<?page no="0"?> S C H W E I Z E R A N G L I S T I S C H E A R B E I T E N S W I S S S T U D I E S I N E N G L I S H Nicole Studer-Joho Diffusion and Change in Early Middle English Methodological and Theoretical Implications from the LAEME Corpus of Tagged Texts <?page no="1"?> Schweizer Anglistische Arbeiten Swiss Studies in English Begründet von Bernhard Fehr Herausgegeben von Andreas Fischer (Zürich), Martin Heusser (Zürich), Daniel Schreier (Zürich) Band 141 <?page no="3"?> Nicole Studer-Joho Diffusion and Change in Early Middle English Methodological and Theoretical Implications from the LAEME Corpus of Tagged Texts <?page no="4"?> Bibliographical information of the Deutsche Nationalbibliothek The Deutsche Nationalbibliothek lists this publication in the Deutsche Nationalbibliografie; detailed bibliographical information is available at http: / / dnb.dnb.de This thesis was accepted as a doctoral dissertation by the Faculty of Arts of the University of Zurich in the spring semester 2012 on the recommendation of Prof. Dr. Daniel Schreier and Prof. Dr. Olga Timofeeva. Published with the support of the Swiss National Science Foundation. © 2014 · Narr Francke Attempto Verlag GmbH + Co. KG Dischingerweg 5 · D-72070 Tübingen All rights, including the rights of publication, distribution and sales, as well as the right to translate, are reserved. No part of this work covered by the copyrights hereon may be reproduced or copied in any form or by any means - graphic, electronic or mechanical including photocopying, recording, taping, or information and retrieval systems - without written permission of the publisher. Internet: http: / / www.francke.de E-Mail: info@francke.de Composition: typoscript GmbH, Walddorfhäslach Printed and Bound by Hubert & Co., Göttingen Printed in Germany ISSN 0080-7214 ISBN 978-3-7720-8539-0 Cover illustration: Martin Heusser, Zürich Cover design: Martin Heusser, Zürich <?page no="5"?> To my Parents Linda and Hans <?page no="7"?> Table of Contents List of Black and White Maps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . XI List of Colour Maps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . XII List of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . XIII List of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . XV List of Abbreviations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . XVI Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . XVII 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 1.1 Early Middle English . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 1.2 Middle English dialectology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 1.2.1 Oakden (1930) and Moore, Meech, and Whitehall (1935) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 1.2.2 The Survey of Middle English Dialects . . . . . . . . . . . . . . 7 1.2.3 The Linguistic Atlas of Late Mediaeval English . . . . . . . 8 1.2.4 The Linguistic Atlas of Early Middle English . . . . . . . . . 10 1.3 Spatial diffusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11 1.3.1 Models of spatial diffusion . . . . . . . . . . . . . . . . . . . . . . . . . . . 12 1.3.2 Spatial diffusion in Middle English? . . . . . . . . . . . . . . . . . . 14 1.4 Outline of the present study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20 1.4.1 Aim and scope . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20 1.4.2 Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21 2. Historical Background and Previous Work . . . . . . . . . . . . . . . . . . . 23 2.1 Reduction of the number of grades in strong verbs . . . . . . . . . . 24 2.1.1 The classes of strong verbs up until the Old English period . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26 2.1.2 Strong verbs in Middle English . . . . . . . . . . . . . . . . . . . . . . 29 2.1.2.1 Shift to weak . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30 2.1.2.2 Hybridization or class mixing . . . . . . . . . . . . . . . . . 33 2.1.2.3 Reduction of the number of grade vowels . . . . . . 34 2.2 OE a ˉ > ME ǭ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37 2.2.1 OE origins of ME a ˉ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37 2.2.1.1 WGmc *ai > OE a ˉ . . . . . . . . . . . . . . . . . . . . . . . . . . . 38 2.2.1.2 WGmc *a ˉ > OE æ ˉ > OE a ˉ . . . . . . . . . . . . . . . . . . . 39 2.2.1.3 Loanwords with a ˉ . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40 2.2.1.4 OE ă . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41 <?page no="8"?> 2.2.1.5 Caveat: Late Old English shortening processes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44 2.2.2 OE a ˉ > ME ǭ : onset, spatial variation and completion . . 46 2.3 Dual forms of the personal pronoun . . . . . . . . . . . . . . . . . . . . . . . . 49 2.3.1 The dual personal pronoun in the Germanic languages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50 2.3.2 The dual in OE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53 2.3.3 The dual in ME . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55 2.4 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59 3. Material and Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60 3.1 The Linguistic Atlas of Early Middle English (LAEME) . . . . . 60 3.2 Methodological considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62 3.2.1 Anchor texts or “ the best available witnesses ” . . . . . . . . . . 62 3.2.2 Coverage of texts in LAEME . . . . . . . . . . . . . . . . . . . . . . . . . 63 3.2.3 Text languages in LAEME . . . . . . . . . . . . . . . . . . . . . . . . . . . 64 3.2.4 Implications for the present study . . . . . . . . . . . . . . . . . . . . 68 3.3 The system of lexico-grammatical tags . . . . . . . . . . . . . . . . . . . . . 70 3.4 Data retrieval . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73 3.4.1 Reduction of the number of grade vowels of strong verbs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73 3.4.2 OE a ˉ > ME ǭ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77 3.4.3 Dual forms of the personal pronoun . . . . . . . . . . . . . . . . . . 80 3.5 Linguistic maps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81 3.5.1 Types of linguistic maps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81 3.5.2 The present maps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84 3.6 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87 4. Reduction of the Number of Grades in Strong Verbs . . . . . . . . 88 4.1 Singular forms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88 4.1.1 Class V . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89 4.1.2 Class IVa . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94 4.1.3 Class IIIc . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95 4.1.4 Class IIIb . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96 4.1.5 Class II . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97 4.1.6 Class I . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98 4.2 2nd singular and plural forms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99 4.2.1 Class V . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100 4.2.2 Class IVa . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102 4.2.3 Class IIIc . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103 4.2.4 Class IIIb . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105 VIII Table of Contents <?page no="9"?> 4.2.5 Class II . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105 4.2.6 Class I . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106 4.3 Verbs of Old Norse origin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107 4.4 Historically weak verbs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108 4.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110 5. OE a ˉ > ME ǭ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112 5.1 General results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112 5.1.1 The South and the North . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113 5.1.2 The East Midlands . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114 5.1.3 The West Midlands . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116 5.1.4 Unlocalized texts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119 5.2 Linguistic environment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122 5.2.1 OE a ˉ + w/ ɣ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122 5.2.2 Shortening before consonant clusters (SHOCC) . . . . . . . 124 5.2.3 Trisyllabic shortening (TRISH) . . . . . . . . . . . . . . . . . . . . . . 127 5.3 Word class and frequency . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129 5.4 <ao> Spellings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132 5.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134 6. The Decline of the Dual in Middle English . . . . . . . . . . . . . . . . . . 136 6.1 Contexts of duals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 138 6.2 1st person dual forms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140 6.3 2nd person dual forms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 144 6.4 The survival of the dual in ME . . . . . . . . . . . . . . . . . . . . . . . . . . . . 150 6.4.1 Preservation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 150 6.4.2 Duration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 151 6.4.3 Consistency . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153 6.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 156 7. Spatial Diffusion in Early Middle English . . . . . . . . . . . . . . . . . . . . 157 7.1 Spatial diffusion in the three case studies . . . . . . . . . . . . . . . . . . . 157 7.1.1 Reduction of the number of grades in strong verbs . . . . 157 7.1.2 OE a ˉ > ME ǭ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159 7.1.3 Dual forms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 160 7.1.4 General observations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 161 7.2 Theoretical implications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163 7.3 Methodological implications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 169 Table of Contents IX <?page no="10"?> 8. Conclusion and Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 170 8.1 Summary of the individual findings from the three case studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 170 8.1.1 Reduction of the number of grades in strong verbs . . . . 170 8.1.2 OE a ˉ > ME ǭ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 172 8.1.3 The decline of the dual . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 174 8.2 Spatial diffusion in Middle English . . . . . . . . . . . . . . . . . . . . . . . . . 174 8.3 Outlook . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 176 9. Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 177 Appendix I - Distribution of Grade Vowel per Text File and Verb Class . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 191 Appendix II - List of Lexels with OE a ˉ in LAEME . . . . . . . . . . . . . . 204 Appendix III - Distribution of <a> and <o> per Text File . . . . . . . . 210 Appendix IV - Linguistic Maps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 215 Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 229 X Table of Contents <?page no="11"?> List of Black and White Maps Map 1.1 Main roads in Medieval England and Wales including rivers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16 Map 1.2 Transport costs to London c.1300 . . . . . . . . . . . . . . . . . . . . . . 17 Map 1.3 Diffusion patterns from London in Late Middle English according to the gravity model . . . . . . . . . . . . . . . . . . . . . . . . . 19 Map 4.4 LAEME Map 107 showing all forms of AFTER with <e> . . 91 Map 4.5 LAEME Map 105 showing forms of AFTER with <a> . . . . . 92 <?page no="12"?> List of Colour Maps Map 3.1 Basic map of all LAEME locations as used in this study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 215 Map 4.1 Past singular forms with the 2nd past grade vowel in verbs of Class V . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 216 Map 4.4 Past singular forms with the 2nd past grade vowel in verbs of Class IVa . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 217 Map 4.5 Past singular forms with the 2nd past grade vowel in verbs of Class II . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 218 Map 4.6 Past plural forms with the 1st past grade vowel in verbs of Class V . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 219 Map 4.7 Past plural forms with the 1st past grade vowel in verbs of class IVa . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 220 Map 4.8 Past plural forms with the 1st past grade vowel in verbs of class II . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 221 Map 5.1 Distribution of <a> and <o> in the LAEME CTT . . . . . . . . 222 Map 5.2 Distribution of <a> and <o> in the West Midlands per period (a) - (f) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 223 Map 5.3 Distribution of <a> and <o> in OE a ˉ + w (a) and OE a ˉ + ɣ (b) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 224 Map 5.4 Distribution of <a> and <o> in GHOST . . . . . . . . . . . . . . . . . . 224 Map 6.1 Spatial distribution of 1st person dual forms . . . . . . . . . . . . 225 Map 6.2 Spatial distribution of 2nd person dual . . . . . . . . . . . . . . . . . . 225 Map 6.3 Forms of the 2nd pl. pers. pron. with g-, ȝ and ᵹ - . . . . . . . 226 Map 6.4 Location of dual forms in LAEME. . . . . . . . . . . . . . . . . . . . . . 227 Map 6.5 LAEME text files with reinforcement of the dual through quantifiers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 227 Map 7.1 Unetymological past grade vowels in class V in the (South) East . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 228 Map 7.2 OE a ˉ > ME ǭ in the South-West Midlands . . . . . . . . . . . . . . 228 <?page no="13"?> List of Tables Table 2.1 Strong verbs from IE to OE verbs (based on Braune and Heidermanns 2004: §§ 29 - 36 and Brunner 1965: §§ 381 - 397) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26 Table 2.2 Strong verbs in Middle English based on Brunner (1963: § 69) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30 Table 2.3 Dual forms of the personal pronoun in the Germanic languages according to Seebold (1984) . . . . . . . . . . . . . . . . . . 51 Table 2.4 Dual forms recorded in Diehn (1901) and Kennedy (1915: 42 - 43) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56 Table 2.5 Dual forms in Middle English (Howe 1996: 138) . . . . . . . . 58 Table 3.1 Sample entry of processed LAEME data through a PHP-script . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73 Table 3.2 ‘ vSpt ’ forms not included in the present studies because of anomalies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75 Table 3.3 Strong verbs in Middle English based on Brunner (1963: § 69) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76 Table 3.4 LAEME Results for BOAT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78 Table 4.1 Frequency of each grade vowel per class in the 1st and 3rd past singular . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89 Table 4.2 The distribution of a and e as singular grade vowel of class V in MSS from the East Midlands . . . . . . . . . . . . . . . . . 93 Table 4.3 The temporal and spatial distribution of the plural grade vowel in the singular for class IIIc . . . . . . . . . . . . . . . . . . . . . . 96 Table 4.4 Frequency of each grade vowel per class in the past plural . