921), Ronjat (1937), the works by Jensen (1972, 1973, 1975b, 1976, 1986, 1994), Paden (1998), and Romieu/ Bianchi (2002). 1 If information is taken from one of these works, it is not referenced specifically in the following, unless there are direct citations. Tables 4.1 to 4.7 are synoptic tables of the information in these grammars. If the information is taken from other sources, this will be indicated directly in the text. <?page no="60"?> 60 4 The Old Occitan Nominal System—an Overview 4.1 Nouns and Adjectives As demonstrated in Chapter 3, the Classical Latin declension system collapsed during the transition to Old Occitan. This resulted in a completely new nominal system, with two genders and two cases, while maintaining two numbers. The following presents the typical declension types of both masculine and feminine nouns. Some nouns have individual, not systematic, declension types. These will not be explored here, as the aim of this chapter is to provide a general overview rather than a detailed study of every case. Furthermore, they are irrelevant to the data presented later in this book. Generally speaking, if the two-case system is intact, noun and adjective declension is quite similar to the Old French declension system (see Chambon (2003: 359); for the Old French nominal system, see Rheinfelder (1976b), L. Wolf/ Hupka (1981), Marchello-Nizia et al. (2020), and Buridant (2022)). However, it should be noted that the two-case system was already in decline in the 13 th century, becoming gradually more facultative (see Thomas 1905), until it eventually disappeared completely (see Jensen 1976: 123; Chambon 2003: 349-350). The following quote summarises the main facts about Old Occitan noun declension in the 13 th century. The variation and optionality of case declension described in this quotation became even more pronounced subsequently (see Jensen 1994: 17-18): En ancien occitan, la flexion des substantifs (i) est optionnelle en cas et (ii) accessible seulement à certains substantifs; (iii) elle appartient au même ordre morphologique que la flexion plurative universelle et (iv) elle ne comporte qu’un cas; (v) les sousclasses flexionnelles ne sont pas disjointes, mais inclusives; enfin, (v[i; MW]) on a affaire à une flexion de mots (formes lexicales libres) et non de thèmes (formes lexicales liées). Par tous ces traits, la flexion substantivale de l’ancien occitan diffère profondément dans son principe et dans son organisation de la déclinaison et des déclinaisons du latin classique. Les modèles de description décalqués du latin classique sont inadéquats (Chambon 2003: 359). Gender Case Singular Plural Gloss Feminine Nominative domna domnas ‘the lady/ ies’ Oblique domna domnas ‘(of) the lady/ ies’ Feminine Nominative flor(s) flors ‘the flower(s)’ Oblique flor flors ‘(of) the flower(s)’ Masculine Nominative murs mur ‘the wall(s)’ Oblique mur murs ‘(of) the wall(s)’ Table 4.1: Nominal declension in Old Occitan. <?page no="61"?> 4.1 Nouns and Adjectives 61 Table 4.1 gives an overview of the prototypical declension in Old Occitan with an intact two-case system. For feminine nouns of the first type that end in a vowel (mainly -a), there is generally no morphological case distinction; only a number marker in the form of the morpheme -s indicates plurality. The singular has no specific marker (see Chambon 2003: 345). For feminine nouns ending in a consonant and thus following the second type, an optional case marking is added to the plural marker -s in the nominative singular. This case marking is realised with the distinct, homonymous morpheme -s. These two endings are identical graphically and phonetically, but cannot be linked morphologically: one means ‘plural’, the other ‘nominative singular’ (see Chambon 2003: 346). However, “[c]ontrairement au pluriel, le cas n’est pas une catégorie de flexion obligatoire des substantifs féminins en ancien occitan, mais une catégorie optionnelle, l’option étant ouverte à certains substantifs féminins seulement” (Chambon 2003: 348). For masculine nouns of this first declension type, which are mainly derived from the second declension class in Latin, an -s is added to mark the nominative singular. A distinct homonymous morpheme -s (the same as that used for feminine nouns) is added to mark the oblique plural for this declension type. According to Chambon (2003: 349), the -s in the nominative singular is optional, the “flexion au pluriel peut être la seule flexion admise par les substantifs masculins” (Chambon 2003: 350). Some nouns, mainly masculine ones, ending in -a, especially loanwords from Greek such as propheta or evangelista (see also Section 8.3), are basically declined according to the first feminine declension type (see Jensen 1976: 79). In addition to these main types, some nouns ending in -s are invariable in case and number regardless of gender, as shown in Table 4.2. Gender Case Singular Plural Gloss Feminine Nominative votz votz ‘voice(s)’ Oblique votz votz ‘voice(s)’ Masculine Nominative temps temps ‘time(s)’ Oblique temps temps ‘time(s)’ Table 4.2: Invariable nouns in Old Occitan. Furthermore, some imparisyllabic nouns have a distinct stem in the nominative singular compared to the oblique singular and both nominative and oblique plural, as shown in Table 4.3. For a list of these imparisyllabic nouns, see Jensen (1975a, 1976: 39-42; 53-76, 1977). In general, it can be said that case marking was lost over time, which led to difficulties in interpretation. For example, distinctions such as the one between cor/ cors ‘heart’ and cor/ cors ‘body’ are no longer clear (see Jensen 1974a). This <?page no="62"?> 62 4 The Old Occitan Nominal System—an Overview Gender Case Singular Plural Gloss Masculine Nominative bar baron ‘the baron(s)’ Oblique baron barons ‘(of) the baron(s)’ Feminine Nominative sor serors ‘the sister(s)’ Oblique seror serors ‘(of) the sister(s)’ Table 4.3: Old Occitan imparisyllabic nouns. development happened even among imparisyllabic nouns, where distinct nominative forms were replaced by new forms built in analogy with the other noun classes, but rarely the other way around. The nominative singular stem expands its functionality to the other cases and numbers (see Jensen (1974b) in the case of coms ‘count’). Regarding the imparisyllabic nouns, Chambon (2003: 353) summarises: (i) que le morphème lexical, en l’occurrence baron, présente deux allomorphes en variation libre: / b’ar/ et / bar’on/ ; (ii) que chacun de ces deux allomorphes peut participer à la flexion plurative seulement, sans option casuelle […](iii) que chacun de ces allomorphes peut également participer optionnellement à la flexion en nominatif-singulier et en nominatif-pluriel (Chambon 2003: 353). Following this description of the noun declension system in Old Occitan, the adjectival declension will be presented briefly, as it works very similarly to the noun declension. The only addition is that, for feminine forms of adjectives, the gender-marking morpheme -a is added. Past participles, which are also susceptible to agreement controllers with respect to gender, case and number, function in a similar way to adjectives. An overview of the classic type of adjectival and participle declension is given in Table 4.4. Gender Case Singular Plural Gloss Feminine Nominative bona bonas ‘good’ Oblique bona bonas ‘good’ Masculine Nominative bons bon ‘good’ Oblique bon bons ‘good’ Feminine Nominative onrada onradas ‘honoured’ Oblique onrada onradas ‘honoured’ Masculine Nominative onratz onrat ‘honoured’ Oblique onrat onratz ‘honoured’ Table 4.4: Adjectival and participle declension in Old Occitan. <?page no="63"?> 4.2 Pronouns and Determiners 63 4.2 Pronouns and Determiners Some pronouns also reflect the gender of the nouns they replace or refer to. These will be presented below. Pronouns without gender agreement (such as personal pronouns in the first and second person, both singular and plural, reflexive pronouns, or relative pronouns) will not be addressed in more detail here. The personal pronouns of the third person are presented in Table 4.5. There are no atonic forms of personal pronouns in the subject function, as Old Occitan is a pro-drop language and this function is therefore already included in the verb ending. Type Function Singular Plural Emphasis feminine celei - masculine celui - Subject tonic feminine ela elas tonic masculine el il Direct Object atonic feminine la las atonic masculine lo los tonic feminine ela, lei elas, lor tonic masculine el, lui els, lor Indirect Object atonic both genders li - tonic feminine lei lor tonic masculine lui lor Table 4.5: Personal pronouns (third person) in Old Occitan, adapted from Paden (1998: 302). Type/ Gender Person Singular Plural Poss. M 1st mieus, mon, mos nos, nostre(-s) 2nd tieus, ton, tos vos, vostre(-s) 3rd sieus, son, sos lor, lur Poss. F 1st mia(-s), ma(-s), mea(-s) nostra(-s) 2nd toa(-s), tua(-s), ta(-s) vostra(-s) 3rd soa(-s), sua(-s), sa(-s) lor, lur Dem. F Proximal esta, cesta, aicesta estas, cestas, aicestas Distal cela, aicela celas, aicelas Dem. M Proximal est, cest, aicest estz, cestz, aicestz Distal cel, aicel cels, aicels Table 4.6: Possessive (poss.) and demonstrative (dem.) pronouns in Old Occitan. <?page no="64"?> 64 4 The Old Occitan Nominal System—an Overview Table 4.6 shows a simplified representation of the demonstrative and possessive pronouns in Old Occitan. For instance, the morphological marking of case is not listed explicitly, as it is parallel to the case marking previously shown for nouns and adjectives in Section 4.1. Also, not all graphical variants are included in the table. The aim is simply to demonstrate the different gender markings so that the examples shown later in Chapters 8 and 9 can be better interpreted. The demonstrative pronoun without reference to a specific noun (and therefore a specific gender), such as lo, o, and ço, is often described as a neuter pronoun (see Chabaneau 1875, 1876, 1878; Schultz-Gora 1938). However, this may be misleading, as it is in fact the default form of the pronoun (on the notion of default gender, see i.e., Corbett/ Fraser (2000)) and does not have the gender value of neuter. Type Gender Case Singular Plural Definite Masculine Nominative lo, le li Oblique lo los Feminine Nominative la, li las Oblique la las Indefinite Masculine Nominative uns, us u, un Oblique u, un uns, us Feminine Nominative una unas Oblique una unas Table 4.7: Old Occitan definite and indefinite articles. Table 4.7 shows the declension system of Old Occitan articles. The indefinite article functions according to the declension system of nouns and may show case marking in the masculine singular nominative, but not for the feminine gender. The situation is more complex regarding definite articles. Not all forms of the articles will be discussed in detail here, as most are unequivocal and their gender value is clear. The only difficult and ambiguous case is the article li. This appears twice in Table 4.7; once as a masculine plural nominative and once as a feminine singular nominative. Clearly, this article has nominative case value. However, the gender value appears to depend on the noun’s number. Therefore, in cases where li is the only agreement target for a certain occurrence, the gender value can only be determined if the noun’s number is clear. Otherwise these occurrences had to be counted as uncertain with respect to gender in Chapters 8 and 9. For a more detailed discussion of the gender value of li in Old Occitan, see Wiedner (2026); for a discussion of the gender and number value of the forms in -i in general, see Sibille (2009). <?page no="65"?> 4.2 Pronouns and Determiners 65 In addition to the bare articles, many contractions of prepositions with the masculine article exist in Old Occitan, but exclusivly with the masculine article; prepositions plus feminine articles do not contract into a portmanteau allomorph. The most common contractions are with de: del and dels; with a: al, als; with per: pel, pels; with vers: vel, vels; and with sus: sul, suls (see Anglade 1921: 212). In addition to these contracted forms with prepositions, the masculine definite article may also be attached to a preceding word ending in a vowel. This occurs with high frequency words such as si, que, e > sil, sils, quel, quels, el, and els (see Anglade 1921: 213). <?page no="67"?> 5 Research Questions As Chapters 3 and 4 have shown, there are several open questions concerning the discussion of grammatical gender in Romance languages (see also Marzo/ Wiedner 2025), especially in Old Occitan. Some of these questions will be explored in the following chapters, with Old Occitan serving as the object language. In general, it has been demonstrated that there is a lack of systematic studies concerning the variation and variability of gender in Old Occitan, both diatopically and micro-diachronically, i.e., with regard to Old Occitan from after 1250 onwards, a topic that will be addressed in this book. However, this is not the only contribution of the research presented here. Adopting a socio-historical, variationist perspective enables the research to go beyond mere descriptions of assignment rules, as presented in Section 3.1, and the different reasons why and how grammatical systems, such as gender systems, can change over time in a given language (see Section 3.5). Instead, this study takes a socio-historically informed approach, systematically incorporating the role of textual transmission and, where possible, the influence of individual scribes. Rather than merely noting the presence of gender variation, the study provides a more nuanced, multifactorial account of variation in medieval languages. By explicitly examining the temporal and spatial distribution of gender marking, the present work integrates sociolinguistic and dialectological perspectives—an approach that constitutes a central contribution and marks the study’s originality. It combines an analysis of the socio-historic context of textuality in Old Occitan with the grammatical changes in nominal morphology resulting from the processes of change from “synthetic” Latin to the more “analytic” Romance languages (e.g., loss of case morphology, retention of number morphology, reduction of declension classes, obligatory nominal determination, see Sections 3.2 and 3.3). Therefore, the investigation considers the impact of scribes’ individual practices and competences on gender assignment. This is achieved by systematically comparing different manuscript versions of the same text, as well as textual segments attributable to different scribes or authors. This allows for a more differentiated understanding of variation in medieval written language. The main research themes and questions derived from the above research gaps can be listed as follows and are all connected to each other. Therefore, these different research areas will not be addressed in isolation, but are listed here individually to provide context for the following chapters: <?page no="68"?> 68 Research Questions 1. The temporal and spatial distribution of gender doublets in Old Occitan, especially in the Languedoc and Provence linguistic areas. 2. The development of the Latin neuter in Old Occitan. 3. The influence of scribes and their personal profiles. With respect to point 1, the main questions concern the distribution of gender doublets within a single region (or even within a single manuscript), and between two regions (in this study, Languedoc and Provence), as well as the distribution with respect to the time of textual production (i.e., the 13 th and 14 th centuries). Loporcaro (2018) provides examples of nouns with divergent gender values in different language areas, such as the term for ‘sea’, which is masculine in Italian (il mare) and feminine in French (la mer). Both of these nouns are inherited from the Latin third declension noun maRe, which was predominantly neuter. Additionally, some examples of gender doublets with semantics that differ according to their respective gender value are given, such as the Italian masculine noun foglio ‘sheet of paper’ vs. the feminine noun foglia ‘leaf’, where the masculine noun is inherited from the Latin neuter singular folium and the feminine noun from the Latin plural folia (see Loporcaro 2018: 14; Marzo/ Wiedner 2025: 203; see also Section 3.5.1 in this book). Building upon this information regarding gender doublets in Romance languages, which differ from one language to another as well as semantic differences with respect to gender, Research Area 1 is concerned with the distribution of gender doublets without semantic differentiation according to time and region in Old Occitan. Are there tendencies of ambiguous gender usage in one area that are not present in another? Do preferences of a specific gender for an ambiguous noun correlate with a certain period of time? Alongside this first research focus is the second (Research Area 2), which concerns the development of formerly neuter nouns in Old Occitan, where the neuter gender no longer exists. As Chapter 8 will show, many of the gender doublets listed in the DOMél (2025) are in fact nouns derived from the third declension class of neuter Latin etyma. Rather than providing an overview of all the neuters inherited from Latin and assigned to one of the remaining genders in Old Occitan, this discussion focuses on those neuters that did not receive a clear gender assignment and are variable in gender (see Chapter 8). The goal of this research area is to gain a more detailed understanding of the nouns that remained gender variable, despite the fact that gender seems to have been fully grammaticalised by the 13 th century, so this kind of variation is not expected. Research Area 3 pertains to the influence of scribes on gender attribution in specific manuscripts. Scribes played a crucial role in the development and standardisation of non-standardised medieval languages (see Selig 2005: 263- <?page no="69"?> Research Questions 69 265), so they should be considered when investigating grammatical features in medieval languages. For example, the presence of a certain noun with different gender markings within a single text written by a single scribe raises interesting questions. Is the respective noun in Old Occitan gender ambiguous, triggering both genders? Or might it be the influence of another language (L1) the scribe knew? Or does the gender variation already exist in the original manuscript or a model? Not all of these questions can be answered in the following due to a lack of information on (most of) the scribes. To shed some light on the scribes’ influence, two case studies are presented in Chapter 9. One study (see Section 9.1) compares three manuscripts of the same text (La Vida de sant Honorat), written by three different scribes. The differences in gender assignment in these manuscripts suggest the influence of the scribes in these specific cases (for more information on the variation in gender of mar in these manuscripts, see Wiedner 2026). The second case study focuses on a single text within a single manuscript (the Chanson de la Croisade contre les Albigeois, see Section 9.2), which was composed by two authors but written by a single scribe. This text has no textual model and addresses one theme: the crusade against the Albigensians. It was composed in the same linguistic area and during the same time period, so some factors that might influence gender attribution are comparable between the two authors. In this research area, therefore, the focus is on two variables that may play an important role in gender assignment choices in this non-standardised language: the first is the influence of the scribe who wrote the respective manuscript and the second is the choice of the author who composed the text. The two case studies in Chapter 9 aim to provide further insight into the influence of these two variables. <?page no="71"?> 6 Constitution of the Corpus COMETA 6.1 Choice of Manuscripts Brunel (1935) lists a total of 357 Old Occitan manuscripts in his Bibliographie des manuscrits litteraires en ancien provençal. This number was recently augmented to 519 through an updated list by Menichetti (2025). As it was not feasible to manually post-correct the transcription of all manuscripts, it was impossible to include all of them in the COMETA (2025) corpus. The COMETA (2025) corpus comprises the following 17 manuscripts, which will be explained more thoroughly in the following sections: • BmC, Cote 34 (13 th century): Roman de Flamenca. This manuscript is written on 139 folios. The text comprises 8,096 verses, which contain a total of 46,928 tokens and 7,849 types. • BmT, Cote 2883 (14 th century): Las Leys d’Amors. This text is written on 154 folios. It comprises 144,920 tokens and 16,090 types. • BnF, Arsenal 6355 (14 th century): La Vida de Santa Enimia. This manuscript of 58 folios has 11,279 tokens and 2,773 types. • BnF, Français 1049 (1343): Trepas du roy Robert de Sicile, comte de Provence ; Libre de vicis et de vertutz, rédaction A ; Barlam et Josaphas. The three Occitan texts comprised in the 205 folios of this manuscript have 144,783 tokens and 11,574 types combined. The major part of these tokens are in part two (113,555 tokens); part one comprises 1,567 and part 3 a total of 29,661. • BnF, Français 2232 (1325): Roman de Philomena (manuscript P). The 134 folios of this manuscript comprise 37,829 tokens and 4,648 types. • BnF, Français 13503 (13 th / 14 th century): Li vida de la benaurada sancta Doucelina, mayre de las donnas de Robaut. This text of 103 folios contains 39,180 tokens and 5,311 types. • BnF, Français 13504 (14 th century): Vie de saint Elzéar ; Traduction d’une vision de Marguerite d’Oingt ; Vie de Sainte Delphine. In this manuscript, 36,844 tokens and 6,694 types are distributed over 61 folios in the three <?page no="72"?> 72 6 Constitution of the Corpus COMETA parts of this text. The first part contains 14,476 tokens, the second 3,695, and the third 18,673. • BnF, Français 13509 (14 th century): Vida de sant Honorat (manuscript R) ; Fragment de sermon, en occitan ; Début d’une prétendue Lettre de Jesus Crist. The 66,625 tokens and 7,825 types of this text are written on 114 folios. It contains three parts, but only the first one is included in this research, as the other ones are too fragmented to be of use for morphosyntactic analysis. • BnF, Français 25425 (1275): Chanson de la Croisade contre les Albigeois. This manuscript has 169 folios and comprises 98,521 tokens and 13,122 types. • BnF, Latin 901 (14 th century): Version occitane des Plaintes de la Vierge. This manuscript contains Occitan text from folio 126 to 129 v o with 1,137 tokens and 532 types. The other folios are written in Latin. • BnF, NAF 6195 (14 th century): La vida de sant Honorat (manuscrit M). This manuscript has 45,457 tokens and 6,100 types on 94 folios. • BnF, NAF 11151 (14 th century): Les sentences arbitrales entre l’abbé de Saint- Géraud et les consuls de cette ville (1280 et 1298) ; traduction occitane de la Mulomedicina ; la recepta del vi. This tripartite manuscript has in total 21,966 tokens and 3,477 types on 33 folios. Part one has 19,135 tokens, part two 2,204, and part three 627. • BnF, NAF 11180 (14 th century): Planh de la Vierge. This text, written on 43 folios, contains 4,636 tokens and 1,500 types. • BL, Additional 10323 (14 th century): La vida de sant Honorat (manuscrit O). This 71-folio-long manuscript has 57,119 tokens and 7,607 types. • BL, Additional 21218 (14 th century): Roman de Philomena (manuscrit B). The 36,584 tokens and 4,771 types of this manuscript are distributed over 63 folios. • BL, Harley 3041 (14 th century): Tenson entre Aicart et Girard. This manuscript contains only one folio in Occitan, comprising 501 tokens and 232 types. • BL, Harley 7403 (14 th century): Traduction occitane de l’Évangile de Nicodème ; traduction occitane (version B) du poème français sur les quinze signes de la fin du monde : Los XV signes que veno ; traduction occitane de la légende du bois de la Croix ; traité de diététique en occitan ; le repentir du pécheur ; doctrinal. The 133 folios of this manuscript contain in total 23,130 tokens <?page no="73"?> 6.1 Choice of Manuscripts 73 and 5,434 types, which are distributed over six parts. The tokens of this text are scattered in the six parts as follows: the first contains 5,142, the second 1,613, the third 3,661, the fourth 2,451, the fifth 7,180, and the sixth 3,083. The following choices were made to get the number of manuscripts down to a feasible amount. The first parameter for reducing the number of manuscripts was accessibility: only manuscripts were used that are available online in a digitised format (mostly from the Bibliothèque nationale de France (BnF), one manuscript from the Bibliothèque municipale de Carcassonne (BmC), and one from the Bibliothèque municipale de Toulouse (BmT)) or that are stored in either the BnF 1 or the British Library (BL). This reduced the number of manuscripts to about one third of the initial number but still left too high a number of manuscripts. The next parameter for reducing the number of manuscripts was the condition of preservation or digitisation of the manuscripts: all manuscripts where most pages are fragmented as well as all badly digitised manuscripts, where the quality was not good enough for automatic transcription of the texts, were categorically excluded. Fragmented manuscripts were excluded, so that the gender assignment within each sentence can be contextualised and causes no problems due to incomplete phrases. In this context, manuscripts with some pages missing in the beginning or at the end are not considered to be fragments. It was mainly important to have manuscripts where the existing pages are complete, to have the syntactic context of each noun. This still left a large number of manuscripts. The next parameter for excluding manuscripts was the region in which they were manufactured: manuscripts written outside of the Occitan speaking area, which were mainly written in the non-occitan-speaking regions of Italy, Catalonia, or northern France, were excluded. The reason for excluding these manuscripts was that there is an additional factor of possible and likely influences on the gender system due to contact constellations with other languages spoken in the respective areas. It is clear that multilingualism and language contact always play an important role in the context of scriptae and also the sociolinguistic context of the scribe and their reasons for writing a certain text in the vernacular (on the use of the vernacular versus Latin in the context of charters, see Selig 1997: 24). However, it is certainly a difference whether there are only some multilingual people and scribes with different mother tongues or if the entire surrounding speaks a different language than the language of the manuscripts. For further reduction, manuscripts from two different dialectal regions (from today’s perspective) were selected in order to compare potential differences in 1 The updated list of manuscripts stored in the BnF was taken from Camps (2010). <?page no="74"?> 74 6 Constitution of the Corpus COMETA gender assignment between two areas: only manuscripts were used that were most likely written in the areas where the Languedocian and the Provençal dialects of Occitan are spoken today. The choice for this region was made due to two main factors: First, these two regions show (or at least have shown) significant differences in gender assignment (see Section 3.3.4.3; Brun-Trigaud/ Sauzet/ Allassonnière-Tang 2023: 18). The second, more pragmatic and methodological reason, is the fact that a considerable number of manuscripts from these two regions were at my disposal and it was therefore possible to create a diatopically comparable and well-balanced corpus (see COMETA 2025). To further reduce the number of manuscripts and to keep the comparability as high as possible, manuscripts from similar time periods from both regions were selected. All the manuscripts used for this study were most likely written in the 13 th or 14 th century. This period was chosen because it was the time directly following the peak of Old Occitan literature, where the literary language employed was rather similar regardless of the region 2 . Only manuscripts from after this time period were chosen, which is described as “a second threshold when vernacular literacy starts to be fully institutionalised, that is, around 1250 in the oc and oïl territories” (Selig 2023: 356). Therefore, the texts used in this research all stem from this second threshold of Old Occitan literacy. In addition to time and area of origin, manuscripts were included that are comparable with respect to contents and text genres. Because the Provençal sub-corpus consisted of fewer manuscripts (only 9 of which fitted the criteria), the texts for the Languedocian sub-corpus were chosen in similar domains. The decision for the final number of manuscripts in the latter area was made while transcribing the manuscripts (see Section 7.1). To keep the two sub-corpora as comparable as possible, the goal was to have the number of tokens per subcorpus as balanced as possible. Therefore, the manuscripts were transcribed alternating between the two regions and the number of tokens was checked regularly. The final number of tokens is 412,210 tokens in the Languedociansub-corpus and 405,229 tokens in the Provençal-sub-corpus. The whole corpus, published under the name of Corpus de l’occitan médiéval comparatif et annoté : Provence et Languedoc (= COMETA 2025), contains a total of 817,439 tokens and 55,645 types. The manuscripts from the BL are taken from the Additional or the Harley collection and the manuscripts from the BnF are from the Français, the Latin or the Nouvelle Acquisition Française (NAF) collections. Regarding the number of all Old Occitan literary manuscripts, COMETA (2025) is constituted of almost 5% according to the list by Brunel (1935). According to 2 For the discussion of the status as Koiné see among others Bec (1963), Chambon (2012), Field (2006), Meliga (2009), or more recently Carles (2024). <?page no="75"?> 6.2 Historical Context of the Manuscripts 75 the updated and augmented list of Old Occitan manuscripts by Menichetti (2025) it still represents a bit more than 3%. When including only the manuscripts written in the respective period of time and regions analysed in this book, the percentage augments to 15% of Brunel (1935) and 12% of Menichetti (2025). With respect to the region, Brunel (1935) lists 73 manuscripts in the 13 th and 14 th centuries written in the Languedocian area (14% of which are included) and 39 in the Provençal, so almost 18% of the Provençal manuscripts of that time are included in COMETA (2025). Menichetti (2025) adds 12 and 14 manuscripts respectively, leaving COMETA (2025) with 12% of the Languedocian manuscripts of that period and 13% of the Provençal ones. 6.2 Historical Context of the Manuscripts For many of the manuscripts, it is unclear who authored and wrote the texts, or exactly where and when they were written. Only very little information is known with respect to some manuscripts, while others are disputed with respect to their origin. This section will therefore focus on those manuscripts for which at least some details are known, and only give a short statement on those with scarcer information. Not all, but a considerable amount, of the bibliographical information on existing secondary literature on the manuscripts has been assembled on the website ARLIMA (2025), which was used as a basis for the following. Additional sources were included to get an exhaustive view of the manuscripts and to complement the bibliographic sources indicated in ARLIMA (2025). 6.2.1 The Roman de Flamenca The Roman de Flamenca is one of the most famous Old Occitan texts. This can be shown by the sheer number of editions and translations. Its editorial history started with a first publication of some passages by Raynouard (1838) and a first literary-historical classification in the same year (see Duval 1838), only 4 years after the discovery of the manuscript in the Municipal Library of Carcassone by the prefect of the Département in 1834 (see Chambers 1966-1967: 489). About thirty years later, the first complete edition by Meyer (1865a) and afterward a second, updated edition by the same editor (Meyer 1901) followed. A considerable number of re-editions followed in the second half of the 20 th century with the editions by Lavaud/ Nelli (1960), Hubert (1962), Limentani (1965), Ryan (1974), Gschwind (1976), Huchet (1989), and Blodgett (1995), followed by a few re-editions in the 21 st century, namely the editions by Manetti (2008) and Zufferey/ Fasseur (2014) as well as the online corpus of the text (Scrivner et al. 2013), that is based on the edition by Meyer (1901). <?page no="76"?> 76 6 Constitution of the Corpus COMETA Additionally, there are translations into Catalan (see Espadaler 2015), English (see Cooke 1955; Scrivner et al. 2013), French (see Meyer 1901; Zufferey/ Fasseur 2014), German (see Kirsch 1989), Italian (see Cocito 1971; Manetti 2008), and Spanish (see Rossell 2009; Covarsí Carbonero 2010). This high number of editions and translations proves the high interest in this text, especially in the 20 th and 21 st centuries. Given this fact, it is not surprising to see a very high number of studies dedicated to this text, trying to get a grasp on its author or scribe. First of all, the question of time of origin has to be addressed. Raynouard (1838: 80), the first researcher to my knowledge to deal with this question, estimates the time of origin of the only manuscript to be between the end of the 13 th and the beginning of the 14 th century, based on paleography. Regarding the time of composition of the original text, he proposes a time in the 13 th century before 1264, because the feast of Corpus Christi, which was instituted by Pope Urban IV in that year, is not mentioned at all in the story. All other Christian holidays are mentioned (see Raynouard 1838: 128-129). He later contradicts this statement as he elaborates that the story itself is based even in the middle or end of the previous century (see Raynouard 1838: 129-132), which might also be the reason for the author not to mention the feast of Corpus Christi. Meyer (1865a: XXI) proposes the composition of the text between 1220 and 1250, which is accepted by Révillout (1875: 5), who tries to determine a precise date for its composition. Based on the liturgical dates and their contextualisation, he concludes the year 1234 as the year in which the story takes place, and therefore proposes the time for the composition of the story around that year (see Révillout 1875: 18). Chambon (2018b) proposes the year 1223 as the date of composition. As for the place of origin of both the text and the author, Bartsch (1866: 191- 192) proposes an origin in the southwestern part of the Old Occitan linguistic area. Siebert (1886: 4-5) contradicts this hypothesis and instead proposes an origin in the northern part of the Old Occitan speaking parts, close to the French speaking areas. In a more systematical analysis of the lexicon in the Roman de Flamenca, Jud (1939: 207) favors the Rouergue as the origin of Flamenca’s author due to a number of regionalisms employed in the text. This origin is supported by other scholars, who analysed regional traits in the language (see Chambon 1992: 165; Zufferey 2016: 298-299). Chambon (2016: 76-77) analyses seven additional lexicalisms that he places in the area of the Basse-Auvergne or the Bourbon. He concludes that the author must have had a connection to someone from that area or that he traveled there frequently (as suggested by Zufferey/ Fasseur 2014: 106). Another interpretation (or an additional reason for these lexical traits) could be the addressed audience from the northern parts of the Auvergne or even north of that (see Chambon 2016: 77). Chambon (2015: 269-270, 2018a: 19) proposes the hypothesis that Daude de Pradas might be the author of the Roman de Flamenca, <?page no="77"?> 6.2 Historical Context of the Manuscripts 77 rather than someone anonymous. Daude de Pradas was a troubadour from the Rouergue, which corresponds well with the linguistic placement of the text. In conclusion to the region and the author of the Roman de Flamenca, it can be said that “it is beyond doubt that the creator of this narrative masterpiece must have been perfectly acquainted both with courtly romance written in langue d’oïl and with the classical tradition, most notably so Ovid” (see Dufter 2025: 209; italics in the original). The last point in the presentation of the Roman de Flamenca will deal with the scribe of the only manuscript, even though there is not much information on them. As mentioned before, the time of production of the only preserved manuscript is dated between the end of the 13 th and the beginning of the 14 th century, based on paleography (see Raynouard 1838: 80). Brunel (1935: 26) and Zufferey/ Fasseur (2014: 110) propose a Provençal origin for the scribe. The quality of the scribe’s work is criticised for rendering the text more complicated and therefore more difficult to understand and interpret: F[lamenca; M.W.] was probably not an easy text when it came fresh from the hands of its author; but the scribe of the only MS extant left it a very difficult text indeed, full of garbled words and lines that will be the despair of editors for years to come. He also omitted verses here and there, as is evident from the rime. Both beginning and ending, I repeat, are lost; in addition, folios are missing in several other places. Thus there is much room for speculation and subjective conjectures, and no editor could hope to satisfy all experts (Chambers 1966-1967: 492). There is no more information on the scribe and the missing of some folios is probably not the copyist’s fault, but rather an effect of the manuscript’s history over the last centuries up until its discovery in 1834. In conclusion, it seems like the text originates from the first half of the 13 th century, while the manuscript was transcribed 50 to 100 years later. The language of both the author (at least to a certain extent) and the copyist seems to be from the Provençal linguistic area, which is why this text is treated as part of the Provençal sub-corpus. 6.2.2 The Chanson de la Croisade contre les Albigeois “Il serait, je pense, inutile de multiplier les observations et les conjectures sur un texte qui a si peu de chances de devenir un sujet de curiosité et d’études” (Fauriel 1837: 658). Contrary to this statement by the first editor of the Chanson de la Croisade contre les Albigeois, this text looks back at a long history of editions and translations. The first edition was the one by Fauriel (1837) cited previously, including a French translation, followed only a year later by Raynouard (1836-1845: volume 1, 225-289), who published some extracts in the <?page no="78"?> 78 6 Constitution of the Corpus COMETA first volume of his Lexique roman 3 . Meyer (1865b) edited and translated the text about thirty years later. In the 20 th century, Anglade (1913) published those extracts of the manuscript that deal with the Battle of Muret on 12 th of September 1213. An edition of the whole manuscript in three volumes by Martin-Chabot (1931, 1957, 1961) followed. Another new edition was published in the end of the 20 th century by Hoekstra (1998), and some extracts were published in 2017 (see Douillet 2017). In addition to the numerous editions and French translations, an English translation (see Shirley 1996), a Catalan translation (see Martines/ Ensenyat 2003), as well as some extracts in contemporary Occitan (see Camprós 1972) were published. As stated above, the entire text of the Albigensian Crusade is currently preserved in one single medieval manuscript (see BnF, Français 25425 1275). There are two further fragments, one of which (the manuscript listed as R) was used by Raynouard (1836-1845) for a publication in his Lexique Roman, and is lost today. The other fragmented version is preserved in the Bibliothèque municipale de Grenoble (shelf-number 1158). Additionally, there are two prose versions: the first one is preserved in two manuscripts (BnF, français 4975 and BmT 608), and the second one in a single manuscript preserved in the Bibliothèque du château in Merville (see Camps 2010: XLI; Raguin 2017). The only complete manuscript that is included in the COMETA (2025) corpus dates back to 1275 based on paleography (or roughly around that year, see d’Heur 1973: 443) and is written by one single scribe (see Lafont 1991: 238). The story was probably written earlier, during the Albigensian Crusade, and the author probably began in 1210 with the composition of the text and went on until 1213 (see Lafont 1962: 45; 52-53). d’Heur (1974: 258) proposes a more dense, continued composition between 1212 and 1213, but it seems clear that the first part was composed during the Crusade. Due to different linguistic and content arguments, it is now considered undisputed that this chanson was composed by two different authors (see Bampa 2017: 93-94; see also the following discussion). According to Meyer (1875: LX), the second part of the poem was composed between the 25 th of June 1218 and the 1 st of August 1219. This suggestion is rejected by Heisig (1926: 3), who proposes a genesis of the second part of the poem later than 1228, because events that happened after 1228 are mentioned in the manuscript. There is a somewhat diverse and polemic discussion of the authorship of the Crusade throughout the last two centuries. This discussion will be presented in detail, because it is quite disputed among scholars whether the text was writ- 3 This publication of an extract is especially interesting, as it is based on a manuscript that is now lost (see d’Heur 1974: 260). <?page no="79"?> 6.2 Historical Context of the Manuscripts 79 ten by one or two different authors, as the following will show. This plays a crucial role for the analysis in Section 9.2. Among the information that can be retrieved directly from the manuscript, the author of the first part of the text names himself right at the beginning as Guilhem and states that he grew up in Tudela, Navarra (see Example 6a). Higounet (1938) proposes a strong relationship to others from Tudela that came to Toulouse, maybe even a blood relationship. This suggests that he might have been in contact with the language of Toulouse from early childhood onward. He furthermore names the place, Mont alba (modern-day Montauban), where he wrote it (see Example 6b). (6) a. El nom del payre e del filh e del sant esperit. Comensa la cansos que maestre .Guilhem. fit Us clercs que en navarra fo a tudela noirit ‘In the name of the father, and the son, and the holy spirit. Begins the song that master Guilhem has made. A cleric who was in Navarra nourished in Tudela’ (BnF, Français 25425 (1275: f o . 1, l. 1-3)) b. Maestre .Guilhem. la fist a mont alba ‘Master Guilhem made it in Montauban’ (BnF, Français 25425 (1275: f o . 6, l. 17)) Further information was given in the now lost manuscript which Raynouard (1836-1845) had at his disposal. In the prologue of this fragment, it is stated that, after 11 years at Montauban, he went to Bruniquel to Count Baudouin and was installed as canon in Saint-Antonin (see Bampa 2017: 91; Raguin 2016: 390-392). For the first two editors, the question of authorship is solved with this information: it is Guilhem de Tudela (see Moncaut 1869: 402-403). There are certain differences in the style, the content and the political position of the first part and the second part of the text, which Fauriel (1837) explains through a change in the author’s perspective on the ideas of the Crusade and with his ideological change, he also changed his style 4 . It was a few decades later that Guibal (1863: 4 The linguistic behavior of the people often serves as delimitation between different ideologies, not only within the text such as the differentiation between Bretons and Allemands (for the dif- <?page no="80"?> 80 6 Constitution of the Corpus COMETA 603-606) proposed the idea of two different authors for this text; the first part still written by Guilhem de Tudela, and the second, larger part by another author who remains anonymous. But he remained rather cautious with this hypothesis. In general, his arguments regard differences in the style, the language, and most of all differences in the political positioning between the two parts. Only a few years later, Meyer (1865b) proposed the following statement on the authorship which is basically the idea that is maintained until today: Guillaume de Tudela, rimeur de profession, établi d’abord à Montauban, puis à Bruniquel, protégé par Baudouin, le frère du comte de Toulouse, entreprend d’écrire en forme de chanson l’histoire de la croisade albigeoise. Il se met à l’oeuvre en 1210, et raconte avec une facilité commune, un talent dont aucune conviction profonde ne vient relever la médiocrité, les faits de la guerre depuis le meurtre de Pierre de Castelnau jusqu’au moment où le roi d’Aragon entre en ligne. Alors, en 1213, il s’arrête et attend les événements. Mais, peu de temps après la bataille de Muret, le comte Baudouin est mis à mort. Dès lors Guillaume disparaît de la scène, laissant son poëme inachevé. En 1218 un Toulousain, ayant pour protecteur l’un des plus vaillants champions du Midi, Roger Bernart, le fils du comte de Foix, enflammé par la victoire que Toulouse vient de remporter sur la croisade, reprend avec les sentiments du patriotisme le plus ardent l’œuvre interrompue de Guillaume. Il la poursuit jusqu’au second siège de Toulouse et s’arrête à son tour, au moment où il allait avoir un nouveau triomphe à raconter, sans que rien annonce la fin de la chanson ni l’intention de la continuer (Meyer 1865b: 422). Directly responding to the results of Meyer’s article, Moncaut (1869: 40) wrote a letter contradicting this statement and going back to the thesis of a single authorship by Guilhem de Tudela. Kraack (1884) went a totally different way in his analysis. He states that Guilhem de Tudela was only the translator (“Umsetzer” Kraack 1884: 21) and proposes an original from the oïl speaking region, with Pons de Mela as the author (see Kraack 1884: 34). Newer research seems to accept the division into two parts written by two different authors, and acknowledges Guilhem de Tudela as the author of the first part (see d’Heur 1974: 241-245), and his authorship is not disputed afterward. With respect to the language employed in the first part of the Albigensian Crusade, the part written by Guilhem de Tudela, Meyer (1865b: 411) proposes a mixed language between Occitan and French. Later, he argues that Guilhem tried to write the Crusade in a language that he did not know well, and that there ferences in language for different ideologies, see Raguin 2014), but also the linguistic structure of the text itself serves to support the political ideology. Therefore, Raguin (2014: 414) states (without specification of any traits) that the first part, which is a more descriptive and conciliating position, also shows a stronger influence of northern Gallo-Romance varieties, while the second part shows a more southern linguistic structure, which underlines the political statement (see Raguin 2014: 427-429). <?page no="81"?> 6.2 Historical Context of the Manuscripts 81 are many mistakes in that part. He argues that Guilhem’s native language was closer to Catalan and Spanish rather than Occitan (see Meyer 1875: II). Martin- Chabot (1931: XVII) asserts that the native language of Guilhem (and therefore also the language spoken in Tudela) was closely related to Aragonese, but that Guilhem wrote in a mixed language between French and Occitan. Contradicting this statement of Guilhem writing in a language other than his mother tongue, Pfister (1970: 193) claims that he most likely wrote in his native language, a mixed language similar to that spoken in other cities south of the Pyrenees. In 12 th and 13 th century Navarra, many immigrants from the Occitan and French speaking areas came to live there, which created an intense language contact scenario, resulting in an Occitan variety strongly influenced by French, Spanish and Catalan. This is the language that Pfister (1970: 193) assumes as Guilhem’s mother tongue. Lafont (1991: 239) supposes that Guilhem wrote most likely in a “langue littéraire qui n’est sans doute pas totalement protégée des variantes dialectales de la communauté «franque» de Tudèle, mais qui est surtout la langue scriptique”. He goes on to argue that the local linguistic traits of the Pays de Foix are solely due to the scribe’s influence and not to a language spoken or written by Guilhem (see Lafont 1991: 239). Santano Moreno (2012: 582) goes back to the language mix, arguing that the original text was closer to the oïl varieties, which is supported by his familiarity with the chansons de geste of that linguistic area on the one hand and through contact with speakers from the north due to the crusades on the other hand. He asserts that the language in the first part was largely adapted and “ha debido sufrir un proceso de eliminación de elementos de oïl por parte de los copistas, sustituidos por las formas provenzales correspondientes” (Santano Moreno 2012: 582). He compares the language of Guilhem with the language of a legal document from Tudela, dated about 20 years after the composition of the first part of the Albigensian Crusade. This shows that many linguistic features attributable to Guilhem correspond very well to the language employed in the document (see Santano Moreno 2012: 585). Parallel to the discussion of whether there was really a second author or not, the question of who might have been the author of the second part began to interest philologists (for a more detailed summary of the different positions and propositions, see Zambon 2016). Diehl (1885: 5-7) proposes it was Guilhem Anelier, the author of the Histoire de la Guerre de Navarre, a text written after 1277. This would also suggest that the second part of the Albigensian Crusade would not have been a direct continuation of the first part but an addition composed 50 to 60 years after the abrupt ending of the first part. Fabre (1907: 264) proposes the famous Troubadour Peire Cardenal as the anonymous author of the second part, based on stylistic and ideological similarities. <?page no="82"?> 82 6 Constitution of the Corpus COMETA This hypothesis is directly rejected by Vossler (1916: 55) based on differences in the lexicon used in the Crusade, which differs significantly from that of Peire Cardenal’s Sirventes. Another proposition is the authorship by one of the people mentioned in the text as maestre Bernat, who Ormières (1972) identifies as Bernardus de Ulmo and who he identifies as the author, based on his political ideas and because of the same profession (notary) that both Guilhem de Tudela and this Bernat practiced. But, as Zambon (2016: 269) argues, neither the political ideas nor the same profession are valid arguments, thus rendering Ormières’ proposition only hypothetical. The next, and first well-grounded proposition regarding the authorship of the second part was proposed by Rostaing (1970). He argues for the authorship by the troubadour and nobleman Gui de Cavalhon (or de Cavaillon), originally from Provence, mainly based on external factors. His idea was taken and strengthened by an article three decades later (see Guida 2003), where the discussion of Gui de Cavalhon as the author is also enriched with philological arguments, such as a comparison of the second part of the Crusade with poems that are with certainty attributable to this troubadour. Zambon (2016: 270-271), while acknowledging this proposition as the first well-grounded and plausible one, gives two arguments against this idea: the first one is with respect to the language, which has a rather regional touch from the area of Toulouse and the County of Foix, which he deems unlikely for an author of Provençal descent. The second argument is with respect to the noun paratge: this noun, mainly used in juridical texts, appears quite often (seven times within 10 verses) in the direct speech of Gui within the text of the Crusade and it appears once in a partimen written by Gui that addresses the same themes (see Guida 2003: 277-278). But, as Zambon (2016: 271) shows, this noun appears only once in that partimen—and this is the only occurrence of this noun in the works of Gui—and with different semantics. He therefore sees it as an argument against this authorship: come si potrebbe spiegare che lo stesso autore, riportando due volte un dialogo con uno stesso personaggio e sullo stesso argomento, non abbia fatto in uno dei due casi la minima allusione alla nozione che nell’altro - come del resto in tutto il poema - appare come la chiave di volta di tutto il discorso? (Zambon 2016: 271). Zambon (2016: 272) uses this noun (paratge) as an argument for another author with the name of Guilhem Anelier, already proposed by Diehl (1885), as he is the only one who also uses this noun frequently (four times) in his Sirventes. Contrary to Diehl’s idea of a genesis of the text in the 1270s, he argues that there are two different Guilhem Aneliers, one who wrote the Histoire de la Guerre de Navarre in the end of the 1270s and another one who wrote the second part of the text here at discussion as well as his Sirventes within the first three <?page no="83"?> 6.2 Historical Context of the Manuscripts 83 decades of the 13 th century. A detailed comparison of the four Sirventes and the extracts from the second part of the Albigensian Crusade do indeed demonstrate undeniable similarities (see Zambon 2016: 273-277). He argues in the end that his hypothesis better correlates with the language employed in the second part, a regional variety from the area of Toulouse. This corresponds to the area in which Guilhem Aneliers grew up, thus making him more likely the author than Gui de Cavalhon (see Zambon 2016: 279-280). Bampa (2017: 93) uses the line “nostr’evesque”, which is used in reference to Folquet de Marseille, as proof of the second author’s profession as a cleric, who was therefore active in the dioceses of Toulouse. When accepting the theory of a twofold authorship, the next question regards the point of transition between the two authors. It is generally assumed that the transition between the two authors is situated somewhere between the 130 th and the 132 nd laisse. Meyer (1865b) proposes the last verse of the 131 st laisse as the last verse of Guilhem’s part, due to differences in the rhyme scheme between the two parts, which, according to him, changes right at that point. He additionally sees a change in political view regarding the Crusade manifested in this laisse (see Meyer 1865b: 410). This statement, mainly regarding linguistic features, has been recently supported in a study by Raguin (2015: 96-100). Another hypothesis is proposed by d’Heur (1974), who sees the 130 th laisse as the point of transition between the two authors. Bampa (2017: 97-98) supports this statement with both linguistic and content-related arguments, while simultaneously rejecting the arguments previously proposed by Meyer (1865b). Bampa (2017) argues that this last verse, ending with “the end of the war” (see Example 7) is also the end of the part written by Guilhem. The name of a potential scribe is written at the end of the next laisse, which is taken as an argument for the change of authorship at this point (see Meyer 1875: XXXIX). However, Bampa (2017: 98-100) specifies that the change might have also been located a bit later during the copying processes from the first manuscript to the preserved one which is included in COMETA (2025). It seems most plausible to follow this last theory for the point of transition, which will therefore be used for the remainder of this study (i.e., for Section 9.2). The following Example (7) displays the last two verses of Guilhem’s text and the first two of his continuator. <?page no="84"?> 84 6 Constitution of the Corpus COMETA (7) Aitant cant la materia. ad enant durara. Tro la guerra er finea. Ans que ia guerra parca. ni sia afinea. I aura mot colp fait. e mota asta brizea ‘As much as the matter it went on before until the war was ended. Before the war had already been contained nor will have ended. There will be many blows made and many spears broken’ (BnF, Français 25425 (1275: f o . 70, l. 6-9)) With respect to the language of the second part of the Crusade, Meyer (1865b: 413) argues that a more pure and correct Occitan is employed than the one used in the first part. This statement is unanimously accepted in the literature. Regarding the style of the second part, the author oriented himself very strongly on the model of Guilhem, thus rendering it a real continuation and not another work (see Bampa 2017: 100-103). 6.2.3 The Vida de Sant Honorat The Vida de Sant Honorat has a long and rather difficult history of edition, as the following will show: The first edition of this text by Sardou (1875) is based only on three of the nine manuscripts (namely the manuscripts G, R, and B 5 ) and leaves out the Latin versions, which is the main point of critique about this first edition (Meyer 1876: 237, 1879: 508). It took over 30 years from Meyer’s critique for a first Latin edition to be published by Munke (1911). Again a decade later, the first part of the Old Occitan version was edited with all nine manuscripts as a basis, but only the description of the manuscripts was published due to the editor’s death a few months after submitting this edition as his thesis (see de Roustan 1924-1925: 255). The edition of the first two books of the Old Occitan version including all manuscripts was elaborated in the 1930s and the beginning of the 1940s and published by Suwe (1943). It took until 2007 for the first (and to this day only) edition of the whole text with all manuscripts to be published (see the edition by Ricketts/ Hershon 2007). 5 The sigla for the manuscripts are taken from de Roustan (1924-1925), Suwe (1943), and Ricketts/ Hershon (2007). <?page no="85"?> 6.2 Historical Context of the Manuscripts 85 Currently, as mentioned above, nine manuscripts of the Old Occitan hagiography La Vida de Sant Honorat are preserved, six of which are stored in the BnF, and the others in the BL, the Bibliothèque municipale de Tours, and the Bibliothèque Méjanes in Aix (see Camps 2010: LIV). This text was written by the Prior of the Abbey of Roquestéron, Raimond Feraud, around 1300. Before that, he was a monk at the monastery on the island of Lérins. The monastery was founded in the fifth century by Saint Honorat, the later bishop of Arles (see Busby 1997: 51), hence the author’s interest in this saint (see Grange 2015: 201-203). Information on the author other than his belonging to the monastery and abbey is scarce, but he was most likely born in Nice around 1245 and died around 1324 or 1325. He probably had an affiliation with the minor nobility of the County of Nice; he was most likely the son of Guillaume Faraud III (see Grange 2015: 203; Ricketts/ Hershon 2007: 119). Other information about his life, such as his Naples expedition with Charles I of Anjou or that he was conferred the position of Prior at the Abbey of Roquestéron by Maria of Hungary, are not confirmed and should therefore be treated with caution (see Grange 2015: 203). The linguistic features of the hagiographic text traced back to the author himself show regional markings of the County of Nice, which corroborates the author’s origin in that area (see Ricketts/ Hershon 2007: 120-138). He based himself on troubadour poems as well as on Old French Chansons de gestes, which is why the Vida is described as a “hagiographical text at the confluence of epic, lyric, and romance” (Busby 1997: 64), partially writing the text in complex rhyme schemes close to the troubadours’ tradition. In other parts, he also uses alexandrines, which are rare in hagiographies, but quite commonly used in epics (see Suwe 1943: CXL; Busby 1997: 52-53). The author clearly knew the most important literary texts of that time, such as the works of Chrétien de Troyes, which certainly influenced his style (see Busby 1997: 53). In the mid-19 th century, Paris (1865: 88) accused the author of introducing elements from other tales: “Tous les récits qui le font intervenir, dans le poëme de Ramon Feraud, sont donc du fait de cet auteur, et il n’a pu les intercaler que par un incroyable anachronisme”. Before going into the genealogy of the different Old Occitan manuscripts, the question of the relation between the Latin and the Old Occitan versions of the life of Saint Honorat will be discussed. Hosch (1877) was the first to address this question, and he came to the conclusion that there was definitively a Latin model at Feraud’s disposal, but that model is not known or transmitted. According to him, the Old Occitan version was the model for another Latin version that was published later on and is known until today in the form of two prints from 1501 and 1511 (see Hosch 1877: 60-61). Meyer (1879: 507) on the other hand argues that both the Old Occitan and the Latin print from the 16 th century go back to a joint Latin model, but they are <?page no="86"?> 86 6 Constitution of the Corpus COMETA independent from each other. But both agree that there is indeed a Latin model that Raimond Feraud must have known when composing and translating his version of the Vida de Sant Honorat (see also Schäfer 1911). It was only later that a Latin version was found and edited (see Cavallin 1952), which was most likely the basis for Feraud’s version of the Vida. Many of the deviations and fictional elements are already in this Latin text from the mid- 13 th century, which thus contradicts the statement by Paris (1865) attributing these elements to Feraud (see Grange 2015: 206). Some differences from the Latin texts and some modifications Feraud introduced into the text can be explained through the historical context of the genesis: “Feraut updated his Latin source to link the heretics expelled by St Honorat to contemporary Cathars” (Grange 2015: 217-218). In general, Feraud remained very faithful to the Latin story and its language, and he only added some parts (see Grange 2015: 204-205). The following section will deal with the genealogy of the Old Occitan versions, with a focus on the three manuscripts included in COMETA (2025). The choice was made to include three manuscripts to be able to compare, at least to some extent, the influence of the respective scribes, according to Research Area 3. These manuscripts were therefore used for the case study in Section 9.1. These three manuscripts are BL, Additional 10323 (14 th century), manuscript O; BnF, Français 13509 (14 th century), manuscript R; and BnF, NAF 6195 (14 th century), manuscript M. The following Figure 6.1 shows the relation between the different Old Occitan manuscripts of the saint’s life as proposed by Ricketts/ Hershon (2007: 46); the manuscripts in circles are the ones included in COMETA (2025). The scribe of manuscript R names himself at the end of the manuscript 6 as Reforsat d’Olieras. He was probably from the département Alpes-Maritimes, near Roquestéron, so the area where Raimond Feraud wrote the text. A short linguistic analysis confirms the origin of this manuscript directly in the Provence and in the 14 th century (see Ricketts/ Hershon 2007: 12-15). The main traits that Ricketts/ Hershon (2007) give as reason for this localisation in the Provence are the graphical realisation <ou> for the close-mid vowel [o], the diphthongisation of the mid-open vowel [ O ], realised as <oa>, insertion of <p> after <m>, and the consistent retention of the prosthetic <e> before s-impurum. Manuscript M is an incomplete manuscript; the beginning and the end are missing, as are some verses within the text. There are eight additional verses that are not part of any other manuscript (see Ricketts/ Hershon 2007: 25-26). The text is one of only two manuscripts of this Vida that is written on paper, and it originates most likely in the middle of the 14 th century in Provence, though there is no precise information on the scribe (see Ricketts/ Hershon 2007: 25-28). 