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99 Table 4.5 Verb forms with the tag ‘ vSpt ’ that historically belong to the weak verbs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109 Table 5.1 General distribution of <a> and <o> among the LAEME text files in total and per word class . . . . . . . . . . . . . . . . . . . . . . . . . 112 Table 5.2 List of those LAEME text files which provide data for the change a ˉ > ǭ but which could not be given a more exact location (ordered by dating) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120 Table 5.3 Distribution of <a> and <o> of lexical items with OE a ˉ plus a non-lengthening consonant cluster . . . . . . . . . . . . . . . . . . . . . 124 Table 5.4 Distribution of <a> and <o> in text files that contain <oa> at least once . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132 Table 6.1 Total number of dual forms per LAEME text file . . . . . . . . 136 <?page no="14"?> Table 6.2 1st person dual forms in the LAEME CTT (P21% + D) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141 Table 6.3 2nd person dual forms in the LAEME CTT (P22% + D) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145 Table 6.4 Reinforcement of the dual through lexical quantifiers . . . . 155 XIV List of Tables <?page no="15"?> List of Figures Figure 5.1 Distribution of <a> and <o> in those 91 LAEME text files that provide data for all three word classes . . . . . . . . . . . . . . 129 Figure 5.2 Distribution of <a> and <o> categories in those 71 LAEME text files that provide data for all four frequency ranges . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131 Figure 7.1 Diachronic distribution of prose and verse texts that contain OE a ˉ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166 <?page no="16"?> List of Abbreviations Angl Anglian C Any consonant CTT Corpus of Tagged Texts DOE Dictionary of Old English EGmc East Germanic eME Early Middle English eLALME Electronic version of LALME eOE Early Old English Gk Greek Gmc Germanic Goth. Gothic HOL Homorganic cluster lengthening IE Indo-European L Latin LAEME Linguistic Atlas of Early Middle English LALME Linguistic Atlas of Late Mediaeval LP Linguistic profile Lith Lithuanian lOE Late Old English ME Middle English MED Middle English Dictionary ML Midlands ModE Modern English N Any nasal NWGmc North-West Germanic OE Old English OED Oxford English Dictionary OFris Old Frisian OHG Old High German OI Old Icelandic OLFrk Old Low Franconian ON Old Norse OS Old Saxon PGmc Proto-Germanic R Any resonant SHOCC Shortening before consonant clusters Skr Sanskrit SWML South-West Midlands TRISH Trisyllabic shortening V Any vowel WGmc West Germanic WS West Saxon <?page no="17"?> Acknowledgements This book would not have been possible without the constant support and encouragement of various people and this is the point to say a heartfelt thank you to everyone who accompanied me on this journey: First and foremost, I would like to thank my supervisor Daniel Schreier, who supported me through all these years with constructive feedback, motivating comments and, most importantly, much friendship and a never-ending sense of humor. I would also like to thank my second reviewer Olga Timofeeva, who arrived in Zurich just in time to offer me her expert opinion from a historical linguistic perspective and who gave me valuable input and stimulating advice for the published version of my thesis. Moreover, I would like to thank Meg Laing and Roger Lass, who read a very early research proposal for this thesis and encouraged me to go through with this no matter what the outcome would be, and David Britain, for early advice on some possible connections along the coast and his inspiring work on spatial diffusion in general. Many thanks also go to Annina Seiler Rübekeil and Dieter Studer-Joho, who read the first draft of this work with the keen interest of medieval scholars and provided valuable comments and suggestions, and to Jennifer Keller, who read the manuscript with the hawk eye of a native speaker before publication. Furthermore, I am grateful for Dieter Studer-Joho ’ s help in writing all the PHP scripts according to my needs and wishes. Invaluable support also came from my work colleagues and friends, who gave me insightful comments and support during numerous coffee and lunch breaks, above all Claudia Rathore (who also shared an office with me and accompanied me through every step of the project), Annina Seiler Rübekeil, Daniela Landert and Martin Mühlheim - one should never underestimate the value of social events and a good meal for the completion of a research project. I also wish to acknowledge the generous financial support of the Swiss National Science Foundation for this publication and I am grateful to the Narr Francke Attempto Verlag and to Andreas Fischer, Martin Heusser and Daniel Schreier, the editors of Schweizerische Anglistische Arbeiten, for giving me the opportunity to publish my work in their series. At the same time I would also like to thank Martin Heusser for doing such a wonderful job in designing the title page. Furthermore, the following publishers have kindly granted me their permission to reprint the maps: Taylor & Francis Books UK and Oxford University Press (Map 1.1), Cambridge University Press (Map 1.2), Versita (Map 1.3), as well as the Ordnance Survey and the compilers of the Linguistic Atlas of Early Middle English (Maps 4.4 and 4.5). <?page no="18"?> On a more personal level I would finally like to take the opportunity to thank my parents Linda and Hans Joho, who have always encouraged me to pursue my academic goals and who have always shown a keen interest in my work; my American host parents Sharon and Bob Rybak, who were most influential in arousing my interest in the English language; and, above all, my loving husband and soulmate Dieter Studer-Joho, with whom I have now shared more than half of my life and whose contributions to this work go way beyond the support already mentioned as he was always there for me whenever I needed him: I treasure every single moment spent with you and look forward to all the adventures that are still ahead of us. XVIII Acknowledgements <?page no="19"?> 1. Introduction The heterogeneity of the surviving Middle English written records provides valuable evidence of a range of different varieties that were spoken in a period famously characterized as “ par excellence, the dialectal phase of English ” (Strang 1970: § 127). This great variety of forms becomes immediately evident to everyone who browses through the contents of the Linguistic Atlas of Early Middle English (LAEME). 1 The very first entry in the so-called “ Tag Dictionary ” serves well to illustrate the type of variation that Middle English texts offer: (1) $&/ cj 2 ⁊ ⁊ + & A ANd Ad An An>d> And And+ Andd Ande Ant An n d Anð A[n]d Ā d Ā t a~ a~d End + ⁊ + +&\+ +and +and+ +end - ⁊ - -. ⁊ - -a+ -an- -and- -ande- -and\- -ant- - Ā t- >A>nd >a>nd >a>nt a a>nd>+ aant ad adn an an+ an>d> an>t> and and+ ande andt ant an[d] anð að a ˉ a ˉ >d> a ˉ d a ˉ n a ˉ t a~ end ent hand ond [ ⁊ ] []nd []nt æd ᵹ e A simple lexical item like the conjunction and is attested in 70 different forms. Some of the forms like adn and að could possibly be attributed to scribal errors, and other forms are unique because of the scribal habit of emending or correcting a text by inserting individual letters (indicated in LAEME by >). However, many of these forms show variation on less incidental levels. Quite frequently the conjunction is represented by the Tironian sign or the ampersand abbreviation, and many of the spelled out forms include upper case initials or nasal strokes. At the same time, many of the forms show spelling variants for the stem vowel or the word final dental consonant, which might be indicative of phonological variation. Most of the lexical items that are collected in LAEME exhibit a similar range of different scribal forms, which are worth closer scrutiny under the aspect of dialectal variation. Any scholar working on language variation in Middle English would agree with Hans Peters, who aptly notes that “ to a dialectologist, Middle English is at once a dream and a nightmare ” (1988: 397). It is a dream because “ our access to variation is direct ” (Milroy 1992: 156); i. e., the extant texts are as rich in variation as shown above because of the lack of a focused or standard variety at the time. This absence of any standard entailed the production of manuscripts 1 At the time of research LAEME was available in version 2.1 (2008 - , compiled by Margaret Laing). Unless stated otherwise this is the version referred to throughout this publication. 2 The tag conventions of LAEME will be explained in section 3.3; The plus and minus signs indicate whether there is a space (-) or no space (+) between two items in a manuscript. <?page no="20"?> written or copied in local varieties (Strang 1970: § 127; Smith 2007: 107 - 8). Furthermore, compared to the Old English period, there is much more textual evidence from the Middle English period. Even in the period right after the Norman Conquest, when English manuscripts are rather scarce, the situation is comparatively better. Horobin and Smith estimate that “ there is more written English surviving in manuscripts from the two centuries after the Conquest than in all the Anglo-Saxon centuries put together ” (2002: 27 - 28). Finally, it is also a dream because, in recent years, many research tools have become available to study this great linguistic variation, such as the Linguistic Atlas of Early Middle English (2008 - ), from which the examples above were taken. However, as Peters notes, this great variability in Middle English can also be a nightmare. Some texts are difficult to locate because information on both author and scribe, as well as their respective origins, is missing. Other texts contain socalled Mischsprachen, i. e. texts which are characterized by the “ persistent cooccurrence of dialect forms whose regional distributions are such that their geographical overlap cannot reasonably be supposed ” (Benskin and Laing 1981: 76). In other words, these are texts which contain a range of forms that are not expected to co-occur within a dialect continuum. Such Mischsprachen arise because of the textual tradition of the manuscripts; i. e., if a certain text is transcribed again and again, some variant forms of the scribe ’ s own inventory might slip in and this, of course, has implications for the localizability of the manuscripts (or parts of them). Other difficulties with Middle English dialectology include the fact that the surviving manuscripts are not evenly distributed across time and space (Laing and Lass 2006: 420 - 1). Again, the situation is not as bad as in the Old English period, but the general coverage of extant English manuscripts, especially for the Early Middle English period, could be better. Even though such problems are omnipresent, it would be a pity not to undertake studies in Middle English dialectology. The abundance of different forms and spellings is evident for the entire Middle English period; however, for anyone interested in language change, the Early Middle English period is of especially high interest, because many changes from Old to Middle English are still ongoing. Moreover, the above-mentioned recent publication of the LAEME provides an ideal tool to study such changes and variation in Early Middle English in general. In the present study I analyze three features whose change is most prominent in the Early Middle English period, and I explore their development across time and space. In the following section I discuss the language situation in Early Middle English before I continue with an overview of the advances in Middle English dialectology over the last hundred years that led up to the publication of LAEME. Section 1.3 looks into models of spatial diffusion and in the final section I present a detailed outline of the present study. 2 1. Introduction <?page no="21"?> 1.1 Early Middle English After the Norman Conquest of 1066 there was a transmission gap of English texts from the period between the end of the eleventh and the middle of the twelfth century, which is due to the Latin and French dominance of the period (Laing and Lass 2006: 419). The only English texts that survive from this period are a few Old English texts that were copied from pre-Conquest material. Latin was introduced as the standard language of government, the nobility spoke Norman French, and both Latin and French, “ the second language of culture ” (Laing 1993: 2), were used to write literary pieces. However, it is important to note that English was never abandoned as a spoken language by a great part of the population and had developed continuously even in the period from which we have no written records. 3 English generally reappears in writing in the middle of the twelfth century and Lass describes it as “ something resembling OE in some ways but quite different in others ” (2006: 59). Most importantly, the West-Saxon standard of the late Old English period had collapsed by then, so the English texts reappearing after the transmission gap show the great dialectal variation described above. Based on an admittedly small sample of medieval sources, Richter (1985) concludes that when the first Middle English texts emerged in the middle of the 12th century, the English aristocracy spoke English as their first language, so Latin and French were mostly confined to the written domain at the time. A similar view is expressed by Strang (1970: 217); Baugh and Cable (1993: 117) also mention the appearance of manuals to teach children French in the middle of the 13th century, which they interpret to mean that at least the children of some upper class families spoke English. The second continuation of the Peterborough Chronicle (MS Oxford, Bodleian Library, Laud Misc. 636) is commonly regarded as the first Middle English text. The end of the Early Middle English period is more difficult to delineate and several dates and events in the first half of the 14th century are mentioned. Some of the defining factors for an early and late period include the increase of French vocabulary after around 1300 (Fisiak 1968: § 0.2), the rise of “ a new national feeling ” because of the Hundred Years ’ War (Nevanlinna et al. 1993: 93), and the emergence of more local documents written in the vernacular (McIntosh et al. 1986: 3). Others simply count Chaucer ’ s birth in c.1343 to distinguish the Early and Late Middle English period (Horobin 3 It is generally assumed that the population at the time was somewhere between 1.5 and 2.5 million (Hinde 2003: 25). It is still a matter of debate as to how universal English was in the OE period - cf. for instance McWhorter (2009: ch9), Schrijver (2009), and the papers in Higham (2007) on the role of Celtic speakers and language shifters in Anglo-Saxon England. 1.1 Early Middle English 3 <?page no="22"?