6 This part is not readable in the digitised version of the manuscript, so it is not possible to insert a caption of the respective part. <?page no="87"?> 6.2 Historical Context of the Manuscripts 87 Y Z T M O 2 C X B R G O 1 K A D Figure 6.1: Genealogy of the manuscripts of the Vida de Sant Honorat, adapted from Ricketts/ Hershon (2007: 46; circle markings are my own). The third manuscript in the COMETA (2025) corpus, manuscript O, was also written in the middle of the 14 th century in the region of Provence. It is more complicated to sort into the genealogy of the manuscripts, as it contains a remarkable number of rubrics that do not exist in any other of the manuscripts (see Ricketts/ Hershon 2007: 28-29). Furthermore, the scribe seems to have had at least two different manuscripts at his disposal when writing the text. The first part of the manuscript (O 1 ) from the beginning to verse 900, is closely related to manuscripts G and R. The second part (O 2 ) follows manuscripts C, T, and M, while maintaining its own character yet sometimes also an orientation on manuscripts G and R (see Ricketts/ Hershon 2007: 44-46). It is not possible to fully attribute this manuscript to one or another tradition of the manuscripts. It is positioned somewhere in between the other two manuscripts of the COMETA (2025) corpus. This fact makes it quite interesting to analyse, but also makes it harder to be certain about influences of the scribe in the attribution of gender. For possible scribal influences, the first part will therefore be considered comparable with manuscript R, and the second part with manuscript M within COMETA (2025), in order to be able to make a statement about the influence of the scribe on gender attribution in the manuscripts. <?page no="88"?> 88 6 Constitution of the Corpus COMETA 6.2.4 The Vida de Sant Enimia The first edition of the entire manuscript 7 by C. Sachs (1857) gives only very little context on the author. It is just specified that he was probably neither the descendant of the viscounts of Marseille as previously suggested by de Nostredame (1575: 189-191) nor that he was Bertrand Carbonel from Marseille as suggested by Crescimbeni (1731, cited after C. Sachs 1857: 1). However, he does not specify any more on the author (see C. Sachs 1857: 1). The Old Occitan version of the Vida de Sant Enimia is transmitted in one single manuscript, the manuscript BnF, Arsenal 6355 (14 th century). The author, or rather translator, of this text reveals himself in the poem on the first folio, where he writes: (8) Trais aquest romans de lati Per rima si com es aysi Maistre bertrans de masselha Ab gran trebalha et ab velha ‘Master Bertrand de Masselha translated this novel from Latin in rhymes as it is here with great work and care’ (BnF, Arsenal 6355 (14 th century: f o . 1 r o , l. 9-12)) From this passage, we can retrieve two pieces of information: First, it is a translation from Latin. Secondly, the name of the translator is Bertran de Masselha. Regarding the first bit, at least two manuscripts of the Latin version from the 12 th century do exist (see Meyer 1898c: 82). In some parts, Bertran’s translation remains close to the Latin original, while he decidedly changes other parts, adding additional miracles or more details on the landscape and more precise information on the area (see Brunel 1916: IV-VII). In the passage seen in Example (8), the translator refers to himself as maistre, which is the reason for Meyer (1898c: 81) to assume that he was a cleric of some 7 It is only according to the author himself the first entire edition of this manuscript, when in reality Bartsch (1856) published an entire edition just one year before. But he does not give any information on the background except the label of the manuscript in the Arsenal library in Paris. <?page no="89"?> 6.2 Historical Context of the Manuscripts 89 sort. Brunel (1916: V) agrees with that, and supposes additionally that he graduated from some university and became a cleric afterward. There was a monk with the same name in the second half of the 13 th century who installed the Pénitence de Sainte-Marie-Madeleine, but it remains unclear whether this was the same person or not. We therefore know the name of the translator who specifies in the manuscript that the Vida is a translation ordered by an unspecified convent (see Example (9)). It was probably ordered by the convent of Sainte-Enimie, a place in the département Lozère, the historical area of Gévaudan close to the river Tarn (see Meyer 1898c: 81; Brunel 1916: 1). Based on the precise description of the landscape in the environment of this convent, which is by far more detailed than the descriptions in the Latin original, Constans (1879: 211, 217) supposes that the translator himself was a monk in this convent, thus having detailed knowledge of the area, and that his surname de Masselha is a reference to his place of birth rather than to the convent he lived in. (9) Ans fo pregatz caramen Daus part lo prior el coven ‘So I was kindly asked by the prior and the convent’ (BnF, Arsenal 6355 (14 th century: f o . 1 v o , l. 7-8)) The translation probably originates from the 13 th century, and the preserved manuscript was probably written in the first half of the 14 th century by a single scribe (see Brunel 1916: XII). We do not know a lot about the translator and, in fact, nothing about the scribe. We only know that the translation was made in the 13 th century and the copy in the first half of the 14 th century. Both are probably from the historical area of Gévaudan (Brunel 1935: 41; Camps 2010: 72). With respect to the language of the manuscript, C. Sachs (1857: 1) states that the forms often appear as if it were Catalan, without giving any examples or reasons to support this statement. He supplements that Bertrand probably did not write this text in Catalan but that the variety of the region of Gévaudan has some traits in common with Catalan, thus making it look like Catalan in some cases. Brunel (1916: VIII-IX) supports this statement by saying that the linguistic traits of this text are clearly from the area of Gévaudan from the beginning of the 13 th century, citing the elision of the n instable and the word-final [s] as examples. He goes on to argue that some traits in the manuscript (e.g., the diphtongations <?page no="90"?> 90 6 Constitution of the Corpus COMETA of <e>) can only be traced back to the scribe of the manuscript, suggesting that it was transcribed in the southern part of the département Lozère, maybe even in the convent of Sainte-Enimie itself (see Brunel 1916: IX-XI). 6.2.5 The Vida de Santa Doucelina The Vida de Santa Doucelina is preserved in only a single manuscript: BnF, Français 13503 (13 th / 14 th century) (see Camps 2010: LII). In total, there are only two editions of the whole text 8 , one by Albanés (1879) and the other one by Gout (1927), who both also translated it into French, and there is also one English translation (see Garay/ Jeay 2001). The text describes the life of Saint Doucelina, the founder of the Beguines of Marseilles, who was born in 1214 or 1215 in Digne and died on the 1 st of September 1274 (see Albanés 1879: XL; Sancho Fibla 2018: 296). The text was composed only a few years after the death of Doucelina; Albanés (1879: XXIII) proposes the years 1286, 1297 or 1315 as possible years of composition, based on rather weak arguments that are rejected by Fibla (2022: 47-48), who proposes a composition of the text between 1320 and 1322. The text is often attributed to Felipa Porcelet, who was born in 1250 and died in 1316 (see Sancho Fibla 2018: 297), the successor of Doucelina as the mother superior of the convent, who either wrote the first version herself or dictated it to someone (see Renan 1885: 527-528). Fibla (2022: 50-55) rejects Felipa’s authorship, suggesting that it was a joint work of the sisters of the Beguines of Marseilles, which better corroborates with the proposed genesis of the text in the beginning of the 1320s. The text was probably written in the house of the Beguines of Marseille and represents the language of that area. The only known manuscript is not the original, but a copy of that text that was produced only years after the composition of the original. Renan (1885: 527) assumes that the copy was “[e]xecutée sans doute à Marseille”, without giving any context on that statement. However, this statement is corroborated by a detailed study on the language of the manuscript by Wehowski (1910). The copyist of the only known manuscript names himself as Jacobus at the end of the manuscript; see the following Example (10). There is no further information on who he might have been. 8 Some extracts are published in Meyer (1874: 142-146), Bartsch/ Koschwitz (1904: 331-334), and Appel (1930: 182-185). <?page no="91"?> 6.2 Historical Context of the Manuscripts 91 (10) Obsecro vos qui hoc legeritis, ut Jacobi peccatoris in orationibus vestris memineritis ‘I beseech you who read this, that you will remember Jacobi the sinner in your prayers’ (translation taken from Fibla 2022: 50) (BnF, Français 13503 (13 th / 14 th century: f o . 102 r o , l. 11-13)) This is the first of the manuscripts in discussion here where there is some information on gender variation. Wehowski (1910: 74) claims a masculine form for maire ‘mother’ in the feminine singular oblique, based on analogy with fraire and paire. In fact, this noun does not display any gender variation, but shows inflection in the nominative singular, sometimes adding the -s, which is not a sign for masculine gender but rather a specificity in the declension of this noun. 6.2.6 The Roman de Philomena The text of the Roman de Philomena, also known as the Old Occitan version of the Gesta Karoli Magni ad Carcassonam et Narbonam, tells a twofold history: the foundation of the monastery of Notre-Dame de la Grasse, attributed to Charlemagne, as well as the conquest of Carcassone and Narbonne (see Fauriel 1847: 373). To my knowledge, only two Old Occitan manuscripts of this text are preserved to this day: manuscript B (BL, Additional 21218 14 th century) and manuscript P (BnF, Français 2232 1325; see also Schneegans 1898: 42) which are both included in COMETA (2025). There are only two editions of this text, the first one in combination with the Latin version by Schneegans (1898), and another one with French translation later by Simonnet (1987). The Old Occitan translations are mostly known under the name Roman de Philomena. Philomena is supposedly the name of the composer of this text. It is said that Philomena was a historian that accompanied Charlemagne on his conquests so that he could write a factual story about it (see Fauriel 1847: 373). This does not seem to be true, given that there are references to institutions and customs only known from the 10 th and 11 th centuries onward, which disproves those theories suggesting a text genesis in Charlemagne’s times. <?page no="92"?> 92 6 Constitution of the Corpus COMETA This circumstance opens the question of whether the Latin version is really the original or whether it might have been composed in another language such as Old Occitan (see Fauriel 1847: 374). There are different theories of a Latin original, but also that there might have been a Gascon or Spanish original that is not preserved until today. Some scholars give contradictory answers to this question, such as Lebeuf, who states that there is a Latin original, while claiming later on “l’original de ce roman paraît avoir été d’abord gascon ou espagnol, et la traduction latine est vraisemblablement du temps de Bernard III , abbé de la Grasse , sous saint Louis” (Lebeuf 1747: 148, cited after Fauriel 1847: 375), thus contradicting his own findings without any proof of either statement. Fauriel (1847: 378-379) argues that the Latin version at his disposal is a direct translation of the Old Occitan version 9 , giving some examples where he finds Old Occitan influences in the Latin version. He corroborates this idea with the fact that the translator of the Latin version (who names himself Guillelmus Paduanus) gives the circumstances of the translation in the epilogue of the manuscript. He states there that he took the story from a very old manuscript that was almost illegible, but he nevertheless managed to read it (see Fauriel 1847: 376). Meyer (1867: 54) contradicts this idea of an Old Occitan original, adding that the author of the Latin version “déclare en termes exprès dans son prologue avoir traduit un ouvrage en langue vulgaire” (Meyer 1867: 54). This makes it also plausible that this person is the author of both texts, thus making it more difficult to be sure about the directionality of the translation. He even questions whether one version is even the translation of the other, or whether they might rather be independent productions of a similar story, given that some parts in the two Old Occitan versions are very similar while the Latin diverges in certain aspects. With this he suggests that the Latin version is mostly independent and situated somewhere in between the two Old Occitan versions. For him, BnF, Français 2232 (1325) is the original text, while the manuscript BL, Additional 21218 (14 th century) is a later version based on the former (see Meyer 1867: 57- 60). Schneegans (1898) agrees that the preserved Latin versions are independent from the two Old Occitan manuscripts as well as from an Old French version. But unlike the previous statements, he argues that both Old Occitan versions are translations of a Latin original that is not preserved anymore (see Schneegans 1898: 41-49). Regarding the language of the Old Occitan manuscript, Fauriel (1847: 377) states that it differs from the Old Occitan language of the troubadours with respect to morphological (e.g., no morphological case marking) or syntactical fea- 9 His statements are only with respect to the manuscript BnF, Français 2232 (1325), as he did not know of the other manuscript. <?page no="93"?> 6.2 Historical Context of the Manuscripts 93 tures (e.g., the use of que in the fixed position between the noun and the verb). He goes on to state that “[n]ous nous laisserions aller volontiers au soupçon que c’était réellement l’idiome de Narbonne et des environs, que nous croyons avoir été la base, le fond du provençal littéraire” (Fauriel 1847: 378), without giving any precise reasons for this assumption. Meyer (1867: 60) asserts that the language of BnF, Français 2232 (1325) shows traces from 13 th century Gascon, without giving any examples or reasons for this statement. Schneegans (1898: 50-75) compares the language in both Old Occitan manuscripts with the language of other sources written in the 13 th and 14 th centuries in the region between Narbonne and Carcassonne. In this study, he shows that most linguistic features found in the text can also be traced in other manuscripts from that time and region. But this analysis has to be considered with caution given that he assumes a genesis of these manuscripts in La Grasse between Carcassonne and Narbonne due to the content of the story with a very local relevance: Das Bedürffnis die Gesta Karoli durch Uebersetzungen zu verbreiten, konnte nur in der Umgegend des Klosters und den an der Erzählung zunächst beteiligten Städten Narbonne und Carcassonne vorhanden sein. Das Buch hat einen zu sehr lokalen Charakter, die Gründungsgeschichte des Klosters nimmt einen zu breiten Raum darin ein, um einen grösseren Kreis von Lesern zu interessieren. So fehlen denn auch Anspielungen auf unsern Text im Mittelalter, ausser in Urkunden, die sich unmittelbar auf die Klostergeschichte beziehen. Es ist daher sehr wahrscheinlich, dass die lateinischen Handschriften und die Uebersetzungen in der Gegend entstanden sind (Schneegans 1898: 50-51). Regarding the time of composition of the manuscripts, Fauriel (1847) argues that the Old Occitan original must have been composed around the end of the 12 th or the beginning of the 13 th century, given that the Latin translation dates from 1237-1255 and the translator states that he used a very old, almost illegible manuscript as the basis. The Old Occitan version at his disposal is a later copy, dating from the 14 th century (see Fauriel 1847: 381). Meyer (1867) agrees with this statement and supplements that there are only modifications in the orthography in this manuscript, but that the manuscript represents language from the beginning of the 13 th century. For the second manuscript that was unknown to Fauriel (1847), Meyer (1867) also gives the 14 th century as the time during which this manuscript was most likely written. In manuscript BnF, Français 2232 (1325) in a fragmented folio in the end, a partial indication of a date can be found, see Example (11): <?page no="94"?> 94 6 Constitution of the Corpus COMETA (11) CXXV.XII. calen (BnF, Français 2232 (1325: f o . 134 v o )) Schneegans (1898: 44) interprets this as “MCC CXXV. xij. calendas” and sees this as the date of composition for this manuscript, thus it would have been written in 1325. Brunel (1935: 39) seems to agree with this when listing the manuscript BnF, Français 2232 (1325) as being written in 1325. He indicates the other manuscript, BL, Additional 21218 (14 th century), as being written in the 14 th century, and both in the area of Narbonne (see Brunel 1935: 5, 39). Although this is only an assumption or a reconstruction by Schneegans (1898), both Brunel (1935) and Camps (2010) accept this as plausible and the language in the manuscript seems to corroborate the date. 6.2.7 The Leys d’amors The text of the manuscript BmT, Cote 2883 (14 th century) is an important grammatical treaty. As it is a grammatical text which specifies rules on how to write and speak, there is also a passage concerning grammatical gender, which will be addressed later in this study (see Example 70). The importance of the text already in medieval times can be shown by the number of different medieval versions of this text. There is the manuscript included in COMETA (version C), which is the second, shortened prose redaction in three parts of the text (version B is a version in verse, which is chronologically in between the two prose versions (see Jeanroy 1949b: 151-152)). The whole text was published in a tripartite edition by Anglade (1919a,b,c) with a fourth volume consecrated to philological studies, a glossary, an index, and an excursus (see Anglade 1920). Before this exhaustive edition of the entire manuscript, it was published in parts by Chabaneau (1885) and Anglade (1917-1918). In addition to the manuscript included in COMETA, at least three other versions of this text were already produced in the 14 th and 15 th centuries, some preserved in more than one manuscript (see Schlieben-Lange 1991: 107; Swiggers 2011: 136). A first edition of some parts of manuscript A was already published in the 17 th century (see Lafaille 1687: Preuves 64-84), which demonstrates a continued importance of the normative text. The further relevance of the text for scholars from the 19 th century onward can be shown by the large number of editions (in addition to the ones mentioned <?page no="95"?> 6.2 Historical Context of the Manuscripts 95 before) of the whole text (see Gatien-Arnout 1841; Fedi 2019) or in extracts (see Chabaneau 1885; Monaci 1913). There are also extracts published in Meyer (1874: 152-158), Bartsch/ Koschwitz (1904: 403-410), Crescini (1905: 395-398), and Appel (1930: 197-201). Additionally, there is an edition of the Old Catalan version of this text (see Gonfroy 1981). The text is attributed to Guilhem Molinier, docteur es lois, who was appointed by the Consistori de la Sobregaya Companhia del Gay Saber to write this treaty on rhetoric and poetry (see Jeanroy 1949b: 139; Hagman 2010: 23). The Consistori are a consortium of seven troubadours founded in 1323 who organised a poetry competition every year for Old Occitan poetry in Toulouse (see Léglu 2008: 383). Guilhem Molinier was part of the Consistori, even though he was not one of the seven troubadours who founded the organisation. It is unclear when and how he became part of the consortium, but he eventually even became the chancellor of the consistory (see Jeanroy 1949b: 141). He was appointed in 1348 to write a final version of the grammatical treaty, which was finished in 1356 (see Swiggers 2009: 588). The date for the beginning of the composition of this text is unclear; Schlieben-Lange (1991: 108) for instance proposes the time from 1332-1356, while Brunel (1935: 76-77) proposes a longer period, already starting in 1325, both without further specification for their assumption. (12) adonx comezero de bocca a mestre guilhem molinier. savi en dreg que el fes e compiles las ditas reglas am cosselh del honorabble e reveren senhor. mossen bartholmieu marc doctor en leys. So they appointed vocally master Guilhem Molinier, erudite in law, that he make and compile the aforementioned rules with the advice of the honorable and respectable sir Bartholmieu Marc, doctor of the law. (BmT, Cote 2883 (14 th century: f o . 4 r o , c. 2, l. 10-15)) <?page no="96"?> 96 6 Constitution of the Corpus COMETA Guilhem Molinier is named as the author right in the beginning of the text and it is specified that he had support or council of another man called Bortholmieu Marc, who was also a jurist, as shown in Example (12). The text of the Leys d’amors is an original text, created for the purposes of the Consistori del Gay Saber, which was nevertheless inspired by previous grammatical treaties. It is certain that the author(s) of the Leys d’amors knew Li livres dou Tresor by Brunetto Latini; the Compendium theologicae veritatis, probably written by Hugo von Strassburg; the Liber consolationis et consilii and the Ars loquendi et tacendi, both written by Albertano da Brescia; the Libellus de muliere forti; the Ars maior by Donatus; the Etymologiae by Sant Isidore; as well as further classical and medieval Latin texts (see Jeanroy 1949b: 219-227; Pasero 1965: 129-131; Marshall 1969a: 40). Furthermore, he knew the two oldest Occitan grammars, the Razos de trobar by Raimon Vidal and the Donatz proensals by Uc Faidit (see Lienig 1890: 5-22). This list of mostly Latin, but also vernacular works on the Ars poetica, which the author knew and had at his disposal when writing the Leys d’amors, demonstrates very well his deep knowledge of prior grammatical texts (see also Léglu 2008). Furthermore, the text often makes references to troubadours of the flourishing period of troubadour poetry that ended about 100 years before the creation of this text. The most referenced and named troubadour is N’At de Mons, a troubadour from the region of Toulouse, who lived in the second half of the 13 th century. He was referenced 18 times in version C of the Leys d’amors and 15 times in version A (see Anglade 1925: 414). With respect to the time of origin, Chabaneau (1885) proposes 1355 as the beginning and 1356 as the time of completion. He determines this time due to the decryption of the following two passages: (13) a. Lay el temps dous plazen e gay. Festa de santa crotz de may. L an de clemens. E de cascu nos las prezens ‘There the time is sweet, pleasant and gay. The celebration of the holy cross in May. In the year of [pope] Clement. And of all us present’ (BmT, Cote 2883 (14 th century: f o . 7 r o , c. 2, l. 17-20)) <?page no="97"?> 6.2 Historical Context of the Manuscripts 97 b. Asert pauzat al reversari. Del mes a mens per nom contrari. Claramen podetz haver l an. Per crotz marc luc e per Johan. ‘Surely pausing to the inversion. From the month to the mind. through the opposite name. Clearly you can determine the year through crossing Marc, Luke, and John’ (BmT, Cote 2883 (14 th century: f o . 12 r o , c. 1, l. 27-31)) Example (13a) is viewed as indicating the time of beginning for this version of the Leys d’amors, which Chabaneau (1885) interprets as 1355. As it was impossible to personally consult this text by Chabaneau, the explanation is cited here as it is in Jeanroy (1949b: 160; italics in the original): “Si l’on isole les lettres à valeur numérale contenues dans Clemens et cascu, et qu’on les dispose dans l’ordre convenable, on obtient le chiffre MCCCLV”. Why Chabaneau chose exactly these two word forms for the decryption of the time of origin seems rather arbitrary without further context. However, since I could not get a more detailed description (and no other time is proposed in the scientific literature), the times are taken as proposed here. For the date of completion in Example (13b), it is clearer, even without the study by Chabaneau, why he chose to interpret it the way he did. The decryption works as follows: “Sert étant tres renversé et mai le contraire de mens, il faut entendre le 3 mai; les lettres à valeur numérale de Crotz, Marc, Luc, Johan donnent MCCCLVI” (Jeanroy 1949b: 160; italics in the original). The end date of the composition in 1356 is almost unanimously accepted. Only Lienig (1890: 3) contradicts this when indicating the death of Guilhem Molinier in 1350, before the generally assumed time of completion. But since he does not give any more details or context for this statement, it is not convincing in comparison to the multiple other scientific texts indicating a final version effectuated by Guilhem Molinier in 1356. No information exists on the scribe of the preserved manuscript of version C. It seems like it was written quite close to the time of composition of that version; the handwriting is clearly from the second half of the 14 th century, and also a <?page no="98"?> 98 6 Constitution of the Corpus COMETA second hand that added a few comments is from the 14 th century (see Anglade 1920: 132; Jeanroy 1949b: 151). 6.2.8 Français 1049 Apart from an initial part in Latin (the Passio Domini Nostri Jhesu Cristi secundum Matheum, […] Marchum, […] Lucham, […] Joannem f o . 1-14 (see ARLIMA 2025: s.v. Français 1049)), the manuscript BnF, Français 1049 (1343) contains three different texts in Old Occitan: the first text has the title Sur le trepas du roy Robert de Sicile, comte de Provence in the manuscript. It is also known as the Conplancha for the death of Robert of Anjou on f o . 14-16. This is followed by version A of the Libre de vicis et de vertutz, also known as the Old Occitan translation of the French Somme le Roi by Frère Laurent from f o . 19-179. The final part of the manuscript contains the Old Occitan version of Barlam et Josaphas from f o . 180-219 (see Brunel 1935: 44; Camps 2010: 10; Fusaroli/ Menichetti 2023: 112). The first Old Occitan text of the manuscript, the Trepas du roy Robert de Sicile, comte de Provence, gives us hints on the time of composition of the text. As the title of the text makes clear, especially the later-attributed title of the Conplancha for the death of Robert of Anjou, this text was composed after his death in the night from the 19 th to the 20 th January 1343 (see Fusaroli/ Menichetti 2023: 113). Radaelli (2018: 58) specifies a genesis of the text after the 19 th of February 1344, the date on which Andrew of Hungary, the husband of Robert’s heiress Joanna I of Naples, was conceded the royal title by Pope Clement IV, while Joanna still maintained the right to rule over the kingdom. As the terminus ante quem for the composition of the text, Radaelli (2018: 58), supported by Zinelli (2019: 22), proposes the 28 th of August in the same year, the date on which the papal delegate Aimeric of Châtelus strengthened Joanna’s position as rightful heiress and ruler of the kingdom of Naples. For the whole manuscript, Fusaroli/ Menichetti (2023: 113) assume a completion before the death of Andrew of Hungary in September 1345, so the whole manuscript would have been written between early 1344 and summer 1345. With respect to the copyist of the manuscript, it is clear that it was entirely written by a single scribe (see Fusaroli/ Menichetti 2023: 114), with only the initials in the manuscript perhaps having been drawn later by someone else (see Radaelli 2016: 51). However, this is not accepted unanimously, as Fusaroli/ Menichetti (2023: 114) propose the copyist as creator of the initials. There are some paleographic traces in the manuscript that lead to the assumption of a notary as the scribe of the text (see Fusaroli/ Menichetti 2023: 114-118). At the end of the third text of the manuscript, the following citation can be found: <?page no="99"?> 6.2 Historical Context of the Manuscripts 99 (14) Iste liber est magister Richardi Lamberti. notarii ḍẹ 10 de Aquis This book belongs to master Richard Lambert, notary of Aix. (BnF, Français 1049 (1343: f o . 179 r o , c. 2, l. 9-11)) Richard Lambert, the person mentioned in Example (14), is possibly the scribe of the manuscript (see Frosini 1999: 123; Fusaroli/ Menichetti 2023: 118). Zinelli (2019: 22) specifies that the time of composition—this notary was active in 1345 and 1346 in Aix-en-Provence—as well as the “caractère provençal” (Zinelli 2019: 22) of the manuscript corresponds well with Richard Lambert, but he only assumes Richard Lambert to be the possessor of the manuscript. Frosini (1999: 123) argues that the handwriting of this note of possession is identical with the handwriting of the rest of the manuscript, which leads her to assume this notary as the scribe of BnF, Français 1049 (1343). Fusaroli (2021: 4) states an origin of this manuscript “intorno al 1344 ad Aixen-Provence”, so one or two years earlier, but she does not give a reason for this assumption. However, since she assumes an origin around that year, the dating proposed by Zinelli (2019) remains plausible. The production of the manuscript was most likely in connection and close relation to the chancellors of Aix-en- Provence (see Fusaroli 2024: 211). The first and shortest of the three texts contains a lament for the death of Robert of Anjou and is preserved only in this manuscript (see Jeanroy 1949a: 645), which was first edited and published in Bartsch (1856: 50-57). The text is constituted of 29 strophes in rhyme schemes, following an Italian model rather than an Old Occitan model (see Radaelli 2018: 15-21). Radaelli (2018: 40) specifies, in an analysis of three lexemes employed in the text, that, from a linguistic point of view, it originates most likely in the area of the kingdom of Arles. This is also strengthened by an analysis of the graphical traits in the text (see Jeanroy 10 The outpointed letters in the manuscripts mean that the scribe (or someone else) wanted to erase the respective letters. Since it was not possible to delete the letters, they were outpointed to signal the errors. They appear as such in my transcriptions, but are not considered for the translations. <?page no="100"?> 100 6 Constitution of the Corpus COMETA 1949a: 650-651). Furthermore, an authorship by someone from Provence can be affirmed by the content of the text, which focuses solely on a Provençal perspective, and it is likely that the author was part of Robert’s entourage (see Jeanroy 1949a: 649-650). It remains unclear exactly who the author of the text was, but scholars agree on the author’s provenance in the Provence (see Radaelli 2018: 51-52). The second Old Occitan text in this manuscript from f o . 19-179 contains version A of the Libre de vicis et de vertutz, also known as the Old Occitan translation of the French Somme le Roi, that was composed in 1279 by the Dominican Frère Laurent (see Fusaroli 2024: 208). This text represents a didactic manual written for use by Christian laymen, with a high success shown by about 100 manuscripts in varieties of Old French and translations in e.g., English, Occitan, Catalan, Dutch, etc. (for a more detailed overview, see Fusaroli 2021). It seems like there were at least three intermediate manuscripts before the manuscript BnF, Français 1049 (1343) was written in Aix-en-Provence (see the stemma in Fusaroli 2024: 215). The text in this manuscript is not a direct translation from Old French into Old Occitan, but a copy. This copy is very close to manuscript 313 of the Bibliothèque municipale in Avignon, written in 1336. It is therefore assumed that the same model was used for these two manuscripts (see Fusaroli 2021: 3-5). The final part of this manuscript contains the only known complete Old Occitan version of Barlam et Josaphas (see Bonnier Pitts 1989: 1). There are at least three editions and translations of this text; the first one is by Heuckenkamp (1912), the second is a dissertation including an edition by Bonnier Pitts (1989). Only recently, a new edition was published by Radaelli (2016). The story is originally based on the life of Siddharta Gautama, the Buddha, that was adapted into a Christian version and translated into many different languages, on the basis of a Latin version, based on a Greek one (see Camps 2010: LII-LIII). It is likely that there was also a previous Old Languedocian version of the text, now lost, that served as basis for the preserved version in this manuscript (see Radaelli 2016: 136-137). The language of the Occitan version is held in “una scripta provenzale del XIV secolo, con tratti riconducibili alla Linguadoca orientale” (Radaelli 2016: 87), which corresponds well with the aforementioned theories on the scribe and the time of creation of the manuscript. This is also supported by the linguistic features of the text, suggesting a place of origin for the copyist somewhere between Aix, Marseille, and Arles (see Radaelli 2016: 129-131). In the study of morphological specificity of the text (see Radaelli 2016: 104-110), no information on gender variation is given. <?page no="101"?> 6.2 Historical Context of the Manuscripts 101 6.2.9 Harley 7403 The manuscript BL, Harley 7403 (14 th century) contains a collection of seven different texts, which will be presented individually here. As information on these texts is not equally available, due to the different interest they received by scholars, the amount of data presented in the following differs greatly. Before going into detail on the seven different texts, information on date and place of writing of this manuscript is given. The language employed in the manuscript suggests a place of origin (of the scribe(s)) somewhere in the area of Narbonne, Toulouse or the Rouergue (see Collura 2014: 70). It is generally assumed to be written near Narbonne, and certainly in the Languedocian linguistic area, most likely written at the end of the 13 th or the beginning of the 14 th century (see Brunel 1935: 7; Zamuner 2005: 759). The text was probably written by one, or maybe two scribes, which can be seen in the internal homogeneity of the writing (see Suchier 1883: 485; Collura 2014: 68-69). The whole manuscript was edited by Suchier (1883). The first text included in this collection is one of the two preserved manuscripts of the Old Occitan version B of the apocryphal Gospel of Nicodemus, also sometimes referred to as the Gospel of Gamaliel (see Hershon/ Ricketts 2007: 136). The name Gospel of Gamaliel is misleading, as there is also another apocryphal gospel with the same name that has some common traits, but is still another text (see Tausend 1980: 4). The text is based on a Greek model, first composed around the 4 th century (see Collura 2014: 7). There are different Latin versions that were translated later, resulting in three main versions. The version of this text included in manuscript BL, Harley 7403 (14 th century) is based on the Latin version, which Collura (2014: 8) calls version B. Due to similarities between the incomplete version in BL, Harley 7403 (14 th century) and the complete one in BnF, Français 1745, Collura (2014) assumes a previous translation into Old Occitan, that served as a model for both the scribe of BL, Harley 7403 (14 th century) and the scribe of BnF, Français 1745. The version included in BL, Harley 7403 (14 th century) was first published as a whole version in Suchier (1883: 1-84), based primarily on the complete manuscript BnF, Français 1745, while sometimes using the manuscript BL, Harley 7403 (14 th century) instead, without commenting on it. This method is criticised by Collura (2014: 11), who proposes a new edition entirely based on both manuscripts. The text starts on the first folio of this manuscript and goes on until folio 27. Only Izquierdo (1997: 134-135) claims the end of this part to be on folio 35, as he includes the Old Occitan version of Los XV signes que veno in his counting. He later clarifies that this part does not really belong to the apocryphon, but rather that there is a strong relationship between the two parts of the manuscript (see <?page no="102"?> 102 6 Constitution of the Corpus COMETA Izquierdo 1997: 136), as it follows directly to the text and is already mentioned in the introduction 11 . Collura (2014: 11) specifies that the presence of the Old Occitan translation of the Old French poem on the Quinze signes de la fin du monde is a specific trait of the two manuscripts of version B. The combination of the Gospel of Nicodemus and the translation of the Quinze signes is often viewed together and attributed the title of Sens e razos d’una escriptura (see Izquierdo 1997: 134; Collura 2014: 14). A direct connection between the two texts, supported by linguistic and manuscript-historic facts (in all versions of this branch, the Gospel of Nicodemus is followed by the poem of the fifteen signs), is mainly assumed (see Suchier 1883; Izquierdo 1997), and only Meyer (1898b: 104) sees the two texts as independent texts that were written one after the other, especially concerning the manuscript BL, Harley 7403 (14 th century). This apocryphal text had a huge success in medieval times, shown by the sheer number of Latin manuscripts known today (there are more than 430 preserved manuscripts, according to Collura 2014: 8) and the number of at least 12 languages it was translated into, e.g., Occitan, Catalan, Belorussian, Low German, Old Norse, Welsh, etc. (see Izydorczyk 2021: 42). This sort of apocryphal religious texts had a flourishing time in the late 13 th and early 14 th century, where many translations of such texts were manufactured (see Collura 2014: 16). The third text in this manuscript also remains in the religious context; it is the Old Occitan translation of the legend of the Lignum crucis, the wood of the cross, from folio 36 to 49. This text is one of the three parts of the Legend of the true cross, which is the united text of three previously unrelated parts (see Collura 2019: 96- 97). The version included in BL, Harley 7403 (14 th century), a translation of the apocryphal Latin text Post peccatum Adae, is one out of three preserved complete versions of the Legend of the wood of the cross, but independent of the other two manuscripts. It has a relation to an Old French version in manuscript Dublin, Trinity College, 951, which suggests that the anonymous translator of this text had both an Old French and a Latin version at their disposal when translating the text (see Collura 2019: 99-101, 2022: 9-10). The following part of this manuscript is filled with a dietary treaty from folio 49 to 62, known as “Occitan Health Advice for the Layperson, formerly the Diététique provençale” (Pfeffer 2024: 5). It represents a complete Old Occitan version of this text that served as basis for the first edition of this text included in Suchier (1883: 201-213). The same editor published another edition based on two additional Old Occitan fragments and another complete version in Old Catalan (see Suchier 1894: 163-164, 173). This text represents an Old Occitan translation of 11 This introductory part, where the translator also specifies that he translated this text from Latin (see Izquierdo 1997: 136), is included only in the other manuscript, and it does not appear in BL, Harley 7403 (14 th century), which is missing the first 1374 verses (see Izquierdo 1997: 134). <?page no="103"?> 6.2 Historical Context of the Manuscripts 103 the Latin model Epistola ad Alexandrum de dieta servanda, part of the Secretum secretorum which is based on an Arabic original, the Sirr-al-‘asrār. It is again an apocryphal text that was considered non-apocryphal until the beginning of the Renaissance (see Zamuner 2007: 166). This text had a huge success in medieval times, as shown by more than 600 preserved manuscripts in many different languages (see Zamuner 2005; Pfeffer 2024). The Old Occitan version does not represent a faithful translation of the Latin model, but nevertheless is oriented on the Latin model (see Pfeffer 2024: 21). There are six known Old Occitan manuscripts, four containing a verse version (the longest and most complete one in BL, Harley 7403 (14 th century)), and two are prose versions (see Pfeffer 2024: 24-28). The text was translated into Old Occitan probably at the end of the 13 th century, maybe around 1288 (see Zamuner 2005: 757-759; Pfeffer 2024: 30). Pfeffer (2024: 40-41) assumes, based on a linguistic analysis of their language, that both the author and the scribe originated from the area around Nîmes, Monpellier, and Narbonne. The next text, from folio 63 to 110, is called Le repentir du pécheur (see Brunel 1935: 7). It was edited and published in Suchier (1883: 214-240) with the title Des Sünders Reue and was also published in parts with the same title in Appel (1930: 151-152). To my knowledge, no more information is available on this text. The last part of the manuscript, from folio 111 to the end, is filled with the Doctrinal, a treatise on morality, ethics and Christian dogma composed by Raimon Castelnou (see Taylor 2015: 505), which is preserved in two manuscripts, one of which is the one included here (see Giannetti 1988: 111). In the end of the version of the Doctrinal that is not preserved in manuscript BL, Harley 7403 (14 th century), because the last 10 verses are missing (see Giannetti 1988: 111), the author of the text names himself. Included in another manuscript of the text, edited by Sutorius (1918), it is written in verse 384: “An soi .j. cauallies laicx de paura eretat Raimon de castelnou soy per nom apellat” (Sutorius 1918: 369). From that quote, it can be assumed that this troubadour was noble (a knight), but there is no more information on this author (see Meyer 1885: 533). The version included in COMETA (2025) (so the one from the manuscript BL, Harley 7403 (14 th century)) was first edited by Suchier (1883: 241-255), without any contextualisation. This text is the only text included in BL, Harley 7403 (14 th century) that is not a translation but a text originally composed in Old Occitan. 6.2.10 Nouvelle Acquisition Française 11151 The manuscript BnF, NAF 11151 (14 th century) is a collection of three different texts. The first one consists of two arbitral verdicts between the Abbey of Saint- Géraut and the consuls of that city. The first verdict dates from 1280 and the second from 1298 (the French title is: “Les sentences arbitrales entre l’abbé de <?page no="104"?> 104 6 Constitution of the Corpus COMETA Saint-Géraud et les consuls de cette ville (1280 et 1298)”, (see Brunel 1935: 70; Camps 2010: 40)). The text was written in Aurillac (see Thomas 1911, 1912), which historically belongs to the Haute-Auvergne and is often seen as a text from the Auvergnat region (see Lodge 1995: 420-422). However, it is seen as belonging linguistically to the Languedocian area: “La zòna d’Orlhac es istoricament en Auvèrnhe mai es de dialècte lengadocian.” (Sumien 2009: 14; see also Alibèrt 1976: XXI). The evolution of Latin ca [ka] is often seen as the isogloss seperating northern from southern Occitan varieties, being palatalised in the North to <cha> [ t>S ], while maintaining the <c> [k] in the southern parts (see Wüest 1995: 441). As [k] before [a] is maintained in Aurillac (see Lodge 1995: 420), it is often seen as Languedocian (for a discussion on the classification of Occitan varieties, see Sumien (2009) and Sibille (2024)). The main part of the manuscript contains the arbitral verdict between the abbot and the consuls of Saint-Géraud, up until today only edited by Rivière (1874: 296-319; 369-394). This editor does not give more context on the manuscript other than specifying the years 1280 and 1298 as dates of the arbitral verdicts. It remains unknown when and by whom the preserved version was written. After this main part of the manuscript containing the arbitral verdicts, four more folios contain two other texts, without any relation to the verdicts (see Thomas 1911: 355-357). The first one from folio 29-32 is a shortened translation of the Mulomedicina, a veterinary text originally composed in Latin by Teodorico Borgognoni, an Italian surgeon. Thomas (see 1911: 355-357) assumes an earlier complete translation into Old Occitan of that text, which is only conserved in extracts in manuscript BnF, NAF 11151 (14 th century). This version of the Mulomedicina is followed by another very short text on folio 32 containing the Recepta del vi, the ‘recipe of the wine’. The two texts are edited in Thomas (1911) on the basis of another manuscript with the same texts, and his analysis is also based on the other manuscript. However, as he confirms only a year later (see Thomas 1912), the manuscript included in COMETA (2025) is very similar to the one he used, so information taken from his article can also be seen as valid for the manuscript BnF, NAF 11151 (14 th century), which he and Brunel (1935: 70) wrongly attribute with the cote NAF 11143 (see Camps 2010: 40). Thomas (1911: 356) dates the manuscript back to the beginning of the 14 th century. Regarding the language employed in the manuscript, Thomas (1911: 357-358) specifies that the language in all three parts of the manuscript is very similar, if not identical. To my knowledge, there is neither information on the author/ translator nor the scribe of this text. <?page no="105"?> 6.2 Historical Context of the Manuscripts 105 6.2.11 Nouvelle Acquisition Française 11180 To my knowledge, there is only one study and edition of the manuscript BnF, NAF 11180 (14 th century), which is the one by Frank (1956). It is listed in the bibliography by Brunel (1935: 71) and the list of manuscripts from the BnF by Camps (2010: 86), but they do not display any further studies on this manuscript. Frank (1956: 32) argues that, based on the script, this text of the Planh de la Vierge is a copy from the end of the 13 th century written by two different scribes. The copy was probably made only shortly after the original version, which is the translation of a Latin original of uncertain authorship. According to the manuscript in discussion here, the Latin text was written by Saint Augustin, but other versions of this text also attribute it to Saint Bernard (see Frank 1956: 32). According to Frank (1956: 33-34), the language of the manuscript is identical to the language of the Vida de Santa Doucelina (see Section 6.2.5), and very similar to the language of Raimon Feraud, the author/ translator of the Vida de Sant Honorat (see Section 6.2.3). From a linguistic point of view, the language of the translator and both scribes seems quite similar. According to Frank (see 1956: 34-36), they are probably all from Marseille. This corresponds on the one hand to the statement that the language employed in this text is identical to the language employed in BnF, Français 13503 (13 th / 14 th century), which was also written in Marseille around the same time, so in a very similar context. On the other hand, this contradicts the statement about strong similarities to the manuscripts of the Vida de Sant Honorat, which were all written in the Provençal linguistic area and represent the Provençal language, which should therefore be different from the language of the manuscript in discussion here. Regarding gender doublets, Frank (1956: 39) only mentions a variation in gender assignment regarding the noun verge, which appears as feminine for the meaning ‘virgin’, and as a masculine noun for the meaning of ‘saint’. The other irregularity he mentions is crida appearing as a feminine noun, which corresponds to the entries in Raynouard (1836-1845) and Levy (1894-1924), so it is not really surprising and therefore not considered to be relevant. 6.2.12 Remaining Manuscripts The remaining manuscripts have not been thoroughly studied yet, which is why we do not have a lot of information about them. Therefore, they are presented all together in one section in alphabetical order, giving as much information as possible. Two of the three remaining manuscripts are stored in the BnF, the other one is stored in the BL. <?page no="106"?> 106 6 Constitution of the Corpus COMETA 6.2.12.1 Harley 3041 The manuscript BL, Harley 3041 (14 th century), stored in the BL as part of the Harley collection, is mainly written in Latin, with only one folio at the end of the manuscript containing Old Occitan text. Therefore, this manuscript represents only a very low number of tokens and thus remains marginal within the whole COMETA (2025) corpus. Horne (1808: 727) describes the Occitan page as follows: “Immediately before this bull are some pious verses on the pains of hell, addressed by the Author to his friend Girard, & written in the Provençal language”. It contains in fact a Tenzone, a sort of disputing text (see Cherchi 2002: 968), between Aicart and Girard, hence the title of the text Tenson entre Aicart et Girard. The statement by Horne (1808) suggests therefore the authorship of a certain Aicart, about whom we do not have any more information. According to Brunel (1935: 6), it was written in the 14 th century in Provence. 6.2.12.2 Français 13504 The manuscript BnF, Français 13504 (14 th century) is divided into three different parts. The first part, from the beginning to the 26 th folio, contains the Vie de saint Elzéar. The hagiographical text goes on from folios 32 to 61 with the life of his wife, the Vie de Sainte Delphine. There is an edition based on the only known manuscript of the hagiographies (the manuscript BnF, Français 13504 (14 th century), included in COMETA (2025)) with a translation of the two hagiographies in Cambell (1963). Between these two hagiographical texts, a translation of one of Maguerite d’Oingt’s visions is inserted. This translation is not a very close translation. The translator or copyist inserted and changed many parts (see Gardette 1957: 229). The two hagiographical texts within this manuscript are in close relation to one another, given that it describes the life of a married couple. It is an Old Occitan translation of a Latin hagiography dated somewhere between 1385 and 1395, and translated into Old Occitan some time later (see Dulong 1928: 32). The two saints lived in the late 13 th and the early 14 th century (Elzéar from 1285 to 1323, Delphine from 1282 to 1360), and were canonised because of their model married life: they married young in 1299 and they both chose to live in chastity and join the Third Order of Saint Francis. They lived a life of prayer, charity, and discipline. Elzéar was sainted already in 1369, while Delphine was beatified in 1694, but the process of officially making her a saint was never completed (see Girard 1912). According to Brunel (1935: 54) and Camps (2010: 73), this manuscript was written in the 14 th century in the region of Albi, so in the Languedocian area, which is why this text is included in this sub-corpus. The language of the text can <?page no="107"?> 6.3 Implications for the Analysis 107 be located somewhere in the northern part of the Tarn, while the manuscript stems from Albi directly (see Dulong 1928: 32). Both time and region for the creation of the text are confirmed by Cambell (1963: 9-25). 6.2.12.3 Latin 901 This manuscript contains only 4 folios of Old Occitan text, which is the Version occitane des Plaintes de la Vierge, while the rest of the manuscript is written in Latin (see Brunel 1935: 61; Camps 2010: 81). There are two editions of this text, one without any contextualisation in Meyer (1874: 131-136), and another with a short introduction in Couture (1862). The latter one describes it as being written in the 14 th century, while the original would go back to about one century before (see Couture 1862: 180). The language is described as follows: “surtout la langue en est appauvrie et mêlée, voisine déjà de l’entière décadence” (Couture 1862: 180), which is why he dates the manuscript to after the end of the flourishing troubadour lyric. He specifies that this text is a shortened translation, and a more concise version of the Latin Stabat. This manuscript was in use in the parish of Albi, but he does not specify where it was written. Later, Brunel (1935: 61) indicates the time of origin in the 14 th century and the place of origin as the region of Albi, which is accepted as such by Camps (2010: 81). They do not specify their reasons for these temporal and local indications, but since it corresponds to the only study on this text, it is assumed as correct for this discussion. 6.3 Implications for the Analysis The constitution of COMETA (2025) as presented in Sections 6.1 and 6.2 has some implications for the analysis that will follow in Chapters 8 and 9. This is linked with the fact mentioned earlier in Chapter 4 that Old Occitan is a nonstandardised language. This means that the data analysed in the following chapters is not comparable to analysis of contemporary standardised languages. All different types of diasystematic variation play a role for the analysis of medieval manuscripts. The first type of variation is diatopic variation. The constitution of the corpus aims at grasping regional variation between Languedocian and Provençal, but also includes supraregional variation that is shared in both analysed areas. An additional factor that is important for the medieval scriptae is also local variation, as shown e.g., by Reichle/ Schöffel (2025). In their analysis of the mots nouveaux 12 found in the Comptes Consulaires de Montfer- 12 The mots nouveaux are those words that are added in the DOMél (2025), that were not included in the previous Old Occitan dictionaries by Raynouard (1836-1845) and Levy (1894-1924) (see Selig/ Reichle/ Schöffel 2023: 270). <?page no="108"?> 108 6 Constitution of the Corpus COMETA rand they showed a considerable amount of lemmata that are used only locally (see Reichle/ Schöffel 2025: 81). It is important to consider this tripartite diatopic variation within the medieval language, which is expected and therefore not surprising: local, regional, and supraregional variation. The high number of localisms found in the Comptes Consulaires is not only due to the locality, but also due to their discourse tradition, which influences the linguistic use (see Selig/ Reichle/ Schöffel/ Wiedner 2026). To keep this variable to a minimum, most manuscripts included in COMETA (2025) were chosen from the domain of religious texts. It should nevertheless be kept in mind that the manuscripts BmC, Cote 34 (13 th century), BnF, Français 25425 (1275), BnF, NAF 11151 (14 th century), BL, Harley 3041 (14 th century), and BmT, Cote 2883 (14 th century) belong to other domains. Variation found only within these manuscripts might therefore be due to their respective discourse tradition. An additional factor that might play a role for linguistic variation is diastratic variation, meaning variation conditioned by the social background of the scribes. As there is only little information on the scribes and no information on their social (nor their linguistic) background, it is not possible to know to which extent this plays a role. From the fact that they know how to write, it is clear that they belonged at least to some extent to an educated elite of the medieval society. In summary, it can be said that idiolectal variation plays a key role in languages that are not (yet) standardised, such as Old Occitan. The scribes are not obligated to follow any scriptural model in their writing process. They are free to write the way they want in that specific moment. They have some influences due to Latin grammar schools and the context where they learned writing, but they are nevertheless free to find their own solutions e.g., for transcribing sounds that did not exist in Latin. It is very complex to know which type of diasystemic variation is key for the variation found within the manuscripts; it might also be the case that the variation is present within the manuscripts and not due to only one of the variables. In the following, variation in gender is considered to be regional if it is present in one of the regions, but absent in the other. It is considered supraregional if it is present in both areas (and in other Occitan speaking areas). The case studies in Chapter 9 aim at demonstrating idiolectal variation due to either the scribe (see Section 9.1) or the author (see Section 9.2). Through these two case studies the variable of the scribe or the author are better graspable and it is therefore possible to get deeper insight into the influence of these two variables on gender usage in the manuscripts. Through quantitative analysis, diatopic perspectivisation, contextualisation in terms of text domains, and the inclusion of case studies, the study avoids a reductionist account of medieval variation and instead offers a nuanced perspective that accommodates both the individuality of specific solutions and the collective tendencies they reflect. <?page no="109"?> 6.4 Intermediate Summary 109 6.4 Intermediate Summary To conclude this presentation of the manuscripts that constitute COMETA (2025), the following Table 6.1 summarises the principal facts about the manuscripts 13 . The manuscripts are divided into two different sub-corpora, according to the provenance of the manuscripts, meaning the place where they were written and the linguistic variety used in the manuscript as described in Section 6.2. Therefore, COMETA (2025) consists of two sub-corpora, the Languedocian and the Provençal sub-corpus. To make it easily visible which texts belong to which subcorpus, the texts from the Languedocian area are in italics, while the remaining texts from the Provençal part are in bold in Table 6.1. 13 Not applicable (NA) is to be understood in the sense that no information on the respective column is available. <?page no="110"?> 110 6 Constitution of the Corpus COMETA Shelf number Time Place Author Scribe Text genre BL, Additional 10323 mid-14 th Provence Raimond Feraud NA Hagiography BL, Additional 21218 14 th Near Narbonne NA NA Historiographic BL, Harley 3041 14 th Provence Aicart (? ) NA Tenzone BL, Harley 7403 late 13 th / early 14 th Near Narbonne Various NA Religious/ dietary BmC, Cote 34 first half 13 th Provence, Rouergue Daude de Pradas (? ) NA Courtly Romance BmT, Cote 2883 1355-1356 Toulouse Guilhem Molinier NA Grammatical Treaty BnF, Arsenal 6355 First half 14 th Gévaudan Bertran de Masselha NA Hagiography BnF, Franç̧ais 1049 1344-1345 Aix en Provence Various Richard Lambert Religious BnF, Franç̧ais 13503 late 13 th / early 14 th Marseille Felipa Porceleta Jacobus Hagiography BnF, Franç̧ais 13504 14 th Albi NA NA Hagiography BnF, Franç̧ais 13509 14 th Provence Raimond Feraud Reforsat d’Olieras Hagiography BnF, Franç̧ais 2232 1325 Near Narbonne NA NA Historiographic BnF, Franç̧ais 25425 1275 Toulouse Guilhem de Tudela/ Anonymous NA Chanson de geste BnF, Latin 901 14 th Albi NA NA Religious BnF, NAF 11151 early 14 th Aurillac NA NA Juridical/ medical BnF, NAF 11180 end 13 th Marseille NA NA Religious BnF, NAF 6195 mid-14 th Provence Raimond Feraud NA Hagiography Table 6.1: Summary of known facts about the manuscripts (bold = Provençal sub-corpus; italics = Languedocian sub-corpus). <?page no="111"?> 7 Technical Creation of the Corpus COMETA Due to issues with existing editions, mentioned in Chapter 1, it was necessary to create a corpus myself. In this section,the tools used for the corpus creation as well as the criteria for the transcription are described (see Section 7.1) before explaining the annotation guidelines (see Section 7.2) for those manuscripts used later in the qualitative analysis (see Chapters 8 to 9). 7.1 Semi-Automatic Transcription using Transkribus At the beginning of this work, no model for automatic Handwritten Text Recognition (HTR) existed. One of the first models is the one by Garces Arias et al. (2023), which is trained to recognise Old Occitan words, but the data is taken from the fiches of the DOM, which were written by one and the same person in the 20 th and 21 st centuries. It is a good model for HTR, but it does not help for the recognition of manuscripts written in the 13 th and 14 th centuries. Therefore, it was necessary to train a separate model for the semi-automatic transcription of the medieval manuscripts. The model used for the transcription of the manuscripts within COMETA (2025) is the model Wiedner (2023), which was trained within the tool Transkribus. This model has a character error rate (CER) of 2.3 % on the training set and 3.6% on the validation set. This is a rather good CER, but not sufficient for this sort of research question, because one of the main sources of errors is the difference between <a> and <o>. The differentiation of these vowels is crucial for an analysis of gender in Romance languages, so all pages were manually post-corrected. For the transcription of the manuscripts, the following editorial choices were made: • The punctuation was preserved as in the original text. • Word segmentation was adjusted according to the DOMél (2025). • No apostrophes or other signs were added. • Capital letters and minuscules were left as is. • Initials were treated as capital letters. • <i> was left as is and not changed into <j>. <?page no="112"?> 112 7 Technical Creation of the Corpus COMETA • <u> and <v> were assigned based on phonological criteria. To illustrate the assignment, some examples are provided here (all screenshots in the following are taken from the manuscript BnF, Français 1049 (1343), from the f o 18 v o and 19 r o ): - → <alcunas>. - → <lur>. - → <perseverar>. - → <serventz>. - → <vilezas>. • Abbreviations were resolved as follows (all screenshots, unless stated otherwise, are taken from the manuscript BnF, Français 1049 (1343), from the f o 17 and 18, both r o and v o ; the abbreviations are listed in alphabetical order): - → <an>/ <am> 1 . - → <christ>. - → <con>/ <com>. - → <contra>. - → <de>. 1 <m> and <n> were inserted according to phonological environment (<m> before bilabials, otherwise <n>), or the scribe’s preference. The same is valid for the following occurrences of an abbreviated nasal consonant. <?page no="113"?> 7.1 Semi-Automatic Transcription using Transkribus 113 - → <Guillem> (BmC, Cote 34 13 th century: f o . 45 v o ., c. 1, l. 5). - → <Ia>. - → <Ihesus>. - → <mer>. - → <na>. - → <nostra>. - → <on>/ <om>. - → <papa> (BL, Additional 21218 14 th century: f o 12 r o ). - → <per>. - → <pre>. - → <pro>. - → <prophetas> (BnF, Français 1049 1343: f o . 197 r o ., c. 1, l. 28). - → <que>. <?page no="114"?> 114 7 Technical Creation of the Corpus COMETA - → <qui>. - → <sancta>. - → <ver>. 