> and Smith 2002). Although it is useful to make the distinction between two periods in Middle English, it becomes evident from these different views that the division in different periods is always arbitrary (cf. Wright and Wright 1928: § 1; Fisiak 1994; Blake 1994; Lass 2000). Among the most important innovations that emerged in the transition period from Old to Middle English, there are certainly i) the loss of inflectional endings and grammatical gender and, as a consequence, ii) the change from a synthetic language to an analytical one, iii) some phonological developments like lengthening and weakening processes or iv) the enrichment in the vocabulary because of foreign influence, just to name a few. For many of these innovations we can only see the effects in whatever written sources have survived from the period, but not the individual stages of their development, because of the transmission gap of English texts in the transition period. Other changes continue or start in the Early Middle English period and their development is well attested in the extant manuscripts, such as the increase of nominal plural forms with -s (cf. Hotta 2009). Many of these changes are still under-studied at this stage, as most Middle English grammars (e. g. Wright and Wright 1928; Mossé 1952; Brunner 1963; Jordan 1974) were published or revised before the recent major projects on Middle English dialectology were available. It is, therefore, necessary to go back to some of these innovations and changes to confirm or revise what has been written in the standard grammars. The Early Middle English period is also of great interest because of the scribal practice at the time. When English re-emerged in the 12th century, the English scribes had to reorganize their familiar Old English, Old French, or Latin writing systems in order to write down their own variety of the language and, due to the lack of a written standard, various local systems emerged (Laing 2006: 6 - 12). This had the implication that the written records from the time provide local evidence for the ongoing linguistic changes: [T]he general policy of medieval scriptoria unquestionably consisted in the establishment of local spelling conventions on the basis of the spoken local dialects [. . .] [M]edieval English spelling, with the well-known exception of late West Saxon standardization, always showed flexibility and readiness to adjust itself to changing phonemic conditions. (Bauer 1986: 207) Also Smith notes that the surviving English texts from the period “ offer fascinating evidence for contemporary pronunciation ” (2007: 107). 4 It is likely that the relationship between writing and pronunciation was more intimate at 4 This contradicts earlier views which suggested that spelling in the Middle English period was “ traditional rather than phonemic ” (Scragg 1974: 206). In a more recent paper Scragg (1992) discusses the philological and linguistic value of spelling variants in the 11th century and redresses some of his earlier views. 4 1. Introduction <?page no="23"?> the time than today: Chaytor (1966: 19) points out that some medieval scribes were actually whispering the words in the process of copying. This means that in theory the surviving texts of the period provide the kind of evidence which in modern dialectological or sociolinguistic studies is elicited through interviews with participants from different areas. However, there is a big difference to the modern material: one has to rely on whatever material happened to survive in the written records and, unfortunately, the surviving texts from the Early Middle English period are unevenly distributed across time and space. Most extant manuscripts originate in the South and South Midlands and practically no early manuscripts exist from the North and North Midlands: Very little written Middle English survives from before the last quarter of the twelfth century. The West Midlands and to a lesser extent the East Midlands and the Southeast are reasonably well represented by texts from the thirteenth century. But almost nothing survives from the North or North Midlands from before 1300, nor from the extreme Southwest. (Laing and Lass 2006: 421) Smith and Horobin (2002: 28) argue that there are only few early texts from the North and the far South-West due to the poorer quality of the land in these areas. It is in the wealthier regions, such as East Anglia, the South-East, or the South-West Midlands in which most texts of the Early Middle period were written, according to Smith and Horobin “ with a very particular, local readership in mind ” (2002: 28). This is further support for the claim that the texts written in English during the period reflect the local variety of their composers. In a study on the Early Middle English period one has to rely on whatever sources have survived to the present day, and it is an unfortunate circumstance that the material is limited to just above a hundred manuscripts that are scattered across an area that includes present-day England. Nevertheless, the extant manuscripts offer a small but valuable treasure trove of texts written in the vernacular, which give a glimpse of various varieties and which enable research into language change across time and space right after the transmission gap of English texts between Old and Middle English. 1.2 Middle English dialectology This richness in variation among the Middle English manuscripts has aroused the interest of scholars for a long time, and this section gives an overview of the most important projects and subsequent publications on Middle English dialectology. There are a number of influential dialectological studies from the 1.2 Middle English dialectology 5 <?page no="24"?> first quarter of the 20th century (e. g. Ekwall 1913; Wyld 1913, 1914, 1921; Hulbert 1921; Serjeantson 1922, 1927). The results of these studies vary because of two main points of disagreement: On the one hand, the surveys included different linguistic features to study the variation and, on the other hand, there was a dispute on whether or not to include place-name evidence. As a consequence, there were discrepancies in the localization of manuscripts, depending on the exact methodology that was employed. A more detailed summary of the earliest developments in the field of Middle English dialectology is provided by Arngart (1949); in sections 1.2.1 - 1.2.4 I only focus on the most influential projects of the last hundred years. 1.2.1 Oakden (1930) and Moore, Meech, and Whitehall (1935) The 1930 s saw the publications of two important studies on Middle English dialectology (Oakden 1930; Moore, Meech, and Whitehall 1935), which for almost 30 years remained unchallenged. Oakden ’ s research is based on literary sources, historical documents and, to some extent, also place-names; however, he stresses the difference in the quality of these sources. What he calls “ genuine texts ” , i. e. the literary and documentary sources, are treated “ as of primary importance ” (1930: 9) while the place-names are only attributed a secondary status. Oakden then draws on an impressive list of 45 features (morphological and phonological features are roughly equal in number) to establish the dialectal boundaries which are presented on a fold-out map. Moore, Meech, and Whitehall ’ s project can be seen as a spin-off to the compilation of the Middle English Dictionary and they published their results as a “ preliminary report of an investigation based exclusively on localized texts and documents ” (1935: 1). Unfortunately, the project remained “ preliminary ” because Samuel Moore died while the article was in press. Moore, Meech, and Whitehall ’ s approach differs from Oakden ’ s in that they reject place-name evidence. Furthermore, they conducted extensive research for localized material before they started their dialectal survey, and they complemented the known material with unpublished documents from individual towns and boroughs from the period before 1450, if available. Moore, Meech, and Whitehall (1935: 8 - 18) discuss a total of eleven features (again, about equal in number with regard to morphology and phonology), which serve as the basis to mark dialect boundaries (1935: map 1) that distinguish ten different dialects (1935: map 2). Since Moore, Meech, and Whitehall did not include a very large amount of different texts and linguistic features, they were criticized as follows: [I]t was clear that their findings were both impoverished and distorted by the small number of texts analysed, the small number of criteria they employed, the discarding 6 1. Introduction <?page no="25"?> of much evidence, and, above all, the use of the written data simply as clues from which the phonemes and morphemes in the spoken language could be reconstructed [. . .] the work covered such a large time-span that no accurate distinction was possible between strictly diatopic (place-to-place) variation, and diachronic (time-to-time) change. (Strang 1970: 225) Oakden ’ s (1930) project was based on a much higher number of linguistic features, but his study also includes material from several periods, which blurs the differentiation between diatopic and diachronic variation. The results of both studies discussed in this section were published as isogloss maps, which fail to show any dialect continua or variant forms in single locations or areas. Nevertheless, both Oakden ’ s (1930) and Moore, Meech, and Whitehall ’ s (1935) contribution to the field of Middle English dialectology have to be recognized. The maps in Moore, Meech, and Whitehall (1935) were reproduced again and again, most notably by Mossé (1952), and as such they are still in the minds of many scholars working on Middle English. 1.2.2 The Survey of Middle English Dialects Some twenty years later two new projects were started, one in Edinburgh by M. L. Samuels and Angus McIntosh (LALME, cf. section 1.2.3) and one in Lund, Sweden by Gillis Kristensson. Kristensson (1967) takes a different approach than the projects discussed in 1.2.1. While Oakden (1935) treats place-names as inferior sources and Moore, Meech, and Whitehall (1935) disregard them completely, Kristensson bases his study entirely on the placenames and family names found in the Lay Subsidy Rolls (complemented by additional material from Durham and Cheshire, for which no Rolls have survived). This approach has the great advantage that the entire area of present-day England is covered and not just those areas with a higher density of surviving manuscripts. Kristensson started with the onomastic material from the North to make it comparable to the first publication of the Survey of English Dialects (Orton et al. 1962 - 1971), which also started with the six northern countries. The other four volumes of the Survey of Middle English Dialects were completed and published much later (Kristensson 1987, 1995, 2001, and 2002). Due to the nature of the onomastic evidence, Kristensson ’ s survey includes phonological features only. For each feature, Kristensson arranged the evidence according to the Old English or Old Norse form of a certain lexical item that contains the features (the individual parts of compounds are treated separately). The place-names and surnames are then listed under the heading of the county with a reference to the year that the document is dated to. These lists are followed by a short discussion each and, finally, some of the features 1.2 Middle English dialectology 7 <?page no="26"?> are presented in feature maps (like those in the SED) that take account of the individual forms. This is a great improvement from the former Middle English dialect maps: The maps reflect the fact that there are differences concerning the distribution of two variants of a variable with regard to different lexical items or with regard to the linguistic context of a variable (e. g. Kristensson 1967: 278, map 12 on different items containing OE a ˉ or ON á ). However, the many symbols sometimes make it difficult to read the maps. 1.2.3 The Linguistic Atlas of Late Mediaeval English Just at about the same time as Kristensson, M. L. Samuels and Angus McIntosh started their revolutionary project on Middle English dialectology in Edinburgh. Their Linguistic Atlas of Late Mediaeval English (McIntosh et al. 1986) includes Middle English texts from about 1325 - 1425 in the South and 1350 - 1450 in the greater parts of the Midlands and the North. Unlike Kristensson they use literary manuscripts and historical documents as source material, as they argue that “ any application of the full range of grammatical and lexical criteria requires a correspondingly full range of text-material in connected prose or verse to supply the answers ” (McIntosh et al. 1986: 4). Their approach is different from Oakden (1930) and Moore, Meech, and Whitehall (1935) insofar as they include many more literary texts and documents and also translations from texts that were originally composed in another dialect, which were dismissed by earlier scholars. Therefore, the LALME contains a higher multitude and a much greater coverage of texts than earlier projects. McIntosh (1963: 4) criticizes the small number of items and the large timespan of Oakden (1930) and Moore, Meech, and Whitehall (1935). McIntosh et al. (1986), therefore, collected their data with the help of questionnaires that include the high number of 280 items. The linguistic profiles (LPs) for each text that emerged from these questionnaires are published in Vol. III of LALME and the results for individual features are compiled in dot maps and item maps. There are 1200 dot maps in the atlas, which chart “ a form or set of forms collected in the LPs ” (McIntosh et al. 1987: 10). The system adopted includes three different sizes of dots to mark the relative frequency of a form in a LP. This was again an improvement from earlier maps, which did not include the dimension of how often a feature was recorded in a location. The item maps, on the other hand, “ display, at the assigned LP locations, all of the various regional forms corresponding to a given item ” (McIntosh et al. 1987: 10). This has the advantage that, at a certain location, all possible competing forms are immediately visible at first sight. These maps, which take up much space, were only completed for a number of sixty items. 8 1. Introduction <?page no="27"?> Not all the texts used for the atlas have a clear provenance, and those texts that could not be localized on the basis of extra-linguistic criteria were placed on the basis of the so-called ‘ fit-technique ’ , which was first introduced by McIntosh (1963). The texts with unknown provenance were fitted on the basis of linguistic features relative to each other within the matrix of the so-called ‘ anchor texts ’ , i. e. documentary texts for which the place of origin is known (McIntosh et al. 1986: 10; Benskin 1991). The method is described as follows by Laing and Lass (2008 - : 1/ 6) 5 : ‘ Fitting ’ is done by comparing, map by map, spellings particular to an unlocalised text with those already placed in the localised matrix. For each map, areas where those or similar spellings are not found are then eliminated, until (in the ideal case) only a single, well-defined location is left where the whole assemblage of spellings could plausibly occur. This method assumes that there are no clear-cut dialect boundaries but rather dialect continua from location to location of these anchor texts; so, by first creating maps of the distribution of distinct linguistic features in these anchor texts, it is possible to later narrow down the exact location of a hitherto unlocalizable text. 6 At an early stage of the project McIntosh was aware of the issues with this methodology of placing texts but redressed the concerns of possible critics as follows: We must be willing to acknowledge that the absolute position of these texts (i. e. their grid-reference in a narrow transcription as it were) may be uncertain. But this need not disturb us so long as we are persuaded that their position relative to one another is topologically sound. We must of course be prepared to make adjustments to the positions of such texts as new evidence about further neighbouring texts becomes available. (McIntosh 1963: 8, emphasis in original) Before the LALME was completed, McIntosh ’ s fit-technique was tested with the material from the Linguistic Survey of Scotland (cf. Benskin 1981). The names and addresses of informants were anonymized and their linguistic output was relocated by applying this fit-technique. The experiment was successful and always proved to place the informants of the study in a relatively small area around the places of their origin. 