7.2 Part of Speech-Tagging using LLMs As at the beginning of the Part of Speech (PoS)-tagging of the texts no tagger for Old Occitan was available, the first text (namely BnF, NAF 6195 14 th century) was pre-tagged with a tagger for Modern Occitan, which included diatopic variation (see Poujade in preparation). This tagger uses the tag-set according to the Universal Dependencies (UD)-standard (see de Marneffe et al. 2021). The following tags appear in COMETA (2025): adjective (ADJ), adposition (ADP), adverb (ADV), auxiliary verb (AUX), coordinating conjunction (CCONJ), determiner (DET), noun (NOUN), numeral (NUM), participle (PART), pronoun (PRON), proper noun (PROPN), punctuation (PUNCT), subjunction (SCONJ), symbol (SYM), and verb (VERB). All UD-Tags appear except INTJ and X. For the tokenisation, this tagger split the contracted pronoun and determiner combination, for instance del or al, automatically into the two constituents de + lo or a + lo respectively. After this first pre-annotation, the text was manually corrected in order to obtain Ground Truth (GT). This text was then used with different prompts and a large language model (LLM) with the goal of finding the best performing one (see Schöffel/ Wiedner, et al. 2025; Schöffel/ Garces Arias, et al. 2025). The next text (namely BL, Harley 7403 14 th century) was then preannotated with the model Phi-4 by Microsoft (see Ollama 2024), and a code written in Python. The temperature was left at the default setting (0.3). The model was prompted with the following prompt (Prompt A, as proposed in Schöffel/ Wiedner, et al. 2025: 346): You are a Medieval Occitan language expert. Analyze the provided text and assign to each word Universal Dependencies Part-of-Speech tags: “ADJ”, “ADP”, “ADV”, “AUX”, “CCONJ”, “DET”, “INTJ”, “NOUN”, “NUM”, “PART”, “PRON”, “PROPN”, “PUNCT”, “SCONJ”, “SYM”, “VERB”, “X”. Do not add or remove punctuation or tokens. <?page no="115"?> 7.2 Part of Speech-Tagging using LLMs 115 Ensure to process token by token. Ensure that the order of words in the text is kept for the output. Return the results as a JSON array of objects, each containing only the ‘word’ and ‘upos’ keys. The output must be only the JSON array without any additional text, explanations, or formatting. Ensure that the JSON array is properly closed. Due to the context window limitations inherent in auto-regressive, decoderonly architectures such as Phi-4, input sequences for PoS tagging were segmented into fixed-size chunks of 50 tokens and processed sequentially. This chunking strategy, while necessary to fit within the model’s maximum context length, introduced notable drawbacks. Most critically, it disrupted the continuity of token-level context, as each chunk was treated as an independent input without access to the tokens preceding or following it. This led to boundary effects, where tagging accuracy degrades at the start and end of each segment due to the model’s inability to attend beyond the local context window. Such context fragmentation is particularly problematic in PoS-tagging, where disambiguation often depends on syntactic and lexical cues extending across multiple tokens. Without cross-chunk attention or memory, models like Phi-4 cannot form cohesive contextual embeddings for boundary tokens, resulting in reduced performance on those positions. These limitations highlight a broader challenge in applying auto-regressive LLMs to structured prediction tasks that require tokenlevel precision and global syntactic consistency. Despite the overall capabilities of the Phi-4 model, several systematic issues emerged. The models introduced (ortho-)graphic noise by omitting, altering, or inserting characters in the input tokens. This behavior posed challenges for maintaining the fidelity of the original text—especially in low-resource Romance language settings, where the model may implicitly bias outputs toward higherresourced related languages. To address these inconsistencies, a reference dictionary was constructed from the original input tokens. During post-processing, the model’s output was automatically compared with this dictionary. In cases where discrepancies were detected, a placeholder label “missing” was inserted to flag the deviation. A similar validation procedure was applied to the UD tags: predicted tags were cross-checked with the set of valid UD labels, and nonconforming outputs were likewise marked as “missing”. This led to a certain number of untagged tokens that had to be annotated manually during the postcorrection phase. <?page no="116"?> 116 7 Technical Creation of the Corpus COMETA After this pre-annotation, the BL, Harley 7403 (14 th century) manuscript was manually post-corrected in order to make it GT and to be able to use it for evaluation of the accuracy of the LLMs. After this step, the remaining manuscripts for the qualitative analysis in Chapter 9 were annotated the same way using the zero-shot prompt with the model Phi-4 (see Ollama 2024). For the post-correction phase, the manuscripts BL, Additional 10323 (14 th century), BnF, Français 13509 (14 th century) as well as the first part of the manuscript BnF, Français 25425 (1275) 2 were only corrected for the nouns, and the other PoS-tags were left as proposed by the model. This means that all tokens were checked, whether or not the respective token is a noun. If it is a noun and the indicated tag was something else, it was changed into NOUN, as well as the other way around; if something was tagged as a NOUN, but is in reality something else, the tag was changed to the correct tag. After this post-correction phase for the nouns, information on gender, number, and case was added manually for each noun within the three manuscripts of the Vida de Sant Honorat (BL, Additional 10323 14 th century; BnF, Français 13509 14 th century; BnF, NAF 6195 14 th century) as well as in the first part of the manuscript BnF, Français 25425 (1275). All remaining manuscripts were pre-annotated with the Occitan tagger of COLaF (Clérice 2020), which performed quite well on Old Occitan texts with a precision of about 80% (see Schöffel/ Wiedner, et al. 2025). The texts in preannotated status are available in COMETA (2025). Because an annotated version of these texts was not necessary for the analysis in this book, these texts were not (yet) manually post-corrected. For the annotation process, the following guidelines were used, adapted in orientation to the UD-standards, with some exceptions explained here: • ADJ: All adjectives, including proper adjectives, but excluding cardinal and ordinal numbers (in contradiction to the UD-standard). • ADP: All prepositions as well as postpositions were included. • ADV: All sorts of adverbs (local, temporal, modal, causal, phrasal), and additionally the negation particles, due to the explanation given later for the PART category. • AUX: The (conjugated) verb forms of eser, estre, aver combined with a participle perfect forming the periphrastic past tense as well as the (conjugated) verb forms of anar combined with infinitive, thus forming the analytic future. Copula verbs were tagged as VERB and not as AUX. 2 The explanation for checking only the nouns in the first part of this manuscript will be given in Section 9.2. <?page no="117"?> 7.2 Part of Speech-Tagging using LLMs 117 • CCONJ: All coordinating conjunctions, mainly e, et, ma, and o. • DET: All determiners, definite and indefinite; additionally pronominal numerals. • NOUN: All nouns were integrated, including nouns derived for instance from verbs after nominalisation due to conversion, but excluding proper nouns. • NUM: All cardinal and ordinal numbers, as well as numeral adverbs, such as premieramen. • PART: In contradiction to the UD-standard, this tag was used for participles, present and perfect. The reason for this choice lies in the fact that participles might function as agreement targets and therefore contain information on the gender of a specific noun, whilst the UD category of PART includes particles such as negation particles which do not give any clue about the gender of nouns. In this category, the texts used for this study here differ from those published in COMETA (2025), which are fully tagged according to UD-standards. • PRON: All pronouns, such as attributive, indefinite, personal, possessive, reflexive, and relative pronouns. • PROPN: All proper nouns, such as names of specific people, places, or objects. • PUNCT: All punctuation. • SCONJ: All remaining conjunctions that subordinate one constituent to another. • VERB: All verbs that do not appear in the form of an auxiliary verb or as a participle. • SYM: Symbols that appear as such in the manuscript and are not cardinal numbers. The only appearing symbol in COMETA (2025) is † . <?page no="119"?> 8 Gender Doublets in COMETA (2025) This chapter presents the quantitative and qualitative data on gender doublets in Old Occitan, which are also listed as such in the DOMél (2025). This approach allows to (in)validate the results of previous studies on gender in Old Occitan, as the DOMél (2025) is the most exhaustive source for Old Occitan lexicography. Due to its composition, it also includes the State-of-the-Art research on gender variation in Old Occitan, as secondary sources were also included in the entries. In the following, all examples will be drawn from the transcription of the manuscripts as they are in COMETA (2025) and cited with screenshots in order to assure the validity of the transcription (on the relevance of including screenshots in the text to assure the validity of transcriptions, see Stratton 2026). The place in the manuscript will be given as precisely as possible, referencing the f o , front (recto (r o )) or back (verso (v o )), the column (c.) (if more than one), and the line (l.). In each example, the noun of interest as well as the respective agreement targets will be marked in bold. In order to facilitate the reading and interpretation of the occurrences, the gender on the respective agreement target will be marked in the translations as follows: masculine with “ M ”, feminine with “ F ”, and agreement targets that do not reflect gender are marked with “ ? ”. For clarity, this annotation will only be made for agreement targets that are directly relevant for the noun in discussion and only for gender, given that case and number should be clear from context, also in the translations. For more details on the nominal system, see Chapter 4. For instance, in the following part taken from Example (15a) “am tot lo comtat pe ioyas.”, tot and lo are the agreement targets that agree with the noun under discussion, comtat. Therefore, these three parts of speech are marked in bold. In the respective translation ‘the M whole M county as a joyful gift’, both agreement targets, the and whole are marked with M in order to make it clear that they appear as masculines in the Old Occitan text. In the following tables, the label NA means it was not possible for the respective occurrences of a certain noun to determine the gender. This is often the case due to the lack of agreement targets or instances, where the agreement targets do not show gender due to e.g., elisions of the vowel before a noun with an initial vowel. This can be shown for instance in the occurrence l’art, where the gender cannot be determined with certainty. To further corroborate or invalidate the findings from the COMETA (2025) corpus and to broaden the picture, additional sources in the form of editions are <?page no="120"?> 120 8 Gender Doublets in COMETA (2025) consulted for nouns that have only a small number of occurrences in COMETA (2025). Furthermore, this will be made for nouns, where the status as a gender doublet is unclear, i.e., if a certain noun only triggers agreement with one gender in the texts of one or both regions. This approach helps to identify potential diatopic or diachronic variation, even though the validity of the information gained from editions has to be seen with caution (see the explanations in Chapter 1). 8.1 Methodology The DOMél (2025) lists a total of 61 gender doublets, three of which will be considered together in the following, as they are two variants of a noun with the same etymon and the same meaning, i.e., serp and serpen ‘snake’, glan and aglan ‘acorn’, and trau and atrau ‘girder, beam, bar, joist’. Only 19 1 of these gender doublets have no occurrences in COMETA (2025) and are therefore not considered. Excluding these 19 lemmata and considering graphical variants 2 of a given gender doublet as one leaves us with 39 different lemmata that have occurrences within the COMETA (2025) corpus. For each of the nouns, information will be given on the etymon, the distribution, and the assignment of gender within the Romance languages as well as the specific data on the nouns within COMETA (2025): In addition to the gender (gen) (masculine (M) and feminine (F) respectively), information on the region (Languedocian or Provençal), the case (nominative (nom) or oblique (obl)) as well as number (num) (namely singular (sg) or plural (pl)) is included in the following tables. As shown in Section 3.3, the number value might also influence the gender value of certain nouns (see also Corbett 2000, 2012; Fedden/ Corbett 2017). As stated in Chapter 4, morphological case marking was no longer consistently implemented, so the case distinction in the following was made based on syntactic and semantic criteria. Furthermore, a list of graphical variants used for the quantitative analysis will be given. The corpus analysis for this chapter was executed with AntConc (2024). Gender doublets in the following are sorted according to their type of etymon (e.g., nouns from the third declension class 1 Namely the following gender doublets: afrau, afrest, alegor, alfa, amalviscle, arange, cambe, carral, cartal, debonaire, diocezi, dot, forceps, lange, led, luminari, malviscle, semenci, and usufructuari. 2 All graphical variants in Old Occitan that are listed in the following sections are taken directly from the dDOM (2025). The lists provided in the respective sections contain only exactly those listed there, but potential case or number markings (such as -s, -z, -ts, -tz, …) were additionally included in the query. Furthermore, other potential graphical variants were searched, even though they are not listed in the dDOM (2025), e.g., coullour. Because none of these additional variants had any occurrences within COMETA (2025), they are not listed in the following. <?page no="121"?> 8.1 Methodology 121 in Latin) and within that group in alphabetical order. In cases, where the state of a certain noun as a gender doublet is unclear from the data within COMETA (2025), or variation in gender is conditioned diatopically or diachronically within this corpus, existing editions and other scientific secondary literature will be consulted (mainly the literature listed in the dDOM 2025), in order to validate or falsify the results obtained by the analysis of COMETA (2025). Since not all secondary literature was available, it is not clear, whether or not all occurrences of a certain noun within this additional literature could be found. Therefore, all numbers in the following are absolute numbers and no percentages will be given. In order to contextualise gender doublets with respect to neighbouring Romance varieties, information on the gender of each noun in other Romance languages will be given. Therefore, the following list of lexicographical (re-)sources was used to gain information on the respective nouns. Unless stated otherwise, the information on the lemmata in each language is taken from the following dictionaries: • All Romance Languages: Dictionnaire Étymologique Roman (DÉRom 2016), Lateinisch-Romanisches Wörterbuch (LRW 1907), and Romanisches Etymologisches Wörterbuch (REW 1935). • Aragonese: Diccionario aragonés-castellano-catalán (DACC 2025). • Asturian: Diccionariu de la Llingua Asturiana (DALLA 2025). • Catalan: Diccionari català-valencià-balear (DCVB 2002) and Diccionari etimològic i complementari de la llengua catalana (DECat 1980-1991). • Corsican: Banque de Données Langue Corse (BDLC 2025). • Franco-Provençal: Glossaire des patois de la Suisse romande (GPSR 2025). • French and Old French: Französisches Etymologisches Wörterbuch (FEW 1922-2003) and Trésor de la langue française informatisé (TLFi 2004). • Friulian: Dizionario Storico Friulano (DSF 2025). • Galician: Dicionario de dicionarios (DdD 2025), which is a collection of data from different Galician dictionaries (see Sánchez-Palomino 2025: 186). • Greek: The Cambridge Greek Lexicon (CGL 2021). • Italian: Lessico Etimologico Italiano (LEI 2025). • Ladin: Etymologisches Wörterbuch des Dolomitenladinischen (EWD 1988- 1998). <?page no="122"?> 122 8 Gender Doublets in COMETA (2025) • Latin: Thesaurus Linguae Latinae (TLL 2025). • Medieval Latin: Mittellateinisches Wörterbuch (MLW 2025) and Glossarium mediæ et infimæ latinitatis (Fresne Du Cange et al. [1678]2009). • Modern Occitan: Multidiccionari occitan: Diccionari occitan de referéncia (dicodòc 2025). • Old Occitan: Dictionnaire de l’occitan médiéval (DOMél 2025). • Portuguese: The etymological dictionaries by Machado (1977) and Cunha (1982). • Romanian: Dicționarul limbii române (DLR 1913-1978). • Romansh: Dicziunari Rumantsch Grischun (DRG 2025). • Sardinian: Dizionario etimologico sardo (DES 1960-1983) and the Studien über den sardischen Wortschatz: 1. Die Familie - 2. Der menschliche Körper (Wagner 1930). • Spanish: Diccionario de la lengua española of the Real Academia Española (DLE 2025). In the following, gender doublets 3 will be addressed according to their etymon, beginning with the nouns inherited from Latin in Section 8.2, followed by Greek and other loanwords in Section 8.3 and Section 8.4, respectively. Within each of the sections, the nouns will be given in alphabetical order. 8.2 Inherited Nouns from Latin Most of the discussed gender doublets are directly inherited from Latin, and only a limited number have other origins. In the following, these nouns will be listed according to their Latin declension class, as there is a correlation between gender and declension class, as illustrated earlier in Section 3.2. Because there are no gender doublets derived from lexemes of the first declension class, this section will start with the second class (see Section 8.2.1), followed by the most 3 Some of the gender doublets listed here were already addressed in two articles (see Wiedner 2026, forthcoming). Therefore, some of the examples for the following gender doublets were also used there: dia (see Section 8.2.4), man (see Section 8.2.1), mar (see Section 8.2.2.12), ost (see Section 8.2.2.16), profeta (see Section 8.3.5), salut (see Section 8.2.2.21), sanc (see Section 8.2.2.22), serpen (see Section 8.2.2.24), and tabor (see Section 8.4). Supplementing the analysis in these articles, the examples will be addressed with respect to the region of the respective text, which will give new insights regarding possible diatopic variation of gender assignment. <?page no="123"?> 8.2 Inherited Nouns from Latin 123 important declension class with respect to gender variation, the third class (see Section 8.2.2). There is only one noun each inherited from the fourth (see Section 8.2.3) and the fifth declension classes (see Section 8.2.4), as well as one noun derived from a Latin adverb (or adjective) and one of unclear etymology, which will be addressed afterward in Section 8.2.5. After the summary on hereditary words from Latin, those nouns loaned from other languages will be presented, divided into Greek loanwords in Section 8.3, followed by one other loanword from Persian in Section 8.4. 8.2.1 Second Declension Class There is only one noun from the second declension class in the list of gender doublets, the noun comtat ‘county’ including the prefixed form of the noun vescomtat ‘viscounty’, inherited from Latin comitatum or the prefixed form vicecomitatum, respectively. Both are neuter nouns from this declension class with the same meaning in Latin as in Old Occitan. This noun is masculine in Aragonese, Catalan, French, Galician, Italian, Medieval Latin, and Spanish, but it is both masculine and feminine in Old French, Old Occitan, Provençal, and Romansh. This noun has only a very low number of occurrences within the COMETA (2025) corpus and shows no sign of gender variation within this corpus. An additional look into other sources listed in the dDOM (2025) confirms the status of this noun as a gender doublet, but in general, only a low number of occurrences with feminine gender could be found. The findings in the COMETA (2025) suggest only masculine use in Languedocian, which can be supplemented with three feminine occurrences by the secondary literature listed in the following. There are no occurrences of this noun in COMETA (2025) for the Provençal linguistic area and in secondary sources only masculine occurrences from Provençal could be found. This noun is therefore a gender doublet in Languedocian, but not in Provençal, as Table 8.1 shows. Lemma M Ldoc F Ldoc M Prov F Prov M Lit F Lit Gender doublet (ves-) comtat X — — — X X X Table 8.1: Gender doublets from the second declension class. For the corpus search, the following graphical variants were used: comtad, comtat, comtatz, comptat, contat, contatz, contaz, coumtat, cumtat, and cumtatz 4 . 4 All these graphical variants were additionally researched in COMETA (2025) with the prefixes <ves-> and <vis->. <?page no="124"?> 124 8 Gender Doublets in COMETA (2025) Languedocian Provençal num+gen+case num+gen gen M sg nom 0 0 0 13 13 obl 13 0 13 pl nom 0 0 0 0 obl 0 0 0 NA sg nom 0 0 0 2 2 obl 2 0 2 pl nom 0 0 0 0 obl 0 0 0 Table 8.2: Gender distribution for comtat/ vescomtat. In the texts included in COMETA (2025), both comtat and vescomtat appear only in the Languedocian sub-corpus and only as a masculine noun or with uncertain gender value (see Table 8.2). It appears in total in four different manuscripts from that area (in BL, Additional 21218 (14 th century), BnF, Français 25425 (1275), BnF, Français 13504 (14 th century), and BnF, Français 2232 (1325)). Two examples of masculine occurrences will be shown in Example (15). (15) a. Rey de Narbona e daria li amb elha la ciutat d albi am tot lo comtat pe ioyas. ‘King of Narbonne and would give him with her the city of Albi and the M whole M county as a joyful gift’ (BnF, Français 2232 (1325: f o . 101 r o ., l. 16-18)) b. E en Wles d encontre qu en tot lo vescomtat. No i a .I. plus ric ome ni de maior barnat. ‘And sir Guilem d’Encontre, that in the M whole M viscountcy there is no (hu)man richer nor of greater nobility’ (BnF, Français 25425 (1275: f o . 29, l. 5-6)) In COMETA (2025), both comtat and vescomtat appear only as masculine nouns or with uncertain gender. Raynouard (1836-1845) lists comtat as a noun with <?page no="125"?> 8.2 Inherited Nouns from Latin 125 both genders, but he gives no examples of feminine agreement and lists vescomtat as only masculine. Levy (1894-1924) adds two examples for the prefixed noun with feminine agreement, both from the Histoire anonyme de la guerre des Albigeois (see Anonyme 1863), where the following two occurrences can be found: “la plus forta vila de sa viscomtat et senhoria” (Anonyme 1863: 7; ‘The strongest town under his F viscounty and seigneury’) and “per tractar ladita viscomtat de Bedierres” (Anonyme 1863: 10; ‘To deal with the F said F viscounty of Bédierres’). Another example for feminine agreement can be found in the DAOA (2009: s.v. “vescomtat”) in a text from 1441: “la vescontat de Murat” ‘The F viscounty of Murat’. To find other instances of feminine use for this noun, the sources listed within the dDOM (2025) were analysed and the indications there were verified with the respective literature. The only additional evidence for comtat as a feminine noun can be found within a single text, where the following passages appear: “E segins la contat de Besencon […] E Bernar la contat de Terascon”(Hackett 1953a: 139- 140; ‘And according to the F county of Besançon […] And Bernar in the F county of Terascon’) and “E laissa sa contat e son païs” (Hackett 1953b: 396; ‘And leaves his/ her county and his/ her land’). Pfister (1970: 347) specifies that the feminine gender can only be found in one manuscript (manuscript O) of this text and he states that this gender appears due to Old French influence, where this noun has mainly a feminine gender. With respect to masculine occurrences of this noun, instances can be found in the following passages: “Senh’En Rassa, aquest comtat” (Gouiran 1985a: 224; ‘Lord Sir Rassa, this M county’) or “ordenet per tot lo comtat de Rossillon” (Lommatzsch 1917: 159; ‘Ordered throughout the M whole M county of Roussillon’). In the DAOA (2009), it is listed as both a masculine and feminine noun with the prefix ves- (see the aforementioned example), but only as a masculine noun without the prefix. Masculine examples are given for both nouns: “jutge del contat de Rodes” (DAOA 2009: s.v. “comtat”; ‘Judge of the M county of Rhodes’) and “terras que sont pausadas infra lo vescontat de Murat” (DAOA 2009: s.v. “vescomtat”; ‘Lands that are situated within the M viscounty of Murat’). Another masculine occurrence from the 13 th century Provence can be found in the following passage: “partiment del comtat” (Nicollet 1910: 32; ‘Division of the M county’). Most of the occurrences of this noun, both with and without the prefix, are masculine in Old Occitan, which corroborates the findings within the COMETA (2025) corpus. In total, only 6 feminine occurrences could be found in further literature, three without the prefix, retrieved from one single manuscript. The other three occurrences for the prefixed form stem from three different texts. Two of the three feminine occurrences are from the Languedocian area, the other one is from Auvergne. This suggests that it is a gender doublet in the Languedocian, but not in the Provençal linguistic area. <?page no="126"?> 126 8 Gender Doublets in COMETA (2025) 8.2.2 Third Declension Class Most nouns in the analysis stem from the third Latin declension class. Table 8.3 gives a summary of the nouns from this declension class as they will be presented afterward, within COMETA (2025) as well as within additional literature. For the most frequent words (e.g., mar or serp), the data from additional literature will not be presented explicitly, but the statement on presence or absence of a given gender for those nouns was nevertheless verified before including this information in the table. These nouns can be divided into four different groups: 1. The first group consists of those nouns that are in fact no gender doublets in Old Occitan, neither in COMETA (2025), nor in further literature. Nouns with only one occurrence of varying gender are included in this group, as it is quite possible and maybe even likely in those cases that it is a writing error (or a mistake in the respective edition, when taken from an edition). This group consists of the following four nouns: color, crin, marme, and mel. 2. The second group is made up of nouns of uncertain status, which show some minimal variation. There are six nouns in this group, namely canal, den, fornatz, labor, lach, and ost. 3. The third group is the one with nouns that are undeniably gender doublets, both during the time and in the areas addressed in this study. This is the major group within COMETA (2025) with 12 of the 25 nouns from the third declension class belonging to this group. For all of these twelve nouns, no variation with relation to time or place of origin could be made; they are all considered to be gender doublets in the entire Old Occitan linguistic area, which is also supported by occurrences in further literature apart from COMETA (2025); see among others Bartsch/ Koschwitz (1904) and Appel (1930). 4. The fourth group consists of those nouns which are in fact gender doublets, but with some special considerations, as they are either only gender doublets in a certain region or only for a specific meaning. This group consists of only three nouns; two show diatopic differences, the other semantic differences. The latter one is salut, that has a fixed gender (feminine) for all meanings but ‘greeting’, for which it appears with both genders. This circumstance corresponds well to what we find in other Romance languages such as Aragonese and Catalan. The first noun that shows diatopic variation, mar, appears only as a gender doublet in Old Provençal, but not in Old Languedocian, where the gender seems fixed as feminine. <?page no="127"?> 8.2 Inherited Nouns from Latin 127 Lemma M Ldoc F Ldoc M Prov F Prov M Lit F Lit Gender doublet aigal — — — — — only in Béarnese — art — X X X X X X canal — X — — minimal X uncertain color — X — X only 1 X unlikely crin — X — — — X — den — X — X minimal X uncertain fornatz — — X — minimal X uncertain glan/ aglan X X — X X X X labor — X — — few few uncertain lach — — — — X few uncertain mar — X X X X X X (but not in Old Languedocian) marme X — X — X only 1 unlikely mel X — X — X only 1 unlikely orde X X X X X X X ost few X few X — X only little variation penche — X — X X X X pols X X — X X X X ret — — — X X X X sal X X X — only 2 X X salut X — X X X X X (only for the meaning ‘greeting’) sanc X X X — X X X (but not in Old Provençal) semen — X — — X X X serp/ serpen X X X X X X X set — X X — few X X trau X X X — X X X Table 8.3: Gender doublets from the third declension class. The masculine gender for Old Provençal could be in relation to the Italo- Romance varieties, where it appears as a masculine noun. Being regionally positioned between feminine use in Old Languedocian in the west and masculine use in Italo-Romance varieties in the east could lead to a longer preservation of the instability of gender assignment. The other word showing diatopic variation is sanc, which could be confirmed as a gender doublet in Old Languedocian, but has only masculine occurrences in the Old Provençal area. <?page no="128"?> 128 8 Gender Doublets in COMETA (2025) 8.2.2.1 aigal The noun aigal ‘canal’ is derived from Latin aalis, which is a masculine noun in Latin with the same meaning as it has in Old Occitan. In Medieval Latin, the gender has changed to neuter agreement. This noun still exists in some Occitan varieties with the same meaning and as a masculine noun. Apart from that, it is mainly preserved in the names of water-bringing winds, usually also as a masculine noun. In Walloon, this noun is preserved as a feminine noun with a new meaning ‘dropsy’. In Romanian, it existed as a feminine noun with the meaning ‘ewer, pitcher’, but it is not used anymore. In Ladin, it still exists as a masculine noun with the same meaning as in Romanian. In the corpus search, the graphical variants aigal, aygual, agau, gau, and their respective plural forms were looked for. As can be seen in Table 8.4, this noun shows no visible variation in gender, as all three occurrences within the COMETA (2025) corpus are of uncertain gender. This is due to the fact that this noun starts with a vowel and the preceding determiners (if existent) appear only in a shortened version with an apocope of the vowel and therefore show no sign of gender. Languedocian Provençal num+gen+case num+gen gen NA sg nom 0 0 0 3 3 obl 0 3 3 pl nom 0 0 0 0 obl 0 0 0 Table 8.4: Gender distribution for aigal. All three occurrences listed in Table 8.4 appear in manuscript BmC, Cote 34 (13 th century), with the meaning of ‘water’ or maybe ‘pouch’, not ‘canal’; (16) gives an example of the way this noun appears in the manuscript. (16) Car per los ueils d aigal descent ‘as/ because water (tears) ran down from the eyes/ as it ran down from the pouches of the eyes’ (BmC, Cote 34 (13 th century: f o . 63 r o , l. 28)) As this example demonstrates, it is not possible to determine the gender of this noun within COMETA (2025), so the statement of the DOMél (2025) attributing two genders to this noun can be neither confirmed nor disproved with the data contained in this corpus alone. A closer look at the dictionaries reveals that <?page no="129"?> 8.2 Inherited Nouns from Latin 129 this noun stayed masculine in all Old Occitan varieties except Béarnese, where it became feminine (see Lespy/ Raymond 1887: s.v. “AGAU, GAU”), due to deglutination l’agau > la gau, la agau (see DOMél 2025: s.v. “aigal”). This can be proved by an Old Béarnese text given in Baldinger (1963: 92-94), where it appears as a feminine noun: “he que l’a dada l’aygue ab aquere agau” (Baldinger 1963: 92; ‘She has given the water by that F canal’) or “peus ponts ni per la agau” (Baldinger 1963: 92; ‘By the bridges nor by the F canal’). The DAOA (2009: s.v. “aigal”) lists it as a masculine noun, but the citation given here does not show a clear gender value: “en l’aygual del moli” ‘In the ? canal of the mill’. It is listed as a masculine noun in Pansier (1927: s.v. “aigal”), but the respective occurrence could not be found (it is indicated as being taken from a text written in 1368, but there is no text from that time in his edition), and therefore cannot be verified. In the DAO (1975-2007), it is also listed just as a masculine noun, but the only reference made there is to Pansier (1927), so again it is not possible to verify. 8.2.2.2 art The noun art ‘art, craft, work’ is derived from the Latin aRs (or the accusative (acc) form aRtem 5 ) of the third declension class with the same meanings as in Old Occitan. This noun was feminine in Latin and in Medieval Latin. It still exists in all Romance languages except Romanian. The feminine gender is maintained in Italian, Sardinian, Spanish, and Portuguese; it is mainly masculine in Aragonese, French, Occitan and Sursilvan (but sometimes also shows feminine agreement), and occurs as both masculine and feminine in Catalan. In Ladin, there is a semantic difference associated with the gender: the feminine noun means ‘art’ in an artistic sense, while the masculine form has the signification of ‘craft, artisanry’. For the corpus search, the following graphical variants were used: art, arte, artes, arts, artz, ars. Table 8.5 gives an overview of the gender distribution for the occurrences of art; Example (17) gives some occurrences in Old Languedocian, and (18) in Old Provençal. 5 It is common knowledge in historical research on Romance languages that most of the nouns in these languages are derived from the accusative form rather than from any other case (see Rheinfelder 1976b: 18). As both the FEW (1922-2003) and the TLL (2025) list the nominative as the lemme, the etyma in the following are cited in the nominative, despite these forms mostly not being the starting point. <?page no="130"?> 130 8 Gender Doublets in COMETA (2025) Languedocian Provençal num+gen+case num+gen gen M sg nom 0 0 0 5 5 obl 0 5 5 pl nom 0 0 0 0 obl 0 0 0 F sg nom 0 0 0 6 18 obl 4 2 6 pl nom 0 0 0 12 obl 6 6 12 NA sg nom 0 0 0 44 46 obl 28 16 44 pl nom 0 0 0 2 obl 1 1 2 Table 8.5: Gender distribution for art. (17) a. De la nobla art de medicina ‘Of the F noble F art of medicine’ (BL, Harley 7403 (14 th century: f o . 49 r o ., l. 6)) b. nos essenha totas las artz. ‘He/ she/ it teaches us all F the F arts’ (BmT, Cote 2883 (14 th century: f o . 23 r o ., c. 1, l. 22)) (18) a. una art ay trobada a la qual non poyra contrastar ‘I found one F art to which F he will not be able to resist’ (BnF, Français 1049 (1343: f o . 207 v o ., c. 1, l. 26-27)) <?page no="131"?> 8.2 Inherited Nouns from Latin 131 b. E cant teodas o auzi leva sus et ansas malas artz fes venir ‘And when Teodas heard this, he stood up and made them/ him/ her come by the power of his F evil F arts’ (BnF, Français 1049 (1343: f o . 207 r o ., c. 1, l. 26-27)) c. Car sa dis .I. libre que parla del art de cavallaria. ‘For this is said in one/ the first book that speaks of/ treats the M art of cavalry/ chivalry’ (BnF, Français 1049 (1343: f o . 66 r o ., c. 1, l. 18-19)) It is unclear whether Example (18c) is really an occurrence of this noun with masculine gender, because it might just as well be an incident of vowel elision of the article before a noun starting with a vowel. Nevertheless, it seems justified to consider this noun as masculine in appearances like this, as the vowel was not elided in other, clearly feminine occurrences of the same noun in the same text (see Example 18a). However, there are further similar instances within this and other texts (especially in the context del art), but without any other agreement target in dependency of art and no occurrence of art with a preceding, not-elided vowel. For these cases, it is therefore not possible to determine the gender with certainty, which is the reason for these occurrences to be counted as NA. This leads to a low number of clearly identifiable masculine occurrences, all five appearing within two texts from Provence: (BnF, Français 1049 1343) with three masculine and three feminine occurrences, and BmC, Cote 34 (13 th century) with two masculine and two feminine occurrences. These texts are therefore the only ones within COMETA (2025) that show internal gender variation for this noun. Other texts do not show such internal variation with respect to art occurrences, <?page no="132"?> 132 8 Gender Doublets in COMETA (2025) all of them being either feminine or of uncertain gender. It is evident that this noun triggers both masculine and feminine agreement, but due to its onset starting with a vowel, it is not really possible to make statements about any potential correlations between region, scribe, period of time, etc. from the mere data of the COMETA (2025) corpus. In order to give more well-grounded information on the status of art as a gender doublet, further sources are necessary. The entry in Raynouard (1836- 1845) lists it only as a feminine noun without any indications of variation in gender, and it is absent in Levy (1894-1924). This corroborates the tendency in the COMETA (2025) corpus, where art is mainly a feminine noun. In order to get a grasp on potential masculine instances of this noun, the fiches of the dDOM (2025) were searched specifically for masculine occurrences. There are only very few sources listed in the dDOM (2025) in which this noun is indexed as being masculine, most of which are not really masculine occurrences, as they often do not appear with an agreement target, e.g., “Que per art cuid’esser peritz” (Dejeanne 1909: 197; ‘That by art he thinks to be skilled’). The second occurrence of art in this text appears with an agreement target, but gender assignment is nevertheless unclear due to the vowel at the onset: “Cascuns si gart, C’ab aital art M’er a viure o a morir” (Dejeanne 1909: 68; ‘Let everyone beware, for with such ? art, it will be for me to live or to die’). The same is valid for the occurrences “e qui belhs motz lassa e lia de belh art s’es entremes” (Beggiato 1984: 110; ‘And whoever lets go and binds beautiful words with beautiful ? art has interposed himself ’) and “quals los pescados de Masselha de l’art de-l’eyssanega devon tenir” (Gleßgen 1989: 147; ‘Which fish from Marseille caught by the ? art of the trawling net must be kept’), where it is indexed as a masculine noun in the glossary (see Gleßgen 1989: 431). Another instance where it is listed, but in fact does not appear with an agreement target, is the following: “aveie sotilessa et art de trobar bos motz e gais sons” (Boutière/ Schutz 1964: 26; ‘He/ she had subtlety and the art of composing fine words and joyful sounds’). The dDOM (2025) indicates that art would be both masculine and feminine in the two occurrences in Kolsen (1910), but in fact both occurrences are of unclear gender: “Que no lor te pro genhs ni artz” (Kolsen 1910: 62; ‘That they do not have enough people nor skills’) and “Ab art et ab fals genh me pren” (Kolsen 1910: 174; ‘With skill and with deceitful folk, he/ she/ they take(s) me’). It would be more justified to list these occurrences as unclear with respect to gender rather than indicating a gender variation. Some masculine occurrences can be found in the literature, for instance “la Ciutat subre los arts de pescar” (Fourquin/ Rigaud 1994: s.v. “Art de pescar s.m.pl.”; ‘The city regarding the M fishing arts’). In the DAOA (2009: s.v. “arts, ars”), one example of masculine agreement can be found in the fixed expression “bache- <?page no="133"?> 8.2 Inherited Nouns from Latin 133 lier es artz” ‘Bachelor of (the M ) Arts’, where es is the contracted form of the preposition en combined with the masculine singular article los. In the glossary of Hackett (1955: s.v. “art”), this noun is indicated as both masculine and feminine, with only one indication of feminine and 6 indications of masculine use. “E fu desus cuberte per art tapine” (Hackett 1953a: 48; ‘And he was covered over by cunning F art’), where the feminine adjective tapine ‘cunning’ as agreement target is the only feminine occurrence listed here. This occurrence has to be considered carefully, given that it appears in a poem and this adjective has to rhyme with sabine (a region in central Italy, the ‘Sabina’) and sarrazine (a term associated with the Muslims in the Middle Ages), so the ending in -e might also be there to keep the rhyme scheme. With respect to the masculine indications, they can mostly be confirmed as such 6 , e.g., “del felon art” (Hackett 1953a: 176; ‘Of the M treacherous M art’) or “porie caler cauz fust li arz” (Hackett 1953b: 246; ‘It could be necessary because it was the M arts’). All other sources in the dDOM (2025) list art as a feminine noun, which can be confirmed for most of the instances, although some also show no gender at all. In general, it can be said that art is in fact a gender doublet, with a strong tendency of being used as a feminine noun. 8.2.2.3 canal This noun with the meanings ‘canal, gutter, tap’ both in Latin and Old Occitan stems from the third declension noun canalis, which already triggered masculine, feminine, and neuter agreement in Latin, which is maintained in Medieval Latin. This noun is inherited into all Romance languages. It is masculine in French, Friulian, Sardinian, and in southern Italian varieties, but feminine in Aragonese, Franco-Provençal, Ladin and northern Italian varieties. It appears both with masculine and feminine agreement in Asturian, Catalan, Galician, Romansh, and Spanish, and belongs to the Genus Alternans in Romanian. For the corpus search, the following graphical variants were used: caal, canal, canals, canalis, canau, chanal, chanau, chenal, chenau, chinau, cinal, and quanal. Languedocian Provençal num+gen+case num+gen gen F sg nom 0 0 0 1 1 obl 1 0 1 pl nom 0 0 0 0 obl 0 0 0 Table 8.6: Gender distribution for canal. 6 Only two instances cannot be seen as masculine instances with certainty due to the onset, where it is not clear whether the vowel might have been elided; both occurrences are “per negun art” (Hackett 1953a: 163, 193; ‘By no ? art’). <?page no="134"?> 134 8 Gender Doublets in COMETA (2025) (19) del mur en la canal ‘from the wall into the F canal’ (BnF, Français 25425 (1275: f o . 62 r o ., l. 40)) Within COMETA (2025) only one example of this noun could be found (see Example 19). This occurrence is feminine, but due to the lack of more data, no valid statement on this noun as a gender doublet can be made based on this data, which agrees with the entries in Raynouard (1836-1845: s.v. “canal”) and Levy (1894-1924: s.v. “canal”), who both list it as only being feminine. In the dDOM (2025), there are a few hints toward masculine occurrences. Two examples from two different poems by Marcabru are cited as masculine, which are actually not clearly attributable to a certain gender, as they appear without an agreement target but with the preposition en without an article or any other potential agreement target: “Jois fora tombatz en canau” (Dejeanne 1909: 91; ‘Joy fell outside into the ? canal’) and “E l’orgoills torn’en canau Per garssos plens de grondill” (Dejeanne 1909: 185; ‘And pride flows back through the ? canal for boys full of reprimands’). There are just two masculine occurrences cited in Fourquin/ Rigaud (1994: s.v. “canal”) which actually show masculine agreement (“lo canal” ‘The M canal’ and “los canals” ‘The M canals’). Despite some variation in gender assignment, it seems to be only minimal and the status of a gender doublet is questionable for this noun. 8.2.2.4 color This noun is derived from the Latin masculine third declension noun colouR with the meaning ‘color’. In Old Occitan, this noun has several more meanings such as ‘glory, complexion’ or ‘opinion’. Inherited forms of this noun exist in all Romance languages except Romanian. It is masculine in Corsican, Friulian, Italian, Ladin, Medieval Latin, and Spanish, feminine in Franco-Provençal, French, Occitan, and Romansh, and shows both agreements in Aragonese, Asturian, Catalan, and Galician. For the corpus search, the following graphical variants were used: collour, color, colore, colors, colos, and qolors. Many occurrences (about one third) of this noun are not with certainty attributable to a specific gender and remain under the category NA. The other occurrences, both singular and plural, and in both regions, can be identified as <?page no="135"?> 8.2 Inherited Nouns from Latin 135 Languedocian Provençal num+gen+case num+gen gen F sg nom 0 0 0 30 43 obl 10 20 30 pl nom 0 0 0 13 obl 1 12 13 NA sg nom 0 0 0 19 24 obl 8 11 19 pl nom 0 0 0 5 obl 4 1 5 Table 8.7: Gender distribution for color. feminine. Occurrences in the Languedocian area are given in the singular in Example (20a) and plural in Example (20b). In the Provençal area, see Example (21a) for the singular and Example (21b) for the plural. It appears in six of the Old Languedocian and six of the Old Provençal manuscripts in COMETA (2025). (20) a. e la sua cara remas de color vermelha ‘and his face remained of vermilion F colour’ (BnF, Français 13504 (14 th century: f o . 25 r o ., l. 29)) b. E donzels ab belas colors ‘And damsels in beautiful F colours’ (BL, Harley 7403 (14 th century: f o . 55 v o ., l. 9)) (21) a. Li viva colors de sa fassa ‘The F vivid F colour of his face’ (BmC, Cote 34 (13 th century: f o . 10 v o ., l. 1)) <?page no="136"?> 136 8 Gender Doublets in COMETA (2025) b. Obradas de mantas colors ‘Works of several F colours’ (BmC, Cote 34 (13 th century: f o . 101 r o ., l. 18)) Color does not appear as a gender doublet in COMETA (2025), as it always triggers feminine agreement in both linguistic areas. Examples (20) and (21) show the agreement in the Languedoc and the Provence respectively. This is coherent with the findings of Raynouard (1836-1845: s.v. “color”) and Levy (1894-1924: s.v. “color”), who list this noun only as feminine. Nevertheless, some masculine occurrences of this noun can be found in the literature. The dDOM (2025) lists 9 sources for masculine occurrences, only one of which can be confirmed: “es de color blanc” (Gorosch 1969: 32; ‘It is white M in colour’). A feminine occurrence of color is found in the same text: “deu esse de meiana color entre negre et vert” (Gorosch 1969: 38; ‘It must be of a medium F colour between black and green’). The other sources indicate a masculine gender for this noun, but a closer look reveals that these occurrences cannot be assigned with certainty to one or another gender, e.g., “Sa pel chanjo de collour” (Sibille 2007: 184; ‘Its skin changes colour’); “han saurs pels en color, so liberals de cor” (Ventura 2010: 12; ‘They have golden-coloured hair, they are generous of heart’); Marshall (1969b: 228, 392) lists it in his glossary as “colors 2823, sb. f. and m., colour”, but the respective occurrence in the text appears without any agreement target. Others are in reality feminine occurrences, e.g., “que chantan de manhtas colors” (Zenker 1900: 111; ‘Who sing in many F colours’). In his glossary, he lists color as a masculine noun with reference to this verse, but without indicating a reason for why he identifies it as a masculine noun. The same applies for Zemp (1978: 403), who lists color as a masculine noun in his glossary, but the two occurrences in the chansons of Cadenet are both feminine: “Quan vostra fresca color” (Zemp 1978: 138; ‘When your F cold F colour’) and “E sa colors plus fresca” (Zemp 1978: 305; ‘And its F coldest F colours’). Similarly, the occurrences with identifiable gender in Ventura (2010) are feminine, e.g., “Ethiopia pren le nom de la color del poble” (Ventura 2010: 38; ‘Ethiopia takes the name from the F colour of the people’) or “ja sia que · ls Maurs haio per sa color aquel nom” (Ventura 2010: 78; ‘Indeed, those Maurs have that name because of their F colour’). Overall, only one de facto masculine occurrence could be found within the Old Occitan literature. It is therefore questionable to see color as a gender doublet in Old Occitan, and rather likely that the one instance of masculine use has other reasons, such as a writing error. <?page no="137"?> 8.2 Inherited Nouns from Latin 137 8.2.2.5 crin Crin, derived from the Latin third declension noun cRinis ‘hair’, later also ‘animal hair’, was predominantly masculine but also existed as a feminine noun. In Old Occitan, it has the signification of ‘women’s hair, horse hair, mountain crest’. It lost the meaning of ‘human hair’, but is still used for ‘animal hair’ in the Romance languages. It is masculine in most Romance languages, but currently also appears feminine in Aragonese, Catalan, Ladin, and Spanish. For the corpus search, the following graphical variants were used: cri, crin, crine, crins, and cris. Languedocian Provençal num+gen+case num+gen gen F sg nom 0 0 0 1 4 obl 1 0 1 pl nom 0 0 0 3 obl 3 0 3 Table 8.8: Gender distribution for crin. (22) a. Le capellans sa crin li tol ‘The sacristan cuts his F / her F / their F hair’ (BmC, Cote 34 (13 th century: f o . 62 v o ., l. 9)) b. Sas belas crins luzens e claras ‘Her F beautiful F , shiny and fair F hair’ (BmC, Cote 34 (13 th century: f o . 20 r o ., l. 17)) c. E com tira sas sauras crins ‘And as he/ she/ they pull(s) his F / her F / their F golden F hair’ (BL, Additional 10323 (14 th century: f o . 60 v o ., c. 1, l. 23)) This noun only appears with feminine agreement in COMETA (2025), as the Examples in (22) show. Raynouard (1836-1845) lists it as a masculine noun but <?page no="138"?> 138 8 Gender Doublets in COMETA (2025) only demonstrates examples with feminine agreement, which is criticised by Levy (1894-1924), who specifies that it exists in Old Occitan only as a feminine noun. In the dDOM (2025), only one source for the masculine agreement is indicated. A closer look at that source reveals that in reality this occurrence (“ab saura cri” Dejeanne 1909: 72; ‘with golden F hair/ fur’) shows a feminine agreement, so this noun de facto only seems to be feminine in Old Occitan. 8.2.2.6 den Derived from the Latin third declension masculine noun dens ‘tooth’, the Old Occitan den means ‘tooth, crag, crest’. The TLL (2025) indicates a potential feminine occurrence, but leaves it with a question mark, so it is unclear whether it also appeared as a feminine noun already in Latin. This noun is preserved in all Romance languages and mainly stayed masculine; only in Gallo-Romance (Franco-Provençal, French, and Occitan), in Catalan, and in Sardinian (but not in Gallurese and Sassarese, where it stayed masculine (see Wagner 1930: 77)) did the gender change and the noun triggers feminine gender. In Aragonese, it is mainly masculine, but appears also as a feminine noun. For the corpus search, the following graphical variants were used: dans, denç, deins, den, dencz, dent, dens, dent, dentz, and denz. As Table 8.9 shows, den only appears as a feminine noun in the COMETA (2025) corpus. Languedocian Provençal num+gen+case num+gen gen F sg nom 0 0 0 5 23 obl 5 0 5 pl nom 0 0 0 18 obl 11 7 18 NA sg nom 0 0 0 4 7 obl 4 0 4 pl nom 0 0 0 3 obl 1 2 3 Table 8.9: Gender distribution for den. The examples in (23) show the agreement in Languedoc, the examples in (24) in the Provence region. This noun can be found in six of the Old Languedocian manuscripts and in three of the Old Provençal manuscripts within COMETA (2025). <?page no="139"?> 8.2 Inherited Nouns from Latin 139 (23) a. et aqui pauseron tropas de reliquias. so es assaber del sanc de sant. Esteve et .Ia. dent de santa columba ‘and here they put (too) many relics. This is to be known of the blood of saint Esteve and one F / the first F tooth of saint Columba’ (BL, Additional 21218 (14 th century: f o . 10 v o ., l. 16-18)) b. La boca la aia eissamentz Que plus belas n auras las dens ‘The mouth would be the same (way), for you will not have (the F ) teeth more beautiful F ’ (BL, Harley 7403 (14 th century: f o . 50 v o ., l. 11-12)) (24) a. Totas son entorn luy e martellan las dens ‘They all are around him and hammer the F teeth’ (BL, Additional 10323 (14 th century: f o . 20 v o ., l. 15)) b. Adoncas fai un ioc cani Que las dens mostra e non ri ‘Then he makes a doglike face showing the F teeth without laughing’ (BmC, Cote 34 (13 th century: f o . 19 r o ., l. 22-23)) Raynouard (1836-1845: s.v. “den”; small caps in the original) lists den as a masculine and feminine noun, and gives one example of masculine agreement: “Leva la man, fer ne l’als dens” ‘Raise your hand, to hit them in the M teeth’. The dDOM <?page no="140"?> 140 8 Gender Doublets in COMETA (2025) (2025) supplements 6 additional sources of den as a masculine noun, where a few can be disproved, e.g., “sieu pels ardien coma fuoc; sas dens de luy eron d’acier” (Manetti 2012: 60-62; ‘His hair burned like fire; his F teeth were of steel’), which is listed as “[dent] s.m. ‘dente’ c.s.pl. dens 513” (Manetti 2012: 126) in the glossary and taken as such in the dDOM (2025). The second source lists it in the glossary as “dens 2099, sb.f. and m., tooth”, but the respective occurrence in the text appears without any agreement target (Marshall 1969b: 228, 392) 7 . The next occurrence is “a caps tor a las dans” ‘He has the head around the F teeth’ which is clearly feminine (Sweetenham/ Paterson 2003: 228), but is indicated in the glossary as “dans (? ), sb.m. ‘teeth’, 567 [r]” (Sweetenham/ Paterson 2003: 303) without an explanation for why this should be a masculine noun in this context. The next source is the translation of the New Testament, where Wunderli (2010: 150; bold and italics in the original) specifies: “dent (=den, Vulg. dens) n.m./ f. ‘dent’ dent per dent Mt. 5/ 38, dentz Mt. 8/ 12, 13/ 42, 13/ 50, 22/ 13, 14/ 51…, denz Mc. 9/ 17”. But all the occurrences of this noun are either of uncertain gender (e.g., “on sera plors et estrenhementz de dentz” (Wunderli 2009: 54; ‘There will be tears and clenching of teeth’)) or feminine (“et estrenh las denz” (Wunderli 2009: 126; ‘And clenches the F teeth’)). The last source where the state of den as a masculine noun can be disproved is Lachin (2004: 615), where the glossary specifies “s.m.: dente III, 48”. The respective occurrence in the text does not show any agreement target and the gender can therefore not be known with certainty: “non intret glirs en boca ni en den” (Lachin 2004: 174; ‘No dormice enters mouth nor tooth’). There is one additional source, a text from the 14 th century, in which the noun really does appear as a gender doublet, appearing as both a masculine and a feminine noun within the same text: “E a las dens grans come ferre […] O el pendrie un deins dal cays […] Che un dent dal cais li arabes” (van der Horst 1974: 124; ‘And at the F teeth great like iron […] Or he would take a M tooth of the dog […] Because a M tooth of the dog scratches him.’) 8 . After this literature review, the general findings within the COMETA (2025) corpus can be supported and it is clear that den does not show many instances of gender variation in Old Occitan. It seems the gender was already fixed by this time. 7 It is the same as we have seen before in Section 8.2.2.4. 8 This quotation might suggest that den is a noun that could qualify for the category of Genus Alternans, but another occurrence reveals that the noun also triggers feminine agreement in the singular in this text: “E avisset sobre la dent” (van der Horst 1974: 127; ‘And had upon the F tooth’). <?page no="141"?> 8.2 Inherited Nouns from Latin 141 8.2.2.7 fam The Old Occitan noun fam ‘hunger, famine’ is inherited from the feminine Latin third declension noun fames. It is often inherited into the Romance languages with two different descendants with a semantic differentiation (e.g., French faim and Spanish hambre, both ‘hunger’ and French famine or Spanish hambruna, both ‘famine’). This noun stayed feminine in all Romance varieties except Sardinian, where it is masculine in most varieties and a gender doublet in Logudorese. Apart from Sardinian, the only indication of this noun showing another gender is found in the DOMél (2025), where it is listed as both masculine and feminine, which will be analysed in the following. For the corpus search, the following graphical variants were used: faim, fam, fame, fami, famine, famne, fams, fan, and fans. Languedocian Provençal num+gen+case num+gen gen F sg nom 0 0 0 0 1 obl 0 0 0 pl nom 0 0 0 1 obl 0 1 1 NA sg nom 0 0 0 59 59 obl 18 41 59 pl nom 0 0 0 0 obl 0 0 0 Table 8.10: Gender distribution for fam. (25) Et ennay suffert motas fams e motas fretz eyssament ‘And from that he has suffered great F hunger and great cold equally’ (BnF, NAF 6195 (14 th century: f o . 12 r o ., l. 17)) As Table 8.10 shows, only one occurrence of fam in COMETA (2025) shows a clearly identifiable gender: it appears as feminine plural (see Example 25). In accordance with this finding, Raynouard (1836-1845) lists fam as a feminine noun, but none of his examples indicate a gender for this noun. Criticising this fact, Levy (1894-1924) gives some clearly feminine and masculine occurrences. The feminine occurrences are “E tanta fam”; “E las fams” (Levy 1894-1924: s.v. “Fam”; italics in the original; ‘And so much F hunger’; ‘And the F famines’); the masculine occurrences are “E dis li, sabens lo fam sieu: Diguas, si tu es filh de Dieu”; “totz los fams” (Levy 1894-1924: s.v. “Fam”, italics in the original; ‘And he said to him, knowing his M hunger: Say, if you are the Son of God’; ‘All the M hunger’). <?page no="142"?> 142 8 Gender Doublets in COMETA (2025) As the state of fam as a feminine noun is not disputed, yet it is unclear whether it also appears as a masculine noun, the focus in the following will be on masculine occurrences. Most of the sources in the dDOM (2025) list it as a feminine noun. With regard to the first occurrence indicated by Levy (1894-1924), taken from the edition of the Breviari d’amors by Azaïs (1862), a note in the dDOM (2025) specifies: “il est dans ce vers du genre masculin, sans doute pour faire rimer l’adjectif possessif, SIEU avec DIEU du v[ers; MW] suivant”. So it appears as a masculine noun, but the grammatical value of masculine for this noun in this context seems doubtful. The second example of a masculine occurrence by Levy (1894-1924) is, according to him, taken from the manuscript BnF, Français 1049 (1343), which is part of the COMETA (2025) corpus. However, this occurrence could not be found in that text, neither on the indicated folio nor with a full-text search; all occurrences of fam in that text are of uncertain gender. This evidence from Levy (1894-1924) therefore cannot be considered as proof of fam being also masculine. The dDOM (2025) lists 9 additional sources indicating fam as either a masculine or an variable noun. Again, most of these sources can be falsified as masculine occurrences, as they have no agreement target and therefore cannot be categorised with a certain gender: e.g., Marshall (1969b: 191; 401) lists it as a masculine and feminine noun, while the occurrence itself shows no sign of gender, as there is no agreement target (there is only “fams” (Marshall 1969b: 191; ‘hunger’)). Zenker (1900: 245) lists it as only a masculine noun with three occurrences in his edition, while all occurrences are of uncertain gender: “d’aisso don ac fam” (Zenker 1900: 89; ‘of that which he had hunger for’), “qu’en sa cort on sui vengutz es fams e vera mermansa” (Zenker 1900: 109; ‘for in his court where I have come are hunger and true poverty’), and “lo renh on nulh set ni fam ni freg non a ni esmai” (Zenker 1900: 133; ‘the kingdom where there is no thirst, nor hunger, nor cold, nor fear’). The same applies for Hündgen (1884: 47), where it is listed in the glossary as “subst. m. obl. sg.”, but in the text itself, it appears as “e qui nos pais, que no murem de fam” (Hündgen 1884: 2; ‘and may he nourish us, so that we do not die of hunger’), showing no sign of gender. It is the same case for the next source, where the entry in the glossary states “fam m.s. ‘fame’ c.9vb, c.10ra, c.14vb” (Piccat 2001: 169), but in fact all occurrences show no sign of gender, see e.g., “calors, freytz, fam, set, e anxietatz” (Piccat 2001: 112; ‘heat, cold, hunger, thirst, and anxieties’). The next example indeed appears with an agreement target, but only with the adjective gran which indicates number but not gender: “grans fams aures am frejas bizas totas ensems” (Pulega 1983: 303; ‘great ? hungers, winds, with cold blasts, all together’). <?page no="143"?> 8.2 Inherited Nouns from Latin 143 Only two instances of masculine agreement can be found in the literature: “tantost el mori pel frech que ac agut e pel fam” (Tausend 1995: 155; ‘soon he died from the cold he had suffered and from (the M ) hunger’) is one example. The other example can be found in a text from the 16 th century, later than the data from COMETA (2025): “Dal fan que ero en my” (Royer/ Duraffour 1928: 83; ‘Of the M hunger that was in me’). The data from COMETA (2025) in addition to the occurrences from the literature confirm that fam is in reality no gender doublet, but only a feminine noun. There are just two occurrences where it even shows masculine agreement, but in one of the contexts it is used as a masculine noun due to the rhyme scheme and not due to gender variability. 8.2.2.8 fornatz This noun with the signification ‘furnace’ is derived from Latin foRnax, a feminine noun that in rare occasions also appeared as a masculine noun. It is a masculine noun in Aragonese, Catalan, Occitan and Romansh, feminine in French, Franco-Provençal, Italian, and Spanish, and it has both genders in Galician. For the corpus search, the following graphical variants were used: affornas, forna, fornaise, fornas, fornatz, fornays, fornaz. Languedocian Provençal num+gen+case num+gen gen M sg nom 0 0 0 1 1 obl 0 1 1 pl nom 0 0 0 0 obl 0 0 0 Table 8.11: Gender distribution for fornatz. (26) lo deneia aysi com fa le fuox l aur que la fina en lo fornatz. ‘purified it like the fire does with the gold that is refined in the M furnace’ (BnF, Français 1049 (1343: f o . 79 v o ., c. 1, l. 3-6)) <?page no="144"?> 144 8 Gender Doublets in COMETA (2025) There is only one occurrence of fornatz in the COMETA (2025) corpus, where the noun triggers masculine agreement (see Table 8.11, Example 26). Contrary to this, Raynouard (1836-1845) lists it exclusively as a feminine noun, giving only one example with certain gender: “en la foRnatz” (Raynouard 1836-1845: s.v. “fornatz”; small caps in the original; ‘In the F furnace’). This classification as only a feminine noun is criticised by Levy (1894-1924), who additionally lists it as a masculine noun, giving two examples: “antre tera he arena e rama per lo fornas”; “quant fasian lo fornas” (Levy 1894-1924: s.v. “Fornatz”; italics in the original; ‘between earth there is sand and branches for the M furnace’; ‘When they built the M furnace’). The dDOM (2025) lists no additional examples for this noun triggering masculine agreement, and all sources but Levy (1894-1924) list it only as a feminine noun in Old Occitan. The two citations in Levy (1894-1924) are both from the same document (edited by Meyer 1898a) written in the 15 th century in the Basses- Alpes, more precisely in Seyne-les-Alpes. Apart from these two occurrences and the one taken from the COMETA (2025) corpus, no more masculine instances could be found. 8.2.2.9 glan/ aglan Derived from the Latin feminine third declension noun glans ‘acorn, mast’, inherited forms of this noun in Old Occitan (glans and aglans 9 ) have the same meanings as in Latin with the addition of ‘gland’. This noun is inherited into all Romance languages (but only rarely used in Spanish, where it is mostly replaced by bellota). This noun is feminine in most of the Romance languages, and only in Aragonese, Ladin, Italian, Occitan (excluding Gascon, where it has both genders, masculine when referring to the acorn, feminine referring to the gland) and Spanish is it used as a masculine noun. In Catalan, two different forms are inherited from the Latin etymon, one with both genders and the meaning of ‘acorn’ ((a)gla(n)) and a masculine form (gland) with the meaning ‘gland’. For the corpus search, the following graphical variants (also without the wordinitial <a->) were used: aglan, agland, aglann, aglans, and aglant. As Table 8.12 shows, this noun has only 6 occurrences within COMETA (2025), most of which are feminine (see Examples (27a and 27b) for Languedocian and Example (28) for Provençal). There is only one occurrence of this noun as masculine (see Example (27c)). It is possible to identify this as a masculine occurrence even though the only agreement target is the Roman numeral “.I.”, because within this specific manuscript BnF, Français 2232 (1325), the scribe differenti- 9 This form is derived from the primary form glan with an agglutinated vowel in the beginning, taken from the feminine article (see DOMél 2025: s.v. “aglan”). <?page no="145"?