5 When quoting from the LAEME documentation available online, I use the format chapter/ page number. In other words, the reference 'Laing and Lass (2008 - : 1/ 6)' refers to chapter 1, page 6 of the pdf files found under the link 'Introduction' on the LAEME webpage: <http: / / www.lel.ed.ac.uk/ ihd/ laeme1/ laeme1_frames.html> (date of access November 2011). 6 A detailed introduction and discussion of the method is provided in Benskin (1991). 1.2 Middle English dialectology 9 <?page no="28"?> The LALME has been revised and was recently re-published by members of the Institute of Historical Dialectology in Edinburgh as an online edition referred to as the eLALME (Benskin, Laing, Karaiskos and Williamson 2013 - ). The online version includes the updated material from the printed edition (with the exception of the Early Middle English texts that are now part of LAEME), but it also allows scholars to generate user-defined maps or fit texts. 1.2.4 The Linguistic Atlas of Early Middle English After LALME was published in 1986, the assistants of the project, Margaret Laing and Keith Williamson, continued with their own projects. Williamson compiled the Linguistic Atlas of Older Scots (LAOS) (2008 - ) while Laing continued to work on the Early Middle English material, later joined by Roger Lass. Their Linguistic Atlas of Early Middle English was made openly available on a website in 2007 and is currently (October 2013) available as version 3.2. The basic aim of LAEME is to “ [map] the history of early Middle English through time and across space ” (Laing and Lass 2006: 430) and it covers the period from about 1150 until about 1325. The material is comparably scarce and LAEME consists of less than a tenth of the number of texts used for LALME (Laing and Lass 2008: - 1/ 9). The earliest text in LAEME (Laing and Lass 2006: 419) is the second continuation of the Peterborough Chronicle. The choice as to which text(s) should represent the latest Early Middle English evidence was less straightforward. The aim of LAEME was to build an atlas that should cover the period before LALME without any chronological gaps between the two atlases. Since there are so few texts from the North and North Midlands, the compilers chose to include the surviving texts from the first half of the 14th century from this area (Laing and Lass 2008 - : 1/ 5). This approach is not necessary for the South, as the LALME already includes some late-13thcentury texts, so Laing and Lass found c.1300 to be sufficient as the latest date for southern manuscripts. Nevertheless, they allowed the exception of Dan Michel ’ s Ayenbite of Inwyt, which was only completed in 1340 (cf. Laing 2008 - : 54). Laing and Lass (2008 - : 1/ 5) diverged from their usual practice for two reasons: First, the text is a rare case in which the author, origin, date of completion, and even the variety in which it was written are known (Dan Michel refers to it as the engliss of kent); second, Dan Michel mentions that he was an old man when he had completed the manuscript in 1340, so Laing and Lass argue that the language very likely represents the Kentish ME variety of the late-13th-century. LAEME, like its predecessor LALME, includes documentary and literary texts (both in verse and prose) and, again, those texts that have no clear provenance are localized on the basis of the fit-technique (cf. section 1.2.3.). 10 1. Introduction <?page no="29"?> There are fewer documentary texts than in the later Middle English period that could be used as anchor texts. This, of course, has to do with the fact that, in the Early Middle English period, many documents were written in Latin (cf. section 1.1). The Proclamation of Henry III (1258) is the notable exception, but other extant documents like wills, writs, or charters written in English are copies of Old English texts and were not always suitable for inclusion in the LAEME corpus (Laing 2000: 105). Therefore, the documentary texts were complemented with a few literary “ anchor ” texts. Such manuscripts were localized on the grounds of non-linguistic evidence like, for example, references to local places in a literary text (Laing and Lass 2008 - : 1/ 10). Furthermore, the texts were also compared to the data for Late Middle English and the much denser distribution of localizable texts in LALME helped to ascertain the LAEME locations (Laing and Lass 2006: 422). While LALME was compiled with the help of a questionnaire, this method was abandoned for LAEME, as the questionnaire would have become too large to handle and valuable data would have been lost (Laing and Lass 2006: 425). Instead, entire manuscripts or representative samples of longer texts were faithfully transcribed. The entire corpus of tagged texts now comprises 167 text files of various lengths; each lexical item in these texts is annotated with a lexico-grammatical tag so that it is possible to search the entire corpus electronically (see section 3.1). The website includes a few dot maps of some selected features (similar to those of LALME) and the latest release of LAEME (version 3.2, Laing 2013 - ) also includes a tool for user defined maps. At the time of research for the present study this option was not available yet. 1.3 Spatial diffusion All the projects described above aimed to map the areal distribution of certain features but with, the exception of LAEME, it is difficult to track the temporal and spatial development of a feature or change. I already mentioned above that the standard grammars on Middle English are all outdated, since they were published before the major Middle English dialect projects of Edinburgh and Lund. And indeed Mossé (1952), Brunner (1963), and Jordan (1974) rarely provide information on the spatial development of a form or a feature. Comments like the following are more common in general introductions and studies on Middle English: Almost everything new begins in the north and percolates down through the east midlands. The southwest midlands and the southeast remain the most conservative. (Lass 2006: 71) 1.3 Spatial diffusion 11 <?page no="30"?> The change [OE / ɑː / > ME / ɔː / ] begins in the south and works northward. (Brunner 1963: § 11.4) It has been noted for a long time that innovations in late Old English and Early Middle English spread from North to South. (Pilch 1997: 450) It is not clear in these rather broad descriptions how exactly the diffusion of innovations in Middle English operates: Do they spread gradually and continuously from north to south or vice versa as suggested by the quotations above? Are there any tendencies as to what path the diffusion of innovations in Middle English takes? By drawing on data from the LAEME (2008 - ) I will tackle such questions in the present study. The ‘ uniformitarian principle ’ claims that “ the same mechanisms which operated to produce the large-scale changes of the past may be observed operating in the current changes taking place around us ” (Labov 1972: 161). When applied to linguistics this suggests that “ the forces operating to produce linguistic change today are of the same kind and order of magnitude as those which operated five or ten thousand years ago ” (Labov 1972: 161). In other words, the diffusion of linguistic innovations, which can be observed and described in detail with data from modern varieties, might have operated similarly in earlier periods. Modern spatial diffusion models were already successfully applied to historical data (e. g. DeCamp 1958; Kitson 1992; Conde Silvestre and Hernández Campoy 2002; Hernández Campoy and Conde Silvestre 2005; Bergs 2005, 2006; Nevalainen and Raumolin-Brunberg 2003), but so far similar studies which concentrate on the Early Middle English material are missing. In the following section I present different types of modern models of spatial diffusion in order to establish the patterns that are observed in the modern data. In section 1.3.2 I discuss possible applications to the Early Middle English data by focusing on the studies by Bergs (2006) and Conde Silvestre and Hernández Campoy (2002), who apply different spatial diffusion models to the Late Middle English period. 1.3.1 Models of spatial diffusion Languages may change because people interact on a daily basis with individuals from elsewhere and through this contact people may adopt certain features that are realized differently in the other location. Britain (2005 b: 996 - 997) discusses four main models for the spatial diffusion of linguistic innovations: - ‘ wave ’ or ‘ contagion ’ diffusion (Trudgill 1986; Bailey et al. 1993; Britain 2002, 2005 a) 12 1. Introduction <?page no="31"?> - ‘ urban hierarchical ’ diffusion (Trudgill 1974, 1983, 1986; Callary 1975; Gerritsen and Jansen 1980; Bailey et al. 1993; Wikle and Bailey 1997; Hernández Campoy 1999, 2003; Britain 2005 a) - ‘ contra-hierarchical ’ diffusion (Trudgill 1986; Bailey et al. 1993; Wikle and Bailey 1997) - ‘ cultural hearth ’ diffusion (Horvath and Horvath 1997, 2001, 2002) The ‘ wave model ’ , which ultimately goes back to Schmidt (1872), proposes that innovations diffuse continuously from one center outwards. It is often compared to the ripple effect caused by a pebble or a raindrop falling into water (e. g. Britain 2002, 2005 b). The diffusion of an innovation very rarely takes place in these ideal concentric circles. More often than not there will be a compression on one side, for instance because of a language border or a natural border that prevents the change from spreading any further or slows it down drastically (cf. Haas 2010: 656). In its basic theory, the wave model is certainly the most simple of all diffusion models and modern research shows that the changes usually spread in a more complex way. The ‘ urban hierarchical model ’ takes into account that the size of a town might be an important factor in the transmission of the change. This model was developed by Trudgill (1974), who refined the wave model and applied the ‘ gravity model ’ , which originally was developed in the field of social geography, to the diffusion of linguistic innovation. His formula includes the population size of two places (i and j) and their distance in order to predict the spread of a linguistic feature: (2) M ij ¼ P i P j ð d ij Þ 2 (Trudgill 1974: 233) M = interaction; P = population size; d = distance; i,j = place The model postulates that “ innovations leap from one place, usually a city, to another place, another city or large town, and then move into the places between, such as towns and villages ” (Chambers and Trudgill 1998: 166). This model was successfully applied again and again, but there are several studies which contradict it (cf. discussion in Britain 2002; Nerbonne and Heringa 2007). Bailey et al. (1993) and Wikle and Bailey (1997) noticed that the opposite was possible as well: their ‘ contra-hierarchical model ’ describes how some changes operate in the other direction, starting in a rural area and moving to towns and cities. It seems, however, that such changes are rather rare. Finally, there is the so-called ‘ cultural hearth model ’ . This model is a combination of wave and hierarchy model and suggests that “ the innovation 1.3 Spatial diffusion 13 <?page no="32"?> gains a foothold in both town and country in one particular region before diffusing to other parts of the country ” (Britain 2005 b: 997). Horvath and Horvath (1997, 2001, 2002) studied / l/ vocalization in Australian and New Zealand English and describe how innovations first gained ground in New Zealand before spreading to Australian towns such as Adelaide and Mount Gambier, which are not the most important ones. It seems that, depending on the variable under investigation, the innovation might spread along different lines. Furthermore, with reference to the spatial distribution of a feature it is also necessary to address questions like the ones proposed by Britain (2005 b: 997): - Is the innovation linguistically identical in its new resting place as it was where it began, with the same social and stylistic status? - Are the linguistic constraints on the variability the same? - Is the traditional form structurally compatible with the change being promoted? - What are the linguistic consequences of the contest between the innovation and the traditional form? Victory for the innovation? Survival of the traditional form? Or a structural compromise between the two? Such questions try to eliminate some of the criticism that the purely spatial model of diffusion had to face. Britain (2005 a: 39) refers to the human geographer Gregory (1985), who raised the following issues: the model does not take into account any social factors in which the innovation is embedded and it puts the focus on the diffusion of the innovation itself rather than on the speakers in a certain place who might adopt or reject an innovation (cf. also Mitzka 1952: 164). By complementing the linguistics analysis with a discussion of such questions that take into account also social and attitudinal data, a much clearer picture emerges on the diffusion of a certain change. 1.3.2 Spatial diffusion in Middle English? The adaptation of any of the diffusion models just mentioned is certainly difficult with regard to Early Middle English, as one has to deal with completely different data sets than those available to sociolinguists examining contemporary spoken language (cf. discussion in Britain 2012). Most importantly, the network of “ informants ” , i. e. the manuscripts in the Middle English period, is less dense and cannot simply be expanded if more data is needed. Nevertheless, I believe that it is useful to draw on such contemporary diffusion models to investigate language change in Early Middle English and, if necessary, adapt or redefine them for these earlier language stages and also 14 1. Introduction <?page no="33"?> take period specific factors into account. This view is supported by the findings of Nevalainen and Raumolin-Brunberg ’ s study on language change in Tudor and Stuart English: “ [I]t is not possible to use the present to explain the past without access to knowledge of past societies in their own right ” (2003: 202). Bergs (2006) also reaches the conclusion that period specific factors have to be included. He takes an approach similar to the present one by testing modern diffusion models with data from LALME (McIntosh et al. 1986). Following the methodology presented in Trudgill (1974: 227 - 228) and Chambers and Trudgill (1998: 176 - 178), Bergs superimposed a honeycomb structure onto the LALME item maps 8 THEM (6) and 9 THEIR (6), both showing the results for East Anglia plus Kent. For each cell, Bergs then divided the number of recorded innovative <th-> forms by the number of traditional <h-> forms. There is a high concentration of <th-> forms in the north around the Wash (from where the innovation spread), around Norwich and in London. Bergs summarizes his findings as follows: “ There appear to be distinctive lines running to and from London northwards, southwards (the coast), and in a north-easterly direction, roughly where Norwich must be ” (2006: 17). He points out that this pattern interestingly coincides with the system of roads and water ways at the time (cf. Map 1.1). Bergs (2006) shows how innovations spread along communication routes like roads and water-ways by what he refers to as the ‘ human thread ’ , i. e. speakers who travel along these main routes and carry the linguistic innovations with them. This is very similar to modern wave models, except that the innovations spread in a star-like fashion from a central location like London along communication routes rather than in concentric circles around a certain location. On Berg ’ s maps (2006: 18 and 19) there are no recorded <th-> forms in a small stretch north of London; however, there is a high concentration of forms in some of the hexagons along the coast. Bergs does not comment on these single concentrations; however, they actually confirm his model of the human thread, as this pattern along the coast coincides with the seaway. A map on transport costs of wheat to London in c.1300 (Keene 2006: 96; cf. Map 1.2) shows that it was much cheaper to transport goods by water along the coast rather than by land from north to south. 1.3 Spatial diffusion 15 <?page no="34"?> Map 1.1: Main roads in Medieval England and Wales including rivers (Griffiths 1996: 183; reprinted in Bergs 2005: 89 and Bergs 2006: 2; based on a map published in English Historical Documents 1189 - 1327, Harry Rothwell © 1975 Eyre & Spottiswoode; reproduced by permission of Taylor & Francis Books UK and Oxford University Press.). 16 1. Introduction <?page no="35"?> Map 1.2: Transport costs to London c.1300 (originally published in Derek Keene. 2000. “ Metropolitan values: migration, mobility and cultural norms, London 1100 - 1700, ” in Laura Wright (ed.). The Development of Standard English, 1300 - 1800. Cambridge: University Press; derived from Campbell et al. (1993: Fig. 7); reproduced by permission of Cambridge University Press). 1.3 Spatial diffusion 17 <?page no="36"?> The different ways of transportation are also taken into consideration by Conde Silvestre and Hernández Campoy (2002) in their study on the applicability of the gravity model to Late Middle English data. Conde Silvestre and Hernández Campoy ’ s methodology differs from Bergs (2006), as they only divide England into nine smaller units corresponding to the Middle English dialect areas as presented in Moore, Meech, and Whitehall (1935): London, East Anglia, the South-East, the South-West, the South-East Midlands, the North-East Midlands, the South-West Midlands, the North-West Midlands and the North. In a next step they gathered the population size of these areas, including separate figures for the larger towns based on Russell ’ s (1949) projection from the poll tax returns in the 14th century. Furthermore, they also added the following factors to calculate the ‘ population potential ’ of a town: seaport (2), primary river (1.7), secondary river (1.5), main road (1.6) and pilgrimage route (1.3) (Conde Silvestre and Hernández Campoy 2002: 161). The figures in brackets refer to the factor weight of a particular location and it is evident that, in Conde Silvestre and Hernández Campoy ’ s model, the importance of seaports and river towns is emphasized. The eighteen towns with the highest potential are then used to calculate the total ‘ Interaction Potential Index (IPI) ’ for each center, which is defined as “ the addition of its different individual interaction potential indexes, obtained using the gravity model formula, with respect to the rest of centres ” (Conde Silvestre and Hernández Campoy 2002: 165). 7 This generates London with the highest percentage, followed by Coventry, Bristol, and King ’ s Lynn or in terms of Middle English dialect areas: London followed by the South-West Midlands and East Anglia. The last step was to include the factor of linguistic similarity. They considered a set of four vowels and six morphological variables typical of late medieval London and calculated whether a feature was present or absent at a certain location. By means of these factors it was possible to calculate the potential patterns of diffusion from London to other areas according to the gravity model (cf. Map 1.3). While both approaches are very intriguing, they both have their disadvantages. In Bergs ’ s (2006) approach there are possible issues with superimposing a honeycomb structure on a set linguistic map, as the size of the hexagons or their respective locations might strongly influence the results. In other words, if the honeycomb structure is moved slightly to any side, it might include different data per cell, which results in different frequencies. Trudgill ’ s 7 They do not use Trudgill's original formula here, but a refined version, which puts more emphasis on the population density of the influencing center: M ij ¼ K P i P j D 2 ij P j P i þ P j (Hernández Campoy 1999: 18). 18 1. Introduction <?page no="37"?> (1974: 227) method worked the other way around. He superimposed the honeycomb structure on a normal map and then randomly selected one place within each hexagon to collect the data. His original method ensures that the amount of data is balanced across the entire area, but the issue with the size of the hexagons or their exact location remains. Map 1.3: Diffusion patterns from London in Late Middle English according to the gravity model (Conde Silvestre and Hernández Campoy 2002: 171; reproduced from Studia Anglica Posnaniensia, 2002, 38, p. 171 by permission of Versita). Conde Silvestre and Hernández Campoy ’ s approach, on the other hand, is problematic because it does not take into account any type of variation within the dialect areas established by Moore, Meech, and Whitehall (1935). Conde Silvestre and Hernández Campoy (2002: 168) present a table with the variants that are attested for a given linguistic variable in the Late Middle English 1.3 Spatial diffusion 19 <?page no="38"?> varieties but, with very few exceptions, it only lists one variant per variety. I already outlined above in section 1.2.1 some of the issues of these early dialectological studies. The maps presented in the LALME (McIntosh et al. 1986) show that the distribution of features in Late Middle English is definitively not as clear-cut as Conde Silvestre and Hernández Campoy (2002: 168) suggest. Nevertheless, both approaches show that if period-specific factors are included, it is possible to apply modern models of spatial diffusion to Middle English data. The surviving material as presented in LAEME is more limited in its scope than the material from the Late Middle English period and, therefore, it is yet to be determined whether similar patterns can also be discerned in the Early Middle English period. 1.4 Outline of the present study 1.4.1 Aim and scope The principle aim of the present study is to analyze diffusion patterns in Early Middle English and for this reason I present three case studies on two morphological and one phonological variable that should help to answer the questions raised in section 1.3. I draw on modern diffusion models and, if necessary, refine them for historical purposes to determine whether diffusion patterns can be recognized in the Early Middle English period at all and, if so, whether they are similar to those described by Conde Silvestre and Hernández Campoy (2002) or Bergs (2006) on the basis of data from LALME. Furthermore, I want to assess whether the same questions as those presented in Britain (2005 b) are useful for the purpose of discussing the historical data or whether they have to be adapted. There is, however, an important caveat that Bergs (2006) does not mention: Many texts in LALME and LAEME were localized through the fit-technique, i. e. only relatively to each other and on the assumption that there is a dialect continuum between two varieties (cf. section 1.2.4). It is, therefore, important not to fall prey to circular reasoning; i. e., instead of only discussing the diffusion patterns that possibly emerge, one also has to turn around the question: On the basis of modern diffusion models, is it possible to confirm the location of those texts in LAEME that are localized through the fit-technique? The analysis might reveal that some LAEME locations have to be moved and localized elsewhere on the basis of new evidence. The study also focuses on the question whether the data from LAEME sheds new light on the variables. Preliminary studies conducted with the 20 1. Introduction <?page no="39"?> corpus indicate that traditional views on Middle English dialectology will have to be re-examined. For instance, Lass and Laing (2005) noticed that Middle English developments of OE / y(: )/ cannot be neatly divided into three clearcut areas with western / y/ , northern/ eastern / i/ , and south-eastern / e/ as is usually presented in traditional handbooks. It appears that a few lexical items, such as OE cyning, always show the grapheme <i> in manuscripts of areas where another reflex would be expected. Moreover, in a later paper on the same problem they state that “ [a] preliminary investigation of putative front rounded vowel reflexes in the SWML does not cohere with the handbook consensus ” (Lass and Laing 2005: 281). Although these are only two examples, it is evident that, through research with LAEME, new insight into Early Middle English dialectology is possible. Therefore, it is a further aim of this study to test whether some the traditional accounts of the variables analyzed have to be revised in the standard grammars. 1.4.2 Structure Chapter 2 presents the variables analyzed in the present study and justifies their choice. The case studies include i) the reduction of strong verbs from four to three grades, ii) the change OE a ˉ > ME ǭ and, finally, iii) the decline of the dual personal pronouns. On the one hand, I explain why I think that these variables are especially suitable for my analysis, and, on the other hand, I present their historical development until the Middle English period. I pay special attention to previous work, if available, on the diatopic variation of the changes. Chapter 3 looks into the method employed in this study and explains how the data is retrieved and how it is turned into the maps. The data for the present study is drawn from the LAEME Corpus of Tagged Texts (CTT). This choice of material has several reasons: First, the LAEME CTT contains transcripts of most surviving texts in English from the Early Middle English period or at least substantial samples, if the texts are too long. Second, the system of tags used in the corpus renders the material easily accessible to anyone working with the corpus, i. e. exact queries corresponding to the three variables under investigation allow the retrieval of large amounts of data for further analysis. Third, all localizable texts or parts of texts are matched with a six-digit national grid reference, and each text is dated within a system of 25 years, which makes it possible to produce detailed maps and to track developments across time and space. Nevertheless, it is still necessary to evaluate the material available through this corpus and to discuss the consequences that some of the points addressed might have for the methodology employed in the present study. For instance, an important issue in this respect 1.4 Outline of the present study 21 <?page no="40"?> is the applicability of the approach taken in the two studies on diffusion in Late Middle English as presented in section 1.3.2. Chapters 4 to 6 illustrate the results for the individual variables. They are presented by means of symbol maps that include a time and relative frequency dimension (cf. section 3.5.2). Since more texts have survived from the East and West Midlands, the focus is on these two general areas, but I also present the results from the North and South, if possible. Furthermore, because the coverage in Early Middle English is patchier than in later stages, Laing argues that “ early Middle English dialects cannot be studied in isolation from later Middle English dialects ” (2003: 118). Therefore, I complement my data if necessary with data from LALME or other studies. In chapters 4 to 6 I also discuss any individual results that emerged from these case studies. The analytical chapters are then followed by chapter 7, where I draw on the results from the individual case studies in order to make general claims and observations on diffusion patterns in Early Middle English and to answer the research questions stated in section 1.4.1. Finally, the conclusion in chapter 8 summarizes the most important findings from this study. 22 1. Introduction <?page no="41"?> 2. Historical Background and Previous Work This chapter looks into the historical background of the three variables that are analyzed in the present study. As explained in the introduction, they include the reduction of grades of strong verbs from four to three, variation of <a> and <o> as reflexes of OE a ˉ (< Gmc ai ) 1 and, finally, the decline of dual forms of the personal pronoun. In choosing three different variables, which will be discussed both individually and in relation to each other, I follow Bergs ’ s approach in his sociolinguistic study on the Paston letters: The comparative perspective is used to demonstrate in how far exactly every single variable requires its own framework, methodology, and approach. But not only does every variable per se require different analytical tools, the specific batches of results also call for specific interpretations. (Bergs 2005: 3) There are several reasons why the choice designated the three variables mentioned above. Most importantly, the three variables are all attested in the earliest Middle English manuscripts and are highly variable in the period under scrutiny. Furthermore, in order to base the research project on firmer grounds, I decided to include both morphological and phonological features, in order to assess whether they demand other methodological or theoretical considerations. Because of the exploratory nature of the present study and in order to detect possible differences in the transmission of the innovations, the three variables also complement each other in other respects. First of all, while items containing OE a ˉ are relatively frequent and likely to be attested in any text, strong past forms and dual forms only occur in certain enviroments and are thus comparably rare. The study of low-frequency variables might provide insights that cannot be gained from the analysis of a high-frequency variable. Second, the dual as a receding feature complements the two Early Middle English innovations that are analyzed in this research project, as it might shed new light on the path the demise of a particular feature takes. At the same time, the appearance of an almost obsolete feature in a text file may be indicative of archaic language and/ or of the transmission history of a text, which might be of importance in the discussion of the results from the other two case studies. Third, while the reduction of the past grade vowels is said to spread from 1 In this study I present the vowels in the traditional 'philological' notation in italics and with macrons to indicate length, a practice which is also adopted by Liebl (2002). If a more exact notation of vowel quality is needed, I rely on phonemic transcriptions in slant brackets. <?page no="42"?