> 8.2 Inherited Nouns from Latin 145 ates the numbers according to the gender of the respective noun, using “.I a .” for feminine nouns. Languedocian Provençal num+gen+case num+gen gen M sg nom 0 0 0 1 1 obl 1 0 1 pl nom 0 0 0 0 obl 0 0 0 F sg nom 0 0 0 3 4 obl 3 0 3 pl nom 0 0 0 1 obl 0 1 1 NA sg nom 0 0 0 1 1 obl 0 1 1 pl nom 0 0 0 0 obl 0 0 0 Table 8.12: Gender distribution for glan/ aglan. (27) a. aysxi co si fos .I a . aglan ‘as if it were one F gland/ acorn’ (BL, Additional 21218 (14 th century: f o . 34 r o ., l. 30)) b. Que valha .Ia. glan ‘That had the value of one F acorn’ (BnF, Français 25425 (1275: f o . 27, l. 18)) c. Baomet no valh .I. aglan pus que aquestz homes ‘Baomet is not worth one M acorn more than these men/ humans’ (BnF, Français 2232 (1325: f o . 53 v o ., l. 14-15)) <?page no="146"?> 146 8 Gender Doublets in COMETA (2025) (28) vengutz aytant que neys de las glans que li porc maniavan ‘There came so many as are born from the F acorns that the pigs ate’ (BnF, Français 1049 (1343: f o . 189 r o ., c. 2, l. 11-12)) Within COMETA (2025), there are only a few occurrences of glan, so a closer look into other text passages is necessary in order to corroborate the findings in the COMETA (2025) corpus suggesting that it might be a gender doublet in Old Occitan. Raynouard (1836-1845) lists it only as a masculine noun and gives some examples accordingly: “Totz non los preze un glant”; “Ni ’l colp d’un aglan” (Raynouard 1836-1845: s.v. “glan”; small caps in the original; ‘I do not value them all even one M acorn’; ‘Nor the blow of an M acorn’). Apart from these occurrences, it seems difficult to be certain about the gender. Most occurrences appear without agreement targets or just with the article, where it is often unclear whether it is the form aglan with elision of the vowel in the article, or the form glan with the feminine article and therefore a feminine noun. The DOMél (2025: s.v. “aglan”), for instance, lists the following occurrence, taken from Brunel (1957: 314), as a masculine occurrence: “mais de l’escorsa de l’aglant met tant com dels dos autres” ‘but of the shell of the ? acorn he puts as much as of the other two’. It is just as well possible that it is actually the feminine form la glant. And even if it really is the form as written in Brunel (1957: 314), it is still unclear which gender this noun has, as it might also be an elision of the <a>. It seems as though glan really is an example of a gender doublet in Old Occitan, even though there are not many instances where the gender of this noun is certain. 8.2.2.10 labor The Old Occitan noun labor with the meanings ‘(field of) work, labour, harvest’, and ‘benefit’ is inherited either from the Latin masculine noun laboR ‘labour, exertion, effort, work, task, struggle, suffering, (di)stress, preoccupation’ from the third declension or from the Latin verb laboRaRe ‘to work, to labour, to produce, to be troubled, to be in distress, to be oppressed’. Because the DOMél (2025) lists both the noun and the verb as a potential etymon, information is given on gender in the Romance languages for nouns inherited from both the noun and the <?page no="147"?> 8.2 Inherited Nouns from Latin 147 verb. It is feminine in Aragonese, contemporary Catalan, Old French, and Spanish, and masculine in Old Catalan, French, Galician, Italian, Ladin, Portuguese, and Romansh; it is not inherited to Romanian. For the corpus search, the following graphical variants were used: labor, labors, labour, laor, laour, lavor, and lavour. Languedocian Provençal num+gen+case num+gen gen F sg nom 0 0 0 3 4 obl 3 0 3 pl nom 0 0 0 1 obl 1 0 1 NA sg nom 0 0 0 1 1 obl 0 1 1 pl nom 0 0 0 0 obl 0 0 0 Table 8.13: Gender distribution for labor. (29) a. per las vinhas destruire. e per l autra labor ‘to destroy the vineyard. and for the other F work’ (BnF, Français 25425 (1275: f o . 193, l. 40)) b. Tu que maniaras las labors de tas mas bonazuratz yest. ‘You who you will eat by the F work of your hands are blessed’ (BmT, Cote 2883 (14 th century: f o . 52 r o , c. 2, l. 17-19)) There are only four instances of labor in the COMETA (2025) corpus (see Table 8.13), three of which are feminine (see examples in (29)), and the other one is of uncertain gender. Raynouard (1836-1845) lists it exclusively as a masculine noun, but all his examples do not clearly show gender, as they all lack agreement targets. Levy (1894-1924) on the other hand lists it only as a feminine noun, which matches the findings in COMETA (2025), but he does not give any examples. <?page no="148"?> 148 8 Gender Doublets in COMETA (2025) Interestingly, most sources in the dDOM (2025), indicate labor as a masculine noun, but only five of the sources can be verified as really displaying labor as a masculine noun: “segon le seu labor” (Wunderli 2009: 415; ‘According to his M / her M work’), “Roma, mal labor fa · l papa” (Hamlin/ Ricketts/ Hathaway 1985: 216; ‘Rome, the pope does bad M work’), “C’ab son labor ditz qu’er pus tost manens” (Guida 1963: 239; ‘With his M / her M work he/ she says that he/ she is rather rich’), “e comandan a bastir lo labor” (Minervini 2010: 68; ‘and they order to build the M work’), and “ton labor e · ls fruz que li terra ge” (Wahl 1980: 142; ‘your M work and the fruits that the land yields’). In the latter text, a feminine occurrence could also be found: “quela labor non aura ja frut” (Wahl 1980: 142; ‘that F labor will no longer bear fruit’). All other indications do not really display gender, as the following examples, listed as masculine evidence (or as masculine and feminine nouns), show: “per que blasma l’autrui labor? ” (Appel 1930: 73; ‘Why does he/ she blame another’s ? work? ’) with autrui as the only potential agreement target. It is in fact an invariable adjective and therefore does not give any information on gender (see Jensen 1976: 21; Raynouard 1836-1845: s.v. “autrui”). The exact same applies for the next indication “Qu’eu no m’esfortz d’altre labor” (Kolsen 1910: 228; ‘That I do not exert myself in other ? work’), where once again the adjective is invariable and therefore does not indicate a gender value. The examples in Bartsch/ Koschwitz (1904: 117) “qu’ab sobresforsiu labor” ‘that with excessive ? work’ and “terra que ses labor grana” ‘Land that grows without work’ again do not show a clear gender assignment; the first example appears with an invariable adjective, the second one without any agreement target: ses is not a pronoun but the preposition with the meaning ‘without’ and grana ‘grow’ is a verb. Additionally, there are some indications of labor as a feminine noun, which corroborate the findings in the COMETA (2025) corpus, even though the following indication does not display a feminine occurrence: “Car trop es grieu qui perlonga labor” (Ernst 1930: 323; ‘For too burdensome is that which prolongs labour’), where no agreement target agrees with labor, as perlonga is a verb with the meaning ‘to prolong’, so the gender remains unclear. There are many more examples showing that labor is indexed as a masculine noun without any legitimate reason in the respective text passages. This noun does show variation in gender assignment, but only two occurrences of this noun as a masculine noun and only three as a feminine one could be found. All other evidence shows it without clear signs of gender. It is noteworthy that many editors seem to take the masculine gender for granted for labor, even though it is often unclear to which gender it belongs. <?page no="149"?> 8.2 Inherited Nouns from Latin 149 8.2.2.11 lach The Old Occitan noun lach ‘milk’ is inherited from the Latin third declension noun lac ‘milk, milky juice of plants, spat, spawn’, which was neuter in Latin. It is masculine in Corsican, Friulian, Franco-Provençal, Galician, Italian, Ladin, Portuguese, Romansh, and Sardinian, feminine in Aragonese, Catalan, Gascon, and Spanish, and varies between masculine and the Genus Alternans in Romanian. The data in VerbaAlpina (2014-2023) indicates the occurrences of this noun with the same morpho-lexical type (lait/ latte) mainly as masculine (648 records vs. 17 feminine indications), which corroborates the information from the dictionaries of the Romance languages spoken in the Alps, which all indicate it as a masculine noun. For the corpus search, the following graphical variants were used: lac, lach, lag, lagh, lags, lagz, lah, laich, laid, lait, laiz, lat, layt, let, and leyt. Languedocian Provençal num+gen+case num+gen gen NA sg nom 0 0 0 5 5 obl 4 1 5 pl nom 0 0 0 0 obl 0 0 0 Table 8.14: Gender distribution for lach. (30) De layt verge noyritz per la vergena mayre ‘From the virgin milk nourished by the virgin mother’ (BmT, Cote 2883 (14 th century: f o . 84 v o ., c. 1, l. 9-10)) For this noun only five occurrences could be retrieved from COMETA (2025), all of which cannot be assigned to a specific gender due to the lack of agreement targets. Raynouard (1836-1845) lists it as both a masculine and a feminine noun, but gives only one example where the gender is explicit: “la layt de cabra” (Raynouard 1836-1845: s.v. “lach”; small caps in the original; ‘The F milk of a goat, Goat’s milk’). Here, layt is a feminine noun. Levy (1894-1924) lists it only as feminine, giving two further examples: “Ni qui popet de bona lait” and “Cum appar els efans popans mala layt” (both examples from Levy 1894-1924: s.v. “Lach”; italics in the original; ‘Nor who drinks good F milk’; ‘The way the children appear that drink bad F milk’). <?page no="150"?> 150 8 Gender Doublets in COMETA (2025) The dDOM (2025) gives many sources where lach appears as a masculine noun, most of which can be confirmed as masculine evidence: “noyrit de sson lagh” (Appel 1930: 186; ‘nourished by his M / her M / its M milk’) or “cobezejatz le lait que en lui crescatz en salut” (Wunderli 2009: 366; ‘Desire the M milk, so that by it you may grow into salvation salvation’). In the DAOA (2009), it is also listed as a masculine noun, but only one of the examples given really shows gender agreement “per lo layt” (DAOA 2009: s.v. “lait”; ‘Through the M milk’). The other examples appear in a list of nouns where no agreement target is included. Another example for lach being a masculine noun is “es facha lach blanc e dous” (Gosman 1982: 523; ‘It is made a white M and sweet M milk’). The only feminine indication in the notes in the dDOM (2025) is “Ni qui popet de bona lait” (Meyer 1895: 27; ‘Nor who drinks good F milk’), which is the same as cited by Levy (1894-1924). Other sources in the dDOM (2025) do not clearly display gender regardless of the (mostly masculine) indication given in the glossaries or on the notes in the dDOM (2025). For instance, the following passages do not show a clear gender value: “Aqui ha pro mel e lah e vi, jaciaisso que no hi aia vinhas” (Hershon/ Ricketts 2009: 242; ‘Here is enough honey and milk and wine, even though there are no vineyards’), “en carns, peyshos, boder, layt, formagges et autres bes copioza” (Ventura 2010: 78; ‘in meats, fish, butter, milk, cheeses, and other abundant goods’), or “e de lag de galina” (Vatteroni 2013a: 280; ‘and from hen’s milk’), where it is listed in the glossary as “lag s. m. ‘latte’” (Vatteroni 2013b: 992). There is only very little evidence of lach being a feminine noun in Old Occitan. In total, only three occurrences could be found, whilst a lot of evidence for this noun being masculine could be found. It seems like it is basically a masculine noun. 8.2.2.12 mar The Old Occitan noun mar ‘sea’ is inherited from the neuter Latin third declension noun maRe ‘sea’. Even being a neuter noun, mare showed already considerable gender variation in Classical Latin, triggering also masculine and feminine agreement. This noun still exists in all Romance languages; it is feminine in Catalan (but “modernament també m.” DCVB 2002: s.v. “mar”), in French, in most Occitan varieties (but it has both genders in Gascon), Old Portuguese, Romanian, and Old Spanish. It is masculine in Galician, Italian, Ladin, Sardinian, and contemporary Portuguese, and it triggers both genders in Aragonese, Romansh and contemporary Spanish. For the corpus search, the following graphical variants were used: ma, maa, mar, mare, mars, mer, and mers. <?page no="151"?> 8.2 Inherited Nouns from Latin 151 Languedocian Provençal num+gen+case num+gen gen M sg nom 0 0 0 67 67 obl 0 67 67 pl nom 0 0 0 0 obl 0 0 0 F sg nom 0 0 0 115 115 obl 26 89 115 pl nom 0 0 0 0 obl 0 0 0 NA sg nom 0 0 0 104 104 obl 32 72 104 pl nom 0 0 0 0 obl 0 0 0 Table 8.15: Gender distribution for mar. As Table 8.15 shows, mar only occurs in the singular in COMETA (2025). It is exclusively feminine in Languedocian (see Examples in (31)), but appears as a gender doublet in Provençal (see Examples (32a) and (32b) for feminine agreement and Examples (32c) and (32d) for masculine occurrences in Provençal). This noun appears in seven of the Old Languedocian and six of the Old Provençal manuscripts within COMETA (2025). (31) a. foro en aquelh pla entre narbona e la mar. ‘They were in that plain between Narbonne and the F sea’ (BnF, Français 2232 (1325: f o . 118 v o ., l. 12-13)) b. Dreg a la mar iran bruian ‘Straight into the F sea they will run roaring’ (BL, Harley 7403 (14 th century: f o . 34 r o ., l. 16)) <?page no="152"?> 152 8 Gender Doublets in COMETA (2025) (32) a. Per terra e per mar salada ‘Towards earth and salted F sea’ (BL, Additional 10323 (14 th century: f o . 15 v o ., c. 1, l. 27)) b. Regardan ves la mar en l illa de lerins ‘They look towards the F sea from the island of Lerins’ (BnF, NAF 6195 (14 th century: f o . 32 r o ., l. 24)) c. Regardan ves lo mar en l ila de lerins ‘They look towards the M sea from the island of Lerins’ (BL, Additional 10323 (14 th century: f o . 23 v o ., l. 27)) d. Tant que li malfachor qu eran luegn en lo mar ‘As much as the raiders who were distant on the M sea’ (BnF, Français 13509 (14 th century: f o . 28 v o ., c. 2, l. 9-10)) The findings within COMETA (2025) suggest that mar could be a feminine noun in Languedocian, but a gender doublet in Provençal. In order to corroborate this tendency, a closer look at the citations from the Languedocian linguistic area within the dDOM (2025) is necessary. There are only two instances of mar as a masculine noun in the Languedocian area, i.e., “la tartanas en los mars”; “quar lai ont franh lo mar” (Fourquin/ Rigaud 1994: s.v. “Mar”; ‘The ducts in the M seas’; ‘For there the M sea breaks’). Both quotations are from Montpellier from the mid- 14 th century. All other sources in the dDOM (2025) indicating mar as a masculine or variable noun are from other regions, e.g., “per lo mar ros passeron” (De Stefano <?page no="153"?> 8.2 Inherited Nouns from Latin 153 1909: 16; ‘Through the M rough M sea they will pass’), taken from a text written in the Piemont area in today’s Italy or “Sieutatz e vilas si bastiron per lo mar” (Haupt 2003: 218; ‘Towns and villages were built by the M sea’), from a text written around 1375 by Bertran Boysset, an autograph who composed and wrote his own texts. He is from Arles (Provence) and wrote his texts there. This seems to confirm the tendency found in the COMETA (2025) corpus suggesting that mar is a feminine noun in Old Languedocian, while it is a gender doublet in other parts of the Occitan linguistic area. To corroborate this, some examples, again from a text by Bertran Boysset, that demonstrate impressively the state of mar as a gender doublet in Old Provençal will be cited. For instance, the following occurrences can all be found within one page: “per la riba del mar […] lo Mar Roge […] las 2 gautas del mar […] aysins fes lo mar […] que vaugan per la mar salada” (Portet 2004: 258; ‘Along the shore of the M sea […] the M Red M Sea […] the two mouths (‘coasts’) of the M sea […] thus the M sea acts […] that go toward the F salty F sea’). 8.2.2.13 marme The Latin third declension neuter noun maRmoR ‘marble’ is inherited in Old Occitan with the same meaning. This noun is feminine in some (mostly) southern Old Italian varieties (e.g., Sicilian, Neapolitan), in Old Portuguese and in Romanian. In the other Romance languages and varieties it is a masculine noun. The FEW (1922-2003) explains the transition to the feminine with analogy to the Latin first declension feminine noun petRa ‘rock’; see also Section 3.5. For the corpus search, the following graphical variants were used: mabre, marbre, marmbre, marme, marmie, marmo, marmor, and marmre. Languedocian Provençal num+gen+case num+gen gen M sg nom 0 0 0 3 5 obl 1 2 3 pl nom 0 0 0 2 obl 2 0 2 NA sg nom 0 0 0 8 11 obl 3 5 8 pl nom 0 0 0 3 obl 0 3 3 Table 8.16: Gender distribution for marme. As Table 8.16 shows, only five occurrences of marme could be attributed with certainty to a gender value, all five being masculine. In Languedocian, it appears in both masculine singular (see Example (33a)) and plural (see Example (33b)), and in Provençal only in the singular (see Example (34)). The occurrences of this <?page no="154"?> 154 8 Gender Doublets in COMETA (2025) noun are distributed over two texts in Old Languedocian and four texts in Old Provençal, but it only shows a clear gender value in one of these four texts. (33) a. E mostrec as els la columpna del marme ‘And he showed them the M marble column’ (BL, Additional 21218 (14 th century: f o . 24 r o , l. 8-9)) b. que nos cove que anem alhs marme ‘That it conveys us that we go to the M marble’ (BnF, Français 2232 (1325: f o . 5 r o , l. 4-5)) (34) e fon pauzatz am gran procession el sepulcre del marme. lo qual le rics homs avia fach per ella ‘And she was put to rest with a great procession in the M marble tomb which the rich man had made for her’ (BnF, Français 13503 (13 th / 14 th century: f o . 80 r o , l. 7-10)) As this noun has only masculine or unidentifiable occurrences within COMETA (2025), it cannot be verified as a gender doublet in Old Occitan with only this corpus. Therefore, other occurrences in the literature were searched. Raynouard (1836-1845) lists marme exclusively as a masculine noun, which corresponds to the findings in COMETA (2025). Levy (1894-1924) gives one additional example of marme as a feminine noun: “Addobar fez la marme dura” (Levy 1894-1924: s.v. “marme”; italics in the original; ‘He prepared the F hard F marble’), an example taken from the Chanson de sainte Foi d’Agen. All sources from the dDOM (2025) that list marme as a feminine noun are referring to this chanson. A look into the facsimile of the manuscript reveals that it is in reality evidence for this noun occurring as a feminine noun (see Example 35). <?page no="155"?> 8.2 Inherited Nouns from Latin 155 (35) Addobar fez la marmre dura ‘He prepared/ decorated the F hard F marble stone’ (Thomas (1925: 26)) Here once again it seems questionable to categorise marme as a gender doublet just because of a single occurrence in another gender. It is more likely that this noun is in fact a masculine noun and the scribe of the Chanson de sainte Foi d’Agen used it once as a feminine noun for an unknown reason. I do not think it is justified to consider the noun a gender doublet in this case. 8.2.2.14 mel The Old Occitan noun mel ‘honey’ is inherited from the Latin third declension noun mel with the same meaning, which was a neuter noun in Latin but was also used with masculine agreement. This noun exists in all Romance languages as a hereditary word. It is feminine in Aragonese, Catalan, Friulian, Ladin, Romanian, and Spanish. It is masculine in Corsican, French, Italian, Portuguese, Occitan, and Sardinian, and shows both genders in Galician. For the corpus search, the following graphical variants were used: meill, mel, melh, mels, meou, meu, miel, and mier. Languedocian Provençal num+gen+case num+gen gen M sg nom 1 0 1 7 7 obl 3 3 6 pl nom 0 0 0 0 obl 0 0 0 NA sg nom 0 0 0 11 11 obl 7 4 11 pl nom 0 0 0 0 obl 0 0 0 Table 8.17: Gender distribution for mel. This noun could be found in both areas, but only as a masculine noun or showing no clear gender marking. All occurrences are in the singular due to the fact that mel is a mass noun, a category of nouns that includes most liquids and uncountable substances, such as salt, water, or blood. Mass nouns usually do not indicate number, as they describe the whole mass that cannot be counted, and therefore exist only in the singular (see Corbett 2000: 80-85). The following examples show some occurrences of mel in COMETA (2025), Examples (36) from <?page no="156"?> 156 8 Gender Doublets in COMETA (2025) the Languedocian and Examples (37) from the Provençal areas. It appears in five texts in the Old Languedocian sub-corpus, but only in one of these with an identifiable gender value. It can be found in two of the texts in the Old Provençal part, in both with masculine agreement. (36) a. E cum en la bresca le mels. ‘And as the M honey in the honeycomb’ (BmT, Cote 2883 (14 th century: f o . 15 v o , c. 2, l. 30)) b. E ab lo mel e ab la cera. ‘And with the M honey and with the wax’ (BnF, Français 25425 (1275: f o . 132, l. 40)) (37) Veray humil es aysi con bona abelha que fa lo mel per garnir son ostal. ‘The truly humble is like the bee that makes (the M ) honey to fill its hive’ (BnF, Français 1049 (1343: f o . 96 r o ., c. 2, l. 13-16)) In accordance with the findings in the COMETA (2025) corpus, Raynouard (1836- 1845) lists mel as a masculine noun. Levy (1894-1924) supplements this information with one feminine occurrence: “E nos devem ly per la resta de la mel” (Blanc 1899: 133; italics in the original; ‘And we owe him for the rest of the honey F ’), from Narbonne. There is only one other example of feminine usage of this noun <?page no="157"?> 8.2 Inherited Nouns from Latin 157 in the dDOM (2025), but it is taken from a poem by Guylem de Cervera, a Catalan poet (and the poem is also in Old Catalan): “Mas hom vol noyrir eylas Per amor de la mel” (Thomas 1886: 55; ‘But one wants to nourish them for the love of (the F ) honey’). All other indications of mel being a gender doublet are in fact only instances where the gender cannot be known due to the lack of agreement targets, e.g., “E sal e meill e farina” (Méjean 1971: 94; ‘And salt and honey and flour’). 8.2.2.15 orde The Latin masculine third declension noun oRdo ‘row, order, rank, succession, class, order of monks’ exists in Old Occitan (and in most Romance languages) as both a hereditary and a loanword, with the same meanings and the additional meanings ‘testament, rule of a religious order, mass, sacrement’. It is a masculine noun in Corsican, Italian, Ladin, Occitan, and Spanish, and feminine in Portuguese and Romanian. In Aragonese, Catalan, and Galician it exists with both genders. For the corpus search, the following graphical variants were used: horde, hordre, orde, ordei, ordein, ordeinh, orden, ordeng, ordenh, ordes, ordi, ordie, ordin, ordo, ordre, ordres, ordy, ouorde, ourde, ourdi, ourdoun, ourdre, and ourdun. Languedocian Provençal num+gen+case num+gen gen M sg nom 0 0 0 8 12 obl 4 4 8 pl nom 0 0 0 4 obl 1 3 4 F sg nom 0 4 4 5 9 obl 1 0 1 pl nom 0 0 0 4 obl 4 0 4 NA sg nom 0 0 0 48 51 obl 12 36 48 pl nom 0 0 0 3 obl 3 0 3 Table 8.18: Gender distribution for orde. It is not possible to clearly attribute the majority of the occurrences of orde in COMETA (2025) to a certain gender, which is due to the initial vowel. Many determiners appear in elided form without a vowel before a noun starting with a vowel and therefore do not give any information on the gender value of this noun. The occurrences where gender value can be determined with certainty are more or less equally distributed over both regions, showing both genders in <?page no="158"?> 158 8 Gender Doublets in COMETA (2025) each of the regions. The Examples in (38) show gender distribution in Languedocian, the Examples in (39) for Provençal. The list of examples always starts with feminine singular (see Examples (38a) and (39a)) and plural (see Examples (38b) for Languedocian; it does not appear as a feminine plural noun in Provençal, so there is no example), followed by the respective masculine occurrences (see Examples (38c) and (39b) for the singular and Examples (38d) and (39c) for the plural). The occurrences of orde can be found in six manuscripts in Old Languedocian, but only in two of those with a clear gender value and only in one of those, namely in BmT, Cote 2883 (14 th century), as a gender doublet. In Old Provençal, it appears in five different manuscripts, again showing an identifiable gender in two of the manuscripts and appearing as a gender doublet only in one, in BnF, Français 13503 (13 th / 14 th century). (38) a. quar devon dir tug li orde le propheta. le papa ‘Because they shall say all the F time le propheta. le papa’ (BmT, Cote 2883 (14 th century: f o . 125 r o ., c. 1, l. 15-16)) b. per que se peco cil que dizo totas las ordes ‘Because he sins who always (all F the F time) says’ (BmT, Cote 2883 (14 th century: f o . 125 r o ., c. 1, l. 13-14)) c. sentensa que haviịan en lo primier orde enayssi. ‘sentence that they had in the M first M order that way.’ (BmT, Cote 2883 (14 th century: f o . 97 r o ., c. 1, l. 29-30)) <?page no="159"?> 8.2 Inherited Nouns from Latin 159 d. cum son diversas gens de diverses ordes e de diversas religios ‘in the same way there are different people from different M [religious] orders and from different religions’ (BmT, Cote 2883 (14 th century: f o . 54 v o ., c. 1, l. 7-8)) (39) a. s avisteron li fraire e tut li orde de marsella 10 ‘the brothers and the F whole order of Marseille came together’ (BnF, Français 13503 (13 th / 14 th century: f o . 79 v o ., l. 13-14)) b. E merce dieu mon orde sai ‘And thanks to god I know my M order’ (BmC, Cote 34 (13 th century: f o . 63 r o ., l. 14)) c. car aquilli amors de dieu ques a liat los ordes per forsa de gran regla. ‘Because that love of god that has tied the M orders by the power of the great rule’ (BnF, Français 13503 (13 th / 14 th century: f o . 56 r o ., l. 1-3)) The noun orde could be verified as a gender doublet within both regions. In Languedocian, the variation exists exclusively within one text (BmT, Cote 2883 10 In this example, the article li shows both its functions (see Chapter 4), masculine plural nominative with li fraire as well as feminine singular nominative with li orde. <?page no="160"?> 160 8 Gender Doublets in COMETA (2025) 14 th century). Interestingly enough, this gender variation in Languedocian appears in the Leys d’amors, which was written by grammarians who tried to normalise the language. It has three more feminine occurrences in Languedocian, all within the manuscript BnF, Français 25425 (1275). In Provençal on the other hand, the variation in gender is distributed over more manuscripts within the COMETA (2025) corpus, so the gender variation is also widely spread in Old Provençal. 8.2.2.16 ost The Old Occitan noun ost ‘army, military service’ is inherited from Latin third declension noun hostis ‘enemy’. This noun was mainly masculine in Latin, but also existed as a feminine noun. In the transition from Latin to the Romance languages, a change in semantics occurred, probably via the intermediate signification ‘adversarial or hostile army’ to the neutral term for any army, hostile or not. This noun still exists in most Romance languages, but in modern French only as an archaism in the literature (as a masculine noun). In middle French, this noun still existed as normal term for the army, showing considerable variation in gender, with the tendency of being used as a feminine noun. It is feminine in Aragonese, Old Italian, Portuguese, Romanian, and Spanish, and triggers both genders in Occitan (see Diez 1882: 416; Meyer-Lübke 1894: 425). For the corpus search, the following graphical variants were used: host, ost, oste, osts, ostz, and osz. Languedocian Provençal num+gen+case num+gen gen M sg nom 0 0 0 1 3 obl 1 0 1 pl nom 0 0 0 2 obl 1 1 2 F sg nom 11 0 11 221 228 obl 202 8 210 pl nom 0 0 0 7 obl 2 5 7 NA sg nom 0 0 0 73 79 obl 62 11 73 pl nom 4 0 4 6 obl 1 1 2 Table 8.19: Gender distribution for ost. As shown in Table 8.19, ost is predominantly feminine in the COMETA (2025) corpus. There are in total only three masculine occurrences, two in Languedocian (see Example (40d) for the singular occurrence and Example (40e) for the plural one), both appearing in manuscript BnF, Français 25425 (1275). The third <?page no="161"?> 8.2 Inherited Nouns from Latin 161 masculine occurrence is in Provençal, see Example (41c). The other occurrences are all either feminine or of unclear gender. Most instances of this noun are in the oblique case. There are only 11 occurrences of this noun in the nominative; all of them are feminine singular and in the Languedocian sub-corpus (see Example 40a). The other examples show the noun in the oblique case in singular (see Examples (40b) for Languedocian and (41a) for Provençal) or in the plural (see Examples (40c) for Languedocian and (41b) for Provençal). Even though this noun has an initial vowel, only 79 instances could not be assigned to a gender value with certainty, while 231 could be allocated a gender value. Ost appears in five Old Languedocian and five Old Provençal manuscripts within COMETA (2025). It shows feminine agreement within all of these manuscripts except BmC, Cote 34 (13 th century), where no gender value could be specified. (40) a. Tota la ost ensems. ‘All F the F army together.’ (BnF, Français 25425 (1275: f o . 187, l. 22)) b. yeu soy le plus fortz d esta ost ‘I am the strongest of this F army’ (BmT, Cote 2883 (14 th century: f o . 127 r o ., c. 2, l. 31-32)) c. Dedins a carcassona aqui las osts atent ‘Inside Carcassonne, there he waits for the F armies’ (BnF, Français 25425 (1275: f o . 7 l. 21)) d. Cassanhol asetia lo ostz e a dins mot aquier ‘The M army besieges Cassanhol and inside there are many archers’ (BnF, Français 25425 (1275: f o . 9, l. 3)) <?page no="162"?> 162 8 Gender Doublets in COMETA (2025) e. D aicestatz ostz de sai. nous volh oimais parler ‘Of these M armies here I do not want to talk further’ (BnF, Français 25425 (1275: f o . 9, l. 23)) (41) a. D aissi faran una gran ost ‘From here onward they will make a F great army’ (BL, Harley 7403 (14 th century: f o . 34 r o ., l. 15)) b. E manda sas ostz environ ‘And he send his F armies around’ (BnF, NAF 6195 (14 th century: f o . 20 r o ., c. 2, l. 7)) c. Cant s avistan los ostzt verratz menar desrey ‘When the M armies gather, you will see them cause disorder’ (BL, Additional 10323 (14 th century: f o . 22 r o ., l. 25)) It is clear that the former Latin masculine noun has the tendency of being used as a feminine noun in Old Occitan, as only three instances of this noun with masculine agreement could be found in COMETA (2025). This matches the indications of both Raynouard (1836-1845) and Levy (1894-1924) who list ost only as a feminine noun. In the dDOM (2025), a certain number of additional sources for the noun with masculine gender can be found, but most of them can be disproved, for instance all occurrences in Wunderli (2009) are feminine, e.g., “E las ostz que so el cel seguian […] et ab la ost de lui” (Wunderli 2009: 352; ‘And the F armies that are in heaven followed […] and with his F / her F / their F army (the F army of him/ her/ them)’). The same applies also for the Fors de Béarn, where the noun has either no agreement target “abans que ost de exir” (Ourliac/ Gilles 1990: 156; ‘Before the army has to set out’) or is used as a feminine noun “oste feyte” (Ourliac/ Gilles 1990: 160; ‘Army formed F ’). Another example where the gender <?page no="163"?> 8.2 Inherited Nouns from Latin 163 remains uncertain, but the noun is nevertheless indexed as being masculine, is Gosman (1982: 622; italics in the original), who lists this noun in the glossary as “host; m: armée”, but the occurrence itself appears with only the gender invariable gran as an agreement target: “mot gran host e poder” (Gosman 1982: 506; ‘very great ? army and power’). The other sources where it is listed as a noun triggering agreement with both genders are all in reference to the Croisade Albigeoise (see Rawls 1983: s.v. “oszt”), which corresponds to the findings in the COMETA (2025) corpus. It seems like ost shows minimal variation, but in the majority of cases is a feminine noun. 8.2.2.17 penche The Old Occitan noun penche ‘comb, hatchel, hackle’ is inherited from the Latin third declension masculine noun pecten ‘comb’. This noun is inherited into all Romance languages. It is masculine in Aragonese, Corsican, French, Galician, Gascon, Italian, Ladin, Portuguese, Romanian, and Spanish, feminine in Catalan and Languedocian, and has both genders in Provençal. For the corpus search, the following graphical variants were used: pecte, penche, penches, penti, piench, pienche, piengos, pinche, and pinti. Languedocian Provençal num+gen+case num+gen gen F sg nom 0 0 0 2 2 obl 1 1 2 pl nom 0 0 0 0 obl 0 0 0 Table 8.20: Gender distribution for penche. (42) E la crotz e la penche. ‘And the cross and the F comb’ (BnF, Français 25425 (1275: f o . 221, l. 29)) <?page no="164"?> 164 8 Gender Doublets in COMETA (2025) (43) E sa penche e so mirail ‘And his F / her F comb and his/ her mirror’ (BmC, Cote 34 (13 th century: f o . 7 r o ., l. 3)) Penche has only two occurrences within COMETA (2025), both feminine singular in the oblique case, one from the Languedocian (see Example (42)), the other from the Provençal area (see Example (42)). Raynouard (1836-1845) lists the noun as only being feminine, thus agreeing with these findings (which are of course not reliable with only two occurrences). But both his examples for this noun are not clearly attributable to a certain gender, which is criticised by Levy (1894-1924). He gives two examples of feminine agreement (which are the same as cited here in Examples (42) and (43)) and adds the statement of penche also appearing as a masculine noun, which he proves with the following two examples: “Item sinq parelhs penches de fer” and “e los pentis una o .II. vegadas l’an” (Levy 1894-1924: s.v. “penche”; italics in the original; ‘Also five equal M iron combs’; ‘and the M combs once or twice a year’). Additional masculine evidence can be found in the following passage: “Et so unitz deves la una partida ab les penches de la vola” (Scinicariello 1982: 197; ‘And they are joined together on one side with the M combs of the palm’). The other sources for feminine occurrences listed in the dDOM (2025) are in reference to the aforementioned citations and therefore do not add any new information. It seems like penche is in fact a gender doublet, appearing with both masculine and feminine agreement only in rare occasions. 8.2.2.18 pols The Old Occitan noun pols with the meaning ‘powder, dust’ is derived from the Latin third declension noun pulvis, which was mainly masculine in Latin, but appeared also as a feminine and neuter noun. It exists in most Romance varieties, but mainly in the hereditary forms of Latin pulveRa. The shorter form exists only in Aragonese as a masculine noun, in Catalan with both genders, in Ladin as a masculine noun and in Occitan as a feminine noun. In Portuguese and Spanish, it exists as a masculine noun. The forms inherited from pulveRa are masculine in Friulian, Logudorese, and Sassarese, and feminine in Aragonese, Catalan, Italian, Occitan, Portuguese, Romanian, Romansh, and Spanish. <?page no="165"?> 8.2 Inherited Nouns from Latin 165 For the corpus search, the following graphical variants were used: plos, pobre, podra, podro, poldra, pols, polvera, polvero, polz, and pous. Languedocian Provençal num+gen+case num+gen gen M sg nom 0 0 0 1 1 obl 1 0 1 pl nom 0 0 0 0 obl 0 0 0 F sg nom (3) 0 (3) 2 (10) 2 (11) obl (5) 2 2 (7) pl nom 0 0 0 (1) obl 0 (1) (1) NA sg nom 0 0 0 1 (5) 7 (11) obl 1 (4) (1) 1 (5) pl nom 0 0 0 6 obl 0 6 6 Table 8.21: Gender distribution for pols. This noun has very few occurrences within COMETA (2025), only three of which show a clear gender value; one masculine occurrence from the Languedocian area (see Example (44a)) and two feminine occurrences from the Provençal area (see both Examples in (45)). It is only by including the forms polvera that we find more occurrences of this noun, all of which are either feminine or of uncertain gender value. These occurrences are included in Table 8.21, as these are forms inherited from the same etymon, but they are in brackets to make clear which of the occurrences are in reality pols. The number without brackets represents the occurrences of pols, and the number in brackets is the combination of both pols and polvera. In the following, some examples will be given that complete the picture for those cases, numbers, and genders which do not appear in the respective sub-corpus of COMETA (2025) with the form pols (see Example (44b) for feminine singular nominative and Example (44c) for feminine singular oblique, both in Languedocian, and Example (45c) for feminine plural oblique in Provençal). (44) a. E de lo pols mi fabrigueron ‘And from (the M ) dust they have created me’ (BL, Harley 7403 (14 th century: f o . 18 v o , l. 10)) <?page no="166"?> 166 8 Gender Doublets in COMETA (2025) b. tornat en polvera. en apres aquela polvera cada dia sia getada ‘turned into dust. afterwards, that F dust will be thrown F every day’ (BnF, NAF 11151 (14 th century: f o . 31 v o ., l. 1-2)) c. Recep de la polvera del solpre ‘Take (the F ) powder of natural sulfur’ (BnF, NAF 11151 (14 th century: f o . 31 r o ., l. 2)) (45) a. En apres formet home de sas mans pres de la pols del sol e formet lo cors. ‘Afterwards he created mankind with his hands, he took (the F ) dust from the ground and created the body.’ (BnF, Français 1049 (1343: f o . 186 r o ., c. 2, l. 15-17)) b. e don a vezer claramens la pols e las lagezas que son dedins. ‘and he showed clearly the F dust and the dirt that are within.’ (BnF, Français 1049 (1343: f o . 80 r o ., c. 2, l. 25-27)) <?page no="167"?> 8.2 Inherited Nouns from Latin 167 c. y deu hom pauzar polveras aspras ‘And man shall deposit bitter F powders’ (BnF, Français 1049 (1343: f o . 104 v o ., c. 2, l. 5-6)) Most instances of this noun are feminine, but if one excludes the forms of polvera, there are only three gender-certain occurrences left, with one masculine occurrence in Languedocian and two feminine occurrences in Provençal. These instances suggest a possible regional gender distribution of this noun. But since there are only very few examples, more evidence is needed to support or disprove this tendency. Raynouard (1836-1845: s.v. “pols”; small caps in the original) classifies it as only a masculine noun, giving one single example with identifiable gender: “Qui ajosta virtut ses humilitat porta lo pols al vent” ‘Whoever adds virtue without humility carries the M dust to the wind’. As this occurrence is from the Provençal area, it contradicts the tendency stated above. There is also some more masculine evidence, e.g., “et ab lo pols de fer limat” (W. Koch 1897: 23; ‘And with the M dust of filed iron’). Furthermore, the example “porta lo pols al vent” (Wahl 1980: 27; ‘Carry the M dust to the wind’) appears in a manuscript from the 14 th century, written in the Auvergnat region (Wahl 1980: 7-9). Corroborating this, the DAOA (2009) lists this noun as masculine, giving an additional masculine occurrence in the Auvergnat region: “Saint Aloys dy que lo pols de la bora luy porta domatge” (DAOA 2009: s.v. “pols”; ‘Saint Aloys says that the M dust of the bora brings him harm’). In the Nouveau Testament de Lyon, probably written in the Languedocian area (see Wunderli 2009: 6-9), this noun appears with feminine agreement: “e secodez la polvera de vostres pes” (Wunderli 2009: 118; ‘and shake off the F dust from your feet’), but only in the longer form. Levy (1894-1924: s.v. “pols”; italics in the original) gives an additional example of this noun with feminine agreement: “Touto la pols gita al vent” ‘Throw all the F dust to the wind’, from a text written in the Hautes-Alpes (see Guillaume 1887: IX). An additional occurrence for feminine agreement can be found in a text from the 14 th century: “la pols de cornu servi ust” (Bondurand 1883: 104; ‘The F dust of the horn serves the tool’). The other feminine evidence listed in the dDOM (2025) refers to citations from the manuscript of Barlaam and Josaphat, cited above in Example (45a), see e.g., <?page no="168"?> 168 8 Gender Doublets in COMETA (2025) “pres de la pols del sol e formet lo cors” (Heuckenkamp 1912: 10; ‘Took of the F dust from the ground and created the body’). There are only very few instances where the form pols shows gender agreement, but these occurrences are more or less equally distributed over both genders. The slight tendency in the COMETA (2025) corpus could not be proven, as there is evidence for both genders in both regions. The form polvera has only feminine occurrences and is therefore not a gender doublet, which justifies the brackets in Table 8.21. 8.2.2.19 ret The Old Occitan noun ret ‘(fishing) net’ is a hereditary word from the Latin neuter third declension noun Rete with the same meaning. From the 4 th century onward, this noun changed gender and triggered feminine agreement. It exists in all Romance languages except Romanian, where only the Latin diminutive form is inherited. This noun is feminine in most Romance languages, e.g., in Italian, Ladin, Portuguese, and Spanish; it is masculine in French, and has both genders in Aragonese, Catalan, and Occitan. For the corpus search, the following graphical variants were used: arret, arretz, reç, recz, red, rei, res, ret, rete, reth, retia, rets, retz, and rez. Languedocian Provençal num+gen+case num+gen gen F sg nom 0 0 0 5 5 obl 0 5 5 pl nom 0 0 0 0 obl 0 0 0 Table 8.22: Gender distribution for ret. This noun only has five occurrences in the COMETA (2025) corpus, all of which are feminine singular in the oblique case and are found in the Provençal area scattered over two texts (BL, Additional 10323 (14 th century) and BnF, Français 13509 (14 th century)); two examples are shown in (46). (46) a. En una ret de pescaria ‘In a F fishing net’ (BL, Additional 10323 (14 th century: f o . 68 r o ., c. 2, l. 6)) <?page no="169"?> 8.2 Inherited Nouns from Latin 169 b. Cant l endeman li pescador tireron la ret contra lor ‘When the next day the fishermen pulled the F net towards themselves’ (BnF, Français 13509 (14 th century: f o . 95 r o ., c. 2, l. 2-3)) From the data in the COMETA (2025) corpus, no gender variation can be proven, so a look into further literature is necessary in order to verify or falsify the label of gender doublet for ret. Raynouard (1836-1845) cites it as both a feminine and masculine noun, giving one example with masculine agreement: “Aissi’ls ten enredatz lo Retz” (Raynouard 1836-1845: s.v. “ret”; small caps in the original; ‘Thus the M net holds them entangled’). Levy (1894-1924) gives an additional example to support the statement by Raynouard (1836-1845): “semblant es lo regnes dels cels al ret mes e la mar” (Levy 1894-1924: s.v. “Ret”; italics in the original; ‘The kingdom of heaven is like a M net that was cast into the sea’). In the same text as the one cited by Levy (1894-1924), ret also appears as a feminine noun: “la recz de lor ruopia” (Salvioni 1890: 71; ‘the net F of their rebellion’) and “donca il meseron la recz” (Salvioni 1890: 132; ‘Then they cast the F net’). There is variation in gender for ret, but only very few instances could be found. But, as the examples from Salvioni (1890) show, it is also a gender doublet within one text. 8.2.2.20 sal Inherited from the Latin masculine and sometimes neuter third declension noun sal ‘salt’, the Old Occitan noun sal means both ‘salt’ and ‘holy water’. It is masculine in Corsican, French, Galician, Ladin, Italian, Portuguese, Romansh, and Sardinian. It is feminine in Aragonese, Catalan, Romanian, and Spanish. For the corpus search, the following graphical variants were used: sa, sal, sale, sals, salt, saou, sau, sel, seu, and ssau. As Table 8.23 shows, only five occurrences for this noun showing an identifiable gender value could be found within COMETA (2025); all are in the oblique singular. Four of these occurrences are from the Languedocian; one masculine (see Example (47b)) and three feminine occurrences (see Example (47a)). It appears in two texts in the Languedocian area, but a gender value could only be ascertained in manuscript BnF, Français 25425 (1275). <?page no="170"?> 170 8 Gender Doublets in COMETA (2025) Languedocian Provençal num+gen+case num+gen gen M sg nom 0 0 0 2 2 obl 1 1 2 pl nom 0 0 0 0 obl 0 0 0 F sg nom 0 0 0 3 3 obl 3 0 3 pl nom 0 0 0 0 obl 0 0 0 NA sg nom 0 0 0 13 13 obl 6 7 13 pl nom 0 0 0 0 obl 0 0 0 Table 8.23: Gender distribution for sal. (47) a. del pebre e de la sal. ‘Of pepper and of (the F ) salt.’ (BnF, Français 25425 (1275: f o . 121, l. 18)) b. Demandan ous e aiga. e estopa el sal. ‘They ask for eggs and water and tow and (the M ) salt’ (BnF, Français 25425 (1275: f o . 124, l. 14)) In the Provençal area, only one masculine occurrence could be found (see Example (48)), even though this noun has occurrences scattered over four Old Provençal manuscripts. There is no evidence for feminine usage of this noun in the Provençal part of the COMETA (2025) corpus. (48) Lo sal espars per miei lo ṣọḷ cap ‘He sprinkles the M holy water in the direction of the head’ (BmC, Cote 34 (13 th century: f o . 49 v o ., l. 17)) <?page no="171"?> 8.2 Inherited Nouns from Latin 171 As the data above show, there is variation in gender for this noun even within a single text (see Examples (47a) and (47b)). However, as there are only five instances where the gender could be determined, these findings are not robust on their own, so a look into further literature was undertaken. Contrary to this tendency in the COMETA (2025) corpus, Raynouard (1836-1845) lists it only as a feminine noun, without giving any examples that show gender. Levy (1894- 1924) lists it as both masculine and feminine, giving the same instances as seen in Examples (47) and (48), and adds additional feminine evidence: “deu mezuratge de la sal” (Levy 1894-1924: s.v. “Sal”; italics in the original; ‘About the measurement of the F salt’). In the DAOA (2009) it is listed as a feminine noun, with three additional feminine instances, e.g., “lay ont ussavo la sal blancha” (DAOA 2009: s.v. “sal”; ‘There where they used the F white F salt’). Other sources from the dDOM (2025) are indicated as displaying the noun as masculine, but do not in fact show a gender value, e.g., “Per so prez’om mais aur que sal” (Kolsen 1910: 376; ‘For that reason, one values gold more than salt’), “e pues metio dins sal” (Piccat 2001: 120; ‘and then he/ she put it into salt’) or “e sal e meill e † farina † ” (Viel 2011: 109; ‘and salt and honey and flour’). There is only one additional source showing both masculine and feminine gender for this noun, both within one verse: “La sal es bona . Mas si lo sal non es sala en qual cosa condiren luy” (Salvioni 1890: 53; ‘The F salt is good F . But if the M salt is not salty, with what will they season it’). All other sources listed in the dDOM (2025) that do really display masculine occurrences are in reference to the two masculine examples within the COMETA (2025) corpus (see Examples (47b) and (48)). There seems to be a little variation in gender, but it is questionable to label sal as a gender doublet with only two masculine and many more feminine occurrences. 8.2.2.21 salut The noun salut is derived from the Latin third declension noun salus ‘health, salvation’. In Old Occitan, this noun kept the same meanings and obtained additionally the significations of ‘salutation, greeting, news/ intelligence, art of poetry’ as well as the denomination of a certain type of coins. This noun was feminine in Latin but quickly became masculine in many Romance languages, e.g., Galician, Occitan, and Spanish, but stayed feminine only in Corsican, Italian, and Sardinian. The FEW (1922-2003) attributes the influence of masculine salut ‘salute, greeting’ as the reason for the change in gender. Even though the Latin etymon already had this meaning, salut might, according to the FEW (1922-2003: s.v. “salus”), have been seen as a nominalisation of the past participle of the verb salutaRe and was therefore used as a masculine noun, subsequently inducing a change in gender for this noun with all mean- <?page no="172"?> 172 8 Gender Doublets in COMETA (2025) ings. In Aragonese and Catalan it is generally feminine, but masculine with the meaning of the salutation. In Romanian, it belongs to the Genus Alternans. In Italian, there is a different word for the meaning of the salutation, masculine il saluto, while for the other meanings the feminine noun la salute is used. For the corpus search, the following graphical variants were used: salu, saluç, salud, salus, salut, salutz, and saluz. Languedocian Provençal num+gen+case num+gen gen M sg nom 0 0 0 0 1 obl 0 0 0 pl nom 0 0 0 1 obl 0 1 1 F sg nom 0 0 0 57 65 obl 22 35 57 pl nom 0 0 0 8 obl 0 8 8 NA sg nom 0 0 0 71 71 obl 30 41 71 pl nom 0 0 0 0 obl 0 0 0 Table 8.24: Gender distribution for salut. As Table 8.24 shows, this noun triggers mostly feminine agreement; only one masculine (oblique plural) occurrence could be found in Provençal, see Example (50c). In the Languedocian area, this noun appears in seven different manuscripts, showing an identifiable gender value in all manuscripts except BnF, Français 2232 (1325). All occurrences with clear gender value in the Old Languedocian sub-corpus are feminine singular (see Examples in (49) for some occurrences). (49) a. avia promes la salut del pobol ‘He had promised the F salvation of the people’ (BL, Harley 7403 (14 th century: f o . 43 r o ., l. 9)) <?page no="173"?> 8.2 Inherited Nouns from Latin 173 b. E nostre cancelier antic. Salut veraya. ‘And our former chancellor. True F salvation’ (BmT, Cote 2883 (14 th century: f o . 6 v o ., c. 2, l. 25-26)) The Provençal sub-corpus also contains occurrences in the plural, one masculine (see Example 50c) and eight feminine ones (see Example (50b)), but most occurrences are feminine singular, see Example (50a). The occurrences with identifiable gender value in the Old Provençal area are distributed over seven manuscripts. (50) a. gran guizardon que recebas per la salut de marma. ‘The great gift you receive for the F salvation of marma’ (BnF, Français 1049 (1343: f o . 209 v o ., c. 1, l. 28-29)) b. Mot an ben entendutz los ditz e las salutz. ‘They understood the sayings and the F salvations very well’ (BnF, Français 13509 (14 th century: f o . 11 v o ., c. 2, l. 27-28)) <?page no="174"?> 174 8 Gender Doublets in COMETA (2025) c. Aytal son los premiers salutz. e las premieras bathalas contra los novels cavalliers de dieu. ‘These are the M first M greetings. and the first battles against the new knights of God’ (BnF, Français 2232 (1325: f o . 7 v o ., l. 5)) As there is only one masculine instance in COMETA (2025), it is again necessary to look into further literature. Raynouard (1836-1845) lists it only as feminine, thus supporting the findings in the COMETA (2025) corpus. Levy (1894-1924) on the other hand lists it as both a masculine and feminine noun for most meanings, and as only feminine for the meaning of ‘health’. For the significations of ‘salvation’ and ‘salutation’, Levy (1894-1924) gives a total of three masculine examples. For ‘salvation’, he lists “per femena fos annunciatz lo salutz del mun” (Levy 1894-1924: s.v. “Salut”; italics in the original; ‘Through a woman was announced the M salvation of the world’) and “Ara devem saber que significan aque[s]tes caus[es] al salut de nostres animes” (Levy 1894-1924: s.v. “Salut”; italics and brackets in the original; ‘Now we must know what these things mean for the F salvation of our souls’). The DAOA (2009) partially agrees with that, listing it as only a feminine noun for the meanings ‘salvation, redemption’. Contrary to Levy (1894-1924), the DAOA (2009) lists it as exclusively masculine for the significations ‘sort of money, health’ with one example: “payet loditz Cayrols per la tara d’un salus” (DAOA 2009: s.v. “salus”; ‘He satisfied the aforementioned Cayrols with the price of one F coin of money’). Further masculine occurrences can be found, e.g., “Nonl dizun neiss un bon salud” (Thomas 1925: 13; ‘They say he is not in (a M ) good M health’). Other editors list it as a masculine noun, while it is not possible to determine the gender of the occurrence in the respective text, e.g., “enperador de Roma, salutz e intrament” (Gosman 1982: 505; ‘Emperor of Rome, salvation and beginning’) or “Tant salut ni tant’amistat” (Klein 1885: 58; ‘So much greeting nor so much friendship’). The same applies for Melani (2016), who lists salut as a masculine noun for the meaning ‘greeting’ and as a feminine noun with the sig- <?page no="175"?> 8.2 Inherited Nouns from Latin 175 nification ‘salvation’, even though both occurrences in the text do not show any clear sign of gender: “cant a midons irai rendre saluç” (Melani 2016: 164; ‘When I go to my lady to render greeting’) and “El es salutz e salvetatz” (Melani 2016: 250; ‘He is salvation and deliverance’; see also the critic in; Chambon 2017: 306). This is also the case for “ja no m’en vengues mais salutz” (Nicholson 1976: 116; ‘You no longer come to me, only greetings’) and “Mon Tort-n’avetz en Narbones man salutz, si tot luenh s’estai” (Nicholson 1976: 81; ‘My wrong, you have sent many greetings to Narbonne, even though I am far away’), where the editor states in the glossary “salut (s.m., salvation, greeting)” (Nicholson 1976: 153; italics in the original), even though no evidence for masculine agreement can be found in either occurrence. The same is the case for supposedly feminine occurrences according to the glossary, where in fact no clear sign of gender can be found, e.g., “Escriu salutz mais de cen” or “En proensa tramet jois e salutz” (Gennrich 1959: 8-9; ‘Write greetings more than a hundred’; ‘To Provence, he/ she sends joy and greetings’). In the glossary of Minervini (2010: s.v. “salut”; “salutz”), it is listed as feminine singular for the meaning of ‘healing’ and as masculine plural for ‘greetings’, even though both occurrences are in fact of uncertain gender: “Salutz al rey” (Minervini 2010: 69; ‘Greetings to the king’) and “el li demandet salut per son payre” (Minervini 2010: 87; ‘He asked him for health for his father’). In the edition of Marcabru’s poems by Dejeanne (1909), salut is listed as being a masculine noun, but all occurrences are either of undeterminable gender (“Que tot jorn demandon salutz” (Dejeanne 1909: 194; ‘That every day they ask for salvation’)) or feminine (“Non trametrai autras salutz” (Dejeanne 1909: 16; ‘I will not send other F greetings’)). The findings in COMETA (2025) as well as in further literature partially support the statements by Levy (1894-1924) who lists both genders for all meanings but ‘health’, where he lists it as only feminine. The latter is true as the noun appears only as a feminine noun for most significations. But the former is not quite supported by these findings, as salut seems to have masculine occurrences only for the meaning of ‘greeting’ and not for the other meanings. There is only one other instance where salut is listed as a masculine noun, for the meaning ‘salvation’, but a look into the poem reveals that it is in fact an occurrence where the gender cannot be identified: “Qe ab se vos guit a salut” (Gouiran 1985a: 432; ‘May it guide you to salvation’). The signification of money is not present within the COMETA (2025) corpus, so no statement on gender value in that context can be made. <?page no="176"?> 176 8 Gender Doublets in COMETA (2025) 8.2.2.22 sanc The Old Occitan noun sanc ‘blood, race’ is derived from the Latin masculine third declension noun sanguis ‘blood’. In Classical Latin, sanguis was an imparisyllabic noun with the genitive sanguinis, but changed from the 4 th century onward to the class of parisyllabic nouns, thus building the genitive as sanguis, and most of the Romance languages inherited the parisyllabic form. The FEW (1922-2003) brings the gender variability of the noun in relation with its new state as a parasyllabic noun in analogy to other parasyllabic nouns in -is, which showed significant variation in gender in Latin (e.g., callis, finis, funis, etc.). The hereditary forms of sanguis are feminine in Aragonese, Spanish, and in some Old Italian varieties (e.g., Old Venetian and Old Tuscan), and masculine in Corsican, French, Italian, Ladin, and Portuguese. It has both genders in Catalan and Galician, with the tendency of feminine use in Catalan and masculine use in Galician, but both genders are seen in these two languages. It belongs to the Genus Alternans in Romanian. For the corpus search, the following graphical variants were used: sac, sags, sainc, sanc, sanch, sancs, sang, sangs, sanh, sanguis, and ssang. Languedocian Provençal num+gen+case num+gen gen M sg nom 0 0 0 61 61 obl 27 34 61 pl nom 0 0 0 0 obl 0 0 0 F sg nom 0 0 0 17 17 obl 17 0 17 pl nom 0 0 0 0 obl 0 0 0 NA sg nom 0 0 0 68 68 obl 44 24 68 pl nom 0 0 0 0 obl 0 0 0 Table 8.25: Gender distribution for sanc. As Table 8.25 shows, sanc is variable in gender in the COMETA (2025) corpus, but only in the Languedocian sub-corpus, where feminine (see Examples (51a) and (51b)) and masculine (see Examples (51c) and (51d)) occurrences could be found, while there are only masculine occurrences in the Provençal sub-corpus (see Examples (52a) and (52a)), distributed over five manuscripts. There are occurrences of this noun in all nine texts of the Languedocian sub-corpus, showing both masculine and feminine agreement in four of these manuscripts (namely BL, Additional 21218 (14 th century), BnF, Français 25425 (1275), BnF, Français <?page no="177"?> 8.2 Inherited Nouns from Latin 177 2232 (1325), and BnF, NAF 11151 (14 th century)). Since sanc is a mass noun denominating a liquid, it is not surprising to find only singular oblique occurrences of the noun. (51) a. Enapres quant la sanc sera restrecha enayssi ‘Afterwards when the F blood will be clotted F that way’ (BnF, NAF 11151 (14 th century: f o . 31 r o ., l. 14)) b. Per que sa sanc benigna. nos sia espandens. ‘For his F blessed F blood may blossom for us’ (BnF, Français 25425 (1275: f o . 239, l. 19)) c. E la carn e lo sancs. e los cervels. ‘And the flesh and the M blood. and the brains’ (BnF, Français 25425 (1275: f o . 233, l. 2)) d. Cant lo drac vi cazer son sanc ‘When the dragon saw his M blood running’ (BnF, Arsenal 6355 (14 th century: f o . 35 v o ., l. 17)) (52) a. pobol de nostre senhor los cals el a rezemut de son sanc precios. ‘people of our lord to whom he redeemed his M precious M blood’ (BnF, Français 1049 (1343: f o . 215 v o ., c. 2 l. 23-25)) <?page no="178"?> 178 8 Gender Doublets in COMETA (2025) b. Car el era del sanc e del linhage. ‘As he was of the M blood/ race and linage’ (BnF, Français 1049 (1343: f o . 15 r o ., c. 2, l. 5)) The data from the COMETA (2025) corpus suggests that sanc might be only masculine in Provençal, supported by Raynouard (1836-1845) who lists it only as a masculine noun, giving no instances of feminine use. Furthermore, the analysis of the COMETA (2025) corpus indicates that sanc is a gender doublet in Languedocian, which is supported by the fact that it appears as a gender doublet even within four manuscripts in that area. This tendency is affirmed by Levy (1894-1924) who only cites feminine examples from texts of the Languedocian or the Gascon areas. Contrary to that finding, in one manuscript of the Vie de sainte Marie Madeleine, written in the Provence, we find a feminine occurrence in manuscript B: “de la sanc” (Chabaneau 1887: 23; ‘Of the F blood’), while it appears as masculine in the other existing manuscript A “de l[a] sanch” (Chabaneau 1887: 10; ‘Of the F/ ? blood’), where the editor added the [a], thus changing the gender value of that occurrence. All other feminine occurrences listed in the dDOM (2025) are from regions other than the Provençal area, e.g., “una gota de la soa sancta preziouza sanc” (Brunel 1953: 17; ‘A drop of his F / her F holy F precious F blood’) in a text from Toulouse or “lhebat la sanc de la mia oferture” (Lespy/ Raymond 1876: 12; ‘He/ She raised the F blood of my offering’) in a text from the Béarn. There are further sources in the dDOM (2025) for sanc being a feminine noun, but in reality, the respective occurrence does not show a gender value, e.g., “se deig en mays qu’ets fanch de carn e sanch” (Riquer 1971: 268; ‘It is said, furthermore, that he/ she/ it is made of flesh and blood’). The findings in the COMETA (2025) corpus in combination with the data taken from other literature and the dDOM (2025) make it clear that sanc is not a gender doublet in Provençal, with only masculine occurrences. Outside the Provençal area, on the other hand, it is a gender doublet, with more masculine than feminine occurrences, but still triggering both genders. This suggests that it is the other way around than what Chabaneau (1887: 23) assumed for the two manuscripts from the Provence: it seems more likely that the mistake is in manuscript B rather than in manuscript A and the correction by Chabaneau (1887: 10) is unjustified. <?page no="179"?> 8.2 Inherited Nouns from Latin 179 8.2.2.23 semen Inherited from the Latin feminine third declension noun sementis ‘seed, sowing’, the Old Occitan noun semen kept the same meaning. Additionally to Occitan, lexemes with the same etymon exist in Aragonese, Catalan, Corsican, French, Italian, Portuguese, and Spanish, and are feminine in all these languages; only in Catalan are some masculine occurrences found in archaic use. For the corpus search, the following graphical variants were used: semen, semences, semenre, semens, sement, and senen. Languedocian Provençal num+gen+case num+gen gen F sg nom 0 0 0 3 3 obl 3 0 3 pl nom 0 0 0 0 obl 0 0 0 NA sg nom 0 0 0 2 2 obl 1 1 2 pl nom 0 0 0 0 obl 0 0 0 Table 8.26: Gender distribution for semen. Only five occurrences of this noun could be found in the COMETA (2025) corpus, all except one in the Languedocian part, scattered over three texts. Three of these occurrences within two different manuscripts (BmT, Cote 2883 (14 th century) and BnF, Français 25425 (1275)) show an identifiable gender value, all feminine singular in the oblique case and from the Languedocian sub-corpus, see both Examples in (53). (53) a. Per que a lor dins falha. lo blatz e la semens. ‘For inside they are missing the wheat and the F seed [supplies]’ (BnF, Français 25425 (1275: f o . 164, l. 39)) b. S espandish per las terras. e brandish la semens. ‘It expands over the lands and paces around the F seed’ (BnF, Français 25425 (1275: f o . 213, l. 33)) Contrary to the findings in the COMETA (2025) corpus, Raynouard (1836-1845) lists semen only as a masculine noun, giving one masculine occurrence: “Lo dols <?page no="180"?> 180 8 Gender Doublets in COMETA (2025) frut dels amars semens de letras” (Raynouard 1836-1845: s.v. “semen”; small caps in the original; ‘The sweet fruit of bitter M seeds of letters’). Levy (1894- 1924) does not give further masculine examples; he just states that semen is also a feminine noun, giving the same examples as here in Example (53). It is listed as masculine in the glossary in Appel (1930: 302), but the occurrence itself does not show a determinable gender: “fazen frug ses semen” (Appel 1930: 93; ‘They make fruit without seed’), just as in Picchio Simonelli (1974: 219): “quar de peccat creys lur semens” ‘Because from sin grows their seeds’ and Caravaggi (1963: 101): “que tu say vieuras ses semen” ‘That you are alive without seeds’. 8.2.2.24 serp/ serpen This noun was already addressed in Wiedner (2026, forthcoming), but without considering potential geographical, syntactical, or scribe-specific variation. Serpen is inherited from the imparisyllabic Latin third declension noun seRpens and serp from the shorter form seRpes, all with the meaning ‘snake’. This noun is deverbal in Latin, a conversion of the present participle seRpens of the verb seRpeRe ‘to crawl’. The shorter form is the present participle, but with an elision of the n. The deverbality in Latin, that caused a strong variation in gender, is certainly not relevant for the speakers of Old Occitan, given that the verb does not exist anymore and the deverbal state is therefore not present in their cognition. For the sake of this study, serpen is used to refer to both forms, well knowing that there may be differences in gender assignment. Because there are no instances of the shorter form in the COMETA (2025) corpus, this is not an issue; all occurrences counted in Table 8.27 are instances of the longer form serpen. In the Romance languages, derivations of at least one of the two Latin etyma continue to exist. The respective nouns are feminine in Aragonese, Portuguese, Sardinian, and Spanish, masculine in French, Italian (but see the archaic form in Italian, feminine la serpe), Ladin, and Romanian, and have both genders in Catalan and Corsican. For the corpus search, the following graphical variants were used: cer, cerp, cerpent, sarpaut, ser, serp, serpe, serpen, serpens, serpent, serpentz, and serps. Table 8.27 shows that serpen exists in both regions with both genders. Most of the occurrences with identifiable gender are from the Provençal sub-corpus; in the Languedocian sub-corpus, 15 instances with clear gender value could be found. These occurrences in Languedocian are all singular and almost equally distributed over the genders: eight are masculine (see Example (54b) for the nominative and Example (54c) for the oblique case), and the remaining seven are feminine, all in the oblique case (see Example (54a)). The occurrences of this noun in the Languedocian area are scattered over five manuscripts. All masculine occurrences are from the manuscript BnF, Arsenal 6355 (14 th century), <?page no="181"?> 8.2 Inherited Nouns from Latin 181 Languedocian Provençal num+gen+case num+gen gen M sg nom 3 15 18 27 29 obl 5 4 9 pl nom 0 0 0 2 obl 0 2 2 F sg nom 0 13 13 29 67 obl 7 9 16 pl nom 0 19 19 38 obl 0 19 19 NA sg nom 2 1 3 16 33 obl 3 10 13 pl nom 0 6 6 17 obl 2 9 11 Table 8.27: Gender distribution for serp/ serpen. where it appears exclusively as a masculine noun (or with unclear gender value). In the other four texts, all occurrences with an identifiable gender value are feminine. So at least within the Languedocian area, the variation in gender within COMETA (2025) correlates with the manuscript. In the Provençal area, this noun can be found in four different manuscripts, in all of which it shows both masculine and feminine agreement (see Example (55)). (54) a. Per dig de la serpent que malazerta fos ‘Through the saying of the F snake that was cursed F ’ (BL, Harley 7403 (14 th century: f o . 100 v o , l. 7-8)) b. [C]ant lo serpens aus lo coniur ‘When the M snake hears the enchantment’ (BnF, Arsenal 6355 (14 th century: f o . 36 v o , l. 11)) <?page no="182"?> 182 8 Gender Doublets in COMETA (2025) c. Adonc preget dyeu humielmen Per la verge contral serpen ‘Then they humbly pray to god for the vergin against the M snake’ (BnF, Arsenal 6355 (14 th century: f o . 32 v o ., l. 15-16)) The remaining (and the majority of the) occurrences of serpen with identifiable gender value are from the Provençal sub-corpus. There is evidence for serpen appearing in all combinations of gender, number, and case with the exception of masculine plural nominative. One example for each of these combinations is given in the following: For feminine singular nominative see Example (55a), for oblique see Example (55b); for feminine plural nominative see Example (55c), for oblique see Example (55d). For masculine singular nominative see Example (55e), for oblique see Example (55f), and for masculine plural oblique see Example (55g). (55) a. La serpent vay liar al coll de sentura. ‘The F snake will join at the hill of sentura’ (BnF, Français 13509 (14 th century: f o . 68 r o ., c. 2 l. 15-16)) b. De la mala serpent que la terra gastava. ‘Of the F bad F snake that ravaged the earth’ (BnF, Français 13509 (14 th century: f o . 68 r o ., c. 2 l. 23-24)) <?page no="183"?> 8.2 Inherited Nouns from Latin 183 c. adonx totas aquellas serpens et aquellas malas bestias ‘Thus all F these F snakes and all these beasts’ (BnF, Français 1049 (1343: f o . 218 r o ., c. 1 l. 27-29)) d. Quant ell vi las serpens trastotas en torn luy ‘When he saw the F snakes all F around him’ (BnF, NAF 6195 (14 th century: f o . 27 v o ., l. 29)) e. Tantost cant esgardavan ves lo serpent pendut eran garitz. Lo serpent del metalh pendut. ‘As soon as they looked at the M hanging M snake they were healed. The M snake hung M from the metal’ (BnF, Français 1049 (1343: f o . 145 v o ., c. 1 l. 21-24)) f. El plan sant hermentari hon trobet lo serpent ‘He mourns saint hermentari where he found the M snake’ (BnF, Français 13509 (14 th century: f o . 68 r o ., c. 2 l. 11-12)) <?page no="184"?> 184 8 Gender Doublets in COMETA (2025) g. Portem la los serpentz de l isla auriana. ‘Let us bring there the M snakes from the island of auriana.’ (BnF, Français 13509 (14 th century: f o . 28 r o ., c. 1, l. 8-9)) The data presented above demonstrates clearly that serpen is a gender doublet in Old Occitan, regardless of the region or the time of the manuscript. There are many examples of both genders within one and the same manuscript in the Provençal area, e.g., both masculine (see Examples (55f) and (55g)) and feminine occurrences (see Examples (55a) and (55b)) in manuscript BnF, Français 13509 (14 th century). It seems like the distribution and variation of gender is relatively free for this noun within the Provençal sub-corpus of COMETA (2025). Variation in gender does not appear within a single manuscript in Languedocian. So in this area, variation in gender may be due to other factors, as for instance micro-diatopic variation: e.g., this noun may be used as a masculine noun in the Languedocian area only in Gévaudan, where the text was composed and copied. In the dDOM (2025), there is no indication for occurrences of both genders within a single text in the Languedocian area, so it remains unclear whether this noun is a gender doublet within the whole Old Languedocian linguistic area, or whether there are differences within this region. The latter would mean that this noun is not a gender doublet in Old Languedocian, but rather that it is masculine in some areas and feminine in others, but not used with both genders within one region. 8.2.2.25 set The Old Occitan noun set ‘thirst’ is inherited from the Latin third declension noun sitis ‘thirst, drought’, which was feminine in Latin. It exists as a feminine noun in all Romance languages; only for Languedocian is it listed as both a masculine and feminine noun. For the corpus search, the following graphical variants were used: cet, say, se, sec, sed, see, seet, sei, sen, ses, set, setz, and sey. As Table 8.28 shows, there are only three occurrences of set with clearly identifiable gender in the COMETA (2025) corpus; one is feminine singular oblique from the Languedocian part of the corpus, see Example (56), the other two are masculine singular oblique from the Provençal part, see both Examples in (57). <?page no="185"?> 8.2 Inherited Nouns from Latin 185 Languedocian Provençal num+gen+case num+gen gen M sg nom 0 0 0 2 2 obl 0 2 2 pl nom 0 0 0 0 obl 0 0 0 F sg nom 0 0 0 0 1 obl 0 0 0 pl nom 1 0 1 1 obl 0 0 0 NA sg nom 0 0 0 35 35 obl 12 23 35 pl nom 0 0 0 0 obl 0 0 0 Table 8.28: Gender distribution for set. (56) La setz los destrenh tant. ‘The F thirst presses them so much’ (BnF, Français 1049 (1343: f o . 218 r o ., c. 2 l. 9-10)) (57) a. Mays apoderaca aquel cet per l autre gran ascedament que avia de vezer nostre senhor. ‘Moreover, he overcame that M thirst through the other great thirst he had to see our Lord’ (BnF, Français 1049 (1343: f o . 217 v o ., c. 2 l. 6-7)) <?page no="186"?> 186 8 Gender Doublets in COMETA (2025) b. trobet ayga et esteys lo gran set que avia. ‘He found water and satisfied the M large thirst he had’ (BnF, Français 1049 (1343: f o . 218 r o ., c. 2 l. 9-10)) As the data from the COMETA (2025) corpus does not give sufficient information to be certain about the gender distribution of set 11 , a closer look into the literature was necessary. Raynouard (1836-1845) lists set as being only feminine, giving one additional example: “Tanta fam, tanta set e tan som” (Raynouard 1836-1845: s.v. “set”; small caps in the original; ‘So much hunger, so much F thirst, and so much sleep’) from a text from Bertrand de Born, a troubadour from Perigord, so from outside the regions included in the COMETA (2025) corpus. Levy (1894-1924) supplements the information of set as a masculine noun, giving the same occurrence as Raynouard (1836-1845) before, but with masculine agreement: “Tanta fam, tant set e tan som” (Levy 1894-1924: s.v. “set”; italics in the original; ‘So much hunger, so much M thirst, and so much sleep’). He additionally cites two masculine occurrences: “E las fams e los setz totz per entiers? ” (Levy 1894-1924: s.v. “set”; italics in the original; ‘And the hungers and the M thirsts all together? ’), and the second one is the same as cited here in Example (57a). The dDOM (2025) gives some more instances of masculine and/ or feminine occurrences of this noun. For feminine use, the examples are the same as above. All other sources for set being feminine can be disproved, as can be seen in the following examples: “que hanc non agront cet ni fam” (Manetti 2012: 36; ‘For never they will have hunger nor thirst’) and “on non auran ni fam ni set” (Manetti 2012: 110; ‘there they will have neither hunger nor thirst’); “qu’oms estrainz agues set ne fam” (Routledge 1977: 126; ‘For strangers have neither thirst nor hunger’). The same is valid for “E cudatz setz M’enoi ni dejunars” (Kolsen 1910: 172; ‘And you think that thirst troubles me, not fasting’), “con han fam ni set” (Gosman 1982: 507; ‘When they have hunger nor thirst’), or “et auras aguda set davans que pro” (Rothwell 1964: 33; ‘And you will have had thirst before hav- 11 As there is only one occurrence per region, it is not enough data to say anything about regional distribution but to mention the slight tendency of set being used as a masculine noun in Provençal and a feminine noun in Languedocian. <?page no="187"?> 8.2 Inherited Nouns from Latin 187 ing enough’). All of these editors list set as a feminine noun in their glossaries, but none of the occurrences in the text actually show gender. The sources indicating masculine evidence in the dDOM (2025) could mainly be disproved 12 , as they do not actually provide examples of masculine use: “nos mes en tal afan per qu’avem set e fam” (Bartsch/ Koschwitz 1904: 21; ‘He/ she/ it puts us in such trouble because we have thirst and hunger’) or “e fam et e sed” (Wunderli 2009: 451; ‘And hunger and thirst’); “Si con ac set, fel e azet” (de Poerck 1957: 532; ‘If one has thirst, [which is] bitter and sour’). There is only one occurrence of masculine use: “non lur pot escantir lo setz” (Zenker 1900: 118; ‘The M thirst cannot scare them’). Some authors list the noun in the glossary as both masculine and feminine, even though those occurrences are in fact of uncertain gender, e.g., Manetti (2008: s.v. “set”) lists it as masculine and feminine, while both occurrences do not show gender assignment: “mor de fam e mor de set” and “de set m’auci e de fam” (Manetti 2008: verses 4035 and 6913; ‘Dies of hunger and dies of thirst’; ‘Of thirst kills me and of hunger’). There is only very little evidence for set as a masculine noun; most occurrences with identifiable gender are feminine, which indicates that masculine use is only marginal. 8.2.2.26 trau The Old Occitan noun trau ‘girder, beam, bar, joist’ is inherited from the Latin feminine third declension noun tRabs with the same meaning. It exists in all Romance languages except Romanian, and is not used anymore in modern French; it disappeared from the written language in the 17 th century. It is masculine e.g., in Aragonese, Catalan, and in most Occitan varieties; only in Provençal does it appear with both genders; and it is feminine e.g., in Friulian, Italian, Ladin, and Spanish. For the corpus search, the following graphical variants were used: tra, trabe, trabes, tran, trau, trauc, traul, and traus. As Table 8.29 shows, almost all occurrences of trau in the COMETA (2025) corpus (19 out of 22) are masculine (see Examples (58b) and (58c) for the Languedocian and both Examples in (59) for the Provençal part). There is only one feminine occurrence in COMETA (2025), a feminine singular oblique occurrence in the Languedocian sub-corpus (see Example (58a)) in manuscript BmT, Cote 2883 (14 th century), where this occurrence is the only one. For the remaining two oc- 12 Or they give the same instances as cited in Levy (1894-1924) before and therefore do not give any additional information, e.g., “e las fams e los setz totz per entiers” (Bartsch/ Koschwitz 1904: 45; ‘And all the hungers and (the M ) thirsts entirely’). <?page no="188"?> 188 8 Gender Doublets in COMETA (2025) Languedocian Provençal num+gen+case num+gen gen M sg nom 0 0 0 14 19 obl 14 0 14 pl nom 0 2 2 5 obl 3 0 3 F sg nom 0 0 0 1 1 obl 1 0 1 pl nom 0 0 0 0 obl 0 0 0 NA sg nom 0 0 0 2 2 obl 2 0 2 pl nom 0 0 0 0 obl 0 0 0 Table 8.29: Gender distribution for trau. currences, it was not possible to determine gender as they appear without any agreement targets. (58) a. E no vezens la gran trau en lo sieu. ‘And not seeing the F great beam in his.’ (BmT, Cote 2883 (14 th century: f o . 33 r o ., c. 2, l. 15-16)) b. preseron lo trau e gueron lo foras ‘they took the M beam and preserved it M outside’ (BL, Harley 7403 (14 th century: f o . 46 v o ., l. 17)) c. troberon lo. una cordada plus breu que los autres traus non eran. ‘They found it one elbow-length shorter than the M other M beams’ (BL, Harley 7403 (14 th century: f o . 45 v o ., l. 10-11)) <?page no="189"?> 8.2 Inherited Nouns from Latin 189 (59) a. Li trau son tuch romputz ‘The M beams are all M broken M ’ (BnF, NAF 6195 (14 th century: f o . 67 v o ., l. 9)) b. li traus son tut romput ‘The M beams are all M broken M ’ (BnF, Français 13509 (14 th century: f o . 65 v o ., c. 1, l. 28)) From the data in the COMETA (2025) corpus, it seems like trau is predominantly masculine in Old Occitan, which will be evaluated in the following with additional data from the literature, where it was searched especially for feminine evidence 13 . Raynouard (1836-1845) lists it as both a masculine and feminine noun, giving one occurrence of feminine use: “E non senton la tRau qu’en lor oill an” (Raynouard 1836-1845: s.v. “trau”; small caps in the original; ‘And they do not feel the F beam that they have in their eye’). Levy (1894-1924) agrees with Raynouard (1836-1845), but gives only one additional feminine example while giving seven masculine occurrences. The feminine evidence is “al cossolat exceptat las traucs de la gran cambra” (Levy 1894-1924: s.v. “trau”; italics in the original; ‘To the consulate, except for the F beams of the great chamber’). Additional feminine occurrences are indicated in the dDOM (2025), as the following examples show: “trebalhava en la obra aquela, una trau que no era be fermada” (Arthur 1992: 153; ‘He was working on that structure, a F beam that was not well fastened F ’) and “dos traulx grandas d’elsi valon” (Pansier 1932: 194; ‘Two large F beams from the little valley of Elsi’). The tendency found in the COMETA (2025) corpus can be supported by additional literature stating that trau is predominantly a masculine noun with very few instances of feminine use; only 3 instances of feminine use could be found and verified. 13 Most of the sources in the dDOM (2025) list it as a masculine noun, for instance “Et el sieu no conois lo trau” (Tortoreto 1981: 223; ‘And he does not know the M beam’). <?page no="190"?> 190 8 Gender Doublets in COMETA (2025) 8.2.3 Fourth Declension Class There is only one noun from the fourth declension class within the list of gender doublets, namely man. This noun is derived from the Latin fourth declension noun manus ‘hand, company, force, violence, finishing touch, handwriting, work’ and inherited into all Romance languages. In Old Occitan, this noun has the significations ‘hand, needle (of a clock), quire of paper, sort, class, work, labour, power, and possession’. As stated in Section 3.2, nouns from the fourth declension class were mainly feminine, which is true also for manus. However, this noun already showed significant gender variation and confusion in Classical Latin times, triggering both masculine and feminine agreement. This is certainly due to the homonym declension endings in -us in both the second and the forth declension classes, which later fell together into one declension class (see Jensen 1976: 14). In most of the contemporary Romance languages, this noun triggers feminine agreement (e.g., in Aragonese, Corsican, French, Friulian, Italian, Ladin, Occitan, Romanian, Sardinian, and Spanish). In Romansh it is still a gender doublet, triggering both masculine and feminine agreement. In medieval times, the same also applied for Old Occitan and Old French, where the noun existed as an agreement controller for both genders. This noun shows considerable variation in gender in Old Occitan; it has the strong tendency to be used as a feminine noun, but it is nevertheless clear that there is variation in gender (see Tables 8.30 and 8.31 for an overview; see also Wiedner (2026: 84-87)). Especially the two examples from the Roman de Flamenca are very interesting, where this noun appears within a single text and only a few verses apart in very similar syntactic environments once as a masculine, the other time as a feminine noun (see Examples (61a) and (61c)). In general, there are too many masculine occurrences for them to be only mistakes in writing or uncertainties due to potential contact with other languages. As there are no other gender doublets from the forth declension class 14 , it is not possible to give a more general statement on gender development for nouns of this declension class. Lemma M Ldoc F Ldoc M Prov F Prov M Lit F Lit Gender doublet man X X X X X X X Table 8.30: Gender doublets from the fourth declension class. 14 This fact is certainly linked with the very small number of nouns already in Classical Latin; only 1.4% of the most frequent nouns belong to this declension class (see Polinsky/ van Everbroeck 2003: 362). <?page no="191"?> 8.2 Inherited Nouns from Latin 191 For the corpus search, the following graphical variants were used: ma, maa, maas, mag, main, mam, man, mand, mandz, manh, mang, mano, mans, mant, mantz, manx, manz, mas, maynh, mo, and mos. Table 8.31 shows that man existed with both genders regardless of the region and the number, but only in the oblique case, which is certainly due to the preferred argument structure, which states that the subject of a phrase is the doer and therefore mainly the agent. Usually, these agents are conscious beings with their own will or forces of nature (see Velupillai 2012: 236), both of which is not the case for the hand, as it is controlled by the human being to whom it is attached. It is not unthinkable of the hand being the subject of a certain phrase, but within the occurrences in COMETA (2025), it only appears as an object. Languedocian Provençal num+gen+case num+gen gen M sg nom 0 0 0 16 30 obl 9 7 16 pl nom 0 0 0 14 obl 7 7 14 F sg nom 0 0 0 180 425 obl 57 123 180 pl nom 0 0 0 245 obl 64 181 245 NA sg nom 0 0 0 28 67 obl 7 21 28 pl nom 0 0 0 39 obl 15 24 39 Table 8.31: Gender distribution for man. Examples in (60) show both masculine and feminine occurrences from Languedocian and Examples in (61) from Provençal. The Examples (60a) and (60b) show the feminine agreement and Examples (60c) and (60d) masculine agreement in Languedocian, always beginning with the singular example, followed by the plural one. The occurrences in this area appear only with the determiner as an agreement target, so there is only anaphoric gender marking for this noun in the Languedocian sub-corpus. (60) a. Mes la man a la espea ‘He puts the F hand on the sword’ (BnF, Français 25425 (1275: f o . 69 r o ., l. 18)) <?page no="192"?> 192 8 Gender Doublets in COMETA (2025) b. son sperit endressat los uelhs e las mas vas lo cel ‘His spirit directed the eyes and the F hands to the sky’ (BnF, Français 13504 (14 th century: f o . 19 v o ., l. 13)) c. E menet son efan pel ma ‘And took his child by the M hand’ (BnF, Arsenal 6355 (14 th century: f o . 28 r o ., l. 8)) d. Que ia la tua obra que fist dels mas ambdos. ‘As already your work that you made with both (the M ) hands’ (BL, Harley 7403 (14 th century: f o . 102 v o ., l. 13-14)) The Examples (61a) and (61b) show feminine agreement and Examples (61c) and (61d) masculine agreement in Provençal. In this sub-corpus, we can also find other agreement targets, in this case post-positioned adjectives that agree to the gender triggered by man (see Examples (61a) and (61c)). It is noteworthy that these two examples are both from the same text written by a single scribe and appear within two folios. In addition, these quotations have very similar structures: both are in the singular oblique case and are preceded by an article (feminine or masculine singular, respectively) and followed by the adjective nuz/ nuda ‘naked, bare’, which appears one time in feminine form, the other time in masculine form. The occurrence in Example (61b) is certainly influenced from Latin, as it is the direct translation of a bible verse in Luke 23: 46: “Pater in manus tuas commendo spiritum meum” (Weber/ Gryson 2007: Evangelium secundum Lucam 23: 46; ‘Father, in your hands I command/ commit my spirit’). <?page no="193"?> 8.2 Inherited Nouns from Latin 193 (61) a. Guillem la ma nuda miret ‘Guillem looked at the F bare F hand’ (BmC, Cote 34 (13 th century: f o . 45 v o ., c. 1, l. 5)) b. Paire en las toas mans comant lo meu esperit ‘Father, in your F hands I command/ commit my spirit’ (BnF, NAF 11180 (14 th century: f o . 25 r o ., l. 5-7)) c. Ab lo man ẹ nuz ‘With the M bare M hand’ (BmC, Cote 34 (13 th century: f o . 43 v o ., l. 20)) d. Tantost con hac los mas lavatz ‘As soon as he had washed M his M hands’ (BmC, Cote 34 (13 th century: f o . 54 v o ., l. 8)) The noun man is clearly a gender doublet in Old Occitan. No regional differences could be found; it appears with both genders in both the Languedocian and the Provençal areas. There is a considerable difference in gender assignment with almost 94% of the occurrences where the gender could be determined to <?page no="194"?> 194 8 Gender Doublets in COMETA (2025) be feminine, but there are nevertheless masculine occurrences in both areas. Even within one and the same text, the gender varies between masculine and feminine, so it seems like man can be counted as a real gender doublet with a strong tendency toward feminine agreement. 8.2.4 Fifth Declension Class Just like the nouns from the second and fourth declension classes, there is only one noun from the fifth declension in the list of gender doublets used for this research, namely dia. This noun has already been discussed in an article (see Wiedner 2026). However, the focus here is on variation with regard to the region, which was not addressed in the article. This noun is derived from dies, which was both masculine and feminine in Latin, with a semantic differentiation, meaning ‘day’ as a masculine noun and ‘date’ as a feminine one. In Old Occitan, this noun has the meaning ‘day, term, time frame, delay’, both masculine and feminine. The semantic difference in combination with different gender agreement was lost in the transition to the Romance languages. This noun is inherited to all Romance languages except French and Italian, where it got supplanted by jour or giorno from Latin diuRnum 15 . This noun is masculine in most Romance varieties 16 such as Catalan, Old French, Romansh, and Spanish. It has both genders in Aragonese, Medieval Latin, Old French, Old Sardinian, and Old Occitan. In Ladin, it also shows both genders, but with a difference in meaning: the masculine noun means ‘day’ while the feminine form has the meaning of ‘daylight’. Lemma M Ldoc F Ldoc M Prov F Prov M Lit F Lit Gender doublet dia X only 1 X only 1 X X X, only 11 th / 12 th centuries Table 8.32: Gender doublets from the fifth declension class. As shown in Table 8.32, this noun appears mainly as a masculine noun within the COMETA (2025) corpus, which can also be corroborated by additional literature. 15 In these regions, it only survived in the days of the week, like French lundi or Italian lunedi, and in some fixed forms like French midi. In standard Italian masculine il dì is still used as an archaic term; furthermore, it is used in some regional varieties. 16 Contrary to the statement in the REW (1935: s.v. “dies”), where it is listed the other way around, and dies is said to be feminine in e.g., Catalan, Italian, or Spanish, and masculine only in Old French and the category “other”. <?page no="195"?> 8.2 Inherited Nouns from Latin 195 It has only two feminine occurrences within the whole corpus, one per region. Further occurrences could be found in other texts that are not part of COMETA (2025). It must be noted that all additional feminine evidence is from the 11 th and 12 th centuries. This circumstance suggests that this noun was a gender doublet in older stages of Occitan, but became stabilised as a masculine noun by the time the texts in the COMETA (2025) corpus were written. A limiting factor to be kept in mind for this statement is the fact that dia often appears with cardinal or ordinal numbers written in the Latin numbering system and not all scribes indicate gender with numbers (only some scribes write .I a ., thus showing feminine agreement) and it is therefore not possible to make valid statements on gender assignment for those cases. For the corpus search, the following graphical variants were used: di, dia, dïa, diaas, dias, dialz, die, dier, dies, dïes, diha, dis, dja, dua, duas, dya, dyas, and dye. Languedocian Provençal num+gen+case num+gen gen M sg nom 0 0 0 105 225 obl 11 94 105 pl nom 0 0 0 120 obl 118 2 120 F sg nom 0 0 0 2 2 obl 1 1 2 pl nom 0 0 0 0 obl 0 0 0 NA sg nom 0 0 0 102 177 obl 61 41 102 pl nom 0 0 0 75 obl 68 7 75 Table 8.33: Gender distribution for dia. (62) a. Ben ves que Ja son filh non layssaria nulha dia ‘You see well that already his/ her son would not let any F day go by’ (BnF, NAF 6195 (14 th century: f o . 8 r o ., c. 3, l. 3)) <?page no="196"?> 196 8 Gender Doublets in COMETA (2025) b. Ben ves que ia sos filltz non laysara nul dia ‘You see well that already his/ her son did not let any M day go by’ (BnF, Français 13509 (14 th century: f o . 10 v o ., c. 2, l. 17-18)) c. las nuetz els dias e maiormens las nuetz. ‘The nights and the M days and mainly the nights’ (BnF, Français 1049 (1343: f o . 199 v o ., c. 2, l. 1-2)) (63) a. .III. dias de la semana ordenadamen deiunava e la .I. dia en pa e en ayga ‘Three days a week he regularly fasted and the F one/ the F first day with bread and water’ (BnF, Français 13504 (14 th century: f o . 46 v o ., l. 1-2)) b. Enimia ab son coven Conoc lo dia de sa fi ‘Enimia with her convent knows the M day of her end’ (BnF, Arsenal 6355 (14 th century: f o . 40 r o ., l. 16-17)) <?page no="197"?> 8.2 Inherited Nouns from Latin 197 c. Ni cal be as tu fait als dias de ta vida ‘Nor what good have you done in the M days of your life? ’ (BL, Harley 7403 (14 th century: f o . 85 r o ., l. 1-2)) As Section 8.2.4 shows, the tendency of the Romance languages to use this noun with masculine agreement can be confirmed by the corpus search in COMETA (2025), as only two examples (Example (63a) for the Old Languedocian and Example (62a) for Old Provençal) could be found. All other occurrences are either masculine or unclear. Levy (1894-1924) cites some examples of dia as a feminine noun, but they all are taken from texts from the 11 th and 12 th centuries, which suggests that this noun may have been a gender doublet in older stages of the language, but it seems the masculine gender has prevailed in the period of the COMETA (2025) corpus (see also Wiedner 2026: 96). 8.2.5 Nouns Derived from Other Latin Word Classes The only two hereditary nouns in the list of gender doublets that were derived from other word classes are semal and sera. The first one is probably derived from the not-documented adjective *semalem (see discussion in Section 8.2.5.1), the other one either from the adverb seRo or the adjective seRa. The latter exists in different forms, with and without the word final -a, but has only three certain feminine instances in the Old Occitan texts. It is therefore questionable to list this noun as a gender doublet and it can rather be considered a masculine noun. Lemma M Ldoc F Ldoc M Prov F Prov M Lit F Lit Gender doublet semal — X — — — X — sera only 1 — X — X only 3 only limited Table 8.34: Gender doublets from other Latin word classes. <?page no="198"?> 198 8 Gender Doublets in COMETA (2025) 8.2.5.1 semal The Old Occitan noun semal ‘dumpster, tub, bucket, container, vessel’ has an unclear etymology. The FEW (1922-2003) states the Greek σάγμα ‘pack-saddle’ as the origin, which was then loaned into the Latin first declension class and from there inherited into the Romance languages, where it still exists. In Old Occitan a metonymic change in meaning occurred, from ‘the saddle’ over ‘the animal wearing the saddle’ to ‘the things the animal carries’; and from there to ‘the containers in which the things are transported’. In most Romance languages, the noun kept the meaning of ‘saddle’ or changed to ‘pack animal’, but the changes in meaning did not go further. The nouns inherited from Greek σάγμα are feminine in all Romance varieties. Only if the theory explained here is valid there are a few exceptions in Catalan and Languedocian, with masculine use for the specific signification that only exists in these two languages/ varieties. Against that etymology, Brüch (1931: 67) suggests the adjective *semalem ‘half-full’ as the origin for the forms in Old Occitan and Catalan, because the noun with this specific meaning exists only in Languedocian and Catalan (see Brüch 1931: 61). He lists only one example in the Occitan-speaking area from outside of the Languedocian area, found in a text written in the Béarn (see Brüch 1931: 63). If Brüch’s theory is true, only the gender in Catalan and Languedocian would be of interest for this research. Brüch (1931: 64-65) gives proof for mainly feminine use in both Catalan and Languedocian, with some rare masculine occurrences, which he leads back to masculine nouns with similar semantics ending in -al, i.e., Catalan barral ‘barrel’. For the corpus search, the following graphical variants were used: semal, semales, semalis, semals, and simal. Languedocian Provençal num+gen+case num+gen gen F sg nom 0 0 0 0 1 obl 0 0 0 pl nom 0 0 0 1 obl 1 0 1 NA sg nom 1 0 1 2 2 obl 0 1 1 pl nom 0 0 0 0 obl 0 0 0 Table 8.35: Gender distribution for semal. As Table 8.35 shows, there are only three occurrences of semal in the COMETA (2025) corpus, two of which are not clearly identifiable concerning gender, and the remaining one is feminine, see Example (64). All occurrences are found in the Languedocian sub-corpus, supporting the findings of Brüch (1931). <?page no="199"?> 8.2 Inherited Nouns from Latin 199 (64) son las plenas semals. ‘are the F full F dumpsters’ (BnF, Français 25425 (1275: f o . 159, l. 9)) There is only one feminine occurrence in COMETA (2025), but no masculine one. Raynouard (1836-1845) does not list this noun; it is first listed in Levy (1894- 1924), who gives one example of masculine use, taken from a charter from Montpellier: “et sos truels (? ) at sos (sic) semals et tota sa ferramenta” (Levy 1894-1924: s.v. semal; italics in the original; ‘And his trowels and his M buckets, and all his tools’). The fact that Levy (1894-1924) added the sic makes it clear that for him, semal is only a feminine noun and this single evidence of masculine use is a writing error for him. Brüch (1931: 64) comments that it may be a writing error triggered by the preceding sos truels, but he goes on to argue that this theory has to be rejected, because there is proof for masculine use of the noun in contemporary Languedocian. I do not think his argument is valid, as the variation in gender may also be a newer phenomenon, and the fact that there is variation in gender in a contemporary variety does not per se prove that there was variation in an older stage of that variety. It seems more likely that semal was not a gender doublet in Old Occitan, but only a feminine noun. 8.2.5.2 sera The Old Occitan noun sera ‘evening’ is derived from the Latin adjective seRum ‘late’. Originally, it was used in combination with time indicating nouns, e.g., seRus vespeR ‘late evening’ or seRa dies ‘late day’. As a consequence of the former Latin term for ‘evening’ (vespeR) being adapted in the context of Christianity for the evening prayer, seRa dies took over, often used without dies and thus being used as a feminine noun. In the Gallo-Romance area, additional forms derived from the Latin adverb seRo ‘in the evening’ exist. Forms inherited from seRa exist as a feminine noun in Corsican, Friulian, Italian, Ladin, and Romanian. In the Gallo-Romance area, the masculine form derived from seRo prevails, and variation in gender is only shown in Old Occitan 17 . In the Ibero-Romance areas, 17 The FEW (1922-2003: s.v. “sero”) sees the high frequency of sera-occurrences in combination with the masculine noun matin ‘morning’ as an explanation for the masculine use of sera in Old Occitan. An additional influence may be the French noun soir which is also masculine. It is not clear why the FEW (1922-2003: s.v. “sero”) does not indicate the etymology with seRo <?page no="200"?> 200 8 Gender Doublets in COMETA (2025) either the form derived from vespeR continued to be used (i.e., Catalan vespre) or it was replaced by a new form from Latin taRde ‘late’, e.g., Spanish tarde. For the corpus search, the following graphical variants were used: sera, seras, sere, seres, sero, and seros. Languedocian Provençal num+gen+case num+gen gen M sg nom 1 0 1 9 9 obl 0 8 8 pl nom 0 0 0 0 obl 0 0 0 NA sg nom 0 0 0 3 3 obl 0 3 3 pl nom 0 0 0 0 obl 0 0 0 Table 8.36: Gender distribution for sera. Table 8.36 shows the gender distribution for sera within the COMETA (2025) corpus. There is only one masculine occurrence in the Languedocian area (see Example (65) with the masculine past participle as the agreement target) and eight masculine occurrences in the Provençal part (see Example (66)); there is no feminine evidence within COMETA (2025). (65) Item que quan sera es devengutz ‘Item of when the evening arrived M ’ (BnF, NAF 11151 (14 th century: f o . 19 r o ., l. 18)) (66) Venguessan as ostal lo sera ‘In the M evening, they came to the house’ (BnF, NAF 6195 (14 th century: f o . 22 v o ., l. 11)) The state of sera as a gender doublet as stated in the DOMél (2025) cannot be confirmed within COMETA (2025). Therefore, a further look into the Old Occitan literature is necessary. Raynouard (1836-1845) lists it as a masculine noun as a reason for masculine use in Old Occitan, even though it is the reason for masculine use in French. <?page no="201"?> 8.3 Greek Loanwords 201 in the forms derived from seRo, and only as a feminine noun for the occurrences with the form sera, contradicting the findings in the COMETA (2025) corpus. He gives two examples for the feminine usage: “Venc si c’una seRa el venc a Javiac” (Raynouard 1836-1845: s.v. “sera”; small caps in the original; ‘It happened that one F evening he came to Javiac’) and “Des la seRa tro lo mati” (Raynouard 1836-1845: s.v. “sera”; small caps in the original; ‘From the F evening until the morning’). Levy (1894-1924) tries to list it as a feminine noun, arguing that forms like “mi colc al sera” ‘I go to bed in the ? evening’ could also be feminine occurrences in analogy to the occurrence “Las maias que · l seras son fachas” (Levy 1894-1924: s.v. “sera”; italics in the original; ‘The tables are set F for the ? evening’). All other occurrences are masculine and the first example given by Levy (1894-1924) remains debatable. Appel (1930: 191) cites a single feminine occurrence of sera in the Vida of Gaubertz de Poicibot, while three other manuscripts of the Vida use it as a masculine noun. One additional feminine occurrence could be found: “alberguet una sera en la ciutat” (Favati 1961: 319; ‘Stayed one F evening in the city’). After looking into the dDOM (2025) and additional literature, a total of only three feminine occurrences could be found; all other occurrences with identifiable gender are masculine. So this noun is predominantly masculine in Old Occitan, which is supported by the findings in the COMETA (2025) corpus. 8.3 Greek Loanwords Table 8.37 gives a summary of gender distribution for the Greek loanwords in the COMETA (2025) corpus and the literature. As before with respect to Table 8.3, for the most frequent word (profeta), the data from additional literature is not presented explicitly, but checked before filling in Table 8.37. It is quite interesting to see that (apart from senebe) all Greek loanwords in this corpus are animate nouns, ending in -a. As discussed in Section 3.2, this vowel is often associated with feminine gender, which contributes to the fact that these Greek loanwords, which mainly designate genuine male concepts, trigger feminine agreement as well (see also Jensen 1973). This is also supported by the findings in Basirat/ Allassonnière-Tang/ Berdicevskis (2021), which suggest a superior role of formal cues over semantic cues in gender assignment. Especially papa ‘pope’ and patriarca ‘patriarch’ are both purely male concepts, as well as evangelista ‘evangelist’, as all four evangelists were men. But also salmista and propheta often designate males, yet the latter sometimes even triggers feminine agreement when it clearly describes a man as shown in Examples (76a) and (76b). The instances of feminine agreement are certainly due to the (morpho-)phonological appearance resembling feminine nouns in -a. It seems like there are some <?page no="202"?> 202 8 Gender Doublets in COMETA (2025) Lemma M Ldoc F Ldoc M Prov F Prov M Lit F Lit Gender doublet evangelista — only 1 only 1 — — — unclear fantasma — only 1 — — — X — papa X only 1 X — X X X patriarca X few X — X only 1 X profeta X X X X X X X salmista X — — — X only 2 X, but limited senebe — — only 1 — X X X Table 8.37: Greek loanwords as gender doublets. (more or less) rare instances where the semantics are more in the background and the (morpho-)phonology is in the foreground. The statement by Jensen (1973: 393) that the default gender for these Greek nouns is feminine, however, cannot be verified; all these nouns apart from evangelista (with one masculine and one feminine occurrence) appear in most cases with masculine agreement in COMETA (2025), which makes masculine the default gender and feminine the exception. It is therefore even more remarkable to see feminine agreement with explicitly named male referents, such as Isaiah in Example (76a) or David in Example (76b). An interesting fact that should not be forgotten is the necessity of the Consistori del Gay Saber to include a metalinguistic commentary (see Example (70)) in their grammar. This means certainly that there was considerable variation (or confusion) in gender assignment at least for papa and profeta, and potentially also for other Greek loanwords. With this comment it is clear that variation in gender for these nouns is not something purely graphical (or even mistakes in writing), but a variation that is so present in the language that the Consistori del Gay Saber felt the need to point this out to their fellow poets, aiming at normalising the language. 8.3.1 evangelista This noun is the first one in the list of gender doublets that is not directly inherited from Latin, but a loanword from Greek, from the noun εὐαγγελιστής ‘bringer of good news’. It was masculine in Greek and adapted as a masculine <?page no="203"?> 8.3 Greek Loanwords 203 noun in Latin, but attributed to the first declension class (with mainly feminine nouns). In most Romance languages (e.g., in Catalan, French, Italian, Romansh, or Spanish), this noun stayed masculine, which goes hand in hand with the semantics—the four evangelists being all men. For the corpus search, the following graphical variants were used: avangelista, avengelista, euangelista, euvangelista, evangelista, and evangeliste. Languedocian Provençal num+gen+case num+gen gen M sg nom 0 0 0 0 1 obl 0 0 0 pl nom 0 0 0 1 obl 0 1 1 F sg nom 0 0 0 0 1 obl 0 0 0 pl nom 1 0 1 1 obl 0 0 0 NA sg nom 4 0 4 4 4 obl 0 0 0 pl nom 0 0 0 0 obl 0 0 0 Table 8.38: Gender distribution for evangelista. (67) Las quatre evangelistas acostadamens ‘The F four evangelists usually’ (BL, Harley 7403 (14 th century: f o . 77 v o ., l. 11-12)) (68) Et el li dis de las prophetas e dels apostols els evangelistas ‘and he told them of the prophets and the apostles and the M evangelists’ (BnF, Français 1049 (1343: f o . 197 r o ., c. 1, l. 28-29)) <?page no="204"?> 204 8 Gender Doublets in COMETA (2025) This noun occurs only very rarely within COMETA (2025), mostly not showing a clearly identifiable gender. Only one occurrence in the plural per region can be attributed to a certain gender: Example (67) shows feminine agreement in Old Languedocian and Example (68) masculine agreement in Old Provençal. For evangelista the dDOM (2025) lists almost exclusively occurrences as a masculine noun; there is only one additional source for a feminine occurrence, which in fact does not display a feminine occurrence: “Auzir, seinor, podet que uos demostra Sans Lux euangelista” (Armitage 1884: 19; ‘Listen, sir, may Saint Luke [the] Evangelist demonstrate to you’). All other sources in the dDOM (2025) indicating a gender value list it as being masculine, but most of them do not have agreement targets reflecting their gender and therefore cannot be assigned a gender with certainty. The only other source really showing masculine agreement has two occurrences of evangelista; the one remains uncertain (“Car, so dis l’avangelista” Heuckenkamp 1912: 23; ‘Because, so says the ? evangelist’), and the other is the same as in Example (68), which seems to remain the only occurrence of evangelista as a masculine noun. There are many cases in which the gender is not clear. The DAOA (2009) lists evangelista as a masculine noun, but provides only one example in which the gender remains uncertain: “la festa de saint Mathieu l’appostol et evangelista” (1444, CCOnsSFlour 49, 67 v , cited after DAOA 2009: s.v. “evangelista”; ‘The feast of Saint Matthew, the apostle and evangelist’). Nevertheless, Raynouard (1836- 1845) lists it exclusively as a masculine noun, but does not give an example of this noun as such. Levy (1894-1924) lists it as a gender doublet, but once again does not show any occurrences as a masculine noun. Another example can be found in Appel (1930: 222), where e.g., “ci cum ditz Sanhz Ioanns evangelista” ‘Here as Saint John [the] Evangelist says’ is listed as being a masculine occurrence. It is in relation to a male name, but does not control any agreement targets. Therefore, no valid statement on gender can be made. The other sources in the dDOM (2025) are all very similar in that respect, as they all appear in combination with a masculine proper noun, but do not give other indications on gender, so they cannot be considered evidence for masculine agreement. Due to this fact, Jensen (1973: 399) states that evangelista actually appears only as a feminine noun in Old Occitan; the only potential masculine occurrence he could retrieve is in reality of uncertain gender (“La locution evangelista benastruc me paraît donc trop ambiguë pour trancher la question de genre” Jensen 1973: 399; italics in the original). 8.3.2 fantasma The noun fantasma ‘phantom, illusion, visionary’ is inherited from the Greek neuter noun φάντασμα ‘appearance, image, imagination, face, wonder, ghost’. <?page no="205"?> 8.3 Greek Loanwords 205 It was taken as a loanword in Latin, stayed a neuter noun, and was integrated into the third declension class (phantasma, phantasmatis). The forms of fantasma are masculine in Italian, Portuguese, Romansh, and Spanish. It is mainly masculine in Catalan, but also exists colloquially as feminine, and it is feminine in Aragonese, Galician, and Romanian. This noun shows both genders in contemporary Occitan and Franco-Provençal. For the corpus search, the following graphical variants were used: fantauma, fantasma, and fantasmas 18 . Languedocian Provençal num+gen+case num+gen gen F sg nom 0 0 0 1 1 obl 1 0 1 pl nom 0 0 0 0 obl 0 0 0 NA sg nom 0 0 0 5 5 obl 3 2 5 pl nom 0 0 0 0 obl 0 0 0 Table 8.39: Gender distribution for fantasma. (69) una fantauma de femna hay vista. ‘You have seen a F phantom of a woman’ (BmT, Cote 2883 (14 th century: f o . 168 v o ., c. 1, l. 27-29)) Fantasma appears exclusively as a feminine noun in the COMETA (2025) corpus. Raynouard (1836-1845) lists it as both a masculine and feminine noun for the signification of ‘ghost, phantom’, and as solely feminine noun for the other meanings. He gives only one example where the gender can be determined with certainty, which is a feminine occurrence (“Totas fansasmas de sa cogitacio” Raynouard 1836-1845: s.v. “fantasma”; small caps in the original; ‘All F illusions of his/ her thought’). Levy (1894-1924) gives an additional occurrence, which is still feminine and is the same as previously shown in Example (69). The dDOM (2025) indicates three more sources for fantasma with masculine agreement: the first one is in Wunderli (2010: 178), where he indicates it in his glossary as masculine. A closer look at the two passages reveal that in reality it is not possible to 18 In addition to these forms, the respective forms with initial <ph> were searched as well. <?page no="206"?> 206 8 Gender Doublets in COMETA (2025) make a statement on the gender: “Que fantauma es? ” (Wunderli 2009: 69; ‘What illusion is it? ’); “cugero se que fo<s> fantauma, e cridero” (Wunderli 2009: 120; ‘I thought it was [a] ghost, and I shouted’). The same applies for the following citations where the glossary also indicates both masculine and feminine agreement: “E ieu me cugiey que fos fantasma e comensiey a dire mas oracios” (Tausend 1980: f o . 121d, l. 27; ‘And I thought that it was [a] ghost and began to say my prayers’) and “albira te si yeu fos fantasma, si no fugira” (Tausend 1980: f o . 122a, l. 2; ‘Notice if I were [a] ghost, I would not flee’). 8.3.3 papa The Old Occitan noun papa ‘pope’ is a Greek loanword from πάππας ‘father, educator (of a boy)’. From the 3 rd century AD onward, it was used in the context of Christianity as an honorary title for bishops and from somewhere between the 6 th or 9 th century onward only for the bishop and patriarch of Rome, the pope. Papa is used as both a masculine and feminine noun in Old French. This is explained in the FEW (1922-2003) via the ending in -a, commonly found in feminine nouns in the Romance languages. Thus, this noun has the phonological appearance of a feminine noun which may have triggered feminine usage of this noun. Papa is used as a masculine noun in most Romance languages, but is also documented with feminine use in Old French and Old Occitan. For the corpus search, the following graphical variants were used: papa, papas, pape, papes, and papo. Languedocian Provençal num+gen+case num+gen gen M sg nom 40 12 52 61 62 obl 9 0 9 pl nom 0 0 0 1 obl 1 0 1 F sg nom 1 0 1 1 1 obl 0 0 0 pl nom 0 0 0 0 obl 0 0 0 NA sg nom 48 6 54 56 57 obl 2 0 2 pl nom 1 0 1 1 obl 0 0 0 Table 8.40: Gender distribution for papa. For the noun papa, only one single feminine occurrence could be found in the COMETA (2025) corpus. But, as the following Example (70) shows, it is part of a meta-linguistic commentary by the authors of the Leys d’Amors, the Consistori del Gay Saber, and therefore cannot be counted as a real feminine occurrence. <?page no="207"?> 8.3 Greek Loanwords 207 The Consistori del Gay Saber observe the usage of both papa and propheta with feminine agreement, which they deem as incorrect and therefore include the following passage in their grammar: (70) per que se peco cil que dizo […] la propheta la papa. quar devon dir tug li orde le propheta. le papa ‘because one makes a mistake who says […] the F , prophet the F pope. Because they shall say all the time the M , prophet the M pope’ (BmT, Cote 2883 (14 th century: f o . 125 r o ., c. 1, l. 13-16)) Apart from this occurrence, all other instances of papa are either masculine or not attributable to a gender with certainty. There is only one occurrence of this noun in the plural, which is probably due to the fact that there is usually only one (recognised) pope at a time, which make instances for plural usage rather rare. The only plural occurrence with a certain gender, found in Languedocian, can be seen in Example (71c). (71) a. lo sant papa. leo fe son sermo ‘The M holy M pope Leo said his sermon’ (BnF, Français 2232 (1325: f o . 7 v o ., l. 5)) b. Parlaran am lo papa ‘They will speak to the M pope’ (BnF, Français 25425 (1275: f o . 7 r o ., l. 2)) <?page no="208"?> 208 8 Gender Doublets in COMETA (2025) c. Crezi qu aytals vertutz passet Als papas que venon apres ‘I believe that these virtues pass on to the M popes that come afterwards’ (BmT, Cote 2883 (14 th century: f o . 111 r o ., c. 2 l. 5-6)) (72) a. Per que pueys le papa leons Am princes et am motz barons ‘Because then the M pope Leon with princes and with many barons’ (BnF, NAF 6195 (14 th century: f o . 19 v o ., c. 1, l. 19-20)) b. Aysi dis com sant eugenis le papa venc ‘So it is told how the M pope Saint Eugenis comes’ (BnF, Français 13509 (14 th century: f o . 1 v o ., l. 7)) Papa appears only as a masculine noun within COMETA (2025), taking the metalinguistic commentary aside. This corresponds with Raynouard (1836-1845) who also lists papa only as a masculine noun. Levy (1894-1924) on the other hand supplements three occurrences of this noun with feminine agreement, e.g., “e mortz e vencutz ei’pretz confondutz. Roma, la vostra papa fai aitals vertutz” (see also Bartsch/ Koschwitz 1904: 223; ‘And dead and defeated, priests are confounded. Rome, your F pope does such deeds’). In the glossary for that chanson, Cavaliere (1938: 602) adds a masculine occurrence just a few verses later within the same text, which in fact does not show a clear gender value “Roma, mal labot fa · l papa, quan tensona” (Cavaliere 1938: 378; ‘Rome, the ? pope does bad work when there is tension’). Jensen (1973: 399) gives an additional example: “mandet sos mesatgiers a la papa” ‘He sent his messengers to the F pope’. There are in fact some instances of papa with feminine agreement, so it can be considered a gender doublet, even though it does not appear as such in the <?page no="209"?> 8.3 Greek Loanwords 209 COMETA (2025) corpus. Because both Levy (1894-1924) and Jensen (1973) do not indicate the time nor the place where and when the respective manuscript was written, it is not possible to know with certainty whether the variation in gender for papa exists only in another area or time and is therefore not present in the COMETA (2025) corpus, or whether there are other reasons. 8.3.4 patriarca Inherited from the Greek masculine noun πατριάρχης ‘progenitor, prime father’, this noun was loaned into ecclesiastical Latin as a masculine noun belonging to the first declension class for the prime fathers of Israel in the New Testament. From the 4 th century onward it was used as an honorary title for the five most distinguished bishops (the bishops of Rome, Antioch, Alexandria, Jerusalem, and Constantinople). It was taken into Old Occitan with this meaning, in connection with the crusades, re-establishing some of these patriarchies being one of the main goals. This noun is masculine in all Romance languages. For the corpus search, the following graphical variants were used: patriarca, patriarcas, patriarcha, and patriarchas. Languedocian Provençal num+gen+case num+gen gen M sg nom 9 1 10 17 19 obl 2 5 7 pl nom 0 0 0 2 obl 0 2 2 F sg nom 0 0 0 2 3 obl 2 0 2 pl nom 1 0 1 1 obl 0 0 0 NA sg nom 0 0 0 3 3 obl 0 3 3 pl nom 0 0 0 0 obl 0 0 0 Table 8.41: Gender distribution for patriarca. Table 8.41 shows the gender distribution for the Old Occitan noun patriarca. In the Languedocian area, most occurrences are singular and most of them are also masculine, both in the nominative (see Example (73d)) as well as in the oblique case (see Example (73e)). However, in the Languedocian sub-corpus, three feminine occurrences could be found, two in the oblique singular (see Examples (73a) and (73b)) and one in the nominative plural (see Example 73c)). Two of the feminine occurrences stem from the manuscript BnF, Français 2232 (1325), in which masculine occurrences can also be found. The third one is in manuscript <?page no="210"?> 210 8 Gender Doublets in COMETA (2025) BL, Harley 7403 (14 th century), where it is the only occurrence of this noun, so no variation within this specific manuscript could be found. All occurrences from Provençal, for which the gender is identifiable, are masculine: singular nominative (see Example (74a)) and oblique (see Example (74b)), as well as both plural oblique occurrences (see Example (74c)). (73) a. Lendema mati .Karles. fe aportar la patriarcha de iherusalem elh avesque de limosi e totz los autres ‘The next morning, Charles made the F patriarch of Jerusalem and the bishop of the Limousin and all the others come’ (BnF, Français 2232 (1325: f o . 96 v o ., l. 14-16)) b. per los defuns elhs van sebelir la patriarcha de iherusalem ‘during the funeral they will bury the F patriarch of Jerusalem’ (BnF, Français 2232 (1325: f o . 102 v o ., l. 12-13)) c. Perque las patriarcas prophetas issament ‘As the F patriarchs and prophets equally’ (BL, Harley 7403 (14 th century: f o . 95 r o ., l. 10-11)) <?page no="211"?> 8.3 Greek Loanwords 211 d. que vic qu elh patriarcha de iherusalem fo mortz hac gran dolor ‘who saw, that the M patriarch of Jerusalem had died, felt great pain’ (BnF, Français 2232 (1325: f o . 91 v o ., l. 12-13)) e. la mort delh patriarcha. ‘The death of the M patriarch.’ (BnF, Français 2232 (1325: f o . 91 v o ., l. 16)) It is noteworthy to see variation in gender within a single manuscript, with patriarca triggering both masculine and feminine agreement in manuscript BnF, Français 2232 (1325), at least in the oblique case (see Examples (73a) and (73b) for feminine agreement and Example (73e) for masculine agreement). Additionally, a look into the other manuscript of the Roman de Philomena reveals that all occurrences of patriarca in that manuscript are in fact masculine, so the variation in gender for this text seems to have a correlation with the scribe of the manuscript. However, as there is another instance of patriarca as a feminine noun (see Example (73c)), this variation in gender is not explainable only through a single scribe. (74) a. Aytals era lo gran patriarcha sant abraham ‘Such was the M great patriarch saint Abraham’ (BnF, Français 1049 (1343: f o . 97 r o ., c. 2, l. 2-4)) <?page no="212"?> 212 8 Gender Doublets in COMETA (2025) b. E venc s en complanher al sant patriarcha. ‘And he comes to complain about it to the M saint M patriarch.’ (BnF, Français 1049 (1343: f o . 131 v o ., c. 2, l. 4-5)) c. ayso es lo matremoni dels patriarchas. ‘Such is the marriage of the M patriarchs’ (BnF, Français 1049 (1343: f o . 156 v o ., c. 1, l. 24-25)) The data presented above makes clear that there is at least some variation in gender assignment for patriarca and that this variation was not caused by a single scribe. This suggests that gender variability was only present in Languedocian but not in Provençal. However, there ar not enough occurrences of this noun to be sure of a diatopic variation. Raynouard (1836-1845) lists it as only a masculine noun, but Levy (1894- 1924) gives additional examples of patriarca as a feminine noun. He cites the same examples that are also in the COMETA (2025) corpus, and one additional example: “Paradis on a Deus plantaz . . las patriarchas . ., las prophetas” (Levy 1894-1924: s.v. “Patriarca”; italics in the original; ‘Paradise where God has planted… the F patriarchs…, the prophets’). Jensen (1973: 399) gives the same examples as Levy (1894-1924), and these are the only evidence for patriarca as a feminine noun listed in the dDOM (2025). There is variation in gender, but in total only four instances of patriarca as a feminine noun could be found, and the main gender for this noun seems to be masculine. 8.3.5 profeta The ancient Greek noun προφήτης ‘interpret, advocate, speaker of God’ was a loanword already in Latin, integrated into the language in the context of Christianity and then transmitted into the Romance languages. This noun was mas- <?page no="213"?> 8.3 Greek Loanwords 213 culine in Greek, but a distinct feminine form προφῆτις also existed. The gender assignment was made with respect to the sex of the person denominated by this noun. In Latin, it was adopted as a masculine noun from the first declension class, but appeared with feminine gender when denominating a woman, and sometimes also as a default gender. In Old French and Old Catalan, this noun was mainly feminine, and in contemporary French and Catalan it is masculine as a default gender. This is also the case in the other Romance languages, e.g., in Aragonese, Galician, Italian, Portuguese, and Spanish. These languages developed a proper feminine form, such as Italian la profetessa or Portuguese a profetisa. For the corpus search, the following graphical variants were used: profet, profeta, profetas, profeto, proffeta, prophet, propheta, prophetas, prophete, prophetes, and propheto. Languedocian Provençal num+gen+case num+gen gen M sg nom 10 16 26 38 43 obl 6 6 12 pl nom 2 0 2 5 obl 2 1 3 F sg nom 1 8 9 10 19 obl 1 0 1 pl nom 3 3 6 9 obl 1 2 3 NA sg nom 1 1 2 7 15 obl 2 3 5 pl nom 7 0 7 8 obl 1 0 1 Table 8.42: Gender distribution for profeta. As stated above, the interesting cases here are those where the noun is used with default gender, referring to no specific person, or those where a specific person is denominated, but the grammatical gender does not correspond to the sex of that person. As described in Chapter 4, the default gender in Old Occitan is masculine, so in general, a masculine agreement could be expected for epicene nouns referring to human beings without a specific referent. Jensen (1973: 393, 1976: 76-77) states that in Old Occitan this rule does not apply for Greek loanwords ending in -a, as he expects feminine to be the default gender for those nouns. Because this noun exists in all genders, cases, and numbers in both regions, the focus will first be on Languedocian and afterward on Provençal. In the Languedocian sub-corpus, we find both feminine and masculine occurrences as a default gender when an unspecified prophet or a group of prophets is referenced. For feminine use in the Languedocian area, see Example (75a) for singu- <?page no="214"?> 214 8 Gender Doublets in COMETA (2025) lar, Examples (75b) and (75c) for plural; for masculine occurrences, see Examples (75d) and (75e) for singular and Examples (75f) and (75g) for plural. It must be specified that all feminine occurrences of this noun in the Languedocian subcorpus can be found in manuscript BL, Harley 7403 (14 th century), where this noun also triggers masculine agreement. In the other two manuscripts (BmT, Cote 2883 (14 th century) and BnF, Français 13504 (14 th century)), it has only indeterminable or masculine occurrences, with the sole exception of a feminine occurrence in the meta-linguistic commentary, shown above in Example (70). This example also displays the only feminine singular nominative occurrence in Languedocian and is therefore in reference to an unspecified prophet. (75) a. dis per la boca de la profeta. prenes l albre real ‘he says through the mouth of the F prophet. take the royal tree’ (BL, Harley 7403 (14 th century: f o . 47 v o ., l. 18 and f o . 48 r o ., l. 1)) b. Segon que las prophetas solian prezicar ‘According to what the F prophets used to say’ (BL, Harley 7403 (14 th century: f o . 116 v o ., l. 17-18)) c. E de part totas las prophetas ‘Apart from all F the F prophets’ (BL, Harley 7403 (14 th century: f o . 9 r o ., l. 5)) <?page no="215"?> 8.3 Greek Loanwords 215 d. D’aquest iutiamen parla le propheta e ditz ‘From this judgment speaks the M prophet and says’ (BmT, Cote 2883 (14 th century: f o . 41 r o ., c. 1 l. 16-17)) e. Dieus es drechura segon lo dig del propheta can ditz. ‘God is rightfulness according to the saying of the M prophet when he says’ (BmT, Cote 2883 (14 th century: f o . 38 v o ., c. 2 l. 29-30)) f. Los autres prophetas apres un pauc ‘The M other M prophets after a while’ (BL, Harley 7403 (14 th century: f o . 28 r o ., l. 3-4)) g. quar promes era estat per los prophetas ‘as it was promised by the M prophets’ (BmT, Cote 2883 (14 th century: f o . 40 v o ., c. 2 l. 2-3)) Just as in the Languedocian sub-corpus, in the Provençal part of the corpus, some instances of the noun with default gender resolution could be found, both in fem- <?page no="216"?> 216 8 Gender Doublets in COMETA (2025) inine (see Example (76c) for singular; (76e) for plural) as well as in masculine form (see Example (76i)). There are also occurrences in the singular with masculine as the default agreement, but as these examples do not add any more information, they are not listed here. In this sub-corpus, this noun appears within 6 different texts, but it only shows internal variation in manuscript BnF, Français 1049 (1343) where it appears both as a masculine and a feminine noun. In the other manuscripts, all occurrences are either masculine (see BL, Additional 10323 (14 th century), BnF, Français 13509 (14 th century), and BnF, NAF 6195 (14 th century)) or feminine (see BmC, Cote 34 (13 th century)). In this sub-corpus, some examples could be found in which the gender does not correspond to the sex of the denominated person. In both Examples (76a) and (76b), the denominated prophet is a man, one time Isaiah, the other time David, but the noun profeta nevertheless triggers feminine agreement in these instances 19 . But this is not the only gender for these two prophets; we can also find masculine agreement for this noun within the same manuscript and with reference to the same two prophets (see Example (76f) with respect to Isaiah and Examples (76g) and (76h) with respect to David). Example (76d) most likely refers to a group of masculine prophets, as it specifies that there is no record of those prophets having wives, which in the temporal context of the genesis of this text in the Middle Ages is a reference to male human beings not having wives. It can therefore be seen as an occurrence of gender-sex discrepancy as well. (76) a. D aquesta razis parla li propheta ysayas ‘From that origin spoke the F prophet Ysayas’ (BnF, Français 1049 (1343: f o . 58 v o ., c. 1, l. 15-16)) 19 As the occurrence in Example (76a) is clearly in the nominative singular, the article li can be identified as a feminine article. <?page no="217"?> 8.3 Greek Loanwords 217 b. dis la propheta david que am mager gaug deu hom complir el mandament de dieu ‘The F prophet David said that one has to fulfill god’s commandments with highest joy’ (BnF, Français 1049 (1343: f o . 