> North to South (Lass 1994 b: 88; Markus 1990: 152), exactly the opposite is the case for the change OE a ˉ > ME ǭ : Brunner (1963: § 11.4), among others, mentions that the change OE a ˉ > ME ǭ starts in the South and then diffuses northwards. In addition, so far nothing is known about the spatial dimensions of the surviving dual forms in Early Middle English. Finally, I believe that the vast amount of data now made available through the LAEME CTT demands a re-examination of any variable, even if it includes such a well-studied change like OE a ˉ > ME ǭ that has been tackled by numerous scholars (most notably Ekwall 1938; Kristensson 1965; Dietz 1978, 1989; Liebl 2002, 2006, 2008). Including a low-frequency variable like the dual personal pronouns allows the assessment of the reliability of the LAEME CTT for a very rare variable. This is necessary as not all texts are transcribed in their entity (Laing and Lass 2008 - : 3/ 2). The three variables analyzed in this study not only require different methodologies and interpretations, each of them also has its individual historical background. In this chapter, the focus of the section on the strong verbs lies on the developments in the past tense formation of strong verbs at the beginning of the Middle English period; with regard to the change a ˉ > ǭ , it is necessary to highlight the sources for the phoneme in Old English, and, finally, since the decline of the dual is a pan-Germanic phenomenon, I extend my survey to other Germanic languages. Furthermore, in all three sections I discuss previous work on the spatial and temporal distribution of the three variables in (Early) Middle English. 2.1 Reduction of the number of grades in strong verbs One of the main features that separate the Germanic language family from other Indo-European languages is the classification of its verbs into a ‘ strong ’ and ‘ weak ’ 2 group based on the two distinct ways of the past and past participle marking: the weak verbs form their past and their past participle by adding a dental suffix, while in the class of the strong verbs the past forms are distinguished by ablaut or reduplication (Lass 1994 a: 151 - 177). The Germanic strong verbs have four ‘ principal parts ’ , i. e. “ basic grades from which the rest of the paradigms can be constructed ” (Lass 1994 a: 153): present, 1st past (1st and 3rd sg.), 2nd past (2nd sg. and pl.) and past participle. This means that, in the classes of strong verbs, there are up to four separate vowel grades as illustrated in (3) by OE be ˉ odan, a strong verb of class II: 2 The strong/ weak distinction goes back to Grimm (1819). 24 2. Historical Background and Previous Work <?page no="43"?> (3) Present 1st Past 2nd Past Past Participle OE e ˉ o e ˉ a u o be ˉ odan be ˉ ad budon boden While this system was relatively stable during the Old English period, the Middle English period saw a series of drastic changes: Some strong verbs shifted to the weak class (cf. Krygier 1994 a); others showed mixed conjugation and, finally, some reduced the system of four grades in Old English to “ no more than three grades per verb ” (Lass 2006: 77) in ME. In this study I am only interested in the third scenario. If the number of grade vowels was reduced, there was usually a merger of the 1st and 2nd past grade vowel, which started in the Early Middle English period (occasionally also the past participle vowel was extended to one of the past vowels). This merger is of interest for the present study, as there is dialectal variation with regard to the reduction of the number of grades: In strong verbs there is a tendency in the South-West and the southern part of the west Midlands for the vowel of the preterite plural or the past participle to spread by analogy to the preterite singular. This is to be contrasted with the practice in Northern dialects, where analogy operates in the other direction. (Brook 1963: 72) Moreover, it appears that there is not only dialectal variation but also diffusion: The northern leveling is reported to have spread southwards in the course of the Middle English period (Brunner 1963: § 69.1; Markus 1990: 152; Lass 1994 b: 88). In this study, I examine whether the rather general claims presented above can be verified and whether there are specific spatial diffusion patterns with regard to this merger. Additionally, I discuss whether some classes of strong verbs are more likely to be affected by this process. While the merger is mentioned in the general handbooks and grammars, a detailed study on this type of simplification of strong verbs through grade vowel mergers is missing so far for the Early Middle English period. Krygier (1994 a) carried out a research project on strong verbs that shifted to the weak class and his findings will be presented in detail (cf. section 2.1.2.1), as similar factors might have triggered the reduction of grades in the system of strong verbs in Middle English. In the following sections I present the development of strong verbs until the Old English period and I discuss the possible scenarios that affected strong verbs in Middle English: shift to the weak group of verbs, hybridization and, most importantly for this study, the reduction of grades by way of extending one past grade vowel to the other. 2.1 Reduction of the number of grades in strong verbs 25 <?page no="44"?> 2.1.1 The classes of strong verbs up until the Old English period Traditionally, strong verbs are divided into six classes according to their ablaut series plus a seventh, originally reduplicating, class. Table 2.1 summarizes the development of strong verbs until the Old English period. While synchronically this looks like a neat system of seven classes that are linked through vowel alternation for past tense and past participle marking, it is necessary to distinguish classes I to V from class VI, on the one hand, and from class VII, on the other hand, because of their different origins. Table 2.1: Strong verbs from IE to OE verbs (based on Braune and Heidermanns 2004: §§ 29 - 36 and Brunner 1965: §§ 381 - 397). Present 1st Past 2nd Past Past Participle I IE e + i ̯ + C IE eiC oiC iC iC Gmc ı ˉ C aiC iC iC OE ı ˉ C a ˉ C iC iC wr ı ˉ tan wra ˉ t writon writen II IE e + u ̯ + C IE euC ouC uC uC Gmc eu/ iuC auC uC uC OE e ˉ o/ u ˉ C e ˉ aC uC oC ce ˉ osan ce ˉ as curon coren III IE e + R + C IE eRC oRC R ̥ C R ̥ C Gmc iRC aRC uRC uRC OE iRC aRC uRC uRC singan sang sungon sungen IV IE e + R IE eR oR e ˉ R R ̥ Gmc eR aR e ˉ R uR OE eR æR WS æ ˉ R/ NonWS e ˉ oR stelan stæl stæ ˉ lon stolen V IE e + C ( ≠ R) IE eC oC e ˉ C eC Gmc eC aC e ˉ C eC OE eC æC WS æ ˉ C/ NonWS e ˉ eC sprecan spræc spræ ˉ con sprecen 26 2. Historical Background and Previous Work <?page no="45"?> Present 1st Past 2nd Past Past Participle VI IE a, o or ə + C IE a, o, ə a ˉ / o ˉ a ˉ / o ˉ a, o, ə Gmc a o ˉ o ˉ a OE a o ˉ o ˉ a bacan bo ˉ c bo ˉ con bacen VII IE V + C IE V reduplication reduplication V Gmc V reduplication reduplication V OE a ˉ e ˉ e ˉ a ˉ ha ˉ tan he ˉ t he ˉ ton ha ˉ ten o ˉ e ˉ o e ˉ o o ˉ flo ˉ wan fle ˉ ow fle ˉ owon flo ˉ wen C = any consonant; V = any vowel; R = any resonant; R ̥ = any syllabic resonant The ablaut series of classes I - Vall reflect the Indo-European e/ o ablaut pattern (cf. Lass 1994 a: 154): (4) Present 1st Past 2nd Past Past Participle IE e-grade o-grade zero-grade zero-grade The distinction into five separate classes is based on their root structure (Bammesberger 1986; Lass 1994 a: 155 - 156; Braune and Heidermanns 2004: §§ 29 - 34). Classes I, II, and III all contain IE e plus a cluster of two consonants or a semi-vowel plus consonant; i. e., ei ̯ C, eu ̯ C and eRC, respectively (cf. Table 2.1). The different qualities in the grade vowels of these three classes in IE, Gmc, and OE simply reflect sound changes that were conditioned by the respective semi-vowel or consonant clusters. The stem vowels of classes IV and V, on the other hand, are followed by a single consonant and their root structure is IE eC ( ≠ R) and eR (cf. Table 2.1). They are also regular in a sense and, at least in the present and 1st past stem, they follow the pattern of classes I to III; however, they differ from the first three classes due to unexpected grade vowels that appear in the 2nd past and past participle forms: (5) Present 1st Past 2nd Past Past Participle I-III e-grade o-grade zero-grade zero-grade IV e-grade o-grade lengthened e-grade zero-grade V e-grade o-grade lengthened e-grade e-grade 2.1 Reduction of the number of grades in strong verbs 27 <?page no="46"?> These forms cannot be explained by regular developments and their origin is still disputed. Bammesberger (1986: 56) and Braune and Heidermanns (2004: § 34, Anm. 2) assume that this inconsistency started in class V in order to avoid clusters like *gb-, which were difficult to pronounce word initially. Other scholars try to find the solution to the problem in Gmc *e ˉ tan or other verbs with word initial e- (e. g. Kortlandt 1992; Bammesberger 1996; Mottausch 2000). Such verbs have a salient syllable structure as they contain a word initial vowel. Mottausch (2000: 54), for instance, assumes that the lengthened e-grade resulted from contraction of the root vowel and *e-, which marked the IE perfect through reduplication of the first syllable. This view has recently been supported by Mailhammer (2007: 67 - 86). Both Mottausch and Mailhammer argue that the e-grade in the past participle appears by analogy to the present stem. It is generally accepted, however, that the lengthened e-grade in the 2nd past of class IV formed by analogy to class V (Mottausch 2000: 45; Braune and Reiffenstein 2004: § 339), as there is evidence that the pattern must have originally developed as regularly as classes I, II, and III. 3 This is apparent in the forms of the preterite-present verbs that correspond to class IV, which show the same ablaut pattern as the strong class III: (6) Present 1st Past 2nd Past Past Participle Strong III helpan healp/ halp hulpon holpen Present sg. Present pl. Pret.-pres. IV sceal/ scal sculon Unfortunately, no preterite-present verbs are attested from class V, so it is not possible to tell whether they originally also followed the regular IE pattern (cf. Bammesberger 1986: 72). The verbs of class VI also distinguish their past forms from the present by vowel alternation; however, the gradation series of class VI cannot be traced back directly to the same Indo-European ablaut patterns. The vowel of the present stem is a (< IE a, o, ǝ ) in all Germanic languages and, because many verbs in this class are not attested in all IE languages, Meid refers to the “ western secondary character ” (1971: 56, my translation) of this category. The only vowel distinction in this class is a for the present forms and the past participle and o ˉ for the past forms (cf. Table 2.1). Again, the origin of this new pattern with lengthened past forms is highly disputed; a comprehensive discussion of the current state of research on the origin of class VI is provided in Mailhammer (2007: 89 - 103). 3 Mailhammer (2007: 83) further assumes an “ underlying unity of classes IV and V ” . 28 2. Historical Background and Previous Work <?page no="47"?> Verbs of class VII initially formed their past tense by reduplication of the first syllable (Brunner 1965: §§ 393 - 399; Hogg and Fulk 2011: § 6.69) and it is still disputed whether the reduplicating verbs are a Germanic innovation (cf. Meid 1971: 69 - 81) or go back to an older class (cf. Bammesberger 1986: 60; Kortlandt 1992: 101). Although some traces of reduplication are still evident in Old English, Old Norse, and Old High German (cf. Bech 1969; Fulk 1987; Van Coetsem 1990; Kortlandt 1991; Mottausch 1998), Gothic is the only Germanic language that fully preserves reduplication as past tense marker (Brunner 1965: § 393 - 4; Braune and Heidermanns 2004: § 178): (7) Present 1st Past 2nd Past Past Participle Goth. haita haíhait haíhaitum haitans Like class VI, the verbs of class VII only distinguish two grades in OE: one vowel for the present and past participle and one vowel for both past grades. The vowel of the present varies in Old English but in the past it is either e ˉ or e ˉ o (cf. Table 2.1). As the present study does not include those cases in which the past participle vowel merges with the 1st or 2nd past tense vowels of the ablaut series, I do not discuss classes VI and VII any further. All classes of strong verbs in Old English contain verbs which divert from the regular ablaut patterns for various reasons but mostly due to sound changes that are conditioned by the respective environments (cf. Campbell 1959; Brunner 1965; Hogg and Fulk 2011). Furthermore, there are a few verbs which shift to other strong classes; these are discussed in more detail in section 2.1.2.2. 2.1.2 Strong verbs in Middle English According to the standard grammars, the ablaut patterns still seem to be intact, at least at the beginning of the Middle English period; e. g., Krygier (1994 b: 68) concludes, on the basis of a study on the strong verb system in the Peterborough Chronicle, that “ one can safely assume that the OE strong verb system did not undergo any significant transformation in the transition period to Middle English. ” Table 2.2 summarizes the gradation series of the strong verbs in Middle English, which differs from Old English only on account of regular sound changes that affected the vowels in the transition period: 2.1 Reduction of the number of grades in strong verbs 29 <?page no="48"?> Table 2.2: Strong verbs in Middle English based on Brunner (1963: § 69). Present 1st Past 2nd Past Past. Part. I ME ı ˉ N/ eME a ˉ , S ǭ N a ˉ , S i i II ME e ˉ ̣ / u ˉ ę ̄ u/ o ǭ IIIa ME N + C ı ˉ a ˉ / ǭ ou/ u ː / ou/ u ː / IIIb ME l + C e a u o IIIc ME r/ h + C e a u o IVa ME regular ę ̄ a (a ˉ ) ę ̄ , e ˉ ̣ ǭ IVb ME niman/ cuman i/ u a/ o ˉ o ˉ u V ME ę ̄ a (a ˉ ) ę ̄ , e ˉ ̣ ę ̄ VI ME a ˉ o ˉ ̣ o ˉ ̣ a ˉ VII ME V e ˉ ̣ e ˉ ̣ V Classes IVa and V also show long vowels in the past singular form and Brunner (1963: § 69) suggests that this is due to lengthening in open syllables. Apart from the two competing grade vowels in the 2nd past grade of class I due to the change a ˉ > ǭ (cf. section 2.2), there is nothing in Table 2.2 that would imply that the grade vowel system of strong verbs was very heterogeneous. At the beginning of the Middle English period and again from about 1550 - 1650 the system of strong verbs had destabilized and there was “ an enormous proliferation of patterns and variants more complex than was the case in Old English ” (Lass 1997: 348). According to We ł na (1991; quoted from Krygier 1994 a) only 78 out of 367 OE strong verbs preserve the original strong paradigm at least to some degree. As I only analyze Early Middle English data in this study, I focus in the following discussion of previous work on the earlier developments of strong verbs. However, where it is relevant, I refer to the studies on the later developments in Early Modern English and/ or Non- Standard English (Price 1910; Lass 1994 b; Cheshire 1994; Anderwald 2009) in order to provide parallels to the Middle English data. 2.1.2.1 Shift to weak Up to this point, most research on the disintegration of the strong verb system has primarily focused on the shift of verbs from the strong to the weak conjugation. Instead of marking the past tense and past participle forms by means of different grade vowels, these verbs adopted the dental suffix from the weak verbs to form new past and past participle forms. According to Michelau (1910: 3), this is a phenomenon that is common to all Germanic languages but had the highest impact in English. Krygier (1994 a: 18 - 25) gives a detailed 30 2. Historical Background and Previous Work <?page no="49"?