85 v o ., c. 1, l. 13-16)) c. e dieus maudis los per la propheta ‘and God cursed them through the F prophet’ (BnF, Français 1049 (1343: f o . 24 v o ., c. 1 l. 20-21)) d. E donx non dis la escriptura que li bonhome ancian e las prophetas agron molhers ‘And thus the scripture does not say that the old ordinary man and the F prophets had wives’ (BnF, Français 1049 (1343: f o . 209 r o ., c. 2 l. 27-29)) <?page no="218"?> 218 8 Gender Doublets in COMETA (2025) e. la predication de las prophetas e dels apostols ti ay dig ‘I have told you the predication of the F prophets and the apostles’ (BnF, Français 1049 (1343: f o . 111 v o ., c. 2 l. 22-26)) f. Ysayas lo propheta vi en esperit aquelas gloriozas nossas ‘Ysayas the M prophet saw in the spirit those glorious weddings’ (BnF, Français 1049 (1343: f o . 71 v o ., c. 1, l. 15-17)) g. On dis lo propheta david el sauteri ‘Where the M prophet David said the psalm’ (BnF, Français 1049 (1343: f o . 83 r o ., c. 1, l. 13-14)) <?page no="219"?> 8.3 Greek Loanwords 219 h. Aysi o dis dieus per lo propheta david. ‘Thus God said it through the M prophet David’ (BnF, Français 1049 (1343: f o . 81 r o ., c. 2, l. 27-29)) i. car dieus o avia promes per los santz prophetas ‘For God had promised it through the M holy M prophets’ (BnF, Français 1049 (1343: f o . 10 v o ., c. 2, l. 8-9)) The data in the COMETA (2025) corpus shows clearly that profeta is a gender doublet regardless of the semantics, triggering both feminine and masculine agreement even for men. It varies in gender in both regions for both numbers and cases, and it seems to vary quite freely. No regularities for a specific gender usage in a specific context could be found. 8.3.6 salmista The Old Occitan noun salmista ‘psalmist’ is a loanword from Greek masculine noun ψαλμιστής ‘one who sings/ writes psalms’. As with most of the other Greek loanwords above, this noun was integrated into Latin in the first declension class, but as a masculine noun. It was taken into the Romance languages without the initial [p], but the onset was later reintroduced in some Romance languages (e.g., French psalmiste vs. Italian salmista). This noun is, with respect to the sex of the referenced person, both masculine and feminine, but the feminine gender is not the default gender for this noun in the Romance languages; it is only used when referring to a woman. For the corpus search, the following graphical variants were used: psalmista, psalmistas, salmista, and salmistas. <?page no="220"?> 220 8 Gender Doublets in COMETA (2025) Languedocian Provençal num+gen+case num+gen gen M sg nom 12 0 12 13 13 obl 1 0 1 pl nom 0 0 0 0 obl 0 0 0 Table 8.43: Gender distribution for salmista. As Table 8.43 shows, this noun appears only in the Languedocian sub-corpus of COMETA (2025); all occurrences are within the manuscript BmT, Cote 2883 (14 th century). Here, it has only singular occurrences and is always a masculine noun. Most instances are in the nominative (see Example (77a)), only one occurrence in the oblique case could be found (see Example (77b)). (77) a. Segon que fa testamoni le psalmista. ‘According to the testimony the M psalmist’ (BmT, Cote 2883 (14 th century: f o . 2 v o ., c. 1 l. 14-15)) b. Et allega lo psalmista coma dessus. ‘And alleges the M psalmist as above.’ (BmT, Cote 2883 (14 th century: f o . 62 r o ., c. 2 l. 22-23)) There is no evidence for this noun as a feminine noun in the COMETA (2025) corpus, which aligns with Raynouard (1836-1845), who lists the noun exclusively as masculine. Levy (1894-1924) on the other hand supplements that the noun also exists with feminine agreement, and gives one example: “Adonc fo adumplit zo que diz la psalmista” (Levy 1894-1924: s.v. “Salmista”; italics in the original; ‘Then was fulfilled what the F psalmist says’). Wahl (1980: 32) adds an additional feminine occurrence: “cum di li salmista” ‘As the F psalmist says’; the other occurrence within this text is not clearly identifiable with respect to its gender, as there is no agreement target. <?page no="221"?> 8.3 Greek Loanwords 221 As the sources above demonstrate, there is some variation in gender, but the main gender for this noun remains masculine, as only two feminine occurrences could be found, and at least 13 masculine instances. 8.3.7 senebe The Old Occitan noun senebe ‘mustard’ is a loanword from the Greek neuter noun σίναπι and is the only Greek loanword in the list of gender doublets addressed in this study that does not designate an animate entity. In ancient Greek, it had the same meaning and was first loaned into Latin and afterward inherited into all Romance languages except Romanian. It was derived from two different forms of the noun, in French, Italian, and Occitan from a form with the stress on the antepenultima and in the other Romance languages from a form with the stress on the penultima. All of these forms are masculine in the respective Romance languages and varieties except Aragonese, where it exists as a feminine noun. For the corpus search, the following graphical variants were used: cerba, cerbe, cheneve, senebe, serbe, and seneva. Languedocian Provençal num+gen+case num+gen gen M sg nom 0 0 0 1 1 obl 0 1 1 pl nom 0 0 0 0 obl 0 0 0 Table 8.44: Gender distribution for senebe. As Table 8.44 shows, only one instance of this noun could be found, a masculine singular oblique occurrence from Provençal, see Example (78). (78) Lo gran del senebe es mot petit. ‘The seed of the M mustard is very small’ (BnF, Français 1049 (1343: f o . 101 v o ., c. 1, l. 26-27)) Because there is only one occurrence of this noun in the COMETA (2025) corpus, no statement on the gender of senebe can be made upon the basis of this corpus. Both Raynouard (1836-1845) and the DAOA (2009) list it as a masculine noun, without giving an example with identifiable gender. Levy (1894-1924) <?page no="222"?> 222 8 Gender Doublets in COMETA (2025) supplements it with one masculine example (“e retenc hi lo senebe a sa taula” (Levy 1894-1924: s.v. “senebe”; italics in the original; ‘And I keep the M mustard at his/ her table’)) and three feminine occurrences: “mala serbe per mel” ‘Bad F mustard for honey’, “cels al gra de la serbe” ‘heaven to the grain of the F mustard’, and “plus del vin en doble que de la cerbe” ‘More than double the wine compared to the F mustard’ (Levy 1894-1924: s.v. “senebe”; italics in the original). Bartsch/ Koschwitz (1904: 199) and W. Koch (1897: 40) give one supposedly masculine example, that is in fact of uncertain gender: “tot lo pezan de dos deniers, de serbe trenta gras entiers” ‘The whole weight is two denarii, of mustard thirty whole grains’. In the dDOM (2025), an occurrence in Wunderli (2009) and Noulet (1880) is cited as masculine, even though that specific occurrence is in fact feminine: “Si co · l gra de la serbe” (Wunderli 2009: 114; Noulet 1880: 118; ‘If as the grain of the F mustard’). As the data presented here shows, senebe is in reality a gender doublet, but due to lack of more occurrences, no statement on a potential regional or temporal distribution of that variation can be made. 8.4 Other Loanwords There is only one other loanword in the list of gender doublets of the DOMél (2025) and present in COMETA (2025), namely the noun tabor (see Table 8.45, also discussed in Wiedner (forthcoming) with respect to four of the five occurrences), a loanword from Persian tabiR ‘kettle, timbal’. It has the same meaning in Old Occitan. The REW (1935) lists a second etymon, the Arabic noun tanbuR ‘string instrument’, which exists in Catalan and Spanish. The FEW (1922-2003) sees only the first one as the etymon and the contact to the latter (through the Ibero-Romance varieties) only as the reason for the addition of the -min tambor. As the directly loaned form does not exist in any Romance varieties except Old Occitan, no statement on the gender outside of the Occitan linguistic area can be made. The loanwords from the Arabic noun tanbuR are masculine in Aragonese, Catalan, and Spanish, which may have had an influence on the gender assignment in Old Occitan. For the corpus search, the following graphical variants were used: tabor, taborn, tabors, tabour, tambor, tambors, and tanbor. Lemma M Ldoc F Ldoc M Prov F Prov M Lit F Lit Gender doublet tabor only 1 only 2 — — — X unclear Table 8.45: Tabor as a gender doublet (Persian loanword). <?page no="223"?> 8.4 Other Loanwords 223 Languedocian Provençal num+gen+case num+gen gen M sg nom 0 0 0 0 1 obl 0 0 0 pl nom 1 0 1 1 obl 0 0 0 F sg nom 0 0 0 0 2 obl 0 0 0 pl nom 2 0 2 2 obl 0 0 0 NA sg nom 0 0 0 1 2 obl 0 1 1 pl nom 0 0 0 1 obl 0 1 1 Table 8.46: Gender distribution for tabor. As Table 8.46 shows, there are only five occurrences of tabor in the COMETA (2025) corpus, three of which show an identifiable gender, all in the Languedocian sub-corpus. There are two feminine plural nominative occurrences (see screenshots of both occurrences in Example (79a)) and one masculine singular nominative occurrence (see Example (79b)). Both occurrences in the Provençal sub-corpus are of uncertain gender, as Example (80) illustrates. (79) a. E las tabors els tempes. ‘And the F drums and the percussion’ (BnF, Français 25425 (1275: f o . 150, l. 12; f o . 212, l. 4)) b. E li corn e las trompas. Els cimbol elh tabor. ‘And the horn and the trumpets. And the cymbals and the M drum’ (BnF, Français 25425 (1275: f o . 113, l. 19)) <?page no="224"?> 224 8 Gender Doublets in COMETA (2025) (80) Grailles e corns, cembolz, tabors ‘Clarions and horns, cymbals, drums’ (BmC, Cote 34 (13 th century: f o . 134 r o ., l. 12)) From the data in the COMETA (2025) corpus, no statement on the gender in the Provençal area can be made, and the informative value of the evidence on the gender in the Languedocian part is only very limited, given that all three occurrences are from one single text, i.e., BnF, Français 25425 (1275). But within this text, it appears as a gender doublet; one occurrence is masculine, the other two are feminine. But due to the limited number of occurrences of this noun, no further statement on the status of tabor with respect to gender can be made. A look into further literature reveals in total only two masculine instances of this noun: Raynouard (1836-1845) lists tabor only as masculine noun, and he gives only one example, the exact same one as cited in Example (79b). One further masculine occurrence can be found in the following quotation: “cum vezem els tabors” (Scinicariello 1982: 72; ‘As we see the M drums’). Contrary to the indications in Raynouard (1836-1845), Levy (1894-1924: s.v. “tabor”) lists it as both a masculine and a feminine noun. In addition to the examples from the manuscript BnF, Français 25425 (1275), he only gives further feminine examples, e.g., one from the Canso d’Antioca: “E las tabors bruis e los grailes cornar” (see also Sweetenham/ Paterson 2003: 204; ‘And the F drums beat and the clairons blow’). Furthermore, this noun is listed in the DAOA (2009: s.v. “tabor”) as a feminine noun with the example “amb la fluta et amb la tabor” ‘With the flute and with the F drum’. Other supposedly feminine occurrences can be disproved: Gouiran (1985b: 966) lists it as a feminine noun, but the occurrences are of unclear gender value: “E corn e taborn e laire” (Gouiran 1985a: 132; ‘And horn, drum, and shawm’) or “Trompas, tabors, seinheras e penos” (Gouiran 1985b: 652; ‘Trumpets, drums, shawms, and pipes’). In the glossary of Martorano (2017: s.v. “tambor”), it is indexed as both masculine and feminine, even though the respective occurrence shows no clear gender: “Trompas ni corn, ni viulas ni tambors” (Martorano 2017: 310; ‘Neither trumpets nor horns, nor viols nor drums’). Information on the gender value of tabor is scarce, given the low number of occurrences in Old Occitan manuscripts. In total, only one masculine instance of this noun could be found, and all other evidence suggests that this noun is in fact a feminine noun, with only one de facto masculine instance. <?page no="225"?> 9 Two Comparative Case Studies This chapter presents two case studies, which aim at grasping differences in gender assignment due to the influence of the scribes (see Section 9.1) or the authors (see Section 9.2). Here, the gender distribution for all nouns within the respective manuscripts will be considered. As mentioned in Chapter 7, the texts used for this analysis were also PoS-tagged and additional information on gender, number, and case were assigned to the nouns in order to compare the gender assignment from one manuscript to another and to take a closer look at variation patterns in combination with other morphosyntactic features. 9.1 Three Manuscripts of the Vida de Sant Honorat For this section, the three manuscripts of the Vida de Sant Honorat in COMETA (2025) were selected for a case study on the comparison of gender assignment within one hagiography in three versions, written by three different scribes (namely the manuscripts BL, Additional 10323 14 th century; BnF, Français 13509 14 th century; BnF, NAF 6195 14 th century). Possible differences in gender for certain nouns can give us an idea on the influence of the scribe. It has to be kept in mind that the three manuscripts combined used at least two different sources for the transcription. The copyist of manuscript BnF, Français 13509 (14 th century) used one source, the scribe of manuscript BnF, NAF 6195 (14 th century) another one, while the scribe of the third manuscript (BL, Additional 10323 14 th century) probably had both models at his disposal when writing the text, using the former model for the first part, the latter for the rest (see Section 6.2.3 for more details). This also explains the overall difference of about 10,000 tokens between the manuscripts; the longest text (BnF, Français 13509 14 th century) has about 66,0000 tokens, the medium-long text in BL, Additional 10323 (14 th century) has about 57,000, and the shortest one (BnF, NAF 6195 14 th century), where some parts are missing, has around 46,000. First, an overview of the general gender distribution within the texts will be given before going into further detail, thus focusing on the qualitative part of the analysis. <?page no="226"?> 226 9 Two Comparative Case Studies 9.1.1 Gender Distribution Among the Manuscripts The difference in size between the manuscripts is also reflected in the number of nouns per text, in direct relation to the total number of tokens, with 10,489 nouns in BL, Additional 10323 (14 th century), 11,978 in BnF, Français 13509 (14 th century) and 8787 in BnF, NAF 6195 (14 th century). Figure 9.1 shows the distribution of nouns per gender within the three manuscripts (the label NA means not clearly attributable). It is evident that there are by far more masculine nouns (between 34% and 37%) than feminine nouns (24% per text) in all three texts. The largest number of nouns are of unclear gender value (between 38% and 42%) since there are either no agreement targets dependent on the noun or the agreement targets show no clear gender value, such as the apocoped article l or the adjective gran, which shows case and number, but not gender. Figure 9.1: Gender distribution among the three manuscripts of the Vida de Sant Honorat. Because nouns without a clearly identifiable gender value do not add any information to this study on gender assignment, those occurrences will not be included in the following anymore, so only the masculine and feminine occurrences are included for the remainder of this analysis. The next two figures will show the gender assignment in the manuscripts with the additional factors case (Figure 9.2) and number (Figure 9.3), while Figure 9.4 gives an overview of the combination of all three factors. <?page no="227"?> 9.1 Three Manuscripts of the Vida de Sant Honorat 227 Figure 9.2: Gender distribution per case among the three manuscripts of the Vida de Sant Honorat. As expected for a Pro-Drop language such as Old Occitan, most of the occurrences are in the oblique case (labelled o), while only a small number of tokens are in the nominative, regardless of the gender or the respective text (see Figure 9.2). The nominative case is labeled with R for French cas régime or the Latin Casus Rectus, in order to avoid confusion, given that the label N in gender analyses is mostly used to label the neuter gender, which does not exist anymore in Old Occitan. Figure 9.3: Gender distribution per number among the three manuscripts of the Vida de Sant Honorat. <?page no="228"?> 228 9 Two Comparative Case Studies Most nouns showing a clear gender value in the three manuscripts are in the singular, the plural is less represented, as seen in Figure 9.3. However, there is no clear correlation between gender and number in these texts, the relation between masculine plural to singular and feminine plural to singular is about the same between the manuscripts. In Figure 9.4 it is evident that the most used combination is singular in the oblique case, regardless of the manuscript or gender. The lower number of occurrences of that combination in the feminine is explainable by the overall lower number of feminine occurrences. From a quantitative point of view, there do not seem to be significant differences in the gender assignment between the three manuscripts. Figure 9.4: Gender distribution per case and number among the three manuscripts of the Vida de Sant Honorat. 9.1.2 Detailed Gender Analysis For this subsection, gender variation of all nouns that appear in the manuscripts was checked (a total of 3,068 unique forms; those with unclear gender value still excluded). The objective is to look for text-internal gender variation, as well as variation between the manuscripts. Table 9.1 gives an overview of the gender doublets found in or among the manuscripts and their quantitative distribution. In this table, different graphical variants of the nouns 1 are presented together with the unmarked form as the 1 For instance different forms due to case or number marking or different graphical variants for some sounds such as the [ ñ ], that appears both as <nh> and as <gn>. This is also valid for different variants of a word, for instance dreg and dreth were taken together, as they represent two graphical variants of a single noun. <?page no="229"?> 9.1 Three Manuscripts of the Vida de Sant Honorat 229 indicated lemma. However, homonymic word forms with gender variation due to their respective etymology and diverging semantics are not listed together and seen as seperate lemmata (e.g., mort which means ‘death’ as a feminine noun and ‘the dead’ as a masculine noun). Lemma Additional 10323 Français 13509 NAF 6195 Fem. Masc. Fem. Masc. Fem. Masc. camp ‘(battle)field’ 0 4 2 0 0 3 caytieu ‘prisoner’ 6 15 8 19 7 12 dia ‘day’ 0 19 0 38 1 12 dreg ‘straight path’ 1 6 1 8 0 7 estaie ‘estate’ 0 5 0 13 1 7 man ‘hand’ 27 1 37 0 25 0 mar ‘sea’ 17 32 41 6 23 10 mayre ‘mother’ 29 1 30 0 22 0 ost ‘army’ 1 1 2 0 2 0 parentat ‘kinship’ 1 1 0 3 0 2 serpent ‘snake’ 18 2 19 6 18 3 voluntat ‘will power’ 20 1 23 0 19 0 Table 9.1: Masculine and feminine occurrences per lemma and text within the three manuscripts of the Vida de Sant Honorat. Table 9.1 shows that there is a certain amount of nouns where gender varies either within all manuscripts or just within a single manuscript (and therefore gender also varies between the three versions of the hagiography). In addition to those nouns listed here, there are also some where gender variation may exist, but it cannot be confirmed due to phonological reasons. For occurrences such as “al honor de dieu” ‘To the ? honor of God’ or “Del honor d aquest segle” ‘Of the ? honor of this century’, it is not totally clear whether it is a masculine occurrence or not. This noun has an onset with an <h>, which was not pronounced (see Anglade 1921: 135), so both the masculine and the feminine article would most likely be in an apocoped form which would mean that these instances are not occurrences of a portmanteau allomorph. Conversely, it may be the case that the presence of the grapheme triggered the scribe to treat this noun as one with an initial consonant, which would make these cases portmanteau allomorphs, thus rendering these occurrences masculine. Therefore, the gender value remains unclear and this noun as well as hora and other similar cases are not listed here. Of the twelve gender-variable nouns, five (namely dia (see Section 8.2.4), man (see Section 8.2.3), mar (see Section 8.2.2.12), ost (see Section 8.2.2.16), and ser- <?page no="230"?> 230 9 Two Comparative Case Studies pent (see Section 8.2.2.24)) have already been discussed in the previous Chapter 8 and thus will not be addressed in more detail here. Most of these nouns (9 out of 12) occur only once or twice in a given gender, while the remaining occurrences appear in another gender. For those nouns with only a low number of occurrences, the examples from the less frequently attested gender will be presented and complemented with one additional example from the more frequent gender in each case. The remaining 3 nouns (caytieu, mar, and serpent) show more variation in gender, and thus the gender seems to be less fixed. For these nouns, it is clear that there is real variation and that it is not merely a single occurrence where it may also be the scribe’s mistake. Camp means ‘field’ or ‘battlefield’ and is often used as a synonym for ‘battle’. It is derived from the Latin noun campus with the meaning ‘plain’, and later also ‘field’ (see TLL 2025: s.v. “campus”). It is a masculine noun from the second declension class and remains masculine in all Romance languages. Both genderidentifiable occurrences in manuscript BnF, Français 13509 (14 th century) are feminine, while the other two manuscripts have only masculine occurrences. The feminine occurrence in Example (81a) appears in both other manuscripts, where the noun triggers masculine agreement (see Example in 81c) while the other feminine occurrence is exclusive to one manuscript. (81) a. Als vazes d aliscamps aqui si fes la camps. ‘At the graves of Aliscamps, here the F battle was fought.’ (BnF, Français 13509 14 th century: f o . 41 r o , c 2, l. 4-5) b. Venteiar e brandir lay hon fey si la camps ‘To wave and to brandish there, where one makes the F battle (field).’ (BnF, Français 13509 14 th century: f o . 41 r o , c 2, l. 4-5) <?page no="231"?> 9.1 Three Manuscripts of the Vida de Sant Honorat 231 c. Als vazes d aliscamps aqui si fes le camps. ‘At the graves of Aliscamps, here the M battle was fought.’ (BL, Additional 10323 14 th century: f o . 29 r o , c 1, l. 23-24) Caytieu means ‘captive, prisoner’ but also ‘miserable person’ or ‘wretched person’. It is derived from the Latin adjective captivus with the meaning ‘imprisoned’. Due to the phonological form in Old Occitan, it may also be that the etymon is a form like *cactivus with the same meaning (see REW 1935: s.v. “captivus”; FEW 1922-2003: s.v. “captivus”). The noun in discussion here is the nominalised form, which was formed in a process of deadjectival conversion. The forms in the Romance languages are all masculine, but there is also a dedicated feminine form for referring to female prisoners (see below, Example 82e). It is variable in gender assignment within all the manuscripts; both genders appear quite frequently in the three manuscripts. Therefore, only two examples per gender will be given to illustrate the variation, starting with the feminine ones in Examples (82a) and (82b), followed by the masculine examples in (82c) and (82d). In the examples listed, the sex of the referenced people is not clear from the context of the hagiography; it is therefore seen as a default gender use in all contexts. There is a distinct feminine form of this noun, caytiva, which is used in non-default contexts with feminine referents, see Example (82e). For this noun, the variation in gender seems to be entrenched. (82) a. E playn si las caytieu, que poyras iamays fayre. ‘And the F miserable ones mourn, what will you ever be able to do.’ (BL, Additional 10323 14 th century: f o . 9 r o , l. 12) <?page no="232"?> 232 9 Two Comparative Case Studies b. Clama si las caytieu ia dieus bon agues facha ‘The F wretched ones cry out God has already done well.’ (BnF, Français 13509 14 th century: f o . 56 r o , c 2, l. 18-19) c. Los caytieus dezastratz que son en l abadia ‘The M ruined captives/ miserable ones who are in the abbey’ (BnF, Français 13509 14 th century: f o . 56 v o , c 1, l. 9-10) d. Onravas los caytieus els aviatz eyssausatz ‘You honored the M captives and those you had expelled.’ (BL, Additional 10323 14 th century: f o . 30 r o , l. 31) e. Que a mantas caytivas as fag tanta bontat. ‘For to many F captive women you have done so much kindness.’ (BnF, Français 13509 14 th century: f o . 73 v o , c 2, l. 9-10) <?page no="233"?> 9.1 Three Manuscripts of the Vida de Sant Honorat 233 The next noun in the list is dreg with the meaning ‘straight path’ 2 or, in this case, since it is about birds, it figuratively means ‘straight flight’. It is derived from the Latin adjective diRectus ‘straight, direct’ and was nominalised in a process of deadjectival conversion. This conversion process most likely already took place in Latin times, since it is inherited as a masculine noun into all Romance languages. The same is valid for the changes in meaning, as it also has the legal meaning in all Romance languages (see FEW 1922-2003: s.v. “directus”). It has a feminine occurrence in two out of the three manuscripts, with one occurrence per text, both in the same passage (see Examples 83a and 83b). This passage is not included in manuscript BnF, NAF 6195 (14 th century). All other occurrences of this noun are masculine, mainly also with another meaning (‘right’ in the legal sense) such as the occurrence in Example (83c). (83) a. D aves la dreg del vignares peyras y rogan mot espes ‘Along the F direct flight of the birds over the vineyards, stones lie thickly scattered.’ (BnF, Français 13509 14 th century: f o . 87 v o , c 2, l. 23-24) b. D aves la dreg del vignares peyras y rogan mot espes ‘Along the F direct flight of the birds over the vineyards, stones lie thickly scattered.’ (BL, Additional 10323 14 th century: f o . 62 v o , c 3, l. 3-4) 2 It has also the meaning of ‘rightfulness’ or ‘right’ in the legal sense (see FEW 1922-2003: s.v. “directus”; DOMél 2025: s.v. “drech”). With the legal meaning, all occurrences are masculine, and only one occurrence with the directional meaning is feminine. <?page no="234"?> 234 9 Two Comparative Case Studies c. Trastot lo dreg del Rey e tota manencia. ‘The M law of the king [is] overturned, and all stability.’ (BnF, Français 13509 14 th century: f o . 35 v o , c 2, l. 19-10) The noun estaie ‘state, estate, condition’ has only one feminine occurrence in the manuscript BnF, NAF 6195 (14 th century) which is listed in Example (84a); all other occurrences of this noun with identifiable gender value are masculine, such as the one in Example (84b). Nevertheless, most of the occurrences of estaie are of uncertain gender value, again due to the onset of this noun, that was strengthened by an initial e-, which was often the case in Vulgar Latin for words beginning with an s impurum (see Rheinfelder 1976a: 182-183). This noun is derived either from the Latin fourth declension masculine noun status ‘position, standing, state, condition, rank’ or from the Latin verb staRe ‘to stand, to keep standing, to stay’. In the latter case, it would be a deverbal nominalisation through conversion from the past participle, which was used also as the past participle for the verb ‘to be’ (see Laurent 1999: 68). Similar conversions are also masculine in Romance languages. But the etymology of this noun is unclear, as it may be an hereditary word or a Romance neologism. (84) a. Leu si es esdevengut en esta fera estaies ‘He has easily fallen into this F fierce F condition.’ (BnF, NAF 6195 14 th century: f o . 32 r o , l. 6) b. E l enfant de roze trayssist Et has conquist lo sant estaie ‘And you betrayed the child of Roze and you have conquered the M holy M state/ place’ (BnF, NAF 6195 14 th century: f o . 85 r o , l. 28) Mayre ‘mother’ has a single masculine occurrence. It is a mismatch between gender and biological sex, as the semantics clearly designate a woman. It is derived <?page no="235"?> 9.1 Three Manuscripts of the Vida de Sant Honorat 235 from the Latin third declension noun mateR ‘mother’, which is also feminine. It still exists in all Romance languages, sometimes with different meanings, such as Portuguese madre ‘abbess’ (see FEW 1922-2003: s.v. “mater”). It is still a feminine noun in all Romance languages. As it is evident that most of the occurrences will be feminine, only the masculine occurrence is given in Example 85. The agreement target here is tantz, which is clearly masculine in this case, as the feminine plural form in this manuscript is tantas (see tantas ientz in the following example), and all other occurrences of tantz in this manuscript are used in reference to masculine nouns. (85) A tantz mayres rent tanto filh tantas ientz gieta de perilh ‘To so many M mothers he brought so many sons, and he relieved so many people from danger’ (BnF, Français 13509 14 th century: f o . 95 v o , c 2, l. 6-7) Parentat ‘kinship’ is derived from the Latin third declension masculine noun paRens ‘parents, relatives’. It is also possible that a masculine derivational form *paRentatus already existed in Vulgar Latin. It probably changed its gender in analogy with other forms in -ate; this makes also a dedicated feminine form *paRentas possible or even probable, even though it is not mentioned in the etymological literature on this noun. It is indicated as gender variable in the Romanische Grammatik by Meyer-Lübke (see 1894: 420-421), but no other indications for gender vairiability of this noun could be found. This noun has a feminine and a masculine occurrence in manuscript BL, Additional 10323 (14 th century), see Example (86). All occurrences of this noun in the other two manuscripts are masculine. (86) a. Que pres molher otra lur grat De totz cels de sa parentat ‘Who took a wife against the will of all his F relatives.’ (BL, Additional 10323 14 th century: f o .48 r o , c 2, l. 3-4) <?page no="236"?> 236 9 Two Comparative Case Studies b. Car aquel sobrenom avia son parentat ‘Because that surname belonged to his M kinship’ (BL, Additional 10323 14 th century: f o . 59 r o , c 3, l. 8) Voluntat is always used as a feminine noun with the exception of a single masculine occurrence in manuscript BL, Additional 10323 (14 th century). This noun is derived from the Latin third declension feminine noun voluntas with the meanings ‘(free) will, wish, desire, intention’. It still exists as a feminine noun in all Romance languages. The following Example (87) shows the masculine occurrence in Example (87a) as well as a feminine occurrence in Example (87b), both taken from the same manuscript. (87) a. E ves ton voluntat. que per beus temporals Non perdas la ricor. dels grans espectals. ‘And it is your M will that for temporal goods you do not lose the richness of the great visions.’ (BL, Additional 10323 14 th century: f o . 34 v o , l. 40-41) b. E non auza dire al cors sant Son dezir ni sa voluntat. ‘And he does not dare to tell the holy body his desire nor his F will.’ (BL, Additional 10323 14 th century: f o . 39 v o , c 3, l. 10-11) It is noteworthy to mention that both nouns ending in <-at> in this case study, namely parentat and voluntat are only gender-variable within one and the same manuscript, BL, Additional 10323 (14 th century). Parentat is used as a masculine noun in the other two manuscripts while voluntat is employed as feminine noun. This case study has shown that there is more variation in gender than previously expected. There are seven more, previously unknown, gender-variable <?page no="237"?> 9.2 The Chanson de la Croisade contre les Albigeois 237 nouns which should be researched in a wider range of texts and potentially added to the list of gender-variable nouns in the DOMél (2025). There are three nouns with considerable gender variation in each of the three manuscripts, namely caytieu, mar, and serpent. The first and the last one have the same tendencies in all three manuscripts analysed here; for caytieus, there is a strong tendency toward masculine usage, while it is the other way round for serpent, where the major part of the occurrences are feminine. Mar is the only noun that shows considerable differences in gender assignment among the three manuscripts. In relative terms, 87% of its occurrences with an identifiable gender are feminine in BnF, Français 13509 (14 th century); in BnF, NAF 6195 (14 th century) it’s still at 70% for feminine occurrences, while the manuscript BL, Additional 10323 (14 th century) shows only 35% of feminine occurrences. Therefore, in the last manuscript, about two thirds of the occurrences are masculine. Camp shows variation in gender between the manuscripts: it retains the same gender within a manuscript, but shows different gender assignments from one manuscript to the other two. It is only feminine in manuscript BnF, Français 13509 (14 th century), while it is masculine in the other two manuscripts. From this data, it cannot be said for certain whether the variation is specifically due to the scribe or due to the model they used. For the rest of the nouns presented in this section, the variation in gender remains minimal. For these nouns, the gender is similar for all manuscripts, with the only difference being that some nouns have one or two occurrences with a different gender. The manuscript with the highest number of nouns that agree with both genders is BL, Additional 10323 (14 th century), where all of the nouns in Table 9.1, with the exception of three nouns (camp, dia, and estaie) occur with both genders. For the other two manuscripts, it is the other way round and most nouns do not appear with different genders. In BnF, NAF 6195 (14 th century), four gender-variable nouns could be determined (caytieu, dia, mar, and serpent), while in manuscript BnF, Français 13509 (14 th century) only the three nouns, that vary in all three manuscripts, show gender variation. 9.2 The Chanson de la Croisade contre les Albigeois As stated in Section 6.2, the manuscript of the Chanson de la Croisade contre les Albigeois was composed by two different authors, one known as Guilhem de Tudela, and the other remains anonymous. The only complete manuscript preserved until today was written in the 1270s in or near Toulouse by one single scribe. If there are differences in gender assignment between the two parts of the text, these may be related to differences in gender usage between the two authors. If there are no differences, this may also be due to the scribe’s normative <?page no="238"?> 238 9 Two Comparative Case Studies tendencies, but also due to a similar use of gender for the respective nouns by the two authors. This will be discussed later in Chapter 10. The first part of the manuscript, attributed to Guilhem, consists of 27,217 tokens in total, while the second, larger part, has almost three times as many tokens (precisely 73,329). Because of the difference in size, these two texts are not totally comparable from a quantitative perspective, and also not qualitatively, as it is two parts of a single text and not two parallel manuscripts of the same text. Therefore, no quantitative comparison between the two parts will be presented, but only a qualitative analysis, in which differences in gender assignment between the two parts will be shown. To do so, the gender of all nouns in the first part of the manuscript was tagged and information on gender, case, and number was added. All nouns with an identifiable gender value 3 were then searched in the second part of the manuscript. Those nouns with differences between the two manuscripts will be presented in the following. Of the 527 nouns with an identifiable gender value, almost 40% (203 in absolute numbers) are not present in the second part of the Croisade and additionally, 62 nouns are of unclear gender value in the second part. Therefore, almost exactly 50% of the nouns from the first part are not comparable regarding their gender attribution between both parts of the text. This can be explained by the differences in the lexicon between the two parts. While both parts have a combined total of 12,328 different graphical types, only 2,141 of these types are shared in both parts. Part one on its own has 5,184 different types, so only little over 40% of the graphical items are shared between both texts, which is a very low number. This corresponds well to the number of nouns that are only present in the first part but do not exist in the second one (and most likely vice versa). For the remaining 262 nouns, their gender could be identified and the nouns could therefore be compared. The majority of the nouns are used with the same gender in both parts of the manuscript: 104 nouns trigger feminine agreement in both parts, while 156 are used as masculine nouns in both parts. Only two nouns show variation between the two parts. These nouns are ost and val. Ost has already been discussed in Section 8.2.2.16, where it was shown that this noun is in fact variable in gender, even though it showed only minimal variation. All masculine occurrences of this noun in the Languedocian sub-corpus were found in the manuscript under discussion here. It is important to note that both masculine occurrences appear in the first part alongside feminine forms, while there are only feminine occurrences in the second part of the manuscript. 3 In total, there were 2,210 tokens with an identifiable gender value, distributed over 912 graphical types. This number could be reduced to 527 by taking the inflected and non-inflected forms together for the count, but all variants were looked for. Homonyms were again not seen as variation in gender. <?page no="239"?> 9.2 The Chanson de la Croisade contre les Albigeois 239 To illustrate this, two examples will be given that were previously shown in Example (40), in which a feminine and two masculine occurrences appear only few verses apart in the first part of the manuscript, see Example (88). The occurrences of ost in the second part of the manuscript are all feminine, see Example (89). (88) a. Dedins a carcassona aqui las osts atent ‘Inside Carcassonne, there he waits for the F armies’ (BnF, Français 25425 (1275: f o . 7 l. 21)) b. Cassanhol asetia lo ostz e a dins mot aquier ‘The M army besieges Cassanhol and inside there are many archers’ (BnF, Français 25425 (1275: f o . 9, l. 3)) c. D aicestatz ostz de sai. nous volh oimais parler ‘Of these M armies here I do not want to talk further’ (BnF, Français 25425 (1275: f o . 9, l. 23)) (89) yeu soy le plus fortz d esta ost ‘I am the strongest of this F army’ (BmT, Cote 2883 (14 th century: f o . 127 r o ., c. 2, l. 31-32)) The second noun with differences in gender assignment between the two parts is val ‘valley’. It is derived from the Latin third declension noun vallis with the same meaning. It was a feminine noun in Latin, which is preserved in all Romance languages. It is masculine in Occitan, Portuguese, and Spanish, while <?page no="240"?> 240 9 Two Comparative Case Studies it preserved the feminine gender in Catalan, French 4 , Italian, and Romanian. The semantic relation to the masculine noun mont is often seen as a reason for the change in gender (see Meyer-Lübke 1894: 428; FEW 1922-2003: s.v. “vallis”), similar to the propositions for the change in gender of mar in relation to terra (see Section 8.2.2.12; see also Section 3.5). In the first part of this manuscript, there is only one occurrence of the noun val with clearly identifiable gender value, i.e., feminine, which is displayed in Example (90a). The two occurrences in the second part are both masculine, as shown in the Examples (90b) and (90c). Both appear in combination with a preposition, contracted with the definite article. As shown before in Chapter 4, this is always a masculine article in front of a noun starting with a consonant, as the feminine article is only apocoped before an onset with a vowel. Even though it is written with an <u>, it was most likely pronounced as a [v] (see Anglade 1921: 146-148) and can therefore be seen as an onset with a consonant. (90) a. Li frances esperonan. com baro natural. al enan que ilh podon. al pendent d una val. ‘The French spurred on, like natural barons, toward the advantage they could gain, on the slope of a F valley.’ (BnF, Français 25425 1275: f o . 52, l. 18-19) b. Entre l mur e las tendas. per la riba del val. Tuit essems esperovan. dreit a l autre portal ‘Between the wall and the tents, along the edge of the M valley. We were all watching, directly toward the other gate.’ (BnF, Français 25425 1275: f o . 123, l. 13-14) 4 There is also a masculine form val in French, which is not really used anymore except in poetic contexts or in place names, e.g., Val d’Aran. It has been supplanted by the feminine form vallée (see FEW 1922-2003: s.v. “vallis”). <?page no="241"?> 9.2 The Chanson de la Croisade contre les Albigeois 241 c. Al val de mont oliu. on eral murs fondutz ‘In the M valley of Mount Olive, where walls were fallen’ (BnF, Français 25425 1275: f o . 150, l. 38) From this comparison of the two parts of the Croisade, composed by two different authors, but written by a single scribe, it becomes clear that there are no strong differences in gender assignment. Only two nouns could be found where the gender varies between the two parts: the noun ost, which shows variation within the first part, but no variation in the second part, where all occurrences are feminine. The other noun is val, which has only one feminine occurrence in the first part of the manuscript and two masculine ones in the second part. This is most likely due to the overall differences in the lexicon of the two authors. <?page no="243"?> 10 Analysis and Discussion After presenting the data in Chapters 8 and 9, the information retrieved from both parts will be connected and an overview of the different aspects in relation to gender variation will be given, divided into extralinguistic (see Section 10.1) and linguistic factors (see Section 10.2). 10.1 Extralinguistic Factors The first two extralinguistic factors regarding gender assignment that will be discussed are diatopic and diachronic variations. This section treats nouns that demonstrate differences in usage either over time or within different linguistic areas. Due to the constitution of COMETA (2025), it was not possible to give any hints at diaphasic variation, as all manuscripts were most likely written in similar circumstances. Following these first two extralinguistic factors, the next two sections are dedicated to the scribes and their potential role in gender assignment in the manuscripts. First, variation between different scribes will be discussed, focusing mainly on the three scribes of the three Vida de Sant Honorat manuscripts within COMETA (2025). The next section focuses on variation within a single scribe’s textual practice, showing the range of variation that can be found in a manuscript written by one scribe. 10.1.1 Diatopic Variation This section will discuss diatopic variation in gender assignment between the two areas Provence and Languedoc. In the quantitative data shown in Chapter 8, three nouns with diatopic variation could be retrieved, namely (ves-)comtat, mar, and sanc. Comtat and its prefixed form vescomtat, have exclusively masculine occurrences in Old Provençal, but both feminine and masculine occurrences in the Old Languedocian area. Only a total of six feminine occurrences of both forms combined were found in Old Occitan, five of which stem from texts produced in the Languedocian area and one from a text written in Auvergne. This clearly demonstrates that this noun is a gender doublet in Languedocian, but not in the wider Occitan-speaking region. Some other nouns ending in -at behave differently in Old Provençal; for instance, the two nouns parentat and voluntat are <?page no="244"?> 244 10 Analysis and Discussion both gender-variable within the manuscript BL, Additional 10323 (14 th century), which was written in Provence (see Section 9.1 for more information). The former tends to be masculine in Old Provençal, while the latter is predominantly feminine. This suggests that the ending -at does not have a clear gender value in Old Provençal, and that the morphology of comtat cannot be the sole explanation for the stabilisation of this noun’s gender. This remains true even though the etymology of the ending -at differs between these three nouns. From a synchronic perspective in Old Occitan, they will most likely be processed as the same ending. Therefore, it is unclear why the gender of comtat is fixed in Old Languedocian but not in Old Provençal. The second noun in this category is mar. This noun seems to be completely variable in terms of gender assignment in Old Provençal, with a high number of both masculine and feminine occurrences in COMETA (2025). In the Languedocian sub-corpus of COMETA (2025), no masculine occurrences of mar could be found. Further literature supports this tendency, with only two masculine occurrences found in a text written in Montpellier in the mid-14 th century. As these are the only two masculine occurrences in Old Languedocian and they are from a city very close to the Provençal linguistic area, it is questionable whether mar should also be listed as a gender doublet in Old Languedocian. These two masculine uses are more likely due to contact with Old Provençal or, maybe, Old Italian than to a variation entrenched in Old Languedocian. The third noun showing diatopic variation in gender assignment is sanc, which seems to be a gender doublet only outside the Old Provençal linguistic area. It has been demonstrated that sanc triggers both feminine and masculine agreement in Languedocian and Gascon, and is therefore a gender doublet in these two linguistic areas at least. In Old Provençal, however, it is only masculine. The only feminine occurrence found in this area stems from manuscript B of the Vie de sainte Marie Madeleine (see Section 8.2.2.22). As there is a difference between the two manuscripts of this text, Chabaneau (1887: 23) argues that there is a writing error in manuscript A and changes the occurrence to feminine in his edition. However, the findings of this book contradict this editorial choice, showing that if there is a writing mistake, it is the feminine occurrence that is incorrect. Therefore, this noun appears to be gender-variable outside the Languedocian area, whereas it is fixed as masculine within that linguistic area. This sort of diatopic variation is likely due to differences in contact languages. The Languedocian area has closer ties to Ibero-Romance languages and the Gascon varieties, while the Provençal area is in closer relation with Italo-Romance varieties. Language contact might have played a role in the case of mar, which is only a gender doublet in Old Provençal and only feminine in Old Languedocian. The masculine agreement in Old Provençal is most likely due to language contact with Italo-Romance varieties, where this noun is also masculine. <?page no="245"?> 10.1 Extralinguistic Factors 245 Both areas are exposed to French influences, but it is unclear to what extent (ves-)comtat in the Languedocian area occurs as a feminine noun due to French influence (see the proposal by Pfister (1970: 347) in Section 8.2.1). This would suggest a stronger French influence in this linguistic area, which can neither be confirmed nor denied based on the COMETA (2025) data. As the substrate language for both areas is Gaulish and information on this language is scarce, it is impossible to determine whether Gaulish varieties might have influenced the gender variation. To assess the influence of the substrate language on gender variation , it would be interesting to compare the two discussed areas here with the Gascon area, which has a Basque substrate. However, due to the constitution of COMETA (2025), which includes no Gascon manuscripts, this study must be left to future research. 10.1.2 Diachronic Variation In the data of the COMETA (2025) corpus, which covers the period from the mid- 13 th to the 14 th century, there is little evidence of diachronic variation in gender assignment. The only noun for which diachronic variation seems to occur is dia (see Section 8.2.4). This noun is only a gender doublet before the time of the manuscripts included in COMETA (2025). Its gender was variable in the 11 th and 12 th centuries, but it was later fixed, triggering only masculine agreement. Within COMETA (2025), only two feminine occurrences could be found, one per region. All other occurrences are either masculine or of uncertain gender value. On the other hand, it is also possible that dia could still trigger both genders later in Old Occitan, but that the gender is not clearly identifiable. This may be due to the frequent combination of ordinal or cardinal numbers written as Roman numerals. As not all scribes mark the gender of these numbers, and only the numbers I and II are marked if the gender is indicated, gender assignment remains uncertain in these contexts. It is not surprising that little diachronic variation is found within the COMETA (2025) corpus. Even though it covers a period of about 150 years, this time span is too short to display long-term changes in the language structure. 10.1.3 Interscribal Variation There are clear differences in gender assignment between the three manuscripts of the hagiography analysed in Section 9.1. However, it is unclear whether this variation is solely due to the scribe or if other factors also influence gender assignment. The most striking difference in gender assignment, which may be due to the scribe’s influence, can be seen with the noun mar. This noun is primarily mas- <?page no="246"?> 246 10 Analysis and Discussion culine (about two thirds of the occurrences) in BL, Additional 10323 (14 th century), while it is mainly feminine in the other two manuscripts. As shown in Section 6.2.3, the manuscript that differs significantly in gender assignment for mar is positioned between the two branches of the manuscripts’ history. It is assumed that the scribe had models from both branches of the manuscript’s history at his disposal, when writing this manuscript. BnF, Français 13509 (14 th century) belongs to the one branche, while BnF, NAF 6195 (14 th century) belongs to the other one. Both manuscripts use mar as a feminine noun 70% or more of the time. Even when considered together, the two manuscripts have only 16 masculine occurrences of mar, whereas 32 masculine occurrences were found in BL, Additional 10323 (14 th century). Therefore, even if the scribe opted for all masculine occurrences in the other two manuscripts (assuming the models had a similar distribution of masculine and feminine usage of this noun as is found in these manuscripts), more than half 1 of the masculine occurrences remain unexplained by the textual models. The masculine usage of mar is likely due to the preferences of the scribe who wrote manuscript BL, Additional 10323 (14 th century). Camp presents a different scenario: it is masculine in the two manuscripts BL, Additional 10323 (14 th century) and BnF, NAF 6195 (14 th century), while it is solely feminine in manuscript BnF, Français 13509 (14 th century). It cannot be said for sure whether this variation in gender is due to the scribe of the latter manuscript or if it is simply a variation in gender that was already present in the model used for that manuscript. This is unclear not only because of the lack of information on how these nouns are used in the models for the respective manuscripts, but also because the scribes remain anonymous and their linguistic background is unknown (see Section 6.2.3) and the influence of their respective backgrounds therefore remains unclear. In order to determine the reason for this gender variation more precisely, it would be necessary to take a closer look at other manuscripts based on the same model used. Only then could it be determined with certainty whether this variation is due to a single scribe. This study is left for future research, as the primary goal of this study was to create a corpus to investigate gender variation and variability in general, rather than variation due to a different scribe. For a more detailed study of the influence of different scribes on gender assignment in the respective Vida de Sant Honorat manuscripts, the remaining six need to be transcribed. 1 Two of the masculine occurrences are in the same passage in both BnF, Français 13509 (14 th century) and BnF, NAF 6195 (14 th century), so there are only 14 different masculine occurrences across these two manuscripts. <?page no="247"?> 10.1 Extralinguistic Factors 247 There are clear differences in the sheer number of gender-variable nouns in each manuscript. These differences may be due to the scribes’ preferences or habits, or perhaps to the specific scribe of each respective model. The manuscript of the Vida de Sant Honorat with the greatest variation with respect to gender is BnF, Français 13509 (14 th century), in which nine of the twelve gender-variable nouns found in this hagiography show both genders. This strongly suggests that the variation in gender is due to a scribe, either of that specific manuscript or of the model. However, for three of the nouns, it is likely that the gender variation was present in general, and not only in the linguistic use of a single scribe. This is because these three nouns (caytieu, mar, and serpent) are variable in all three manuscripts of the hagiography, regardless of the specific scribe. For mar and serpent, it could also be shown in Section 8.2.2.12 and Section 8.2.2.24, respectively, that these nouns are generally variable in gender in Old Occitan as well as across Romance languages. Overall, it can be concluded that there is clearly variation in gender between the three manuscripts of the Vida de Sant Honorat. The reason for this variation cannot be determined for all nouns. However, for mar, the specific gender assignment in manuscript BL, Additional 10323 (14 th century) is due to the preferences or practices of that specific scribe, while gender variation of that noun in general appears to be entrenched in Old Occitan. All different Old Occitan manuscripts of this hagiography are based on the same model at some point, with some intermediate steps, as shown in Figure 6.1. As there is variation between the different manuscripts in the COMETA (2025) corpus, this variation must be due to the influence of a scribe. It is unclear whether it is the scribe who wrote the model for either branch or whether it was the scribe of a specific manuscript who was responsible for the respective gender variation (or maybe both). However, it is clear that the differences in gender assignment between the manuscripts originate somewhere in the copying process, meaning the scribe plays a key role in gender assignment in medieval written language. 10.1.4 Intrascribal Variation The most notable finding is that 14 out of the 17 manuscripts in the COMETA (2025) corpus show internal variation in gender assignment. This variation is present to a different extent, but is to be seen in all of them. This strong gender variation appears to be entrenched in this time period, regardless of the scribe. The lack of variation in the remaining three manuscripts (BL, Harley 3041 (14 th century), BnF, Latin 901 (14 th century), and BnF, NAF 11180 (14 th century)) can easily be explained: These manuscripts have a low number of tokens: 501, 1,137, and 4,636 respectively, totalling only 6,274 tokens. <?page no="248"?> 248 10 Analysis and Discussion An example of variation within a single scribe’s text can be seen in manuscript BmC, Cote 34 (13 th century), where the noun man occurs in both the masculine and feminine forms within only two folios of the manuscript, in very similar contexts. These examples were already shown in (61a) and (61b). However, they will be shown again in Example (91) to illustrate the variability in gender assignment for this noun within the text of a single scribe. (91) a. Guillem la ma nuda miret ‘Guillem looked at the F bare F hand’ (BmC, Cote 34 (13 th century: f o . 45 v o ., c. 1, l. 5)) b. Ab lo man ẹ nuz ‘With the M bare M hand’ (BmC, Cote 34 (13 th century: f o . 43 v o ., l. 20)) Regarding gender assignment in the two parts of the Croisade, it was shown that there is little difference between the two parts of the manuscript. This could be due to the scribe normalising gender assignment. However, another, more likely explanation is that the two parts differ so much lexically that they do not have enough shared lexical material to display any possible differences in gender assignment. The third, but most unlikely explanation is that the two authors’ language is so similar that they did not differ in their gender assignment apart from the small variations shown in Section 9.2. This explanation is highly improbable, as a comparison of the types used in the different parts of the manuscript shows that the two authors only overlap slightly in their use of lexical types, whith each author using many types not present in the other part. 10.2 Linguistic Factors This section will deal with the linguistic factors that contribute to the gender variability of certain nouns, especially morphology and semantics, as well as a combination of these two factors. Apart from the factors below, the following additional linguistic factors were checked for potential influences on gender assignment: type of agreement (prenominal versus postnominal), type of <?page no="249"?> 10.2 Linguistic Factors 249 agreement target (article, adjective, pronoun, participle), number (singular versus plural), case (nominative versus oblique), and if applicable, rhyme structure. As none of these factors triggered any apparent differences in gender assignment (apart from maybe rhyme structure, see below), they will not be discussed further. More precisely, there is no indication of the presence of a similar gender system to that found in Asturian (see Section 3.3.3.2), nor of a Genus Alternans. Another factor that plays a role in gender variation, but which cannot be clearly attributed to any of the aforementioned factors, is text type. For example, texts in rhyme form are susceptible to, or even encourage, gender variation. As was demonstrated for art (see Section 8.2.2.2) and fam (see Section 8.2.2.7), the agreement targets of these nouns were changed to a different gender in order to fit the rhyme scheme. As most of the texts in COMETA (2025) are prose, it was not possible to systematically check for gender variability in poetry. Nevertheless, it is evident that rhymes contribute to gender variability. In order to grasp the importance of this factor, a larger corpus of poetry would be necessary. 10.2.1 Morphological Variation Based on the presented data, some common morphological features were found among a majority of nouns that are variable in gender. In general, all gender doublets discussed in this book have a rather simple morphology without any suffixes. This can easily be explained by the fact that suffixes mainly have a fixed gender (see Bauer 2010; Spencer 2015), meaning that words formed through suffixation are usually not gender variable. Many of the nouns analysed here are even simplicia without any affixes, or have only inflectional morphology for number or case marking. An example of the differences in gender assignment between a simple word form and a more complex one (regarding phonology, not morphology) can be seen with pols. This noun is usually used alongside its longer form, polvera; however, the findings in Section 8.2.2.18 clearly suggest that these two forms should be considered separately. The short form pols, which is invariable in case and number, triggers both masculine and feminine gender. This can be explained by the lack of morphological (and also semantic) cues which would contribute to a preference for the one gender. The longer form polvera, on the other hand, is not a gender doublet, and appears only with feminine agreement. This is certainly due to the ending in -a, which strongly indicates feminine gender in Old Occitan. Furthermore, historical morphology plays an important role. In this case, the original Latin declension class is crucial in determining the gender. While most nouns inherited from the first, second, and fourth declension classes have a clear gender value in Old Occitan, most of the gender-variable nouns are inherited from the third declension class. The fifth declension class would also <?page no="250"?> 250 10 Analysis and Discussion be important for variability in gender assignment, but due to the low overall number of nouns in this class, only one gender-variable noun remains (dia), which was not even gender-variable anymore at the time of the manuscripts in the COMETA (2025) corpus. Of all the nouns discussed in this book, only two gender-variable nouns are derived from a second declension class noun (namely (ves)comtat and camps), and only one is from the fourth declension class (man). There are no gender-variable nouns inherited from the first declension class 2 . Additionally, there are four gender-variable nouns that may not have been inherited from Latin nouns but were derived from other word classes through conversion, namely the nouns caytieu, dreg, estaie, and sera. However, it is unclear when this conversion process took place and it is also possible that these nouns were inherited as nouns and converted already in Latin (see, for example, the discussion in Section 9.1.2; see also serpen, Section 8.2.2.24). All other hereditary nouns with variable gender are inherited from the third declension class, demonstrating the important role of declension class in gender assignment during the transition from Latin to Romance languages. Another case where gender variability originates in the Latin word formation process concerns the two nouns in the list of gender doublets that end in -al, namely aigal (see Section 8.2.2.1) and canal (see Section 8.2.2.3). Both of these stem from Latin nouns ending in -lis. In Latin, the suffix -lis was used to form adjectives describing a relation to or a quality of something (see Rubenbauer/ Hofmann 1995: 19-20). Therefore, the noun aalis is a deadjectival conversion of the former adjective derived from aa ‘water’ with the meaning ‘watery, related to water’. These adjectives belong to the third declension class in Latin and therefore have no distinct feminine or masculine form, both genders have the same ending. These deadjectival nouns behave similarly to other third declension nouns and are predisposed to gender variation, a tendency that is further reinforced by their adjectival use with different genders already in Latin. 10.2.2 Semantic Variation There is one noun in the list of gender doublets where the gender variation found in COMETA (2025) is always in combination with semantics, i.e., salut (see Section 8.2.2.21). In general, this noun is feminine in Old Occitan for meanings such as ‘salvation, health, intelligence’ and so on. Only when this noun refers to salutation or means ‘greeting’ is it variable in gender, triggering both masculine and feminine agreement. This is the only noun addressed in this study where 2 With the exception of Greek loanwords adopted into the first declension class and inherited into Old Occitan; however, they are treated as loanwords rather than hereditary words. Furthermore, they represent cases where semantic and formal assignment rules play a role, so they will be discussed next in Section 10.2.3. <?page no="251"?> 10.2 Linguistic Factors 251 semantics alone seem to distinguish between a feminine noun and a gender doublet. The fact that salut is not variable in gender apart from when it means the greeting may have another cause besides semantics: It is also possible that the gender-variable form with the meaning of ‘greeting’ has a different etymology (or folk etymology, see the discussion in Section 8.2.2.21), and is therefore (considered) a homonymous form of the invariable feminine salut. In this case, the etymon of the noun designating the greeting would be the past participle of the Latin verb salutaRe ‘to greet’. This would be analogous to the situation in other Romance languages such as Italian saluto or Friulian, French, and Catalan salut, where the dedicated form is assumed to be derived from the past participle (see REW 1935: s.v. “salutare”). In the case of folk etymology, as proposed in the FEW (1922-2003), the perceived etymon would be the past participle of the Old Occitan verb saluer ‘to greet’. In either case, the variability in gender would not be surprising as past participles may take both genders and variability in gender can be expected (see Marzo/ Wiedner 2025: 204). 10.2.3 Morphosemantic Variation The category of combined variation mainly concerns the loanwords from Greek inherited in Romance languages. Most of the gender-variable nouns loaned from Greek designate men, as e.g., papa, profeta, or evangelista. In a semantic gender assignment system, these nouns should all be masculine. However, since Old Occitan has a formal assignment system with a semantic basis rather than a purely semantic one, the gender assignment cannot be made solely on the basis that these nouns have the semantic property of designating male humans. From a formal point of view, these nouns closely resemble feminine nouns ending in -a, as there are only a small number of masculine nouns with this ending, most of which are Greek loanwords. In this case, there is clearly a mismatch between semantics and morphological form. It seems as though the variation in gender is free; it was impossible to identify any semantic or formal cues that would trigger masculine use in one case and feminine use in another. The choice of gender appears to be totally variable. One theory on a factor that may influence the choice of gender for these nouns is the following: for a particular speaker or scribe in a given context, the semantics of the noun designating specifically a man are paramount, and the noun is therefore used with masculine agreement. In another context, this aspect of the semantics is less important when the noun simply designates a person or group of people. In this context, formal cues override semantics and the noun is used with feminine agreement. This may be due to a hypercorrection in relation to the Latin gender system, where the morphological and phonological cue of the <?page no="252"?> 252 10 Analysis and Discussion singular word final -a was strongly correlated with feminine gender. In these cases, gender assignment is based on formal criteria in relation to the Latin system. This orientation creates the possibility of solely formal assignment in cases where semantic criteria are not as important. <?page no="253"?> 11 Conclusion This study first introduced the fundamental concepts of gender and gender doublets (see Chapter 2), before demonstrating different scenarios of gender systems in languages around the world, particularly in Latin and Romance languages (see Chapter 3). This was followed by Chapter 4, which provided an overview of the Old Occitan nominal system. Following these theoretical chapters, Chapter 5 presented the research objectives, which guided the subsequent chapters. Based on these objectives, it became clear that existing editions and corpora would not meet the needs of the present study, so a specific corpus was created. Chapter 6 presented in detail the manuscripts selected for inclusion in the COMETA (2025) corpus, providing an overview of known facts about the scribes, authors, contexts of creation, contents, and editorial history. This offered insights into the importance of the manuscripts for scholars. All manuscripts in the COMETA (2025) corpus originate from the Languedocian and Provençal linguistic areas between 1250 and 1400. Following this, Chapter 7 presented the technical aspects of corpus creation: it was built using Transkribus (see Wiedner 2023) for semi-automatic transcription as well as LLMs to annotate the transcriptions according to an adaptation of the UD tagset (see Schöffel/ Wiedner, et al. 2025; Schöffel/ Garces Arias, et al. 2025). The following two chapters presented the corpus analysis. Chapter 8 presented an analysis of 39 gender doublets in the COMETA (2025) corpus focusing on diatopic variation between Old Languedocian and Old Provençal. Chapter 9 presented two case studies. The first explored differences in gender assignment between three manuscripts of the Vida de Sant Honorat (see Section 9.1). The second case study, in Section 9.2, dealt with differences in gender assignment between the two parts of the Croisade, the first part composed by Guilhem de Tudela, and the second by an anonymous author. The findings of these two chapters were then discussed in Chapter 10. As demonstrated in Chapter 8, the methodological approach adopted in this study shows that editions of medieval texts are not always reliable because they often contain errors or normative tendencies regarding morphosyntactic features such as gender. Furthermore, it is important to bear in mind that glossary entries are not always faithful to the occurrences in the text, with many examples showing one gender indicated in the glossary and a different gender (or no clear gender) present in the text itself. Therefore, the respective manuscripts should be consulted to support or disprove the findings extracted from the edi- <?page no="254"?> 254 Conclusion tions. This is corroborated by the discovery of seven gender-variable nouns within the three manuscripts of the Vida de Sant Honorat, analysed in Section 9.1. This variation in gender is not indicated in the editions and would therefore not have been found in the context of this research had only the existing editions of the manuscripts been used. However, further investigation is needed for some of these nouns to confirm or disprove the status of a gender doublet, and to ensure that these occurrences are not writing errors, e.g., concerning maire. In light of these issues encountered when using editions, it is necessary to verify the occurrences listed in Chapter 8 that are taken from editions against the original manuscripts. This was not feasible in the context of this research, but would be an important next step in order to further confirm the tendencies found here. The following sections revisit the research areas outlined in Chapter 5: Research Area 1 concerned the temporal and spatial distribution of gender doublets in Old Occitan, particularly in Languedocian and Provençal. In total, three nouns were found to exhibit differences in gender assignment between the two regions: (ves-)comtat, mar, and sanc. Additionally, the noun dia showed diachronic variation in Old Occitan, as its gender became fixed by the mid-13 th century despite it clearly being a gender doublet previously. Four nouns listed in the DOMél (2025) were shown not to be gender doublets, and the status of six further nouns is uncertain, as they show only minimal variation. Conversely, seven additional nouns were identified that should be validated and potentially added to the list of gender-variable or ambiguous nouns (see Section 9.1 for details). Further examination of other gender doublets is required, in particular those listed in the DOMél (2025) that have no occurrences in COMETA (2025). Furthermore, additional case studies are required to gain a more detailed understanding of gender variation in Old Occitan and other medieval Romance vernaculars. Regarding the development of the Latin neuter nouns in Old Occitan (Research Area 2), it was found that they behave similarly to neuters in the standard Romance languages (except Romanian). Most neuter nouns from the second declension class became either masculine, derived from the singular form of the noun, or feminine, derived from the plural form. These fused with nouns from the second or first declension class, respectively, based on morphophonological similarities. Neuter nouns derived from the third Latin declension class show greater gender variation in Old Occitan as evidenced by the extensive list of gender doublets from the third declension class presented in Section 8.2.2. For nouns where the ending does not provide much information about gender (see also Section 3.2), the gender was already somewhat variable in Latin. This variability was inherited into the Romance languages. This is also evident in the gender difference for nouns such as maRe, which is masculine in Italian but <?page no="255"?> Conclusion 255 feminine in French. This study therefore demonstrates that both Old Occitan and Occitan behave similarly to other Romance languages in their treatment of Latin neuters. This emphasises the importance of morphological cues for consistency (or the lack thereof) in gender assignment (see also the results of the computer simulation in Wiedner/ Schöffel 2026). As expected, due to the time period analysed here, no instances of Genus Alternans could be found. In order to find potential evidence of alternating gender in Old Occitan, older texts would need to be analysed (see Loporcaro 2018; Wang 2023, 2025). Research Area 3 concerned the influence of scribes and their personal profiles, as well as the authors’ influence on gender assignment. Regarding the latter, it could not be determined whether the two authors of the Croisade Albigeoise influenced gender assignment, probably because of the substantial differences in their respective lexicons. Nevertheless, it was shown that scribes played an important role in gender assignment within different manuscripts of the same text. For example, the gender assigned to the noun mar in the Vida de Sant Honorat depends on the scribe; while the gender remains unchanged, the scribe clearly influences the preference for either masculine or feminine use. Furthermore, it was demonstrated that gender variation is a reality in Old Occitan and is not solely dependent on a single scribe. Evidence of gender variation within a manuscript written by a single scribe was found in 14 out of the 17 manuscripts in the COMETA (2025) corpus. The other three manuscripts probably contain too few tokens (fewer than 6,300 in total) for gender variation to become apparent. Nevertheless, it would be important to know more about the sociolinguistic background of these scribes to better contextualise these findings. However, the fact that 14 different scribes employed at least some nouns with both genders within a single manuscript demonstrates that the variation is not solely due to a single scribe with a different linguistic background. This finding underscores the importance of taking manuscript variation seriously and interpreting it as such, rather than automatically classifying anything that appears unexpected from a modern, norm-based perspective as an error. The meta-linguistic commentary by the Consistori del Gay Saber in Example (70) corroborates the presence of gender doublets in Old Occitan in general. This commentary criticises the feminine use of semantically masculine Greek loanwords such as papa and profeta. This part of the Leys d’amors strongly suggests that the gender variation found in the manuscripts is not merely due to scribal errors or rare confusions, but is most probably a feature of the spoken language used by speakers (at least those of a certain educated elite). Further investigations into gender distribution in other Romance languages would be very interesting, particularly in Ibero-Romance languages such as Aragonese and Catalan, given that these languages still show considerable variation today. <?page no="256"?> 256 Conclusion Another interesting area for future research concerns gender assignment in word formation processes involving deverbal conversion from the past participle (see Marzo/ Wiedner 2025: 202). Research in other Romance languages on gender assignment for deverbal nominalisations due to conversion from the past participle suggests that greater gender variation may be expected (for Old Italian, see Marzo/ Wiedner 2025: 202-204), a phenomenon that should therefore be further investigated. As the COMETA (2025) corpus is too small to display this sort of variation, further research will be necessary (and probably fruitful), as soon as larger corpora for Old Occitan become available. As shown by Marzo/ Wiedner (2025), the study of gender assignment in word formation through deverbal conversion from the past participle is surprisingly limited, as research on gender in Romance languages generally does not address conversion at all. This is all the more remarkable given that gender assignment in deverbal noun conversion is one of the most challenging questions in the study of conversion in Romance languages. Unlike nominalisation through derivation, where derivational affixes can account for the gender of the derived nouns, gender assignment in cases of conversion remains of great interest when studying gender systems in the Romance languages. The findings of this analysis raise the question of whether gender has always been an important feature throughout the history of Occitan and, therefore, of other Romance languages. The high degree of variation, to some extent without visible regularities, suggests that it does not matter whether a noun is used with masculine or feminine agreement in order to be understood. This observation is further supported by authors deliberately changing the gender of nouns to fit the rhyme scheme in poems, as shown for art (see Section 8.2.2.2) and fam (see Section 8.2.2.7). To confirm this, it would be necessary to look at a broader range of nouns within the context of rhymed texts to ensure that this flexibility in gender assignment also applies to nouns that are not gender-variable in other contexts. This would provide evidence for the relatively low importance of gender in Old Occitan. Furthermore, this interpretation is further supported by the fact that a fixed gender is not consistently applied to animates and human beings, as demonstrated by Greek loanwords, for example, where a mismatch between grammatical gender and biological sex can be observed. While it seems important to have some sort of determiner (if syntactically or semantically necessary), the gender does not seem to play a crucial role in Old Occitan for it to function as a means of communication. This is supported by psycholinguistic research showing that even L1 speakers of languages with grammatical gender do not acquire them with 100% accuracy (see Ayoun 2018: 138). If native speakers do not fully master the gender system of their L1, variation in gender will inevitably arise. However, the fact that gender is a feature present in many languages suggests <?page no="257"?> Conclusion 257 that it is important, perhaps to facilitate access to nouns in the mental lexicon and aid in recognising these nouns (see Vigliocco et al. 2005; Cubelli et al. 2011; Boutonnet/ Athanasopoulos/ Thierry 2012; Pavlidou/ Alvanoudi 2014). Gender variation seems to be independent of the type of gender system found in a given language. It exists in Old Occitan to a greater extent than in standardised languages, suggesting that standardisation processes can eliminate (or reduce) gender variability. Since Old Occitan is a non-standardised medieval language, variation is present on many levels, such as graphical representations or morphosyntactic categories, such as gender. The latter is regulated to a certain extent diatopically, but variability emerges in the written form and writing process, due to exposure to numerous influences, such as the scribe’s personality and linguistic background, the respective scripta, language contact and multilingualism, text genre and discourse tradition, and the context of writing (e.g., text copying versus text composition). Much of this variation cannot be explained by diatopy, diachrony, or semantics; it seems to exist independently of these factors. This finding demonstrates that two forms with the same meaning can coexist without leading to communicative difficulties. It also underscores the importance of studying variation in medieval texts that cannot be accounted for by a single, clear explanation. Through the thorough contextualisation of manuscripts and different case studies, I have ensured that no reductionist view of medieval variation is presented here. Instead, I employed an approach that captures both the distinctiveness of individual patterns and the broader collective tendencies underlying them. <?page no="259"?> Bibliography Primary sources BL, Additional 10323 (14 th century): La vida de sant Honorat (manuscrit O). BL, Additional 21218 (14 th century): Roman de Philomena (manuscrit B). BL, Harley 3041 (14 th century): Tenson entre Aicart et Girard. BL, Harley 7403 (14 th century): Traduction occitane de l’Évangile de Nicodème ; traduction occitane (version B) du poème français sur les quinze signes de la fin du monde : Los XV signes que veno ; traduction occitane de la légende du bois de la Croix ; traité de diététique en occitan ; le repentir du pécheur ; doctrinal. BmC, Cote 34 (13 th century): Roman de Flamenca. BmT, Cote 2883 (14 th century): Las Leys d’Amors. BnF, Arsenal 6355 (14 th century): La Vida de Santa Enimia. BnF, Français 1049 (1343): Trepas du roy Robert de Sicile, comte de Provence ; Libre de vicis et de vertutz, rédaction A ; Barlam et Josaphas. BnF, Français 13503 (13 th / 14 th century): Li vida de la benaurada sancta Doucelina, mayre de las donnas de Robaut. BnF, Français 13504 (14 th century): Vie de saint Elzéar ; Traduction d’une vision de Marguerite d’Oingt ; Vie de Sainte Delphine. BnF, Français 13509 (14 th century): Vida de sant Honorat (manuscript R) ; Fragment de sermon, en occitan ; Début d’une prétendue Lettre de Jesus Crist. BnF, Français 2232 (1325): Roman de Philomena (manuscript P). BnF, Français 25425 (1275): Chanson de la Croisade contre les Albigeois. BnF, Latin 901 (14 th century): Version occitane des Plaintes de la Vierge. BnF, NAF 11151 (14 th century): Les sentences arbitrales entre l’abbé de Saint- Géraud et les consuls de cette ville (1280 et 1298) ; traduction occitane de la Mulomedicina ; la recepta del vi. BnF, NAF 11180 (14 th century): Planh de la Vierge. BnF, NAF 6195 (14 th century): La vida de sant Honorat (manuscrit M). COMETA = Wiedner, Marinus (ed.) (2025): COMETA : Corpus de l’occitan médiéval comparatif et annoté: Provence et Languedoc. <https: / / zenodo.org/ records/ 15300719> [visited on 04/ 08/ 2026]. <?page no="260"?> 260 Bibliography Secondary sources Académie Française ( 9 1980-2024): Dictionnaire de l’Académie française. <https: / / www.academie-francaise.fr/ le-dictionnaire/ portail-numerique-dudictionnaire-de-lacademie-francaise> [visited on 04/ 08/ 2026]. Académie française (2020): Le Covid-19 ou La Covid-19. <https: / / www.academiefrancaise.fr/ le-covid-19-ou-la-covid-19> [visited on 04/ 08/ 2026]. Acquaviva, Paolo (2020): “Gender as a Property of Words and as a Property of Structures”. Catalan Journal of Linguistics 19: 49-74. Aikhenvald, Alexandra Y. (2000): Classifiers: A Typology of Noun Categorization Devices. Oxford: Oxford University Press. Aikhenvald, Alexandra Y. ( 2 2006): “Classifiers and Noun Classes: Semantics”. In: Brown, Keith (eds.): Encyclopedia of Language and Linguistics. Oxford: Elsevier, 463-471. Aikhenvald, Alexandra Y. (2012): “Round Women and Long Men: Shape, Size, and the Meanings of Gender in New Guinea and Beyond”. Anthropological Linguistics 541: 33-86. Aikhenvald, Alexandra Y. (2016): How Gender Shapes the World. Oxford: Oxford University Press. Aikhenvald, Alexandra Y. (2025): A Guide to Gender and Classifiers. Oxford: Oxford University Press. Aikhenvald, Alexandra Y. (2026): Noun Categorization: A Comprehensive Typology (LINCOM Studies in Language Typology 35). Munich: LINCOM. Aikhenvald, Alexandra Y./ Mihas, Elena (2019): Genders and Classifiers: A Cross- Linguistic Typology. Oxford: Oxford University Press. Albanés, L’abbé Joseph-Hyacinthe (1879): La vie de sainte Douceline fondatrice des Béguines de Marseille. Composée au treizième siècle en langue provençale. Publiée pour la première fois, avec la traduction en français et une introduction critique et historique. Marseille: Étienne Camoin. Alibèrt, Loís ( 2 1976): Grammatica occitana: segon los parlars lengadocians. Toulouse/ Barcelona: Institut d’estudis occitans/ Institut d’estudis catalans. Allassonnière-Tang, Marc et al. (2021): “Expansion by Migration and Diffusion by Contact is a Source to the Global Diversity of Linguistic Nominal Categorization Systems”. Humanities & Social Sciences Communications 8: 1-6. Anglade, Joseph (1913): La bataille de Muret (12 septembre 1213) d’après la ”Chanson de la Croisade”. Texte et traduction. Toulouse/ Paris: Privat/ Picard. <?page no="261"?> Bibliography 261 Anglade, Joseph (1919a): Las leys d’amors. Manuscrit de l’Académie des jeux floraux. Tome I (Bibliothèque méridionale, 1re série 17). Toulouse/ Paris: Privat/ Picard. Anglade, Joseph (1919b): Las leys d’amors. Manuscrit de l’Académie des jeux floraux. Tome II (Bibliothèque méridionale, 1re série 18). Toulouse/ Paris: Privat/ Picard. Anglade, Joseph (1919c): Las leys d’amors. Manuscrit de l’Académie des jeux floraux. Tome III (Bibliothèque méridionale, 1re série 19). Toulouse/ Paris: Privat/ Picard. Anglade, Joseph (1920): Las leys d’amors. Manuscrit de l’Académie des jeux floraux. Tome IV. Études, Notes, Glossaire et Index (Bibliothèque méridionale, 1re série 20). Toulouse/ Paris: Privat/ Picard. Anglade, Joseph (1921): Grammaire de l’ancien provençal ou ancienne langue d’oc: Phonétique et morphologie. Paris: Librairie Honoré Champion. Anglade, Joseph (1925): “Le troubadour N’At de Mons et les Leys d’amors”. Romania 51: 414-422. Anglade, Joseph (1917-1918): “Poésies religieuses inédites du XIVe siècle en dialecte toulousain tirées des Leys d’amors”. Annales du Midi 29-30: 1-48. Anonyme (ed.) (1863): Histoire anonyme de la guerre des Albigeois. Nouvelle édition publiée pour la première fois séparément, revue et corrigée sur l’édition des Bénédictins, sur celle de M. du Mège et sur le manuscrit de Toulouse avec un glossaire, des fragments de langue romane, depuis le XI e siècle jusqu’à nos jours et une introduction par un indigène. Toulouse: Bompard. AntConc = Anthony, Lawrence (ed.) (2024): AntConc (Version 4.3.1). <https: / / www.laurenceanthony.net/ software/ antconc/ > [visited on 04/ 08/ 2026]. Appel, Carl ( 4 1930): Provenzalische Chrestomathie mit Abriss der Formenlehre und Glossar. Leipzig: O. R. Reisland. ARLIMA = (2025): Les Archives de littérature du Moyen-Âge. <https: / / www.arlima.net/ > [visited on 04/ 08/ 2026]. Armbruster, Karl (1888): Geschlechtswandel im Französischen: Masculinum und Femininum. Heidelberg: Malsch & Vogel. Armitage, Frederick (1884): Sermons du XIIe siècle en vieux provençal, publiés d’après le ms. 3548 B de la Bibliothèque nationale. Heilbronn: Henninger. Armstrong, Nigel et al. (eds.) (2001): La langue française au féminin : Le sexe et le genre affectent-ils la variation linguistique ? Paris: L’Harmattan. <?page no="262"?> 262 Bibliography Arthur, Ingrid (1992): Miracles que Dieus ha mostratz per sant Frances apres la sua fi. Version occitane de la Legenda Maior Sancti Francisci - Miracula de saint Bonaventure. Édition et étude de la langue (Monografier utgivna av Kungliga Humanistiska Vetenskaps-Samfundet i Uppsala 4). Stockholm/ Uppsala: Almqvist & Wiksell International. Ayoun, Dalila (2018): “Grammatical Gender Assignment in French: Dispelling the Native Speaker Myth”. Journal of French Language Studies 28(1): 133-148. Ayoun, Dalila/ Maranzana, Stefano (2020): “Italian Grammatical Gender: A Corpus Study from a Second Language Acquisition Perspective.” Italica 97(2): 306-334. Azaïs, Gabriel (ed.) (1862): Le Breviari d’amor de Matfre Ermengaud suivi de sa lettre à sa sœur. Tome second. Béziers: Secrétariat de la Société archéologique, scientifique et littéraire de Béziers. Azofra Sierra, María Elena (2019): “Morfosintaxis diacrónica”. In: Ridruejo, Emilio (eds.): Manual de lingüística española. Berlin/ Boston: De Gruyter, 96- 132. Bach, Xavier (2024): “Flexion des constituants du syntagme nominal”. In: Esher, Louise/ Sibille, Jean (eds.): Manuel de linguistique occitane. Berlin/ Boston: De Gruyter, 313-328. Baldinger, Kurt (1963): “Textes anciens gascons tirés du Livre Vert de Bénac (Cartulaire des vicomtes de Lavedan), publié et annoté par Gaston Balencie, 1910”. In: Straka, Georges (eds.): Les anciens textes romans non littéraires : leur apport à la connaissance de la langue au Moyen-Âge. Colloque international organisé par le Centre de Philologie et de Littérature romanes de l’Université de Strasbourg du 30 janvier au-4 février-1961. Paris: Klincksieck, 82-94. Balnat, Vincent (2020): “Unter Beobachtung: Corona-Wortschatz im Deutschen und Französischen”. Nouveaux Cahiers d’Allemand: Revue de linguistique et de didactique 38(2): 139-159. Bampa, Alessandro (2017): “La transition entre les deux parties de la Chanson de la Croisade albigeoise”. Romania 135: 90-113. Bartsch, Karl (ed.) (1856): Denkmäler der Provenzalischen Literatur. Stuttgart/ Tübingen: H. Laupp. Bartsch, Karl (1866): “Besprechung von Paul Meyers Le roman de Flamenca, Paris 1865”. Jahrbuch für romanische und englische Literatur 7: 188-205. Bartsch, Karl/ Koschwitz, Eduard (eds.) (1904): Chrestomathie provençale (X e -XV e siècles). Marburg: N. G. Elwert’sche Verlagsbuchhandlung. <?page no="263"?> Bibliography 263 Basirat, Ali et al. (2021): “An Empirical Study on the Contribution of Formal and Semantic Features to the Grammatical Gender of Nouns”. Linguistics Vanguard 1: 20200048. Bauer, Brigitte (2010): “Word Formation”. In: Maiden, Martin et al. (eds.): The Cambridge History of the Romance Languages. Cambridge: Cambridge University Press, 532-563. BDLC = (2025): Banque de Données Langue Corse. <https: / / bdlc.univ-corse.fr> [visited on 04/ 08/ 2026]. Bec, Pierre (1963): La langue occitane. Paris: Presses universitaires de France. Becquer, Annie et al. (1999): Femme, j’écris ton nom… : Guide d’aide à la féminisation des noms de métiers, titres, grades et fonctions. Paris: La Documentation française. Beggiato, Fabrizio (1984): Il trovatore Bernart Marti (Subsidia al Corpus des troubadours 10 = Studi, testi e manuali 12). Modena: Edizioni dell’Orso. Benincà, Paola/ Penello, Nicoletta (2010): “Morfologia nominale”. In: Renzi, Lorenzo/ Salvi, Giampaolo (eds.): Grammatica dell’italiano antico (Strumenti II). Bologna: Il Mulino, 1389-1421. Blanc, Alphonse (ed.) (1899): Le livre de comptes de Jacme Olivier, marchand narbonnais du XIV e siècle. Publié avec une introduction, un glossaire, des notes et des tables. Tome seconde 1 re partie. Paris: Alphonse Picard et fils. Blodgett, Edward D. (1995): The Romance of Flamenca. Edited and translated by E. D. Blodgett. New York/ London: Garland. Blom, Alderik H. (2023): “Gaulish in the Late Empire (c. 200-600 CE)”. In: Mullen, Alex/ Woudhuysen, George (eds.): Languages and Communities in the Late- Roman and Post-Imperial Western Provinces. Oxford: Oxford University Press, 129-154. Bondurand, Édouard (1883): “Fragment de recettes médicales en langue d’oc”. Romania 12: 100-104. Bonnier Pitts, Monique (1989): Barlam et Jozaphas, roman du XIVe siècle en langue d’oc: BN, fr. 1049. Édition critique, traduction, notes et commentaires (Langue et littérature d’oc 5). Paris: Presses de l’Université de Paris-Sorbonne. Boutière, Jean/ Schutz, Alexander Herman (eds.) ( 2 1964): Biographies des troubadours: textes provençaux des XIIIe et XIVe siècles. Édition refondue, augmentée d’une traduction française, d’un appendice, d’un lexique, d’un glossaire et d’un index (Les classiques d’oc 1). Paris: A.-G. Nizet. Boutonnet, Bastien et al. (2012): “Unconscious Effects of Grammatical Gender During Object Categorisation”. Brain Research 1479: 72-79. <?page no="264"?> 264 Bibliography Brüch, Josef (1931): “Altprov. semal „Zuber, Bütte” und kat. samal „Eimer, Kiepe””. Zeitschrift für französische Sprache und Literatur 54, 1/ 2: 60-67. Brun-Trigaud, Guylaine/ Sauzet, Maguelone (2023): “Variations de genre et métaplasme dans les parlers francoprovençaux”. In: Aquino-Weber, Dorothée et al. (eds.): ”Coum’on étèila que kòoule… Come una stella cadente… Comme un étoile filante…” Mélanges à la mémoire de Federica Diémoz. Geneva: Droz. Brun-Trigaud, Guylaine et al. (2023): “Variation du genre des substantifs dans les dialectes gallo-romans. Étude exploratoire”. Géolinguistique 23. Brunel, Clovis (1916): Bertran de Marseille, La vie de sainte Énimie, poème provençal du XIII e siècle (Les classiques français du Moyen Âge, 17). Paris: Champion. Brunel, Clovis (1935): Bibliographie des manuscrits litteraires en ancien provençal. Paris: Librairie E. Droz. Brunel, Clovis (1953): “Le sermon en langue vulgaire prononcé à Toulouse par Saint Vincent Ferrier le Vendredi Saint 1416”. Bibliothèque de l’École des chartes 111: 5-53. Brunel, Clovis (1957): “Recettes médicales de Montpellier en ancien provençal”. Romania 78: 289-327. Buridant, Claude (2022): Grammaire du français médiéval XI e -XIV e siècles : Nouvelle édition intégralement revue et élargie. Strasbourg: Éditions de Linguistique et de Philologie (ELiPhi). Busby, Keith (1997): “Hagiography at the Confluence of Epic, Lyric, and Romance: Raimon Feraut’s La vida de sant Honorat”. Zeitschrift für romanische Philologie 113(1): 51-64. Cambell, Jacques (1963): Vies occitanes de saint Auzias et de sainte Dauphine (Bibliotheca Pontificii Athenaei Antoniani 12). Roma: Pontificium Athenaeum Antonianum. Camprós, Carles (1972): La cançon de la crosada, tròces causits, presentas amb una introduction e de nòtas (Los pichons classics occitans 10). Montpellier: Centre d’Estudis Occitans. Camps, Jean-Baptiste (2010): Les manuscrits occitans à la Bibliothèque nationale de France. Cappellaro, Chiara (2015): “Plurality and Gender in a Central Calabrian Dialect: The Emergence of an Innovative Type of Romance Genus Alternans in Bocchiglierese”. Archivio Glottologico Italiano 100(2): 208-231. Caravaggi, Giovanni (1963): Vangeli provenzali dell’infanzia (Testi e manuali di filologia romanza 47). Modena: Società Tipografica Editrice Modenese (STEM-Mucchi). <?page no="265"?> Bibliography 265 Carles, Hélène (2024): “Émergence de l’occitan et développement des usages à l’époque médiévale”. In: Esher, Louise/ Sibille, Jean (eds.): Manuel de linguistique occitane. Berlin/ Boston: De Gruyter, 113-158. Carles, Hélène/ Glessgen, Martin (2019): “L’élaboration scripturale du francoprovençal au Moyen Âge”. Zeitschrift für romanische Philologie 135(1): 68-157. Cavaliere, Alfredo (ed.) (1938): Cento liriche provenzali. Testi, versioni, note, glossario. Introduzione di Giulio Bertoni ”La lirica dei Trovatori”. Bologna: Zanichelli. Cavallin, Samuel Johan (1952): “Sermo de Vita S. Honorati”. In: Vitae sanctorum Honorati et Hilarii: episcoporum Arelatensium (Skrifter utgivna av Vetenskapssocieteten i Lund 40). Lund: Vetenskapssocieteten i Lund. CGL = Diggle, James et al. (eds.) (2021): The Cambridge Greek Lexicon. Cambridge: Cambridge University Press. Chabaneau, Camille (1875): “Notes sur quelques pronoms provençaux”. Romania 4: 338-347. Chabaneau, Camille (1876): “Sur quelques pronoms provençaux. Notes supplémentaires”. Romania 5: 232-235. Chabaneau, Camille (1878): “Sur lo, pronom neutre en provençal”. Romania 7: 329-330. Chabaneau, Camille (1885): Origine et établissement de l’Académie des jeux floraux, extraits du manuscrit inédit des Leys d’amors publiés avec une introduction, des notes et une table alphabétique des poètes de l’École de Toulouse. Toulouse: Privat. Chabaneau, Camille (1887): Sainte Marie Madeleine dans la littérature provençale. Recueil des textes provençaux en prose et en vers relatifs à cette sainte publié avec introductions et commentaires. Paris: Maisonneuve frères et Ch. Leclerc. Chambers, Frank M. (1966-1967): “Flamenca”. Romance Philology 20(4): 489-500. Chambon, Jean-Pierre (1992): “Le choronyme alverg (Flamenca v. 1173)”. Nouvelle revue d’onomastique 19-20: 163-167. Chambon, Jean-Pierre (2003): “La déclinaison en ancien occitan, ou : comment s’en débarrasser? : Une réanalyse descriptive non orthodoxe de la flexion substantivale”. Revue de Linguistique Romane 67 (267-268): 343-364. Chambon, Jean-Pierre (2012): “Développement et problèmes actuels des études occitanes”. Revue de Linguistique Romane 76: 199-210. Chambon, Jean-Pierre (2015): “Un auteur pour Flamenca? ” Cultura neolatina 75: 229-271. <?page no="266"?> 266 Bibliography Chambon, Jean-Pierre (2016): “Sur la langue de Flamenc(h)a: quelques diatopismes septentrionaux”. In: Boillat, Stefania Maffei/ Corbellari, Alain (eds.): L’aventure du sens. Mélanges de philologie provençale en l’honneur de François Zufferey (Travaux de linguistique romane. Philologie et édition de textes). Strasbourg: Éditions de linguistique et de philologie (ELiPhi), 67-81. Chambon, Jean-Pierre (2017): “Sur une nouvelle édition du troubadour Daudé de Pradas”. Revue de Linguistique Romane 81: 277-312. Chambon, Jean-Pierre (2018a): “Glanures troubadouresques: possibles attributions de deux coblas et du Roman de Flamenca (Garin d’Apchier, Daude de Pradas)”. Cultura Neolatina 78(1-2): 9-20. Chambon, Jean-Pierre (2018b): “Sur la date de composition du roman de Flamenca”. Estudis romànics 40: 349-355. Chatterjee, Ahan et al. (2026): “Lost in Translation? Exploring the Shift in Grammatical Gender from Latin to Occitan”. In: Hamilton, Sil et al. (eds.): Proceedings of the 6th International Conference on Natural Language Processing for the Digital Humanities. San Diego, USA: Association for Computational Linguistics, 276-296. Cherchi, Paolo ( July 2002): “Review of Il genere “tenzone” nelle letterature romanze delle Origini, edited by Matteo Pedroni and Antonio Stäuble”. Speculum 773: 968-970. Chircu-Buftea, Adrian (2011): Précis de morphologie romane. Cluj-Napoca: Casa Cartii de Stiinta. Clérice, Thibault (2020): Pie Extended, an Eextension for Pie with Pre-Processing and Post-Processing. <https: / / doi.org/ 10.5281/ zenodo.3883589> [visited on 04/ 08/ 2026]. Cocito, Luciana (1971): Il romanzo di Flamenca. Testo, introduzione, traduzione e note. Genoa: Tilgher. Collura, Alessio (2014): Sens e Razos d’una Escriptura: édition et étude de la traduction occitane de l’Evangelium Nicodemi. Montpellier/ Trent: Université Paul-Valéry Montpellier-3/ Università degli Studi di Trento. Collura, Alessio (2019): “Approssimazioni alla “Leggenda del Legno della Croce” nell’Occitania medievale”. In: Mariani, Daniela et al. (eds.): «Tra chiaro e oscuro». Studi offerti a Francesco Zambon per il suo settantesimo compleanno. Trent: Università degli Studi di Trento, Dipartimento di Lettere e Filosofia, 95-112. Collura, Alessio (2022): Apres que Adam fon gitatz de Paradis. Una leggenda occitana sul legno della Croce. Edizione critica a cura di Alessio Collura. Alessandria: Edizioni dell’Orso. <?page no="267"?> Bibliography 267 Constans, Léopold (1879): “Quelques mots sur la topographie du poëme provençal intitulé: Vie de sainte Énimie”. Revue des langues romanes 16: 209- 217. Cooke, Franklin Osborne (1955): Le Roman de Flamenca. Translated for the First Time Completely into English Verse, from the Thirteenth Century Old Provençal, with a Critical Introduction. Boulder: University of Colorado. Corbett, Greville G. (1991): Gender (Cambridge Textbooks in Linguistics). Cambridge: Cambridge University Press. Corbett, Greville G. (2000): Number (Cambridge Textbooks in Linguistics). Cambridge: Cambridge University Press. Corbett, Greville G. (2003): “Agreement: Terms and Boundaries”. In: Griffin, William Earl (eds.): The Role of Agreement in Natural Language. Proceedings of the 2001 Texas Linguistics Society Conference, Austin, Texas. Austin, Texas: Texas Linguistics Society, 109-122. Corbett, Greville G. (2006a): Agreement (Cambridge Textbooks in Linguistics). Cambridge: Cambridge Universitiy Press. Corbett, Greville G. (2006b): “Gender, Grammatical”. In: Brown, Keith (eds.): Encyclopedia of Language & Linguistics (Second Edition). Oxford: Elsevier, 749- 756. Corbett, Greville G. (2007): “Gender and Noun Classes”. In: Shopen, Timothy (eds.): Language Typology and Syntactic Description: III: Grammatical Categories and the Lexicon. Cambridge: Cambridge University Press, 241-279. Corbett, Greville G. (2008): “Systems of Nominal Classification I: Gender Oppositions”. In: Cruse, D. Alan et al. (eds.): 2. Halbband: Ein internationales Handbuch zur Natur und Struktur von Wörtern und Wortschätzen. Berlin/ New York: De Gruyter Mouton, 986-994. Corbett, Greville G. (2012): Features (Cambridge Textbooks in Linguistics). Cambridge: Cambridge University Press. Corbett, Greville G. (2013a): “Number of Genders”. In: Dryer, Matthew S./ Haspelmath, Martin (eds.): The World Atlas of Language Structures Online (v2020.4). Zenodo. <https: / / doi.org/ 10.5281/ zenodo.13950591> [visited on 04/ 08/ 2026]. Corbett, Greville G. (2013b): “Sex-based and Non-Sex-Based Gender Systems”. In: Dryer, Matthew S./ Haspelmath, Martin (eds.): The World Atlas of Language Structures Online (v2020.4). Zenodo. <https: / / doi.org/ 10.5281/ zenodo.13950591> [visited on 04/ 08/ 2026]. Corbett, Greville G. (2013c): “Systems of Gender Assignment”. In: Dryer, Matthew S./ Haspelmath, Martin (eds.): The World Atlas of Language Structures <?page no="268"?> 268 Bibliography Online (v2020.4). Zenodo. <https: / / doi.org/ 10.5281/ zenodo.13950591> [visited on 04/ 08/ 2026]. Corbett, Greville G. (2014): “Gender Typology”. In: Corbett, Greville G. (eds.): The Expression of Gender (The Expression of Cognitive Categories 6). Berlin/ Boston: De Gruyter Mouton, 87-130. Corbett, Greville G./ Fedden, Sebastian (2016): “Canonical Gender”. Journal of Linguistics 52(3): 495-531. Corbett, Greville G./ Fraser, Norman M. (2000): “Default Genders”. In: Unterbeck, Barbara et al. (eds.): Gender in Grammar and Cognition (Trends in Linguistics: Studies and Monographs 124). Berlin/ New York: Mouton de Gruyter, 55-97. Corbett, Greville G./ Fraser, Norman M. (2010): “Gender Assignment: A Typology and a Model”. In: Senft, Gunter (eds.): Systems of Nominal Classification (Language, Culture and Cognition 4). Cambridge: Cambridge University Press, 293-325. Couture, Léonce (1862): “La complainte de Notre-Dame sur la Passion de Notre Seigneur Jésus-Christ. Poésie inédite, en langue d’oc, du quatorzième ou du quinzième siècle”. Bulletin du Comité d’histoire et d’archéologie de la province ecclésiastique d’Auch 3: 180-192. Covarsí Carbonero, Jaime (2010): Roman de Flamenca. Traducción y edición a cargo de Jaime Covarsí Carbonero. Murcia: Editum. Ediciones de la Universidad de Murcia. Crescini, Vincenzo ( 2 1905): Manualetto provenzale per uso degli alunni delle facoltà di lettere. Introduzione grammaticale, crestomazia e glossario. Verona/ Padua: Drucker. Cubelli, Roberto et al. (2011): “The Effect of Grammatical Gender on Object Categorization”. Journal of Experimental Psychology. Learning, Memory, and Cognition 37: 449-60. Cunha, Antônio Geraldo da (ed.) (1982): Dicionário etimológico Nova Fronteira da língua portuguesa. Rio de Janeiro: Nova Fronteira. D’Achille, Paolo/ Thornton, Anna M. (2003): “La flessione del nome dall’italiano antico all’italiano contemporaneo”. In: Maraschio, Nicoletta/ Salani, Teresa Poggi (eds.): Italia linguistica anno Mille - Italia linguistica anno Duemila. Atti del XXXIV congresso internazionale di studi della SLI. Rome: Bulzoni, 211-230. d’Heur, Jean-Marie (1973): “Notes sur l’histoire du manuscrit de la Chanson de la Croisade albigeoise et sur quelques copies modernes”. Annales du Midi 85: 443-450. d’Heur, Jean-Marie (1974): “Sur la date, la composition et la destination de la Chanson de la Croisade albigeoise de Guillaume de Tudèle”. In: De Caluwé, <?page no="269"?> Bibliography 269 Jacques et al. (eds.): Mélanges d’histoire littéraire, de linguistique et de philologie romanes offerts à Charles Rostaing par ses collègues, ses élèves et ses amis. Tome 1. Liège: Association des romanistes de l’Université de Liège, 231-255. DACC = Martínez Tur, Feliciano/ Rodés Orquín, Francho (eds.) (2025): Diccionario aragonés castellano catalán. Estudio de Filolochía Aragonesa (Edicions Dichitals de l’Academia de l’Aragonés, EDACAR 14). Zaragoza: Academia de l’Aragonés. DALLA = Academia de la Llingua Asturiana (ed.) (2025): Diccionariu de la Llingua Asturiana. <https: / / www.diccionariu.alladixital.org/ > [visited on 04/ 08/ 2026]. DAO = Baldinger, Kurt (1975-2007): Dictionnaire onomasiologique de l’ancien occitan. 10 fascicles. Tübingen/ Berlin: Niemeyer/ De Gruyter. DAOA = Philipe, Olivier (2009): Dictionnaire d’ancien occitan auvergnat, Mauriacois et Sanflorain (1340-1540) (Beihefte zur Zeitschrift für Romanische Philologie 349). Tübingen: Niemeyer. Darcos, Xavier (2021): Drôle de genre. <https: / / www.academie-francaise.fr/ drolede-genre> [visited on 04/ 08/ 2026]. DCVB = Alcover, Antoni Maria/ Moll, Francesc de Borja (2002): Diccionari catalàvalencià-balear. Catalans, Institut d’Estudis (eds.). <https: / / dcvb.iec.cat/ > [visited on 04/ 08/ 2026]. DdD = Santamarina, Antón (ed.) (2025): Dicionario de dicionarios. Santiago de Compostela/ Vigo. <https: / / ilg.usc.gal/ ddd/ > [visited on 04/ 08/ 2026]. dDOM = (2025): digitales Dictionnaire de l’occitan médiéval. <https: / / dienste.badw.de: 9999/ dom/ db/ > [visited on 04/ 08/ 2026]. de Dardel, Robert (1965): Recherches sur le genre roman des substantifs de la troisième déclinaison. Geneva: Droz. de Marneffe, Marie-Catherine et al. (2021): “Universal Dependencies”. Computational Linguistics 47(2): 255-308. de Nostredame, Jehan (1575): Les Vies des plus célèbres et anciens poètes provensaux, qui ont floury du temps des comtes de Prouence. Recueillies des oeuures de diuers Autheurs nommez en la pages suyuante , qui les ont escrites, & redigees premierement en langue Prouensale, & depuis mises en langue Françoyse par Iehan de nostre Dame Procureur en la Cour de Parlement de Prouence. Par lesquelles est monstrée l’ancienneté de plusieurs Nobles maisons tant de Prouence, Languedoc, France, que d’Italie, & d’ailleurs. Lyon: Alexandre Marsilij. de Poerck, Guy (1957): “Un poème religieux quercinois du XIIe s. Eu aor Damrideu”. In: Mélanges de linguistique et de littérature romanes à la mémoire <?page no="270"?> 270 Bibliography d’István Frank, offerts par ses anciens mâitres, ses amis et ses collègues de France et de l’étranger. Saarbrücken: Universität des Saarlandes, 512-545. de Roustan, René Flachaire (1924-1925): “Les manuscrits du poème de Raimon Feraut sur la vie de saint Honorat de Lérins”. Le Moyen Âge. 2e série 26: 255- 284. De Stefano, Antonino (1909): La Noble Leçon des Vaudois du Piémont. Paris: Honoré Champion. DECat = Coromines, Joan (1980-1991): Diccionari etimològic i complementari de la llengua catalana. Barcelona: Curial Edicions Catalanes. <https: / / decat.iec.cat/ > [visited on 04/ 08/ 2026]. Dejeanne, Jean-Marie-Lucien (1909): Poésies complètes du troubadour Marcabru (Bibliothèque méridionale, 1 re série, 12). Toulouse: Éditions Edouard Privat. DÉRom = Buchi, Éva/ Schweickart, Wolfgang (eds.) (2016): Dictionnaire Étymologique Roman. Berlin/ Boston. <http: / / www.atilf.fr/ DERom> [visited on 04/ 08/ 2026]. DES = Wagner, Max Leopold (1960-1983): Dizionario etimologico sardo. Wiesbaden: Kraus. dicodòc = Lo Congrès (ed.) (2025): Multidiccionari occitan: Diccionari occitan de referéncia. <https: / / dicodoc.eu/ oc/ > [visited on 04/ 08/ 2026]. Diebowski, Jessica (2021): “Grammatical Gender in Spanish”. In: Gender Acquisition in Spanish. Berlin/ Boston: De Gruyter Mouton, 15-58. Diehl, Rudolph (1885): Guillem Anelier von Toulouse, der Dichter des zweites Theils der Albigenserchronik. Marburg: N. G. Elwert’sche Verlagsbuchhandlung. Diez, Friedrich ( 5 1882): Grammatik der Romanischen Sprachen. Zweiter Theil. Bonn: Eduard Weber’s Verlag. Dixon, Robert M. W. (2010): Basic Linguistic Theory: Volume 1, Methodology. Oxford: Oxford University Press. DLE = Real Academia Española (ed.) (2025): Diccionario de la lengua española. <https: / / dle.rae.es/ > [visited on 04/ 08/ 2026]. DLR = Iordan, Iorgu et al. (1913-1978): Dicționarul limbii române. Bucharest: Editura Academiei Române. DOMél = (2025): Dictionnaire de l’occitan médiéval. <http: / / www.dom-enligne.de/ > [visited on 04/ 08/ 2026]. Donaldson, Bryan/ Sibille, Jean (2024): “Histoire interne de la langue”. In: Esher, Louise/ Sibille, Jean (eds.): Manuel de linguistique occitane. Berlin/ Boston: De Gruyter, 193-230. <?page no="271"?> Bibliography 271 Douillet, Christian (2017): “Quelques vers retrouvés de la Chanson de la Croisade albigeoise”. Heresis : revue d’hérésiologie médiévale 2: 7-38. DRG = De Planta, Robert et al. (eds.) (2025): Dicziunari Rumantsch Grischun. <https: / / online.drg.ch/ > [visited on 04/ 08/ 2026]. DSF = Vicario, Federico (ed.) (2025): Dizionario Storico Friulano. <https: / / www.dizionariofriulano.it/ > [visited on 04/ 08/ 2026]. Dufter, Andreas (2025): “Grice Meets Flamenca: On the Art of Interpreting Minimal Utterances”. In: Bleortu, Cristina et al. (eds.): Discursos tradicionales y particulares. Festschrift für Johannes Kabatek zum 60. Geburtstag. Frankfurt am Main/ Madrid: Vervuert Verlagsgesellschaft, 209-216. Dulong, Marthe (1928): La vie provençale de sainte Delphine (1284-1360) et le procès de sa canonisation. Paris: École des chartes. Duval, Amaury (1838): “Le roman de Flamenca, par un anonyme”. Histoire littéraire de la France, ouvrage commencé par des religieux bénédictins de la congrégation de Saint-Maur, et continué par des Membres de l’Institut (Académie des Inscriptions et Belles Lettres). Tome XIX. Suite du treizième siècle: 776-789. Ernst, Willy (1930): “Die Lieder des provenzalischen Trobadors Guiraut von Calanso”. Romanische Forschungen 44: 255-406. Espadaler, Anton M. (2015): Flamenca. traducció, pròleg i notes. Barcelona: Universitat de Barcelona. EWD = (1988-1998): Etymologisches Wörterbuch des Dolomitenladinischen. Volume 1: A - B, volume 2: C, volume 3: D - H, volume 4: I - M, volume 5: N - R, volume 6: S, volume 7: T - Z, volume 8: Indizes. Kramer, Johannes et al. (eds.). Hamburg: Buske. Fabre, Césaire-Antoine (1907): “Les Provençalistes du Velay et M. Camille Chabaneau”. In: Mélanges Chabaneau (Romanische Forschungen 23). Erlangen: F. Junge, 257-273. Faraoni, Vincenzo (2013): “Manifestazioni del neutro italo-romanzo nella documentazione notarile altomedievale”. In: Rainer, Franz et al. (eds.): Actes du XXVIIe Congrès international de linguistique et de philologie romanes, 29-43. Faraoni, Vincenzo/ Loporcaro, Michele (2014): “Manifestazioni residuali di sistemi trigeneri in còrso settentrionale (2016)”. In: Retali-Medori, Stella (eds.): Lingue delle isole, isole linguistiche, Corte, 22 September 2014 - 23 September 2014. Corte: Edizioni dell’Orso, 31-61. Faraoni, Vincenzo et al. (2013): “Manifestazioni del neutro nell’italo-romanzo medievale”. In: Casanova Herrero, Emil/ Calvo Rigual, Cesáreo (eds.): Actas del XXVI congreso internationcal de linguística y de filología románicas, Valencia 2010. Berlin: De Gruyter, 171-182. <?page no="272"?> 272 Bibliography Fauriel, Claude-Charles (1837): Histoire de la croisade contre les hérétiques albigeois écrite en vers provençaux par un poëte contemporain (Collection de documents inédits sur l’histoire de France. Première série: Histoire politique). Paris: Imprimerie royale. Fauriel, Claude-Charles (1847): “Guillaume, auteur des Gesta Caroli Magni ad Carcassonam”. In: Académie des Inscriptions et Belles Lettres (eds.): Histoire littéraire de la France, ouvrage commencé par des religieux bénédictins de la congrégation de Saint-Maur, et continué par des Membres de l’Institut (Académie des Inscriptions et Belles Lettres). Tome XXI. Suite du treizième siècle, depuis l’année 1296. Suppléments. Paris: Académie des Inscriptions et Belles Lettres, 373-382. Favati, Guido (1961): Le biografie trovadoriche, testi provenzali dei secc. XIII e XIV. Edizione critica. Bologna: Libreria Antiquaria Palmaverde. Fedden, Sebastian/ Corbett, Greville G. (2017): “Gender and Classifiers in Concurrent Systems: Refining the Typology of Nominal Classification”. Glossa: A Journal of General Linguistics 2 (1): 34. Fedi, Beatrice (2019): Las leys d’amors. Redazione lunga in prosa. Edizione critica (Archivio romanzo 35). Florence: SISMEL-Edizioni del Galluzzo. FEW = von Wartburg, Walther (ed.) (1922-2003): Französisches etymologisches Wörterbuch: eine Darstellung des galloromanischen Sprachschatzes. 25 Bände. Bonn et al.: Klopp et al. <https: / / lecteur-few.atilf.fr/ index.php/ > [visited on 04/ 08/ 2026]. Fibla, Sergi Sancho (2022): “Reading in Community, Writing a Community: Douceline’s Vida and the Beguines of Roubaud”. In: Nieto-Isabel, Delfi I./ Milian, Laura Miquel (eds.): Living on the Edge. Transgression, Exclusion, and Persecution in the Middle Ages (Studies in Medieval and Early Modern Culture 83). Berlin/ Boston: Medieval Institute Publications, 23-58. Field, Thomas (2006): “Troubadour Performance and the Origins of the Occitan ‘Koine’”. Tenso 21: 36-54. Formentin, Vittorio/ Loporcaro, Michele (2012): “On the Fourth Grammatical Gender in Old Romanesco”. Lingua e Stile: Rivista di storia della lingua italiana 47(2): 221-264. Fourquin, Noël/ Rigaud, Philippe ( 2 1994): De la nave au pointu. Glossaire nautique de la langue d’oc (Provence, Languedoc) des origines à nos jours. Saint-Tropez/ Toulon: Objectif Mer et Capian Méditerranée. Frank, Isidore (1956): “Un opuscule de piété provençal du XIII e siècle (version marseillaise du Planctus beatae Mariae attribué à saint Augustin et saint <?page no="273"?> Bibliography 273 Bernard)”. In: Estudios dedicados a Menéndez Pidal 6. Madrid: Consejo Superior de Investigaciones Científicas, 31-64. Fresne Du Cange, Charles du et al. ([1678]2009): Glossarium mediæ et infimæ latinitatis. Digital edition of the 1883-1887 Favre edition, hosted by École nationale des chartes and Sorbonne. <http: / / ducange.enc.sorbonne.fr/ > [visited on 04/ 08/ 2026]. Frosini, Giovanna (1999): “Dall’Oriente all’Occidente. Il romanzo di Barlaam e Iosafas: circolazione e utilizzazione dei testi”. I Quaderni del M.Æ.S. - Journal of Mediæ Ætatis Sodalicium 21: 113-143. Fusaroli, Federica (2021): “La traduzione della Somme le roi tra Occitania e Catalogna: primi sondaggi”. Mot so razo 20: 1-18. Fusaroli, Federica (2024): “La tradizione manoscritta del Libre de vicis et de vertutz”. In: Menichetti, Caterina et al. (eds.): La littérature occitane médiévale dans sa tradition manuscrite: témoins, traditions, corpora. Rome: Viella, 207- 232. Fusaroli, Federica/ Menichetti, Caterina (2023): “Produzione e pubblico di una miscellanea didattico-morale: nuove proposte per il manoscritto BnF, fr. 1049”. Romania 141: 111-152. Garay, Kathleen/ Jeay, Madeleine (2001): The Life of Saint Douceline, a Beguine of Provence. Woodbridge: Boydell/ Brewer. Garbo, Francesca Di/ Miestamo, Matti (2019): “The Evolving Complexity of Gender Agreement Systems”. In: Garbo, Francesca Di et al. (eds.): Grammatical Gender and Linguistic Complexity. Volume II: World-Wide Comparative Studies (Studies in Diversity Linguistics 28). Berlin: Language Science Press, 15-60. Garces Arias, Esteban et al. (2023): “Automatic Transcription of Handwritten Old Occitan Language”. In: Bouamor, Houda et al. (eds.): Proceedings of the 2023 Conference on Empirical Methods in Natural Language Processing. Singapore: Association for Computational Linguistics, 15416-15439. Gardani, Francesco (2013): Dynamics of Morphological Productivity: The Evolution of Noun Classes from Latin to Italian (Brill’s Studies in Historical Linguistics 8). Leiden: Brill. Gardette, Pierre (1957): “Marguerite d’Oingt. Pour une nouvelle édition de ses œuvres”. In: Mélanges de linguistique et de littérature romanes à la mémoire d’István Frank offerts par ses anciens maîtres, ses amis et ses collègues de France et de l’étranger (Annales Universitatis Saraviensis. Philosophie 6). Saarbrücken: Universität des Saarlandes, 225-235. Gatien-Arnout, Adolphe-Félix (1841): Las flors del gay saber estier dichas las leys d’amors = Les fleurs du gai savoir, autrement dites lois d’amour. Traduits <?page no="274"?> 274 Bibliography par MM. d’Aguilar et d’Escouloubre (Monumens de littérature romane, publiés sous les auspices de l’Académie des jeux floraux, avec l’appui du Conseil municipal de la ville de Toulouse, et du Conseil général du département de la Haute-Garonne). Toulouse: Paya. Gennrich, Friedrich (1959): Lo gai saber : 50 ausgewählte Troubadourlieder; Melodie, Text, Kommentar, Formenlehre und Glossar (Musikwissenschaftliche Studienbibliothek 18-19). Darmstadt: self-published. Giannetti, Andrea (1988): Raimon de Castelnou. Canzoni e dottrinale (Biblioteca di filologia romanza 33). Bari: Adriatica. Girard, Pierre (1912): Saint Elzéar de Sabran et la bienheureuse Delphine de Signe. Paris: Librairie Saint-François. Gleßgen, Martin-Dietrich (1989): Lo Thesaur del Hospital de Sant Sperit. Edition eines Marseiller Urkundeninventars (1399-1511) mit sprachlichem und geschichtlichem Kommentar unter besonderer Berücksichtigung des Rechtswortschatzes (Beihefte zur Zeitschrift für romanische Philologie 226). Tübingen: Niemeyer. Gonfroy, Gérard (1981): La rédaction catalane en prose des Leys d’amors. Édition et étude critique des trois premières parties. Poitiers: Université de Poitiers. Gorosch, Max (1969): La notomia de Anric de Mondavilla. Oxford: Oxford University Press. Gosman, Martin (1982): La Lettre du Prêtre Jean. Les versions en ancien français et en ancien occitan. Textes et commentaires (Mediaevalia Groningana 2). Groningen: Bouma’s Boekhuis bv. Gouiran, Gérard (1985a): L’amour et la guerre. L’œuvre de Bertran de Born. Tome 1. Aix-en-Provence/ Marseille: Presses de l’Université de Provence. Gouiran, Gérard (1985b): L’amour et la guerre. L’œuvre de Bertran de Born. Tome 2. Aix-en-Provence/ Marseille: Presses de l’Université de Provence. Gout, Raoul (1927): La vie de sainte Douceline. Texte provençal du XIIe siècle, traduction et notes (Ars et Fides 8). Paris: Bloud et Gay. GPSR = Gauchat, Louis (ed.) (2025): Glossaire des patois de la Suisse romande. <https: / / portail-gpsr.unine.ch/ > [visited on 04/ 08/ 2026]. Grafström, Åke (1968): Étude sur la morphologie des plus anciennes chartes languedociennes. Stockholm: Almqvist Wiksell. Grandgent, Charles Hall (1905): An Outline of the Phonology and Morphology of Old Provençal. Boston: Heath. Grandi, Nicola (2011): “Renewal and Innovation in the Emergence of Indo-European Evaluative Morphology”. Lexis 6: Journal in English Lexicology. <?page no="275"?> Bibliography 275 Special issue: “Diminutives and Augmentatives in the Languages of the World” : 5-25. Grandi, Nicola/ Körtvélyessy, Lívia (2015): “Introduction: Why Evaluative Morphology”. In: Grandi, Nicola/ Körtvélyessy, Lívia (eds.): The Handbook of Evaluative Morphology. Edinburgh: Edinburgh University Press, 3-20. Grange, Huw (2015): “Fraudsters or Patriots? The Authors of the Lives of St Honorat”. Romania 133: 201-218. Gschwind, Urs (1976): Le roman de Flamenca, nouvelle occitanie du treizième siècle. Texte établi et commenté (Romanica Helvetica 86). Bern: Francke. Guibal, Georges (1863): Le poëme de la croisade contre les Albigeois, ou l’Épopée nationale de la France du sud au XIIIe siècle: étude historique et littéraire. Paris: Chauvin. Guida, Saverio (1963): Jocs poetici alla corte di Enrico II di Rodez («Subsidia» al «Corpus des troubadours» 8 = Studi, testi e manuali 10). Modena: Mucchi. Guida, Saverio (2003): “L’autore della seconda parte della Canso de la crotzada”. Cultura Neolatina 63: 255-282. Guillaume, Paul (1887): Istoria Petri et Pauli. Mystère en langue provençale du XV e siècle publié d’après le manuscrit original sous les auspices de la société d’études des Hautes-Alpes. Paris: Maisonneuve. Haase, Martin (2000): “Reorganization of a Gender System: The Central Italian Neuters”. In: Unterbeck, Barbara et al. (eds.): Gender in Grammar and Cognition (Trends in Linguistics: Studies and Monographs 124). Berlin/ New York: Mouton de Gruyter, 221-236. Hackett, W. Mary (1953a): Girart de Roussillon. Chanson de geste. Volume 1 (Société des Anciens Textes Français 93). Paris: A. & J. Picard. Hackett, W. Mary (1953b): Girart de Roussillon. Chanson de geste. Volume 2 (Société des Anciens Textes Français 94). Paris: A. & J. Picard. Hackett, W. Mary (1955): Girart de Roussillon. Chanson de geste. Volume 3 (Société des Anciens Textes Français 95). Paris: A. & J. Picard. Hagman, Roy (2010): “Grammar of An Uncodified Language: The Old Occitan Grammar of Guilhem Molinier”. Tenso 25(1): 23-35. Hamlin, Frank P. et al. (eds.) ( 2 1985): Introduction à l’étude de l’ancien provençal. Textes d’étude (Publications romanes et françaises 96). Geneva: Droz. Harris, James W. (1991): “The Exponence of Gender in Spanish”. Linguistic Inquiry 22(1): 27-62. Haspelmath, Martin (2014): “Comparative Syntax”. In: Carnie, Andrew et al. (eds.): The Routledge Handbook of Syntax. London: Routledge, 490-508. <?page no="276"?> 276 Bibliography Haupt, Hans-Christian (2003): Le Roman d’Arles dans la copie de Bertran Boysset (Romanica Helvetica 121). Tübingen/ Basel: Francke. Heinemann, Sabine/ Neumann-Holzschuh, Ingrid (2009): “Historische Aspekte geschlechtsspezifischer Sprache in der Romania”. In: Ernst, Gerhard et al. (eds.): Romanische Sprachgeschichte / Histoire linguistique de la Romania (Handbücher zur Sprach- und Kommunikationswissenschaft 3). Berlin/ New York: De Gruyter Mouton, 2378-2390. Heisig, Karl (1926): Studien zur Chanson de la Croisade contre les Albigeois. Wrocław: Volkswacht-Buchdruckerei. Hershon, Cyril P./ Ricketts, Peter T. (2007): “La tradition occitane de l’Évangile de Gamaliel, éditions et commentaires”. La France latine. Revue d’études d’oc, nouvelle série 144: 132-327. Hershon, Cyril P./ Ricketts, Peter T. (2009): “Las Merevilhas de la terra de Ybernia”. La France Latine. Revue d’études d’oc, nouvelle série 148: 233-297. Heuckenkamp, Ferdinand (1912): Die provenzalische Prosa-Redaktion des geistlichen Romans von Barlaam und Josaphat, nebst einem Anhang über einige deutsche Drucke des XVII. Jahrhunderts. Halle (Saale): Niemeyer. Higounet, Charles (1938): “À propos de Guillaume de Tudèle”. Annales du Midi 50: 377-379. Hockett, Charles Francis ( 4 1962): A Course in Modern Linguistics. New York: Macmillan. Hoekstra, Dirk (1998): Huit ans de Guerre albigeoise: édition avec notes et commentaires de la version en ancien occitan offerte par le manuscrit de Merville. Groningen: Rijksuniversiteit Groningen. Horne, Thomas Hartwell (1808): A Catalogue of the Harleian Manuscripts in the British Museum: With Indexes of Persons, Places, and Matters. Vol. 2 (Great Britain Record Commission Publications 2). British Museum. Department of Manuscripts (eds.). London: G. Eyre & A. Strahan. Hosch, Siegfried (1877): Untersuchungen über die Quellen und das Verhältniss der provençalischen und der lateinischen Lebensbeschreibung des Hl. Honoratus. Berlin: Donny. Hualde, José Ignacio (2020): “Ibero-Romance II: Astur-Leonese, Spanish, Navarro-Aragonese”. In: Loporcaro, Michele (eds.): Oxford Research Encyclopedia of Linguistics. Oxford: Oxford University Press. Hubert, Merton Jerome (1962): The Romance of Flamenca, a Provençal Poem of the Thirteenth Century. English verse translation by Merton J. Hubert; revised Provençal text by Marion E. Porter. Princeton: Princeton University Press for the University of Cincinnati. <?page no="277"?> Bibliography 277 Huchet, Jean-Charles (1989): Flamenca, roman occitan du XIIIe siècle. Texte établi, traduit et présenté. Paris: Union générale d’éditions. Hündgen, Franz (ed.) (1884): Das altprovenzalische Boëthiuslied unter Beifügung einer Uebersetzung, eines Glossars, erklärender Anmerkungen sowie grammatischer und metrischer Untersuchungen. Oppeln: Eugen Franck’s Buchhandlung. Izquierdo, Josep (1997): “The Gospel of Nicodemus in Medieval Catalan and Occitan Literatures”. In: Izydorczyk, Zbigniew (eds.): The Medieval Gospel of Nicodemus: Texts, Intertexts, and Contexts in Western Europe (Medieval & Rennaissance. Text & Studies 158). Tempe: Medieval Institute Publications. Izydorczyk, Zbigniew (2021): “The Gospel of Nicodemus in Medieval European Vernaculars: Textual Sources and Translatorial Strategies”. In: Rojszczak- Robińska, Dorota et al. (eds.): Zródła staropolskich apokryfów. Pytania, problemy, perspektywy (Biblioteczka Poznanskich Studiów Polonistycznych Seria Jezyznawczna / 58). Poznań: Wydawnictwo PSP, 11-76. Jeanmaire, Guillaume (2014): “Le genre grammatical en français : approche basée sur l’usage”. Romanistisches Jahrbuch 64: 61-105. Jeanroy, Alfred (1949a): “Complainte sur la mort de Robert d’Anjou, comte de Provence et roi de Naples”. In: Histoire littéraire de la France, ouvrage commencé par des religieux bénédictins de la congrégation de Saint-Maur, et continué par des Membres de l’Institut (Académie des Inscriptions et Belles Lettres). Tome XXXVIII. Suite du quatorzième siècle. Paris: Imprimerie nationale, 645- 652. Jeanroy, Alfred (1949b): “Les Leys d’amors”. In: Histoire littéraire de la France, ouvrage commencé par des religieux bénédictins de la congrégation de Saint- Maur, et continué par des Membres de l’Institut (Académie des Inscriptions et Belles Lettres). Tome XXXVIII. Suite du quatorzième siècle. Paris: Imprimerie nationale, 139-233. Jensen, Frede (1972): From Vulgar Latin to Old Provençal. Chapel Hill: University of North Carolina Press. Jensen, Frede (1973): “Désaccord entre genre et flexion: Les substantifs masculins à désinence féminine en provençal”. Revue des langues romanes 80: 393- 404. Jensen, Frede (1974a): “Provençal cor and cors: A Flexional Dilemma”. Romance Philology 28: 27-31. Jensen, Frede (1974b): “Retour sur la flexion du provençal coms-comte”. Zeitschrift für romanische Philologie 905-6: 497-598. Jensen, Frede (1975a): “Les imparisyllabiques masculins en ancien provençal”. Romania 96: 459-480. <?page no="278"?> 278 Bibliography Jensen, Frede (1975b): “Les substantifs féminins collectifs en ancien provençal”. Romania 96382: 268-275. Jensen, Frede (1976): The Old Provençal Noun and Adjective Declension. Odense: Odense University Press. Jensen, Frede (1977): “Les imparisyllabiques féminins en ancien provençal”. Romania 98: 370-379. Jensen, Frede (1986): The Syntax of Medieval Occitan. Berlin/ Boston: Niemeyer. Jensen, Frede (1994): Syntaxe de l’ancien occitan (Beihefte zur Zeitschrift für romanische Philologie 257). Tübingen: Niemeyer. Josh. (2021): “Expresso & Tschianti”. In: Teilzeitromantik. Hamburg: Warner Music. Jud, Jakob (1939): “Observations sur le lexique du roman de Flamenca”. In: Mélanges A. Duraffour. Hommage offert par ses amis et ses élèves, 4 juin 1939. Paris/ Zurich/ Leipzig: Droz/ Niehans, 204-210. Kilarski, Marcin (2013): Nominal Classification: A History of Its Study from the Classical Period to the Present. Amsterdam: Benjamins. Kirsch, Fritz P. (1989): Flamenca, ein altokzitanischer Liebesroman. übersetzt, mit Einführung, Erläuterungen und Anmerkungen. Kettwig: Phaidon. Klein, Otto (ed.) (1885): Die Dichtungen des Mönchs von Montaudon neu herausgegeben (Ausgaben und Abhandlungen aus dem Gebiete der romanischen Philologie VII). Marburg: N. G. Elwert’sche Verlagsbuchhandlung. Koch, Peter/ Oesterreicher, Wulf (2011): Gesprochene Sprache in der Romania: Französisch, Italienisch, Spanisch. Berlin/ New York: De Gruyter. Koch, Wilhelm (1897): Beiträge zur Textkritik der Auzels cassadors von Daude de Pradas. Münster: Johannes Bredt. Kolsen, Adolf (1910): Sämtliche Lieder des Trobadors Giraut de Bornelh mit Übersetzung, Kommentar und Glossar. Erster Band Texte mit Varianten und Übersetzung. Halle (Saale): Niemeyer. Konishi, Toshi (1993): “The Semantics of Grammatical Gender: A Cross-Cultural Study”. Journal of Psycholinguistic Research 22(5): 519-534. Kraack, Ludwig (1884): Über die Entstehung und die Dichter der Chanson de la croisade contre les Albigeois (Ausgaben und Abhandlungen aus dem Gebiete der romanischen Philologie 15). Marburg: N. G. Elwert’sche Verlagsbuchhandlung. Krefeld, Thomas (2003): “Methodische Grundlagen der Strataforschung”. In: Ernst, Gerhard (eds.): Romanische Sprachgeschichte (Handbücher zur Sprach- <?page no="279"?