> account of possible reasons for the shift in English and in this section I summarize the most important findings of his study. The shift of strong verbs to the weak conjugation is discussed in most language histories and grammars; however, usually the shift is mentioned but not explained. Krygier (1994 a: 18) notes that if an explanation is given, very often it is just a short reference to the principle of analogy that operates because of the high frequency of weak verbs (e. g. Brook 1958; Williams 1975). According to Brinton and Arnovick (2011: 218) maybe 75 percent of all OE verbs were weak and, while the vast number of weak verbs certainly had a triggering effect, I agree with Krygier (1994 a: 20) that it definitively cannot be the only cause; after all, weak verbs had existed alongside strong verbs for several centuries ever since the Germanic languages developed a separate system of strong and weak verbs. According to Krygier, it was Kaluza who first tried to find other explanations than analogy for the cause of the shift. Kaluza (1901: 150) suggests that the close proximity of the grade vowel of the present and the two past grades in some classes was one of the triggering factors for the shift. Similarly, Michelau (1910: 81) notes a connection of the shift to the pattern of the gradation series. Krygier ’ s confirms that “ verbs with fewer vowel alternants were more likely to undergo the shift in OE and in the transition period ” (1994 a: 249). He adds, however, that similarities or differences in the grade vowels did not play a role in the other periods, since the changes in the ablaut system during the Middle English period were too extreme in general. Michelau (1910: 68 - 69) proposes that deverbal weak verbs might have had a first influence, as the distinction between the original strong verb and the derived weak verb was supposedly already blurred in Old English. However, Krygier is cautious about this claim, because a quantitative study to support it would be difficult “ [f]or if one is unable to establish with certainty the accurate number of Old English strong verbs, establishing the number of Old English weak verbs from the same roots is even less feasible ” (1994 a: 20). Krygier still takes the factor into consideration in his study but concludes that weak verbs derived from parallel strong verbs do not seem to have any “ systemic influence on the shift ” (1994 a: 249). Long (1944: 267) also mentions parallel weak and strong verbs as an influencing factor and, apart from native parallel forms, she points to Old Norse verbs like ON hengja and ON flýja that apparently led to the shift of the parallel verbs ME hang(e) and ME flee(n). Also, Taylor (1994: 157) considers it plausible that the contact with the Scandinavians triggered the shift towards the weak conjugation of some strong verbs. Similarly, the language contact between English and Anglo-Norman after the Norman Conquest is mentioned as a highly influential factor (e. g. Michelau 1910; Strang 1970). 2.1 Reduction of the number of grades in strong verbs 31 <?page no="50"?> According to Strang (1970: 276), the Old Norse and Old English strong verb classes were so similar that the Old Norse verbs could easily be conjugated according to the native Old English gradation system and, therefore, she believes that they did not have an effect. However, she continues that in the 12th and 13th century there was a large intake of French loans, which then almost exclusively conformed to the weak past formation. This results in a vast number of weak verbs in Middle English and, because the weak past formation is the lesser marked one anyway, more and more strong verbs shifted to the weak class. Nevertheless, Krygier (1994 a: 250) concludes that “ Old Norse can be assigned the shift-triggering role with higher certainty ” , but it did not change the system significantly because of the close affinities of the two languages. He continues that it was only possible to analyze the influence of Old Norse because forms from texts of the former Danelaw could be compared with forms of the area that did not have this close contact with Scandinavian settlers. The contact with Anglo-Norman was of a different nature and, although the phase of the contact was longer, Krygier explains that “ no pockets free of its domination could be isolated for comparison, as the French-English bilingualism was societal and not geographical ” (1994 a: 250). For this reason he was unable to establish any influence through the Anglo- Norman contact. Another factor that is frequently mentioned is the group of strong verbs with a root-final dental plosive, which had a similar syllable structure as the weak verbs with the dental suffix (e. g. Michelau 1910; Brunner 1960 - 62). However, Krygier (1994 a: 21) raises some doubts about the role these verbs might have played, since some of these have not shifted to the weak class until today. Among these are such verbs as bite, fight, or ride. Nevertheless, Krygier ’ s (1994 a: 249) own research has confirmed this root structure as a statistically significant factor for the transition period from Old to Middle English. Recently, Branchaw pointed out that many of these dental-final verbs “ ablauted with vowels of a low degree of perceptual distinctness ” (2010: 100), which might have added to the triggering factor of the dental. In the period of the 13th and 14th century another root structure appears to have been shift-conducive: Krygier (1994 a: 249) points out that, in this later period, especially verbs with stem-final sonorant plus consonant clusters were prone to shift in this period. Because Krygier (1994 a) was mainly interested in the causes for the shift, the study falls short of information about the spatial development of the shift to weak. The only reference to diatopic variation is found in Taylor (1994). She notes that the strong verbs that shifted to the weak class are mainly found in the East and West Midlands. However, because her conclusions are based on 6 texts of the 13th century and 19 texts of the 14th century only, her conclusions have to be taken with caution. 32 2. Historical Background and Previous Work <?page no="51"?> 2.1.2.2 Hybridization or class mixing There is very little research on the phenomenon of ‘ hybridization ’ or class mixing. In the available handbooks and grammars, the mixing of verb classes is hardly mentioned and it seems that more research is necessary to fully grasp to what extent verbs shifted from one class to another. Lass (2006: 77), for instance, only mentions ModE break and speak as two verbs that underwent a hybridization process during the Middle English period. These two verbs originally belonged to class V but show the past and past participle vowel of class IV in Modern English. Some verbs already shifted to other strong verb classes during the Old English period. Krygier (2001: 55), by summarizing Campbell (1959: §§ 307 - 320), lists the following verbs: (8) I > II te ˉ on, þe ˉ on, wre ˉ on II > VII cne ˉ odan, sne ˉ owan IIIc > I frignan V > IV brecan, cnedan, drepan, hlecan VI > VII waxan VII > IIIa fo ˉ n, ho ˉ n Brunner (1965: § 383) notes that ð ı ˉ on/ ðe ˉ on, t ı ˉ on/ te ˉ on, and wr ı ˉ on/ wre ˉ on changed to class II only in the West Saxon dialect; in the Anglian dialect the original gradation series of these verbs was preserved. This means that both ablaut patterns have to be expected in the data of the present study. The case of ð ı ˉ on is especially interesting, as it originally belonged to class III (*þinhan > *þ ı ˉ han > ð ı ˉ on) and in Old English the past plural ðungon and the past participle ðungen are still preserved in accordance with this class (Brunner 1965: § 386.1, Anm. 2). There are a few other isolated forms that can be found in some manuscripts that show some variant forms in Old English. However, apart from the shift of the contracted verbs of class I to class II, no regular shift is evident for the Old English period. Hybridization in Middle English is most prominent in class V, where several verbs like ME speke(n), wreke(n), trede(n), and weue(n) are reported to form the past participle in allegiance with class IV with o. According to Brunner (1963: § 69, note 12) this phenomenon is first present in 14th century manuscripts from the East but later also elsewhere. Furthermore, there are a few verbs of class VI as well, namely ME swerie(n)/ swere(n), hebbe(n), and waxe(n)/ wexe(n) that frequently show grade vowels of class IV. In addition to these more regular changes, there are also a few minor forms that need to be mentioned. First, OE st ı ˉ gan > ME stie(n) may show past singular forms with the respective grade vowel of class II, i. e. steigh (Brunner 2.1 Reduction of the number of grades in strong verbs 33 <?page no="52"?> 1963: § 69, note 2). Less frequently, past plural and past participle forms with the grade vowel of class II can also be found, i. e. stuge and istoge. Furthermore, ME bere(n) frequently shows forms with o ˉ in the entire past, i. e. boor, booren, which Brunner (1963: § 69, note 9) explains by analogy to class VI, but they could also be explained by analogical extension of the past participle vowel to the past stems. When discussing the results of the grade vowel mergers in the past tense in chapter 4, it will be necessary to bear these shifts in mind as they might account for unexpected vowels in the two past tense grades. 2.1.2.3 Reduction of the number of grade vowels As already mentioned above, in Middle English there was a general constraint of “ no more than three grades per verb ” (Lass 2006: 77) and the grade vowels of the four original principle parts of strong verbs started to merge by extension of one past vowel to the other. There is general agreement on the analogical extension of the singular past vowel to the plural in the North (Long 1944: 263; Brunner 1963: § 69.1; Brook 1963: 72; Nielsen 1985: 45; Iglesias-Rábade 2003: § 197); however, with regard to the South and the Midlands there are differing observations. Iglesias-Rábade (2003: § 197) mentions the leveling of the singular under the past plural grade vowel in the South and Midlands. Brunner (1963: § 69.1) and Brook (1963: 72) point out the extension of the 2nd past or past participle grade vowel to the singular; however, according to Brunner this was only the case in classes II to V. Finally, Long (1944), Nielsen (1985: 43), and Lass (1994 b: 88) only mention the western leveling of the past participle grade which is found in both the singular and the plural past grade. The possible processes are illustrated in (9) with the principle parts of class II: (9) Present 1st Past 2nd Past Past Participle OE e ˉ o e ˉ a u o ME e ˉ / ɛː / e ˉ / e ː / u o North e ˉ e ˉ e ˉ o South/ ML e ˉ u u o South/ ML e ˉ o o It is possible that the different views on the process in the South and South- West Midlands can be explained by temporal differences. Long (1944) and Nielsen (1985) only analyze Late Middle English verbs and Lass (1994 b: 88) explains that the leveling under the past participle vowel starts a century later than the northern shift. An examination of the data from LAEME is necessary 34 2. Historical Background and Previous Work <?page no="53"?> to establish whether there ever was a leveling under the plural vowel and whether this happened as early as the shift of the singular to the plural in the North. According to Bülbring (1889: 116 - 117), for instance, all the apparent occurrences of the plural past vowel in the singular can be explained by phonological processes or by analogy to other classes (i. e. hybridization), as none of the original plural grade vowels in any verb have survived until Modern English. Furthermore, Long (1944: 263), who collected about 15,000 strong verbs in forty manuscripts dated to 1400 - 1475, does not only refer to dialectal differences but also concludes that there are class specific tendencies: In classes I and III the singular vowel spreads to the plural and past participle, whereas in classes II and IV the past participle vowel spreads to the past tense forms. However, these are just general observations; Long ’ s thorough discussion of the individual verbs within each class shows that the picture is much more complex than her final conclusion might suggest. With regard to the dating of the change, there is little evidence other than vague references to the different centuries of the Middle English period. Lass (2006: 77) generally remarks that the reduction of grades by analogical extension of some grade vowels to others starts in the 13th century and increases gradually. Elsewhere, he notes that the loss of the past vowel distinction starts in the North and diffuses southwards at a considerable speed until the early 15th century (Lass 1994 b: 88). 4 Similarly, Markus (1990: 152) comments that the leveling starts in the Early Middle English period and spreads from north to south until the 15th century, but Brunner (1963: § 69.1) mentions that the leveling of the past grade vowels under the singular vowel started in the North, where the vowel of the 1st and 3rd singular past tense forms had already been extended to that of the 2nd singular past forms before the appearance of the first Middle English written records. Since there are already Middle English manuscripts in the 12th century it looks at first as if Brunner, on the one hand, and Lass and Markus, on the other, disagree with regard to the dating of the change; however, as already mentioned in chapter 1, there are hardly any Middle English texts from the North earlier than 1300 (Laing 2000: 100), so Brunner does not really contradict Lass and Markus. Brunner continues that the change is later in the Midlands (in the course of the 13th and 14th century) but notes that it was again the 2nd singular past vowel which was leveled out earlier than the plural forms. With regard to the South, Brunner notes that, as late as the 15th century, the distinction of separate grade vowels even for the 2nd singular past was still preserved, but he also 4 Görlach (1996: 163), following Gburek (1986), notes that the leveling under the past singular vowel is also evident in Scots. 2.1 Reduction of the number of grades in strong verbs 35 <?page no="54"?> acknowledges that a few leveled forms are found without commenting on whether the singular or plural vowel was more common. Long (1944: 263) and Mossé (1952: § 8) have argued that the leveling of forms in the past grades of strong verbs might have been triggered by analogy to classes VI and VII, which historically already had only one grade vowel for the entire past, as is evident in Table 2.1. Berndt (1956: 267), on the other hand, notes the loss of a distinct 2nd singular past inflection in the northern texts of Late Old English. As a consequence the 2nd singular past form became structurally more similar to the 1st and 3rd singular past form, but with a different grade vowel. According to Berndt it was this similarity of forms, which was then leveled out, that might have triggered the merger. Brunner (1963: § 69.1) also notes that the leveling begins in the 2nd person singular forms. Similarly, Lass comments on the fact that the beginning of the merger “ goes hand in hand with the loss of suffixal -en as a plural marker ” (1994 b: 88), so the loss of number marking made the 1st and 2nd past grade vowel redundant. It is very likely that such conditions favored the leveling of the past forms under one of the two vowels; however, there are also other factors that have to be taken in account. Bybee and Slobin conclude in a seminal study on the formation of irregular verbs in Modern English that, on the one hand, the phonological properties of individual verbs play a role but that “ frequency of usage is an important variable in the learning of irregular past-tense forms ” (1982: 271). This was recently confirmed by Lieberman, Michel, Jackson, Tang and Nowak (2007) and Branchaw (2010). Branchaw, however, concludes that “ the single most important factor in preserving the strong inflection of a verb was the perceptual ease of distinguishing the vowels of the principal parts ” (2010: 101). In other words, if the vowels of the principle parts were too similar, they were more likely to merge. To sum up, the descriptions suggest that there was gradual diffusion of the past form mergers under the singular vowel through the East Midlands which, according to traditional views, is attested in the written sources from the 13th century onwards, which is the century that provides the most texts for LAEME. Furthermore, the extension of the plural vowel to the singular is evident in the West and South, and it is yet to be determined through the data from LAEME whether spatial diffusion can also be observed with regard to this merger. 