> Bibliography 279 und Kommunikationswissenschaft 22.1). Berlin/ New York: De Gruyter, 555- 568. Lachin, Giosuè (2004): Il trovatore Elias Cairel. Modena: Muchi. Lafaille, Germain (1687): Annales de la ville de Toulouse depuis la réünion de la Comté de Toulouse à la Couronne : avec un abrégé de l’ancienne histoire de cette ville et un Recueil de civers titres et actes pour servir de Preuves ou d’éclaircissement à ces Annales. Première Partie. Toulouse: G.-L. Colomyez. Lafont, Robert (1962): “Composition et rythme épiques dans la seconde partie de la « Chanson de la croisade albigeoise »”. Revue de langue et littérature d’Oc 9: 42-56. Lafont, Robert (1991): La Geste de Roland, Tome 2: Espaces, Textes, Pouvoirs. Paris: L’Harmattan. Lambelet, Amelia (2012): L’apprentissage du genre grammatical en langue étrangère : à la croisée des approches linguistiques et cognitives. Fribourg: Université de Fribourg. Lampitelli, Nicola (2014): “The Romance Plural Isogloss and Linguistic Change: A Comparative Study of Romance Nouns”. Lingua 140: 158-179. Laurent, Richard (1999): Past Participles from Latin to Romance. Berkeley: University of California Press. Lavaud, René/ Nelli, Rémy (1960): Les Troubadours: Jaufre, Flamenca, Barlaam et Josaphat. Bruges/ Paris: Desclée de Brouwer. Ledgeway, Adam (2009): Grammatica diacronica del napoletano (Beihefte zur Zeitschrift für romanische Philologie 350). Tübingen: Niemeyer. Léglu, Catherine (2008): “Languages in Conflict in Toulouse: Las leys d’amors”. The Modern Language Review 103(2): 383-396. LEI = Pfister, Max et al. (eds.) (2025): Lessico Etimologico Italiano. <https: / / online.lei-digitale.it/ > [visited on 04/ 08/ 2026]. Lespy, Vastin/ Raymond, Paul (1876): Récits d’Histoire Sainte en béarnais. Traduits et publiés pour la première fois sur le manuscrit du XV e siècle. Tome Premier. Pau: Léon Ribaut. Lespy, Vastin/ Raymond, Paul (1887): Dictionnaire béarnais ancien et moderne. 2 volumes. Montpellier: Imprimerie centrale du Midi. Levy, Emil (1894-1924): Provenzalisches Supplement-Wörterbuch. Berichtigungen und Ergänzungen zu Raynouards Lexique roman, 8 Bände. Leipzig: O. R. Reisland. Lienig, Paul (1890): Die Grammatik der provenzalischen Leys d’amors, verglichen mit der Sprache der Troubadours. Erster Teil: Phonetik. Wrocław: Koebner. <?page no="280"?> 280 Bibliography Limentani, Alberto (1965): Las novas de Guillem de Nivers (Flamenca). Introduzione, scelta e glossario. Padua: Franco Muzzio Editore. Lodge, R. Anthony (1995): “Okzitanische Skriptaformen II. Auvergne”. In: Holtus, Günter et al. (eds.): Lexikon der Romanistischen Linguistik, Bd. II/ 2. Die einzelnen romanischen Sprachen und Sprachgebiete vom Mittelalter zur Renaissance. Berlin/ New York: Niemeyer, 420-424. Lommatzsch, Erhard (1917): Provenzalisches Liederbuch: Lieder der Troubadours mit einer Auswahl biographischer Zeugnisse, Nachdichtungen und Singweisen. Berlin: Weidmannsche Buchhandlung. Loos, Theodor Albert (1884): Die Nominalflexion im Provenzalischen (Ausgaben und Abhandlungen aus dem Gebiete der Romanischen Philologie, Heft XVI). Marburg: N. G. Elwert’sche Verlagsbuchhandlung. Loporcaro, Michele (2015): “The Impact of Morphology on Change in Agreement Systems”. Trends in Linguistics. Studies and Monographs 287: 103-126. Loporcaro, Michele (2017): “Composti V+N e genere grammaticale in rumeno”. In: D’Alessandro, Roberta et al. (eds.): Di tutti i colori: Studi linguistici per Maria Grossmann. Utrecht: Utrecht University Repository, 197-202. Loporcaro, Michele (2018): Gender from Latin to Romance. History, Geography, Typology. Oxford: Oxford University Press. Loporcaro, Michele (2022): “Aspetti grammaticali della transizione latinoromanza: la morfosintassi nominale”. Medioevo romanzo 1: 5-41. Loporcaro, Michele/ Paciaroni, Tania (2011): “Four-Gender Systems in Indo- European”. Folia Linguistica 45(2): 389-433. Loporcaro, Michele/ Pedrazzoli, Dafne (2016): “Classi flessive del nome e genere grammaticale nel dialetto di Agnone (Isernia)”. Revue de Linguistique Romane 80: 73-100. Loporcaro, Michele/ Silvestri, Giuseppina (2016): “Residui del genere neutro come categoria morfosintatticamente funzionale nell’italo-romanzo meridionale”. Actes du XXVIIe Congrès international de linguistique et de philologie romanes. Ledgeway, Adam et al. (eds.): 277-290. Loporcaro, Michele et al. (2014): “The Third Gender of Old Italian”. Diachronica 31(1): 1-22. Loporcaro, Michele et al. (2021): “Contact-Induced Complexification in the Gender System of Istro-Romanian”. Journal of Language Contact 14(1): 72-126. Loporcaro, Michele et al. (2021): “Il genere in movimento: mutamento in corso nel dialetto di Molfetta (Bari)”. Vox Romanica 80(1): 55-104. <?page no="281"?> Bibliography 281 LRW = Körting, Gustav ( 3 1907): Lateinisch-Romanisches Wörterbuch. Etymologisches Wörterbuch der romanischen Hauptsprachen. Paderborn: Ferdinand Schöningh. Lust, Barbara C. (2006): Child Language: Acquisition and Growth (Cambridge Textbooks in Linguistics). Cambridge: Cambridge University Press. Machado, José Pedro (ed.) ( 3 1977): Dicionário etimológico da língua portuguesa. 1: A - B; 2: C - E; 3: F - L; 4: M - P; 5: Q - Z. Lisbon: Livros Horizonte. Maiden, Martin (2010): “Morphological Persistence”. In: Maiden, Martin et al. (eds.): The Cambridge History of the Romance Languages. Cambridge: Cambridge University Press, 155-214. Maiden, Martin (2016): “The Romanian Alternating Gender in Diachrony and Synchrony”. Folia Linguistica 50 (Historica 37): 111-144. Maiden, Martin et al. (2021): The Oxford History of Romanian Morphology. Oxford: Oxford University Press. Malkiel, Yakov (1951): “Lexical Polarization in Romance”. Language 27(4): 485- 518. Manetti, Roberta (2008): Flamenca. Romanzo occitano del XIII secolo (Studi, testi e manuali, Nuova serie, 11). Modena: Mucchi Editore. Manetti, Roberta (2012): La passione di santa Margherita. Testo occitanico del XIII secolo (Firenze, Biblioteca Medicea Laurenziana, Ashburnham 105). Edizione con versione a fronte, introduzione, note e glossario (Carrefours. Medioevo 3). Florence: Alinea. Marchello-Nizia, Christiane et al. (eds.) (2020): Grande grammaire historique du français. Berlin/ Boston: Mouton de Gruyter. Marshall, John H. (1969a): “Observations on the Sources of the Treatment of Rhetoric in the Leys d’amors”. The Modern Language Review 64(1): 39-52. Marshall, John H. (1969b): The Donatz proensals of Uc Faidit. Oxford: Oxford University Press. Martin-Chabot, Eugène (1931): La chanson de la croisade albigeoise. Tome Ier: La chanson de Guillaume de Tudèle (Classiques de l’histoire de France au Moyen Âge 13). Paris: Champion. Martin-Chabot, Eugène (1957): La chanson de la croisade albigeoise. Tome II: Le poème de l’auteur anonyme (Classiques de l’histoire de France au Moyen Âge 24). Paris: Les Belles Lettres. Martin-Chabot, Eugène (1961): La chanson de la croisade albigeoise. Tome III: Le poème de l’auteur anonyme (Classiques de l’histoire de France au Moyen Âge 25). Paris: Les Belles Lettres. <?page no="282"?> 282 Bibliography Martines, Vicent/ Ensenyat, Gabriel (2003): La cançó de la croada contra els albigesos. Barcelona: Proa. Martorano, Antonella (2017): Pons de Chaptoil, Poesie. Corpus des troubadours-6 (Corpus des troubadours - Éditions, vol. 3). Florence: Edizioni del Galluzzo. Marzo, Daniela (2013a): “Italian Verb to Noun Conversion : the Case of Nouns in -a Deriving from Verbs of the 2nd and 3rd Conjugation”. Linguística 8: 69-87. Marzo, Daniela (2013b): Polysemie als Verfahren lexikalischer Motivation: Theorie und Empirie am Beispiel von Metonymie und Metapher im Französischen und Italienischen (Tübinger Beiträge zur Linguistik 537). Tübingen: Narr. Marzo, Daniela/ Wiedner, Marinus (2025): “Remarks on Grammatical Gender in Romance”. In: Linzmeier, Laura et al. (eds.): Parla, e sie breve e arguto. Festschrift für Maria Selig / Studies in Honor of Maria Selig (ScriptOralia 147). Tübingen: Narr, 201-207. Matasović, Ranko (2004): Gender in Indo-European. Heidelberg: Winter. Matasović, Ranko (2017): “Agreement in Role and Reference Grammar: A Typology of Possible Targets”. Suvremena lingvistika 43(84): 157-171. Matthews, Peter H. ( 2 1991): Morphology (Cambridge Textbooks in Linguistics). Cambridge: Cambridge University Press. McConnell-Ginet, Sally (2014): “Gender and its Relation to Sex: The Myth of ‘Natural’ Gender”. In: Corbett, Greville G. (eds.): The Gxpression of Gender (The Expression of Cognitive Categories 6). Berlin/ Boston: De Gruyter Mouton, 3- 38. Meisenburg, Trudel (1996): Romanische  Schriftsysteme  im  Vergleich:   eine  diachrone  Studie (ScriptOralia 82). Tübingen: Narr. Méjean, Suzanne (1971): La chanson satirique provençale au Moyen-Âge. Paris: Nizet. Mel’čuk, Igor A. (2013): “The Notion of Inflection and the Expression of Nominal Gender in Spanish”. Studies in Language 37(4): 736-763. Melani, Silvio (2016): Per sen de trobar : l’opera lirica di Daude de Pradas (Publications de l’Association Internationale d’Études Occitanes 11). Turnhout: Brepols. Meliga, Walter (2009): “Philologie et linguistique de l’occitan médiéval”. In: La voix occitane. Actes du VIIIe Congrès de l’Association Internationale d’Etudes Occitanes. Bordeaux, 12-17 octobre 2005. Pessac: Presses Universitaires de Bordeaux, 23-51. <?page no="283"?> Bibliography 283 Menichetti, Caterina (2025): “Le Répertoire critique des manuscrits littéraires en ancien occitan : modèles de description et état de la documentation”. Revue de Linguistique Romane 89: 67-122. Messalles, Mar Massanell i (2020): “5.1 Word Classes, Inflectional Categories and Paradigms”. In: Argenter, Joan A./ Lüdtke, Jens (eds.): Manual of Catalan Linguistics. Berlin/ Boston: De Gruyter, 129-164. Meyer, Paul (1865a): Le roman de Flamenca. publié d’après le manuscrit unique de Carcassonne, traduit et accompagné d’un glossaire. Paris/ Béziers: Franck/ Delpech. Meyer, Paul (1865b): “Recherches sur les auteurs de la Chanson de la croisade albigeoise”. Bibliothèque de l’École des chartes Vingt-sixième année. Tome premier. Sixième série: 401-422. Meyer, Paul (1867): “Recherches sur l’épopée française”. Bibliothèque de l’École des chartes. Revue d’érudition consacrée spécialement à l’étude du moyen âge. Vingt-huitième année. 6th ser. 3: 28-63. Meyer, Paul (1874): Recueil d’anciens textes bas-latins, provençaux et français accompagnés de deux glossaires. Paris: Librairie Franck. Meyer, Paul (1875): La chanson de la croisade contre les Albigeois, commencée par Guillaume de Tudèle et continuée par un poète anonyme, éditée et traduite pour la Société de l’histoire de France. Tome premier: Texte, vocabulaire et table des rimes. Paris: Renouard. Meyer, Paul (1876): “Compte rendu de: La Vida de sant Honorat, légende en vers provençaux par Raymond Féraut, troubadour niçois du XIIIe siècle, publiée pour la première fois en son entier par A.-L. Sardou (Nice, 1875)”. Romania 18: 237-251. Meyer, Paul (1879): “La vie latine de saint Honorat et Raimon Féraut”. Romania 8: 481-508. Meyer, Paul (1885): “Notice de quelques manuscrits de la collection Libri à Florence”. Romania 14: 485-548. Meyer, Paul (1895): Guillaume de la Barre. Roman d’aventures par Arnaut Vidal de Castelnaudari. Publié pour la première fois d’après le manuscrit unique appartenant à M le duc d’Aumale. Paris: Librairie de Firmin Didot et C ie . Meyer, Paul (1898a): “Documents linguistiques des Basses-Alpes”. Romania XXVII: 336-441. Meyer, Paul (1898b): “Vies latine et provençale de Sainte: Légendes pieuses en provençal: Versions provençales d’èvangiles apocryphes. Évangile de Nicodème”. In: Histoire littéraire de la France, ouvrage commencé par des religieux bénédictins de la congrégation de Saint-Maur, et continué par des Mem- <?page no="284"?> 284 Bibliography bres de l’Institut (Académie des Inscriptions et Belles Lettres). Tome XXXII. Suite du quatorzième siècle. Paris: Imprimerie nationale, 102-108. Meyer, Paul (1898c): “Vies latine et provençale de Sainte: Légendes pieuses en provençal: Vie de sainte Enimie par Bertran de Marseille”. In: Histoire littéraire de la France, ouvrage commencé par des religieux bénédictins de la congrégation de Saint-Maur, et continué par des Membres de l’Institut (Académie des Inscriptions et Belles Lettres). Tome XXXII. Suite du quatorzième siècle. Paris: Imprimerie nationale, 80-89. Meyer, Paul ( 2 1901): Le Roman de Flamenca: publié d’après le manuscrit unique de Carcassonne. Traduit et accompagné d’un vocabulaire. Paris: Librairie Emile Bouillon. Meyer-Lübke, Wilhelm (1883): Die Schicksale des lateinischen Neutrums im Romanischen. Halle (Saale): E. Karras/ Niemeyer. Meyer-Lübke, Wilhelm (1894): Grammatik der romanischen Sprachen. Band 2: Formenlehre. Leipzig: O. R. Reisland. Mihatsch, Wiltrud (2021): “Collectives in the Romance Languages”. In: Loporcaro, Michele (eds.): Oxford Research Encyclopedia of Linguistics. Oxford: Oxford University Press. Minervini, Vincenzo (2010): Il Libre de Sidrac di Bertran Boysset (Biblioteca della ricerca. Philologica 6). Fasano: Schena. MLW = Bayerische Akademie der Wissenschaften (ed.) (2025): Mittellateinisches Wörterbuch, digitalisierte Fassung im Wörterbuchnetz des Trier Center for Digital Humanities, Version 01/ 23. <https: / / www.woerterbuchnetz.de/ MLW> [visited on 04/ 08/ 2026]. Monaci, Ernesto (1913): Las Flors del gay saber, estier dichas las Leys d’amors (Testi romanzi per uso delle scuole 31). Rome: Loescher. Moncaut, Cénac (1869): Lettre à M. Paul Meyer, professeur à l’École des chartes, sur l’auteur de la Chanson de la croisade albigeoise en particulier et sur certains procédés de critique en général. Paris: Aubry. Mota, Maria Antónia (2016): “Morfologia nas interfaces”. In: Martins, Ana Maria/ Carrilho, Ernestina (eds.): Manual de linguística portuguesa. Berlin/ Boston: De Gruyter, 156-177. Müller, Daniela/ Wiedner, Marinus (2022): “Le schwa dans l’accent français méridional”. SHS Web of Conferences. Actes du 8e Congrès Mondial de Linguistique Française, 4-8 juillet 2022, Université d’Orléans. 138: 08006. Munke, Bernhard (1911): Die Vita sancti Honorati nach drei Handschriften herausgegeben von Bernhard Munke. Nebst Untersuchungen über das Verhältnis zu Raimon Feraut von Wilhelm Schäfer und über die Ortsnamen beider Texte von <?page no="285"?> Bibliography 285 Adolf Krettek. Mit zwei Faksimile und zwei Landkarten (Beihefte zur Zeitschrift für romanische Philologie 32). Halle (Saale): Niemeyer. Nicholson, Derek E. T. (1976): The Poems of the Troubadour Peire Rogier. Manchester: Manchester University Press. Nicollet, François-Napoléon (1910): “Le provençal d’Arles au XIIIe siècle. Mots tirés d’une enquête faite en 1268-1269”. Annales de Provence 7 : 47-64. Noulet, Jean-Baptiste (1880): “Observations sur le Leudaire de Saverdun, publié par M.-Pasquier”. Revue de Linguistique Romane 17: 117-118. Núñez-Méndez, Eva (2026): “Analyzing Historical Gender Alternation in Ambiguous and Inanimate Spanish Nouns”. Letras Hispanas: Revista de Literatura y de Cultura 22 (1). Ogneva, Anastasiia (2020): “Gender Agreement Hierarchy in Common Gender and Epicene Nouns in Spanish”. Borealis - An International Journal of Hispanic Linguistics 9(1): 279-292. Okan, Yasmina et al. (2010): “Relación entre pensamiento y lenguaje: cómo el género gramatical afecta a las representaciones semánticas de los objetos”. Boletín de AELFA 10(3): 52-55. Ollama (2024): Phi-4: A Large Language Model. <https: / / ollama.com/ library/ phi4> [visited on 04/ 08/ 2026]. Ordóñez, Inés Fernández (2006): “Del Cantábrico a Toledo: el ”neutro de materia” hispánico en un contexto románico y tipológico”. Revista de Historia de la Lengua Española 1: 67-118. Ordóñez, Inés Fernández (2007a): “Del Cantábrico a Toledo: el ”neutro de materia” hispánico en un contexto románico y tipológico (continuación)”. Revista de Historia de la Lengua Española 2: 29-81. Ordóñez, Inés Fernández (2007b): “El neutro de materia en Asturias y Cantabria. Análisis gramatical y nuevos datos.” In: Delgado Cobos, Inmaculada/ Puigvert Ocal, Alicia (eds.): Ex admiratione et amicitia. Homenaje a Ramón Santiago. Madrid: Ediciones del Orto, 395-434. Ormières, Paul (1972): “Identification de l’Anonyme de la Chanson de la Croisade”. Bulletin de la Commission Archéologique de Narbonne 34: 153-160. Ourliac, Paul/ Gilles, Monique (eds.) (1990): Les Fors Anciens de Béarn. Paris: Éditions du Centre National de la Recherche Scientifique. Paciaroni, Tania (2017): Grammatica dei dialetti del Maceratese. Fonetica e morfologia. Zurich: University of Zurich. <?page no="286"?> 286 Bibliography Paciaroni, Tania/ Loporcaro, Michele (2018a): “Il genere in ripano: manifesto ma sub condicione”. In: Antonelli, Roberto et al. (eds.). Strasbourg: Société de linguistique romane, 580-593. Paciaroni, Tania/ Loporcaro, Michele (2018b): “Overt Gender Marking Depending on Syntactic Context in Ripano”. In: Audring, Jenny et al. (eds.): Non- Canonical Gender Systems. Oxford: Oxford University Press, 147-175. Paciaroni, Tania et al. (2013): “Persistenza del neutro nell’italo-romanzo centromeridionale”. Vox Romanica 72(1): 88-137. Paden, William D. (1998): An Introduction to Old Occitan (Introductions to Older Languages). New York: Modern Language Association of America. Paden, William D. (2003): “Declension in Twelfth-Century Occitan: On Editing Early Troubadours, With Particular Reference to Marcabru”. Tenso 18: 67-115. Panhuis, Dirk (2015): Lateinische Grammatik. Berlin/ Boston: De Gruyter. Pansier, Pierre (1927): Histoire de la langue provençale à Avignon du XIIe au XIXe siècle. Avignon: Aubanel frères. Pansier, Pierre (1932): “Supplément au lexique provençal-français de la langue comtadine du XII e au XIX e siècle”. In: Histoire de la langue provençale à Avignon du XII e au XIX e siècle. Avignon: Aubanel Père, 153-196. Paris, Gaston (1865): Histoire poétique de Charlemagne. Paris: Librairie A. Franck. Pasero, Nicoló (1965): “Sulle fonti del libro primo delle Leys d’amors”. Studj romanzi 34: 125-128. Pavlidou, Theodossia-Soula/ Alvanoudi, Angeliki (2014): “Grammatical Gender and Cognition”. In: Major Trends in Theoretical and Applied Linguistics 2. London: De Gruyter Open Poland, 109-124. Pfeffer, Wendy (2024): Medieval Diet and Medicine: Occitan Health Advice for the Layperson (Medical Traditions 10). Berlin/ New York: De Gruyter. Pfister, Max (1970): Lexikalische Untersuchungen zu Girart de Roussillon (Beihefte zur Zeitschrift für romanische Philologie 122). Tübingen: Niemeyer. Piccat, Marco (2001): La versione occitana dello Pseudo Turpino : Ms. Londra B. M. Additional 17 920 (Beihefte zur Zeitschrift für romanische Philologie 308). Tübingen: Niemeyer. Picchio Simonelli, Maria (1974): Lirica moralistica nell’occitania del XII secolo: Bernart de Venzac. Modena: Mucchi. Pinkster, Harm (2015): The Oxford Latin Syntax Volume I. Oxford: Oxford University Press. <?page no="287"?> Bibliography 287 Platz, Ernest (1918): “Recherches sur la formation du genre et la superposition verbale d’après l’« Atlas linguistique de la France »”. Archivum Romanicum, nuova rivista di filologia romanza II: 133-184. Polinsky, Maria/ van Everbroeck, Ezra (2003): “Development of Gender Classifications: Modeling the Historical Change from Latin to French”. Language 79(2): 356-390. Portet, Pierre (2004): Bertrand Boysset, la vie et les œuvres techniques d’un arpenteur médiéval (v. 1355-v. 1416). Paris: Le Manuscrit. Poujade, Clamenca (in preparation): La linguistique outillée à l’épreuve de la variation : Ressources pour l’analyse des parlers occitans de l’Ariège. PhD Thesis in preparation at the Université de Toulouse. Pulega, Andrea (1983): I sermoni in verso e l’«Arlabecca». Bergamo: Istituto universitario di Bergamo. Radaelli, Anna (2016): Il Libre de Barlam et de Josaphat e la sua tradizione nella Provenza angioina del XIV secolo. Rome: Viella. Radaelli, Anna (2018): “Tra finzione e realtà: la conplancha per Roberto d’Angiò, una voce per un re immaginato”. Lecturae Tropatorum 11: 1-69. Raguin, Marjolaine (2014): “Langues et identités dans la partie anonyme de la Chanson de la Croisade albigeoise”. In: Le Pottier, Jean et al. (eds.): Le temps de la bataille de Muret (12 septembre 1213), Actes du 61 e Congrès de la Fédération historique Midi-Pyrénées (Muret, 13-15 septembre 2013). Toulouse: Fédération historique de Midi-Pyrénées, 413-431. Raguin, Marjolaine (2015): Lorsque la poésie fait le souverain: Étude sur la Chanson de la Croisade albigeoise (Nouvelle bibliothèque du Moyen Âge 115). Paris: Honoré Champion. Raguin, Marjolaine (2016): “Problèmes de transmission textuelle et d’interprétation dans l’épique: le cas du prologue de la Chanson de la Croisade albigeoise. Édition critique synoptique”. Medioevo romanzo 40: 371-396. Raguin, Marjolaine (2017): “Les traditions manuscrites de la Chanson de la Croisade albigeoise et de ses remaniements en prose : quelle lecture pour quelle réécriture ? ” In: De Riquer, Isabel et al. (eds.): Actes du XXVII e Congrès international de linguistique et de philologie romanes (Nancy, 15-20-juillet-2013). Section-14-: Littératures médiévales. ATILF, Nancy, 157-168. Rainer, Franz (2016): “Italian”. In: Müller, Peter O. et al. (eds.): Word-Formation. An International Handbook of the Languages of Europe. Volume 4 (Handbücher zur Sprach- und Kommunikationswissenschaft / Handbooks of Linguistics <?page no="288"?> 288 Bibliography and Communication Science 40/ 4). Berlin/ Boston: De Gruyter Mouton, 2712- 2731. Rawls, Diane Nelson (1983): An Etymological Glossary for ”La Chanson de la Croisade Albigeoise”. Ann Arbor: The university of New Mexico. Raynouard, François-Juste-Marie (1838): “Notice de Flamenca, poëme provençal, manuscrit de la Bibliothèque de Carcassonne, n° 681”. Notices et extraits des manuscrits de la Bibliothèque du Roi et autres bibliothèques 13(2): 80-132. Raynouard, François-Juste-Marie (1836-1845): Lexique roman ou Dictionnaire de la langue des troubadours comparée avec les autres langues de l’Europe latine, 6 volumes. Paris: Silvestre. Reichle, Elisabeth/ Schöffel, Matthias (2025): “Lokalität, Regionalität und Überregionalität im Wortschatz der spätmittelalterlichen Verwaltungsschriftlichkeit von Saint-Flour”. In: Linzmeier, Laura et al. (eds.): Parla, e sie breve e arguto. Festschrift für Maria Selig / Studies in Honor of Maria Selig (ScriptOralia 147). Tübingen: Narr, 77-85. Reimann, Paul (1882): Die Declination der Substantiva und Adjectiva in der Langue d’Oc bis zum Jahre 1300. Gdańsk: Groening. Renan, Ernest (1885): “Philippine de Porcellet. Auteur présumé de la Vie de Sainte Douceline”. In: Académie des Inscriptions et Belles Lettres (eds.): Histoire littéraire de la France, ouvrage commencé par des religieux bénédictins de la congrégation de Saint-Maur, et continué par des Membres de l’Institut (Académie des Inscriptions et Belles Lettres). Tome XXIX. Suite du quatorzième siècle. Paris: Imprimerie nationale, 526-546. Révillout, Charles (1875): “De la date possible du roman de Flamenca”. Revue des langues romanes 8: 5-18. REW = Meyer-Lübke, Wilhelm ( 3 1935): Romanisches Etymologisches Wörterbuch. Heidelberg: Carl Winters Universitätsbuchhandlung. Rheinfelder, Hans ( 5 1976a): Altfranzösische Grammatik. 1. Teil Lautlehre. Munich: Hueber. Rheinfelder, Hans ( 2 1976b): Altfranzösische Grammatik. 2. Teil Formenlehre und Syntax. Munich: Hueber. Rice, Curt (2006): “Optimizing Gender”. Lingua 116: 1394-1417. Ricketts, Peter T./ Hershon, Cyril P. (2007): La vida de Sant Honorat (Publications de l’Association internationale d’études occitanes 4). Turnhout: Brepols. Riquer, Martín de (1971): Guillem de Berguedà. 2. Edición crítica, traducción, notas y glosario (Scriptorium Populeti 6). Abadía de Poblet: Peter Lang. <?page no="289"?> Bibliography 289 Rivière, Hippolyte-Ferréol (1874): Histoire des institutions de l’Auvergne-: contenant un essai historique sur le droit public et privé dans cette province. Tome II. Paris: A. Marescq, aîné. Romieu, Maurice/ Bianchi, André (2002): Iniciacion a l’occitan ancian dètz e nòu tèxtes de l’Edat Mejana comentats = Initiation à l’ancien occitan. Pessac: Presses Universitaires de Bordeaux. Romieu, Maurice/ Bianchi, André ( 4 2025): La lenga del trobar précis de gramatica d’occitan ancian = La langue du trobar. Pessac: Presses Universitaires de Bordeaux. Roncaglia, Aurelio (1965): La lingua dei trovatori: Profilo di grammatica storica del provenzale antico (Officina romanica, vol. 2). Rome: Edizioni dell’Ateneo. Ronjat, Jules (1937): Grammaire istorique des parlers provençaux modernes 2. Partie, Morphologie et formation des mots. Montpellier: Société des Langues Romanes. Rossell, Antoni (2009): El Roman de Flamenca. Novela occitana del siglo XIII. Guadalajara: Editorial Arlequín. Rostaing, Charles (1970): “Le vocabulaire courtois dans la deuxième partie de la Chanson de la Croisade des Albigeois”. In: Mélanges de linguistique, de philologie et de littérature offerts à Albert Henry (Travaux de linguistique et de littérature 1/ 8). Paris: Klincksieck, 249-263. Rothwell, William (1964): “Liber divini amoris”. Proceedings of the Leeds Philosophical and Literary Society. Literary and Historical Section 11: 29-59. Routledge, Michael J. (1977): Les poésies du Moine de Montaudon. Montpellier: Publications du Centre d’Etudes Occitanes de l’Université Paul Valéry. Rovai, Francesco (2012): “Between Feminine Singular and Neuter Plural: Reanalysis Patterns”. Transactions of the Philological Society 110(1): 94-121. Royer, Louis/ Duraffour, Antonin (eds.) (1928): Mystère du XVIe siècle en dialecte embrunais publié avec une introduction et des notes et suivie d’une esquisse philologique et d’un glossaire. Gap: Imprimerie Louis Jean. Rubenbauer, Hans/ Hofmann, Johann Baptist ( 12 1995): Lateinische Grammatik. Bamberg: Buchner. Ryan, John Leonard (1974): The Romance of Flamenca. Albuquerque: University of New Mexico. Sachs, Carl (1857): La vie de sainte Énimie von Bertran von Marseille, in provenzalischer Sprache zum ersten Male vollständig herausgegeben. Berlin: Weidmannsche Buchhandlung. <?page no="290"?> 290 Bibliography Sachs, Hermann (1886): Geschlechtswechsel im Französischen: Ein Versuch der Erklärung desselben. I. Ursprüngliche Neutra. Berlin: Trowitzsch u. Sohn. Salvioni, Carlo (1890): “Il Nuovo Testamento valdese, secondo la lezione del Codice di Zurigo”. Archivio glottologico italiano 11: 1-308. Sánchez-Palomino, María-Dolores (2025): “Lexicography”. In: Sousa, Xulio/ Seoane, Ernesto González (eds.): Manual of Galician Linguistics. Berlin/ Boston: De Gruyter, 173-192. Sancho Fibla, Sergi (2018): “Li vida de Doucelina de Dinha, de Felipa Porcelleta. Imaginería, prácticas devocionales y legitimación de la vida beguina en el Mediterráneo”. In: Corral Díaz, Esther (eds.): Voces de mujeres en la Edad Media : Entre realidad y ficción. Berlin/ Boston: De Gruyter, 296-308. Santano Moreno, Julián (2012): “La lengua de Guilhem de Tudela”. In: Bellone, Luca et al. (eds.): Filologia e linguistica. Scritti in onore di Anna Cornagliotti. Alessandria: Edizioni dell’Orso, 569-589. Sardou, Antoine-Léandre (1875): La vida de sant Honorat, légende en vers provençaux par Raymond Féraut, troubadour niçois du XIIIe siècle, publiée pour la première fois en son entier par les sons et aux frais de la société des lettres, sciences et arts des Alpes-Maritimes. Avec de nombreuses notes explicatives. Nice: Caisson et Mignon. Sarradilla Castaño, Ana (2023): “Femenino y masculino: la larga andadura de la categoría del género en español desde la Edad Media hasta el siglo XXI”. Vox Romanica 82(1). Schäfer, Wilhelm (1911): Das Verhältnis von Raimon Ferauts Gedicht ”La vida de sant Honorat” zu der ”Vita sancti Honorati”. Halle (Saale): Karras. Schafroth, Elmar (2013): “La féminisation des noms de métier et des titres dans trois langues romanes (français, italien, espagnol): convergences et divergences”. In: Karyolemou, Marilena/ Pavlou, Pavlos (eds.): Language Policy and Planning in the Mediterranean World. Newcastle upon Tyne: Cambridge Scholars Publishing, 103-120. Schafroth, Elmar (2015): “Sexe et genre”. In: Polzin-Haumann, Claudia/ Schweickard, Wolfgang (eds.): Manuel de linguistique française. Berlin/ Boston: De Gruyter, 334-364. Schlieben-Lange, Brigitte (1991): “Grammatikographie und Lexikographie / Grammaticographie et lexicographie”. In: Holtus, Günter et al. (eds.): Band V/ 2 Okzitanisch, Katalanisch. Berlin/ New York: Niemeyer, 105-126. Schneegans, Friedrich Ed. (1898): Gesta Caroli Magni ad Carcassonam et Narbonam. Lateinischer Text und provenzalische Übersetzung mit Einleitung (Romanische Bibliothek XV). Halle (Saale): Niemeyer. <?page no="291"?> Bibliography 291 Schöffel, Matthias/ Garces Arias, Esteban, et al. (2025): Unveiling Factors for Enhanced POS Tagging: A Study of Low-Resource Medieval Romance Languages. arXiv: 2506.17715 [cs.CL] . Schöffel, Matthias/ Wiedner, Marinus, et al. (2025): “Modern Models, Medieval Texts: A POS Tagging Study of Old Occitan”. In: Hämäläinen, Mika et al. (eds.): Proceedings of the 5th International Conference on Natural Language Processing for Digital Humanities, 334-349. Schön, Ilse (1971): Neutrum und Kollektivum das Morphem -a im Lateinischen und Romanischen (Innsbrucker Beiträge zur Sprachwissenschaft). Innsbruck: Institut für Vergleichende Sprachwissenschaft der Universität Innsbruck. Schultz-Gora, Oskar (1938): “Vermischte Beiträge zum Altprovenzalischen”. Zeitschrift für romanische Philologie 58: 63-80. Schultz-Gora, Oskar ( 6 1973): Altprovenzalisches Elementarbuch. Heidelberg: Winter. Scinicariello, Sharon Guinn (1982): A Critical Edition of Books I-VII of the Elucidari de las proprietatz de totas res naturals. Chapel Hill: University of North Carolina at Chapel Hill. Scotto, Silvia Carolina/ Perez, Diana I. (2020): “Relatividad lingüística, gramáticas de género y lenguaje inclusivo: algunas consideraciones”. Análisis Filosófico 40(1): 5-39. Scrivner, Olga et al. (2013): Le Roman de Flamenca: An Annotated Corpus of Old Occitan, Old Occitan Digital Collection. <https: / / cl.indiana.edu/ ~obscrivn/ Flamenca.html> [visited on 04/ 08/ 2026]. Seibane, Sara Gómez (2010): “Diacronía del neutro de materia en el occidente Vizcaíno: características de los antecedentes y contextos sintácticos de aparición”. Boletín de la Real Academia Española 90(302): 217-239. Selig, Maria (1997): “Zu den Anfängen der altokzitanischen Urkundenschriftlichkeit”. Zeitschrift für Literaturwissenschaft und Linguistik 27(4): 24- 44. Selig, Maria (2005): “Schreiberprofile und Sprachstandardisierung. Bemerkungen zur mediävistischen Korpuslinguistik”. In: Schrott, Angela/ Völker, Harald (eds.): Historische Pragmatik und Historische Varietätenlinguistik. Göttingen: Universitätsverlag Göttingen, 255-268. Selig, Maria (2023): “Discourse Traditions in the Early Romance Period (with a Focus on Gallo-Romance Varieties)”. In: Winter-Froemel, Esme/ Octavio de Toledo y Huerta, Álvaro S. (eds.): Manual of Discourse Traditions in Romance. Berlin/ Boston: De Gruyter, 353-368. <?page no="292"?> 292 Bibliography Selig, Maria/ Reichle, Elisabeth/ Schöffel, Matthias/ Wiedner, Marinus (2026): “Scripturalisations et traditions discursives dans les comptes consulaires de Montferrand (1259-1385)”. In: de Aguirre, Jennifer Gabel et al. (eds.): «Solazoxa festa, stradolze e stracomplia» Festschrift für Raymund Wilhelm zum 65. Geburtstag mit Beiträgen zu Diskurstraditionen und mittelalterlicher Philologie (Studia Romanica 249). Heidelberg: Winter, 157-170. Selig, Maria et al. (2023): “New Entries in the Dictionnaire de l’occitan médiéval: Some Preliminary Remarks on Methodological and Historical Aspects”. In: Pomino, Natascha et al. (eds.): From Formal Linguistic Theory to the Art of Historical Editions: The Multifaceted Dimensions of Romance Linguistics. Göttingen: Brill/ Vandenhoeck/ Ruprecht Verlage, 263-280. Shirley, Janet (1996): The Song of the Cathar Wars: A History of the Albigensian Crusade. Aldershot/ Brookfield: Scolar Press/ Ashgate. Sibille, Jean (2007): La passion de Saint André : drame religieux de 1512 en occitan briançonnais; édition critique suivie d’une étude linguistique comparée. Paris: Champion. Sibille, Jean (2009): “Les formes en -i issues du nominatif pluriel de la 2ème déclinaison latine, en occitan”. In: Fréchet, Claudine (eds.): Langues et cultures de France et d’ailleurs. Lyon: Presses Universitaires de Lyon, 233-250. Sibille, Jean (2024): “Les dialectes occitans”. In: Esher, Louise/ Sibille, Jean (eds.): Manuel de linguistique occitane. Berlin/ Boston: De Gruyter, 423-472. Siebert, Gustav (1886): Sprachliche Untersuchung der Reime des provenzalischen Romans Flamenca. Marburg: Universitäts-Buchdruckerei (R. Friedrich). Simonnet, Émile Jean (1987): Édition critique du Roman de Notre-Dame de Lagrasse. Paris: Université de Paris-IV-Sorbonne. Skårup, Povl (1997): Morphologie élémentaire de l’ancien occitan. Copenhagen: Museum Tusculanum Press. Spencer, Andrew (2015): “Derivation”. In: Müller, Peter O. et al. (eds.): Word- Formation: An International Handbook of the Languages of Europe. Berlin/ Boston: De Gruyter Mouton, 301-321. Spiewok, Wolfgang (1975): “Semantische Konsequenzen morphologischer Dubletten beim deutschen Substantiv”. Kalbotyra 26(3): 125-135. Stich, Dominique (1998): Parlons francoprovençal : Une langue méconnue. Paris/ Montreal: L’Harmattan. Stratton, James M. (2026): “Issues of Faithfulness and Sampling in Historical Corpora: Challenges and Recommendations”. Journal of English Linguistics 54(2): 83-108. <?page no="293"?> Bibliography 293 Suchier, Hermann (1883): Denkmäler provenzalischer Literatur und Sprache. Halle (Saale): Niemeyer. Suchier, Hermann (1894): “Provenzalische Diätetik. Auf Grund neuen Materials herausgegeben”. In: Festschriften der vier Fakultäten zum zweihundertjährigen Jubiläum der vereinigten Friedrichs-Universität Halle-Wittenberg den 3. August 1894. Halle (Saale): Buchdruckerei des Waisenhauses. Sumien, Domergue (2009): “Classificacion dei dialèctes occitans”. Lingüistica occitana 7: 1-55. Sutorius, Blanche (1918): “Le Doctrinal de Raimon de Castelnou. (Reproduction diplomatique du ms. Libri 105)”. Archivum Romanicum 2: 369-383. Suwe, Ingegärd (1943): La Vida de sant Honorat, poème provençal de Raimond Féraut publié d’après tous les manuscrits. Uppsala: Lundequist. Sweetenham, Carol/ Paterson, Linda (2003): The Canso d’Antioca : an Occitan Epic Chronicle of the First Crusade. Aldershot: Ashgate. Swiggers, Pierre (2009): “Guillaume Molinier (Guilhem Molinier)”. In: Stammerjohann, Henning (eds.): Lexicon grammaticorum. A Bio-Bibliographical Companion to the History of Linguistics. Tübingen: Niemeyer, 588-589. Swiggers, Pierre (2011): “La (relative) standardisation de l’ancien occitan appréhendée à travers les premières descriptions grammaticales”. In: Dessì Schmid, Sarah et al. (eds.): Koineisierung und Standardisierung in der Romania. Heidelberg: Carl Winter Universitätsverlag, 133-149. Tausend, Monika (1980): Eine altprovenzalische Version des Nicodemus-Evangeliums, mit Anmerkungen zur Laut- und Formenlehre. Munich: Unpublished Magisterarbeit. Tausend, Monika (1995): Die altokzitanische Version B der «Legenda aurea». Ms. Paris, Bilbl. nat., n. acq. fr. 6504 (Beihefte zur Zeitschrift für romanische Philologie 262). Tübingen: Niemeyer. Taylor, Robert A. (2015): Bibliographical Guide to the Study of the Troubadours and Old Occitan Literature. Kalamazoo: Medieval Institute Publications, Western Michigan University. Thomas, Antoine (1886): “Les Proverbes de Guylem de Cervera, poème catalan du XIII e siècle”. Romania 15: 25-110. Thomas, Antoine (1905): “Le nominatif pluriel asymétrique des substantifs masculins en ancien provençal”. Romania 34: 353-363. Thomas, Antoine (1911): “Traduction provençale abrégée de la Mulomedicina de Teodorico Borgognoni, suivie de Recettes pour le vin”. Romania 40(159): 353-370. <?page no="294"?> 294 Bibliography Thomas, Antoine (1912): “Un manuscrit provançal retrouvé”. Romania 41(164): 612-614. Thomas, Antoine (1925): La chanson de sainte Foi d’Agen, poème provençal du XIe siècle édité d’après le manuscrit de Leide avec fac-similé, traduction, notes et glossaire (Les classiques français du Moyen Âge 45). Paris: Champion. Thornton, Anna M. (2001): “Some Reflections on Gender and Inflectional Class Assignment in Italian”. In: Schaner-Wolles, Chris et al. (eds.): Naturally! Linguistic Studies in Honour of Wolfgang Ulrich Dressler Presented on the Occasion of his 60th Birthday. Turin: Rosenberg & Sellier, 479-487. Thornton, Anna M. (2003): “L’assegnazione del genere in italiano”. In: Miret, Fernando Sánchez (eds.): Actas del XXIII Congreso Internacional de Lingüística y Filología Románica. Tübingen: Niemeyer, 467-481. Thornton, Anna M. (2004): “Conversione”. In: Grossmann, Maria/ Rainer, Franz (eds.): La formazione delle parole in italiano. Tübingen: Niemeyer, 500-553. Thornton, Anna M. (2009): “Constraining Gender Assignment rules”. Language Sciences 31(1): 14-32. TLFi = ATILF - CNRS (ed.) (2004): Trésor de la langue française informatisé. Nantes. <http: / / atilf.atilf.fr/ tlf.htm> [visited on 04/ 08/ 2026]. TLL = (2025): Thesaurus linguae latinae. Berlin/ Boston: De Gruyter. <https: / / publikationen.badw.de/ de/ thesaurus/ lemmata> [visited on 04/ 08/ 2026]. Tortoreto, Valeria (ed.) (1981): Il trovatore Cercamon. Edizione critica con introduzione, testo, glossario ecc. (Subsidia al «Corpus des Troubadours» = Studi, testi e manuali 7 = 9). Modena: Società Tipografica Editrice Modenese (STEM- Mucchi). van der Horst, Cornelius H. M. (1974): Blandin de Cornouaille. Introduction, Édition Diplomatique, Glossaire. The Hague/ Paris: Mouton. Vatteroni, Sergio (2013a): Il trovatore Peire Cardenal. Volume I (Subsidia al “Corpus des troubadours”, nuova serie, 17 = Studi, testi e manuali, nuova serie, 12). Modena: Mucchi. Vatteroni, Sergio (2013b): Il trovatore Peire Cardenal. Volume II (Subsidia al “Corpus des troubadours”, nuova serie, 17 = Studi, testi e manuali, nuova serie, 12 2). Modena: Mucchi. Velupillai, Viveka (2012): An Introduction to Linguistic Typology. Amsterdam/ Philadelphia: Benjamins. Ventura, Simone (2010): Cultura enciclopedica nell’Occitania dei trovatori: il libro XV dell’«Elucidari de las proprietatz de totas res naturals». Edizione critica, <?page no="295"?> Bibliography 295 traduzione e commento (Corpus des troubadours 1, Éditions 1). Florence: Edizioni del Galluzzo. VerbaAlpina = Krefeld, Thomas/ Lücke, Stephan (eds.) (2014-2023): VerbaAlpina. Der alpine Kulturraum im Spiegel seiner Mehrsprachigkeit. Munich. <https: / / www.verba-alpina.gwi.uni-muenchen.de/ > [visited on 04/ 08/ 2026]. Viel, Riccardo (2011): Troubadours mineurs gascons du XIIe siècle: Alegret, Marcoat, Amanieu de la Broqueira, Peire de Valeria, Gausbert. Édition critique bilingue avec introduction, notes et glossaire (Les classiques français du Moyen Âge 167). Paris: Champion. Vigliocco, Gabriella et al. (2005): “Grammatical Gender Effects on Cognition: Implications for Language Learning and Language Use”. Journal of Experimental Psychology: General 134(4): 501-520. Vossler, Karl (1916): Peire Cardinal: Ein Satiriker aus dem Zeitalter der Albigenserkriege (Sitzungsberichte der Bayerischen Akademie der Wissenschaften, Philosophisch-Philologische und Historische Klasse). Munich: Verlag der Königl. Bayerischen Akademie der Wissenschaften. Wagner, Max Leopold (1930): Studien über den sardischen Wortschatz: 1. Die Familie - 2. Der menschliche Körper. Geneva: Leo S. Olschki. Wahl, Angelika (ed.) (1980): Die altprovenzalische Übersetzung des Liber Scintillarum mit Einleitung und Glossar (Romanica Monacensis 14). Munich: Wilhem Fink. Wang, Ziwen (2023): Pérdida del género neutro del latín al hispanorromance altomedieval. Una reconstrucción panrománica. Barcelona: Tesi. Wang, Ziwen (2025): “Genus Alternans in the Early History of Ibero-Romance: Textual Evidence from Early Medieval Iberian Peninsula”. Transactions of the Philological Society 123(2): 1-26. Weber, Robert/ Gryson, Roger ( 5 2007): Biblia sacra: iuxta Vulgatam versionem adiuvantibus B. Fischer, I. Gribomint, H.F.D. Sparks, W. Thiele recensuit et brevi apparatu critico instruxit. Stuttgart: Deutsche Bibelgesellschaft. Wehowski, Else (1910): Die Sprache der Vida de la benaurada sancta Doucelina. Lautstand, Formen und einige syntaktische Erscheinungen: Eine Studie zum Dialekt von Marseille (Romanische Studien 8). Berlin: Emil Ebering. Wiedner, Marinus (2023): Old Occitan handwriting. (Modell-Nr. 52822, CER=3,51%), PyLaia-Modell for handwritten Occitan from the 13th and 14th century. <https: / / readcoop.eu/ model/ old-occitan-handwriting/ > [visited on 04/ 08/ 2026]. Wiedner, Marinus (2026): “Doublons de genre en occitan médiéval : huit études de cas sur corpus”. In: Wissner, Inka/ Dufter, Andreas (eds.): Aspects de la <?page no="296"?> 296 Bibliography variation en diachronie. Regards sur la Galloromania (Beihefte zur Zeitschrift für romanische Philologie 506). Berlin/ Boston: De Gruyter, 69-101. Wiedner, Marinus (forthcoming): “Les doublons de genre en occitan médiéval à l’exemple de serpen et tabor ”. In: Selig, Maria/ Müller, Daniela (eds.): Actes du XIV ème congrès de l’association internationale d’études occitanes. Munich: Open Publishing LMU. Wiedner, Marinus/ Schöffel, Matthias (2026): “Simulating the Evolution of Grammatical Gender from Latin to Old Occitan: A Computational Approach Using LSTM with Attention”. AI-Linguistica. Linguistic Studies on AI-Generated Texts and Discourses 3 (1). Windisch, Rudolf (1973): Genusprobleme im Romanischen: Das Neutrum im Rumänischen (Tübinger Beiträge zur Linguistik 31). Tübingen: Gunter Narr Verlag. Wolf, Johanna (2024): “In Chunks We Trust . . . the Problem of Gender Assignment in Foreign Language Learning of French”. International Journal of Bilingualism 29(2): 474-490. Wolf, Lothar/ Hupka, Werner (1981): Altfranzösisch: Entstehung und Charakteristik: Eine Einführung. Darmstadt: Wissenschaftliche Buchgesellschaft. Wüest, Jakob (1995): “Les scriptae occitanes IV. Languedoc”. In: Holtus, Günter et al. (eds.): Lexikon der Romanistischen Linguistik, Bd. II/ 2. Die einzelnen romanischen Sprachen und Sprachgebiete vom Mittelalter zur Renaissance. Berlin/ New York: Niemeyer, 441-450. Wunderli, Peter (2009): Le nouveau testament de Lyon (Bibliothèque de la ville A.I.54 / Palais des arts 36). Volume 1. Introduction et édition critique (Romanica Helvetica 128). Tübingen/ Basel: Francke. Wunderli, Peter (2010): Le nouveau testament de Lyon (Bibliothèque de la ville A.I.54 / Palais des arts 36). Volume 2. Analyse de la langue, Lexique et Index des noms (Romanica Helvetica 131). Tübingen/ Basel: Francke. Zambon, Francesco (2016): “Una nuova ipotesi sull’autore della seconda parte della Canzone della Crociata albigese”. Romance Philology 70: 267-281. Zamuner, Ilaria (2005): “Per l’edizione critica dei volgarizzamenti provenzali dell’Epistola ad Alexandrum de dieta servanda”. In: Latry, Guy (eds.): Actes du VIIe Congrès de l’Association Internationale d’Études Occitanes (A.I.E.O.) Bordeaux: Presses Universitaires de Bordeaux, 739-759. Zamuner, Ilaria (2007): “Les versions françaises de l’Epistola ad Alexandrum de dieta servanda: mise au point”. In: Galderisi, Claudio/ Pignatelli, Cinzia (eds.): La traduction vers le moyen français. Actes du II e colloque de l’AIEMF (Associ- <?page no="297"?> Bibliography 297 ation internationale pour l’étude du moyen français), Poitiers, 27-29 avril 2006. Turnhout: Brepols, 165-187. Zapf, Miriam (2024): Gender, Sprache, Kognition: Eine linguistische Untersuchung zu gender-inklusivem Sprachgebrauch im Spanischen (Linguistik in Empirie und Theorie/ Empirical and Theoretical Linguistics). Berlin/ Heidelberg: Springer. Zemp, Josef (1978): Les poésies du troubadour Cadenet. Bern/ Frankfurt am Main/ Las Vegas: Peter Lang. Zenker, Rudolf (1900): Die Lieder Peires von Auvergne. Kritisch herausgegeben mit Einleitung, Uebersetzung, Kommentar und Glossar. Erlangen: Fr. Junge. Zinelli, Fabio (2019): “Le Barlaam occitan est-il une traduction du catalan ? Les versions occitane et italiennes à la lumière du ms. Vic, Arxiui Biblioteca Episcopal, 174”. Romania 137: 19-80. Zufferey, François (2016): “Les régionalismes dans les textes littéraires: une contribution à leur tradition manuscrite et à l’histoire culturelle”. In: Glessgen, Martin/ Trotter, David (eds.): La régionalité lexicale du français au Moyen Âge. Volume thématique issu du colloque de Zurich (7-8 sept. 2015), organisé sous le patronage de la Société de linguistique romane (Travaux de linguistique romane. Lexicologie, onomastique et lexicographie). Strasbourg: Éditions de linguistique et de philologie (ELiPhi), 289-300. Zufferey, François/ Fasseur, Valérie (2014): Flamenca, Texte édité d’après le manuscrit unique de Carcassonne (Le livre de poche, 32551. Lettres gothiques). Paris: Librairie générale française. <?page no="299"?> Language Index Aragonese 81, 121, 123, 126, 129, 133, 134, 137, 138, 143, 144, 147, 149, 150, 155, 157, 160, 163, 164, 168, 169, 172, 176, 179, 180, 187, 190, 194, 205, 213, 221, 222, 255 Asturian 5, 15, 43-45, 121, 133, 134, 249 Basque 245 Béarnese 127, 129 Catalan 23, 36, 43, 44, 76, 78, 81, 89, 95, 100, 102, 121, 123, 126, 129, 133, 134, 137, 138, 143, 144, 147, 149, 150, 155, 157, 163, 164, 168, 169, 172, 176, 179, 180, 187, 194, 198, 200, 203, 205, 213, 222, 240, 251, 255 Corsican 40, 121, 134, 149, 155, 157, 163, 169, 171, 176, 179, 180, 190, 199 Diyari 29 Dyirbal 29, 30 English 20, 29, 76, 78, 90, 100 Finnish 56 Franco-Provençal 5, 45, 47-49, 121, 133, 134, 138, 143, 149, 205 French 5, 19-24, 30, 36, 45-49, 54-57, 68, 76-78, 80, 81, 90, 91, 98, 100, 103, 121, 123, 129, 133, 134, 138, 141, 143, 147, 150, 155, 160, 163, 168, 169, 176, 179, 180, 187, 190, 194, 199, 200, 203, 213, 219, 221, 227, 240, 245, 251, 255, 303 Friulian 121, 133, 134, 149, 155, 164, 187, 190, 199, 251 Galician 121, 123, 133, 134, 143, 147, 149, 150, 155, 157, 163, 169, 171, 176, 205, 213 Gallurese 138 Gascon 55, 92, 93, 144, 149, 150, 163, 178, 244, 245 Gaulish 55, 56, 245 German 13, 19, 20, 30, 32, 54, 76, 102 Greek 56, 61, 100, 101, 121-123, 198, 201, 202, 204, 206, 209, 212, 213, 219, 221, 250, 251, 255, 256 Hausa 33 <?page no="300"?> 300 Language Index Italian 13, 23, 36, 38-40, 43, 49, 54, 57, 68, 76, 99, 104, 121, 123, 129, 133, 134, 143, 144, 147, 149, 150, 155, 157, 163, 164, 168, 169, 171, 172, 176, 179, 180, 187, 190, 194, 199, 203, 205, 213, 219, 221, 240, 251, 254, 305 Kaingang 29 Ladin 121, 128, 129, 133, 134, 137, 144, 147, 149, 150, 155, 157, 163, 164, 168, 169, 176, 180, 187, 190, 194, 199 Languedocian 16, 47, 51, 74, 75, 101, 104, 106, 107, 109, 110, 120, 123-125, 128, 130, 133, 135, 137, 138, 141, 143-145, 147, 149, 151-153, 155-161, 163- 165, 167-170, 172, 176, 178-181, 184-188, 191, 193, 195, 198-200, 203, 205-207, 209, 212-215, 220, 221, 223, 224, 238, 243-245, 253, 254, 305 Latin 5, 6, 16, 22, 23, 27, 28, 30, 33-36, 38-40, 43, 50-52, 54-56, 60, 61, 67, 68, 72, 73, 84-86, 88, 89, 91-93, 96, 98, 100-108, 121-123, 125, 127-129, 131, 133-135, 137-139, 141, 143-147, 149-151, 153, 155, 157, 159-165, 167- 169, 171, 173, 175-177, 179-181, 183-185, 187, 189-195, 197-200, 202, 203, 205, 209, 212, 213, 219, 221, 227, 230, 231, 233-236, 239, 249-255, 306 Latvian 56 Limilngan 33 Livonian 56 Logudorese 141, 164 Medieval Latin 122, 123, 128, 129, 133, 134, 194 Occitan 5, 45-47, 49, 71-74, 78, 80, 81, 84, 96, 100, 102, 104, 106, 108, 114, 116, 122, 128, 129, 134, 138, 143, 144, 150, 153, 155, 157, 160, 164, 168, 171, 179, 187, 190, 195, 198, 205, 221, 222, 239, 243, 255, 256 Old French 49, 54-56, 60, 85, 92, 100, 102, 121, 123, 125, 147, 190, 194, 206 Old Italian 39, 40, 49, 153, 160, 244, 256 Old Languedocian 59, 100, 126, 127, 129, 135, 138, 151, 153, 154, 156, 158, 161, 172, 184, 197, 204, 243, 244, 253 Old Occitan 6, 13-17, 22, 24, 28, 30, 49-51, 55, 56, 59-65, 67-69, 71, 74-76, 84- 86, 88, 91-93, 95, 98-104, 106-108, 111, 114, 116, 119, 120, 122, 123, 125, 126, 128, 129, 133, 134, 136-138, 140, 141, 144, 146, 149, 150, 153- 155, 157, 160, 162-164, 168, 169, 171, 176, 179, 180, 184, 187, 189, 190, 193, 194, 197-201, 204, 206, 209, 213, 219, 221, 222, 224, 226, 227, 231, 243-245, 247, 249-251, 253-257, 305 Old Provençal 126, 127, 129, 135, 138, 151, 153, 154, 156, 158, 160, 161, 170, 173, 197, 204, 243, 244, 253 Persian 123, 222, 306 <?page no="301"?> Language Index 301 Portuguese 36, 43, 122, 129, 147, 149, 150, 153, 155, 157, 160, 163, 164, 168, 169, 176, 179, 180, 205, 213, 235, 239 Provençal 16, 36, 47, 74, 75, 77, 82, 100, 105, 107, 109, 110, 120, 123-125, 128, 130, 133, 135, 137, 138, 141, 143-145, 147, 149, 151-153, 155-158, 160, 161, 163-165, 167, 168, 170, 172, 173, 176, 178-182, 184-188, 191-193, 195, 198, 200, 203, 205, 206, 209, 210, 212, 213, 215, 220, 221, 223, 224, 244, 253, 254, 305 Qafar 32 Romanian 5, 15, 36, 37, 40, 49, 52, 122, 128, 129, 133, 134, 147, 149, 150, 153, 155, 157, 160, 163, 164, 168, 169, 172, 176, 180, 187, 190, 199, 205, 221, 240, 254, 305 Romansh 36, 122, 123, 133, 134, 143, 147, 149, 150, 164, 169, 190, 194, 203, 205 Russian 30-32, 303 Sardinian 36, 122, 129, 133, 138, 141, 149, 150, 155, 169, 171, 180, 190, 194 Sassarese 138, 164 Spanish 19, 23, 36, 39, 43, 44, 49, 56, 57, 76, 81, 92, 122, 123, 129, 133, 134, 137, 141, 143, 144, 147, 149, 150, 155, 157, 160, 163, 164, 168, 169, 171, 176, 179, 180, 187, 190, 194, 200, 203, 205, 213, 222, 239 Sursilvan 129 Swahili 28, 31, 305 Tamil 29 Vulgar Latin 40, 43, 234, 235 <?page no="303"?> List of Figures 3.1 Gender assignment rules in Russian, taken from Corbett (1991: 41). ........................................................................................... 32 3.2 The gender system in Old Italian, from Loporcaro/ Faraoni/ Gardani (2014: 7). ............................................................................ 40 3.3 Gender system for the dialects of San Giovanni in Fiore and Bocchigliero = Old Romanesco/ Old Salentino, from Loporcaro (2015: 112). ................................................................................ 41 3.4 Four-gender system in Agnonese, from Loporcaro (2018: 140). ..... 42 3.5 Five-gender system in Poggio S. Romualdo, from Loporcaro (2018: 280). .......................................................................................... 42 3.6 Preand post-nominal agreement in Asturian, taken from Loporcaro (2018: 160)..................................................................... 44 3.7 Summary of the two concurrent gender systems of Central Asturian, as proposed in Loporcaro (2018: 172). ............................... 45 3.8 Variation in gender of 165 nouns compared to French with the data of the ALF, taken from Brun-Trigaud/ Sauzet/ Allassonnière- Tang (2023: 18). ......................................................................... 46 3.9 Syncretism in the plural formation of French, from Loporcaro (2018: 41)................................................................................... 47 3.10 Romance gender systems in Europe, adapted from Loporcaro (2018: 60)................................................................................... 53 6.1 Genealogy of the manuscripts of the Vida de Sant Honorat, adapted from Ricketts/ Hershon (2007: 46; circle markings are my own). .... 87 9.1 Gender distribution among the three manuscripts of the Vida de Sant Honorat. ............................................................................. 226 9.2 Gender distribution per case among the three manuscripts of the Vida de Sant Honorat. ................................................................. 227 9.3 Gender distribution per number among the three manuscripts of the Vida de Sant Honorat......................................................... 227 9.4 Gender distribution per case and number among the three manuscripts of the Vida de Sant Honorat......................................................... 228 <?page no="305"?> List of Tables 3.1 Morphological gender assignment in Swahili, taken from Corbett (1991: 47). ........................................................................... 31 3.2 Correlation between gender and declension class in Latin, from Maiden (2010: 167). .................................................................... 35 3.3 The gender system in Romanian, from Loporcaro (2018: 92). ........ 37 3.4 Simplified gender system of Modern Italian. ................................ 38 3.5 Phonological gender assignment in French, from Corbett (1991: 59). ........................................................................................... 48 4.1 Nominal declension in Old Occitan. ............................................ 60 4.2 Invariable nouns in Old Occitan.................................................. 61 4.3 Old Occitan imparisyllabic nouns. .............................................. 62 4.4 Adjectival and participle declension in Old Occitan. .................... 62 4.5 Personal pronouns (third person) in Old Occitan, adapted from Paden (1998: 302). ...................................................................... 63 4.6 Possessive (poss.) and demonstrative (dem.) pronouns in Old Occitan...................................................................................... 63 4.7 Old Occitan definite and indefinite articles. ................................. 64 6.1 Summary of known facts about the manuscripts (bold = Provençal sub-corpus; italics = Languedocian sub-corpus). .......................... 110 8.1 Gender doublets from the second declension class. ...................... 123 8.2 Gender distribution for comtat/ vescomtat. ................................... 124 8.3 Gender doublets from the third declension class. ......................... 127 8.4 Gender distribution for aigal. ..................................................... 128 8.5 Gender distribution for art. ........................................................ 130 8.6 Gender distribution for canal. ..................................................... 133 8.7 Gender distribution for color. ...................................................... 135 8.8 Gender distribution for crin. ....................................................... 137 8.9 Gender distribution for den......................................................... 138 8.10 Gender distribution for fam. ....................................................... 141 8.11 Gender distribution for fornatz. .................................................. 143 8.12 Gender distribution for glan/ aglan. ............................................. 145 8.13 Gender distribution for labor. ..................................................... 147 <?page no="306"?> 306 List of Tables 8.14 Gender distribution for lach........................................................ 149 8.15 Gender distribution for mar. ....................................................... 151 8.16 Gender distribution for marme. .................................................. 153 8.17 Gender distribution for mel. ....................................................... 155 8.18 Gender distribution for orde........................................................ 157 8.19 Gender distribution for ost.......................................................... 160 8.20 Gender distribution for penche. ................................................... 163 8.21 Gender distribution for pols. ....................................................... 165 8.22 Gender distribution for ret. ......................................................... 168 8.23 Gender distribution for sal.......................................................... 170 8.24 Gender distribution for salut....................................................... 172 8.25 Gender distribution for sanc. ...................................................... 176 8.26 Gender distribution for semen. .................................................... 179 8.27 Gender distribution for serp/ serpen.............................................. 181 8.28 Gender distribution for set. ......................................................... 185 8.29 Gender distribution for trau........................................................ 188 8.30 Gender doublets from the fourth declension class. ....................... 190 8.31 Gender distribution for man. ...................................................... 191 8.32 Gender doublets from the fifth declension class. .......................... 194 8.33 Gender distribution for dia. ........................................................ 195 8.34 Gender doublets from other Latin word classes. ........................... 197 8.35 Gender distribution for semal. .................................................... 198 8.36 Gender distribution for sera. ....................................................... 200 8.37 Greek loanwords as gender doublets. .......................................... 202 8.38 Gender distribution for evangelista.............................................. 203 8.39 Gender distribution for fantasma. ............................................... 205 8.40 Gender distribution for papa....................................................... 206 8.41 Gender distribution for patriarca. ............................................... 209 8.42 Gender distribution for profeta.................................................... 213 8.43 Gender distribution for salmista.................................................. 220 8.44 Gender distribution for senebe. ................................................... 221 8.45 Tabor as a gender doublet (Persian loanword). ............................. 222 8.46 Gender distribution for tabor. ..................................................... 223 9.1 Masculine and feminine occurrences per lemma and text within the three manuscripts of the Vida de Sant Honorat. ...................... 229 <?page no="307"?> Orbis Romanicus Studia philologica Monacensia Edunt Andreas Dufter et Bernhard Teuber Bisher sind erschienen: Band 1 David Klein Medienphantastik Phantastische Literatur im Zeichen medialer Selbstreflexion bei Jorge Luis Borges und Julio Cortázar 2015, IV, 209 Seiten €[D] 58,- ISBN 978-3-8233-6986-8 Band 2 Benjamin Meisnitzer Das Präsens als Erzähltempus im Roman Eine gedruckte Antwort auf den Film 2016, 306 Seiten €[D] 68,- ISBN 978-3-8233-8001-6 Band 3 Elisabeth Kruse La recepción creadora de la tradición mística en la lírica de Dámaso Alonso ¿Un poeta metafísico moderno? 2016, 277 Seiten €[D] 58,- ISBN 978-3-8233-6995-0 Band 4 Anna Marcos Nickol El teatro de la guerra Raum, Krieg und Theater bei Juan Benet 2016, 300 Seiten €[D] 69,- ISBN 978-3-8233-8050-4 Band 5 Kurt Hahn Mentaler Gallizismus und transkulturelles Erzählen Fallstudien zu einer französischen Genealogie der hispanoamerikanischen Narrativik im 19. Jahrhundert 2017, 414 Seiten €[D] 78,- ISBN 978-3-8233-8052-8 Band 6 Martina Bengert Nachtdenken Maurice Blanchots „Thomas l’Obscur“ 2017, 340 Seiten €[D] 68,- ISBN 978-3-8233-8045-0 Band 7 Johanna Vocht/ David Klein / Gerhard Poppenberg (Hrsg.) (Des)escribir la Modernidad - Die Moderne (z)erschreiben: Neue Blicke auf Juan Carlos Onetti 2019, 233 Seiten €[D] 68,- ISBN 978-3-8233-8101-3 Band 8 Thomas Scharinger Mehrsprachigkeit im Frankreich der Frühen Neuzeit Zur Präsenz des Italienischen, seinem Einfluss auf das Französische und zur Diskussion um das françois italianizé 2018, 719 Seiten €[D] 138,- ISBN 978-3-8233-8160-0 Band 9 Laura Linzmeier Kontaktinduzierter Lautwandel, Sprachabbau und phonologische Marker im Sassaresischen 2018, 625 Seiten €[D] 108,- ISBN 978-3-8233-8141-9 Band 10 Martina Bengert/ Iris Roebling-Grau (Hrsg.) Santa Teresa Critical Insights, Filiations, Responses 2019, 360 Seiten €[D] 68,- ISBN 978-3-8233-8246-1 <?page no="308"?> Band 11 Jörg Dünne/ Kurt Hahn/ Lars Schneider (Hrsg.) Lectiones difficiliores - Vom Ethos der Lektüre 2019, 664 Seiten €[D] 88,- ISBN 978-3-8233-8258-4 Band 12 Christoph Hülsmann Initiale Topiks und Foki im gesprochenen Französisch, Spanisch und Italienisch 2019, 329 Seiten €[D] 68,- ISBN 978-3-8233-8301-7 Band 13 Mattia Zangari Tre storie di santità femminile tra parole e immagini Agiografie, memoriali e fabulae depictae fra Due e Trecento 2019, 150 Seiten €[D] 58,- ISBN 978-3-8233-8360-4 Band 14 Manfred Bös Transzendierende Immanenz Die Ontologie der Kunst und das Konzept des Logos poietikos bei dem spanischen Dichter Antonio Gamoneda 2020, 395 Seiten €[D] 88,- ISBN 978-3-8233-8340-6 Band 15 Johanna Vocht Onettis Santa María(s): Machträumliche Spannungsfelder zwischen biologischer Reproduktion und künstlerischer Produktion 2022, 281 Seiten €[D] 68,- ISBN 978-3-8233-8425-0 Band 16 Aurelia Merlan (ed.) Romanian in Migration Contexts 2024, 337 Seiten €[D] 78,- ISBN 978-3-8233-8467-0 Band 17 Felix Bokelmann Varianzphänomene der Standardaussprache in Argentinien Indizien aus Sprachproduktion und -perzeption 2021, 392 Seiten €[D] 78,- ISBN 978-3-8233-8490-8 Band 18 Denis Heuring Verdrängen und Erinnern im Theater Bürgerkrieg und Diktatur im spanischen Drama nach 1975 2023, 445 Seiten €[D] 92,- ISBN 978-3-8233-8507-3 Band 19 Veit Lindner Wege, Lichtung, Horizont: Konstellationen des ‚Essayistischen‘ in María Zambranos Claros del bosque und Octavio Paz’ El mono gramático 2021, 314 Seiten €[D] 68,- ISBN 978-3-8233-8529-5 Band 20 Sebastià Moranta Mas Discursos lingüísticos e identitarios en Mallorca y en la República de Moldavia Una investigación contrastiva de los conflictos entre catalán y español en Mallorca y entre rumano y ruso en Moldavia desde el enfoque del análisis crítico del discurso, la teoría sociolingüística y los estudios culturales (noch nicht erschienen) ca. 320 Seiten €[D] 78,- ISBN 978-3-8233-8536-3 Band 21 Anke Grutschus Stimmenvielfalt im Monolog Formale und funktionale Aspekte von Redewiedergabe in spanischsprachigen Stand-up-Acts, Predigten und wissenschaftlichen Vorträgen 2022, 453 Seiten €[D] 88,- ISBN 978-3-8233-8557-8 <?page no="309"?> Band 22 Daniel Graziadei / Florencia Sannders (Hrsg.) Macedonio Fernández: Nicht jedes Wachen ist das mit den offenen Augen Eine Übersetzung aus dem argentinischen Spanisch mit philologischer und philosophischer Einführung 2022, 197 Seiten €[D] 58,- ISBN 978-3-8233-8555-4 Band 23 Martha Guzmán Morirse, salirse, comerse y otros pseudorreflexivos sin motivación argumental De su presente e historia en español y francés 2025, 333 Seiten €[D] 88,- ISBN 978-3-8233-8574-5 Band 24 Anna Waldschütz Zwischen Eros und Agape Das paulinisch-augustinische Liebeskonzept beim Arcipreste de Hita 2023, 248 Seiten €[D] 68,- ISBN 978-3-8233-8606-3 Band 25 Mattia Zangari Pazze di Lui - Pazze di Lui - Mad for Him: Hagiographic Stereotypes, Mental Disturbances and Anthropological Implications of Female Saintliness in Italy and Abroad from the 13 th to the 20 th Century 2024, 231 Seiten €[D] 68,- ISBN 978-3-381-11111-4 Band 26 Sebastian Ortner Latinisierende Syntax im Französischen des 16. Jahrhunderts Der Accusativus cum Infinitivo in Jean Calvins Institution de la religion chrestienne und zeitgenössischen Texten 2025, 556 Seiten €[D] 109,- ISBN 978-3-381-10771-1 Band 27 Marinus Wiedner Grammatical Gender in Old Occitan 2026, 306 Seiten €[D] 88,- ISBN 978-3-381-16511-7 <?page no="310"?> ISBN 978-3-381-16511-7 This study treats variation in grammatical gender in Old Occitan. By integrating corpus-linguistic, typological, and sociolinguistic methods, the research succeeds in tracing the dynamics of gender-systems in usage. For this sake, a corpus was created based on 17 manuscripts from the Languedoc and Provence regions in France, written in the 13 th and 14 th centuries. This enables the author to grasp the variation present in medieval languages, and to avoid bias from the normalisation of later editions. With this approach, the investigation of gender variation is based on a solid empirical foundation and thus demonstrates the kinds of new insights that become attainable when linguistic data are approached from a historically informed, discourse-tradition-aware, text-pragmatic, and norm-historical perspective.