36 2. Historical Background and Previous Work <?page no="55"?> 2.2 OE a ˉ > ME ǭ Of special interest in the Early Middle English period is the phoneme OE a ˉ , which was raised and rounded to ME ǭ in all dialects but the northern (Jordan 1974: § 44; Mossé 1952: § 27; Brunner 1963: § 11.4; Iglesias-Rábade 2003: 326 - 327). 5 The following example from Lass (2006: 64) highlights variation of <a> and <o> spellings for a ˉ > ǭ in two 13th century texts (one each from the West and East): (10) La ȝ amon A (#277), Worcestershire OE ba ˉ n “ bone ” : <a> 4 x, <o> 2 x OE hla ˉ ford “ lord ” : <a> 12 x, <æ> 1 x, <o> 1 x OE la ˉ þ “ loath ” : <a> 11 x, <ae> 11 x, <oa> 1 x, <o> 2 x Vices and Virtues (#64), SW Essex OE ga ˉ st “ ghost ” : <a> 26 x, <o> 11 x OE ga ˉ stlice “ ghostly ” : <a> 10 x, <o> 1 x OE ha ˉ lig: <a> 130 x, <o> 14 x The <a> spellings are dominant in both texts and Lass notes that “ [t]his looks like an early stage of diffusion, typically variable and lexically specific ” (2006: 64). An examination of all the material of the LAEME CTT is necessary to determine whether this type of variation is evident in all texts and whether some items are more advanced than others as Lass ’ s comment implies. In section 2.2.1, I establish the historical background of the variable up until the Middle English period, which is necessary in order to define exactly which instances of OE a ˉ are included in this study. Furthermore, I highlight the findings of earlier studies concerning the dialectal distribution of the change in Middle English in section 2.2.2. 2.2.1 OE origins of ME a ˉ ME ǭ developed from lOE a ˉ , which had several possible origins (cf. Liebl 2002): First and foremost, it was the regular output of the change WGmc ai > OE a ˉ . 6 Furthermore, OE a ˉ had also developed through retraction of OE æ ˉ 5 This change is so important that even today it is referred to as forming “ [o]ne of the major isoglosses separating the 'true North' (northern England and Scotland) from the rest of Britain ” (Lass 1997: 67). 6 I follow Lass (1976: 129 - 134) and Hogg (1992: § 2.13), who take OE <a> to represent a short or long low back unrounded vowel in all dialects, usually transcribed as / ɑ , ɑː / . For a different view Hogg refers to Kuhn (1961: 524), who transcribes the phoneme as / a, a ː / , but nevertheless notes with regard to the short variant that, “ [p]honetically, this sound was probably low back unround ” (Kuhn 1961: 527). 2.2 OE a ˉ > ME ǭ 37 <?page no="56"?> (< WGmc a ˉ ) in some positions and there are also some loanwords that contain a vowel of the same quality as OE a ˉ . Sections 2.2.1.1 and 2.2.1.2 look into the historical development of these processes. In addition to the historically long vowels, there are also a few lengthening processes that affected OE ă in some positions in the course of the OE period and the resulting vowel merges with the originally long vowel. I outline the historical development of OE ă with subsequent lengthening in section 2.2.1.4 and I explain why OE a ˉ through lengthening processes of OE ă have been excluded from the present project. Finally, in section 2.2.1.5, I discuss some late Old English shortening processes, which might have affected OE a ˉ and subsequently the change ME a ˉ > ǭ . 2.2.1.1 WGmc *ai > OE a ˉ Gmc *ai (< IE *ai/ oi) was regularly monophthongized to OE a ˉ in stressed and secondary-stressed syllables (Campbell 1959: § 134; Brunner 1965: 73; Prins 1972: § 2.17; Hogg 1992: § 5.7 - 5.9; Lass 1994 a: 39 - 41): (11) Gmc *stain > OE sta ˉ n but OHG stein, ON steinn, Goth. stains Gmc *aik > OE a ˉ c but OHG eih, OI eik Eichner notes that this change is a “ shibboleth for the identification of the English language ” (1990: 323, my translation), as it took place in all Old English dialects and in any context. In this respect the change is different from a similar development which affected Old Frisian. According to Eichner, monophthongization also took place in Old Frisian, but the change was conditioned: Gmc *ai changed to OFris a ˉ only in open syllables before back vowels and w, elsewhere it was monophthongized to e ˉ . Hogg (1992: § 5.9) and Bammesberger (2006: 178) therefore assume an independent development in OFris. Hogg attributes this monophthongization process to “ compensatory lengthening for the loss of i ” (1992: § 5.7) without providing an explanation for this loss of i. A different view is advanced by Bammesberger (2006). Based on the runic inscription on the Caistor-by-Norwich roe-deer astragalus ( ᚱᚨᛇᚺᚨᚾ raïhan) where ᚨ appears both before ᛇ and the nasal ᚾ , he argues that the first element of the diphthong must have been rounded and the second centralized ([å ǝ ] in Bammesberger ’ s notation), which he regards as a suitable precursor for the monophthong a ˉ . 7 7 This is again in opposition to Luick (1964: § 122), Pinsker (1963: § 20.3), and Eichner (1990: 322), who propose a sequence ai ̯ > ae ̯ > æe ̯ > æa ̯ > a ˉ (Pinsker also on the basis of the raïhan inscription on the Caistor-by-Norwich astragalus). 38 2. Historical Background and Previous Work <?page no="57"?> The dating and relative chronology of this monophthongization process have not been fully resolved yet. The most widespread view (e. g. Campbell 1959: § 132; Hogg 1992: § 5.9; Stiles 1995: 197) holds that the change must have taken place after the shift of WGmc a ˉ > OE æ ˉ (cf. section 2.2.1.2), as the two vowels have not merged. Furthermore, it must have been monophthongized before ‘ first fronting ’ set in (cf. section 2.2.1.4), as otherwise the first element of the diphthong would have undergone the development of the short vowel, namely the fronting of Gmc *a > OE æ, which, according to Hogg (1992: § 5.9) would have resulted in a different vowel: Gmc *ai > OE *æi > OE æ ˉ . 8 Lass, on the other hand, speculates that the change “ might not be sequential, but implicational ” (1994 a: 40) and offers two possible solutions: On the one hand, the fronting of NWGmc a ˉ might have led to a drag chain shift, in which it triggered the change WGmc ai > OE a ˉ in order to fill the empty vowel space. On the other hand, it is also possible that the diphthong began to monophthongize first and then pushed the old a ˉ into a new (empty) position. Hogg (1992: § 5.7) points out that in the Old English period there were already some lexical items in which the reflexes of Gmc *ai alternated between OE a ˉ and o ˉ . These include words such as a ˉ , o ˉ ‘ ever ’ or na ˉ , no ˉ ‘ never ’ and compounds containing these words like a ˉ wiht, o ˉ wiht or a ˉ hwæðer/ a ˉ wðer, o ˉ hwæðer/ o ˉ wðer (cf. also Brunner 1965: § 73, Anm. 1). Hogg (1992: §§ 6.4(1) and 6.5) assumes that a low stress level in these forms was likely to have been responsible for the <o> spellings, since in unstressed positions OE a ˉ would have been shortened, and <o> spellings for unfronted Gmc *a in unstressed prefixes are not unusual. Hogg notes that some degree of rounding might have been involved in the change in all instances, which would be in line with Bammesberger (2006). 2.2.1.2 WGmc *a ˉ > OE æ ˉ > OE a ˉ One of the features that separates NWGmc from EGmc is the lowering and retraction of PGmc e ˉ 1 9 to NWGmc a ˉ , which is retained in OS, OHG, and ON (Lass 1994 a: 25): (12) Gmc *de ˉ 1 di > OS da ˉ de, OHG ta ˉ t, ON dáð but Goth. dêds Gmc *me ˉ 1 no ˉ þ > OS ma ˉ nuth, OHG ma ˉ nod, ON mánaðr but Goth. menoþs 8 Luick (1964: §§ 121 - 2), reiterated by Pinsker (1963) and Eichner (1990), and also Fulk (1998 b: 145) suggest a different sequence, namely that the monophthongization of */ ai/ took place later than first fronting. Hogg (1992: § 5.9 n.1) rejects this view as “ implausible ” . 9 The exact quality of PGmc e ˉ 1 is highly debated and Lass (1994 a: 26) renders it as / e ː / . As the discussion is not relevant for the present study, I will not go into further detail. 2.2 OE a ˉ > ME ǭ 39 <?page no="58"?> NWGmc a ˉ did not retain its quality in Anglo-Frisian and the respective OE and OFris cognates are illustrated in example (13): (13) WGmc *da ˉ di- > OFris de ˉ de, OE dæ ˉ d (WS)/ de ˉ d (NonWS) WGmc *ma ˉ no ˉ þ- > OFris mo ˉ nath, OE mo ˉ naþ This example shows that both in OE and OFris there are two different reflexes of WGmc *a ˉ (< Gmc *e ˉ 1 ). First, in pre-nasal positions, the vowel was possibly nasalized and, as a consequence, rounded and raised to o ˉ (Campbell 1959: § 127; Brunner 1965: § 80; Prins 1972: 2.13; Hogg 1992: 3.22). According to Lass (1994 a: 38), the vowel eventually denasalized again and merged with IE *o ˉ and Gmc *o ˉ (< IE *a ˉ ). Therefore, OE bro ˉ þor, fo ˉ t and mo ˉ naþ all show the same vowel (Lass 1994 a: 41). Sometime after WGmc *a ˉ was nasalized in the relevant positions; elsewhere, it was fronted to OFris e ˉ and OE æ ˉ (WS)/ e ˉ (NonWS), respectively (Campbell 1959: § 128; Brunner 1965: § 62; Prins 1972: 2.13; Hogg 1992: § 3.22; Lass 1994 a: 39). At first, the development of Gmc *a ˉ does not seem to be relevant for the present study but, in the West Saxon dialect, a following w caused the retraction of æ ˉ > a ˉ (Hogg 1992: § 5.21). Furthermore, this retraction affected the former ‘ verba pura ’ of class VII in all dialects, i. e. bla ˉ wan ‘ blow ’ , cna ˉ wan ‘ know ’ , cra ˉ wan ‘ crow ’ and sa ˉ wan ‘ sow ’ . The reason for the appearance of a ˉ in these verbs has not yet been fully resolved (Brunner 1965: § 63 Anm. 2; Hogg 1992: § 5.21). As the vowel in these verbs is known to have changed to ME ǭ at some stage, I include them in the present study. 2.2.1.3 Loanwords with a ˉ Liebl (2002: 36 - 38) lists three possible sources for OE a ˉ in loanwords: ON á, ON ǫ ́ and L a ˉ . According to Liebl (following Luick 1964: § 383.1), ON á is of the same quality as OE a ˉ , as it also changes to ME ǭ in place-name elements like skáli ‘ hut ’ or vrá ‘ nook ’ . ON ǭ is the result of u-umlaut of ON á in West Norse and, according to Liebl (2002: 37; following Björkman 1900 - 1902: 85 - 86 and Luick 1964: § 383.3), it appeared as a ˉ in Middle English past plurals like wa ˉ ren, ga ˉ ven, and ga ˉ ten, which were then also subject to change to ME ǭ . Finally, Latin loanwords which were borrowed after the fronting of Gmc a ˉ > OE æ ˉ show a vowel which is of the same quality as OE a ˉ : L pa ˉ lus > OE pa ˉ l ‘ pole ’ , L pa ˉ pa > OE pa ˉ pa ‘ pope ’ (Campbell 1959: § 504). These items will be considered in the present study if they appear in the data. 40 2. Historical Background and Previous Work <?page no="59"?> 2.2.1.4 OE ă Gmc *a was affected by similar changes as its long counter-part. ‘ First Fronting ’ , also known as ‘ Anglo-Frisian Brightening ’ , caused the fronting of the low back vowel WGmc * ɑ to OE æ (Campbell 1959: § 31; Brunner 1965: § 49; Prins 1972: § 2.8; Hogg 1992: §§ 5.10 - 5.15; Lass 1994 a: 42), both in stressed and unstressed positions (Hogg 1992: § 6.2): (14) Gmc *dagaz: OE dæg, OFris deg but Goth dags, ON dagr, OHG tag. First fronting would, of course, not regularly yield OE a ˉ even after lengthening processes; however, Gmc *a retained its quality or was restored in some linguistic environments early in the Old English period: A following nasal inhibited first fronting and in some special contexts eOE æ retracted to ă . 10 There is general agreement that Gmc a was nasalized before nasal consonants at a relatively early stage and that first fronting was blocked in these environments (Campbell 1959: § 130; Brunner 1965: § 79; Prins 1972: 2.9; Hogg 1992: § 5.4 - 6; Lass 1994 a: 42). Early OE textual evidence shows that Gmc a is variably spelled <a> and <o> in words like mann/ monn, land/ lond, or nama/ noma. Brunner (1965: § 79) summarizes the spatial and temporal variation of OE <a> and <o> before nasals as follows: (15) Early West Saxon <a>/ <o> Kentish (9th C) and Anglian mostly <o> Kentish and West Saxon (both 10th C) increasingly <a> Late West Saxon almost exclusively <a> At the end of the Old English period this sound merged with other cases of OE a in all areas except for the West Midlands (Hogg 1992: § 5.5). 11 In the Middle English period <o> spellings for OE a plus nasal were still attested in texts from Herefordshire, Staffordshire, Shropshire, most of Warwickshire, Cheshire, Derbyshire, South Lancashire, and also Worcestershire (Jordan 1974: § 5). Most scholars agree that first fronting was categorical except when conditioned by a following nasal as described above (Luick 1964: § 164; Campbell 1959: §§ 131 and 139, footnote 1; most prominently Hogg 1992: 10 In West Mercian dialects OE a and æ are fronted further to æ and e, respectively. This process is referred to as 'second fronting' and is described in detail by Hogg (1992: § 5.87 - 5.92). 11 Hogg (1992: § 5.5) considers it possible that this sound may have “ achieved phonemic status in WMerc, that is, / ɔ / , ” but continues that “ it seems unlikely that such phonemicization occurred elsewhere, even in Nbr. ” (cf. Hogg 1982). For a different view cf. Kuhn (1961). 2.2 OE a ˉ > ME ǭ 41 <?page no="60"?> § 5.10 - 5.15). 12 There are certain environments in which the old quality of the vowel was restored: in all dialects before w unless followed by i or j, in the second class of weak verbs where the thematic element changed from Gmc *-o ˉ j- > OE -i-, before liquids plus consonant in those Old English varieties in which breaking did not take place, i. e. before lC in Anglian and rC in Northumbrian. Furthermore, a was also restored before a back vowel in the following syllable (Brunner 1965: §§ 10, 50; Hogg 1992: § 5.10). In addition to these environments, Hogg (1992: § 6.4) adds three groups of words that remained unaffected by first fronting: “ (a) non-lexical items or forms of the copula wesan; (b) the originally secondary-stressed elements of obscured compounds or prefixed nouns; (c) unstressed verbal prefixes. ” Hogg explains that it is not unusual to find <o> spellings for regular OE a in these contexts. However, the environments just described are only of interest for this study if the vowels were later lengthened. The first lengthening process of lOE ă is the so-called ‘ homorganic cluster lengthening ’ (HOL), which affected stressed vowels followed by a consonant cluster that complies with the following rule: liquid or nasal plus any voiced consonant which is articulated in the same place, unless followed by a third consonant (Campbell 1959: § 283 - 285; Brunner 1963: § 8; Prins 1972: §§ 2.49 - 2.56; Jordan 1974: § 22; Hogg 1992: §§ 5202 - 5204; Smith 2007: 110): (16) OE l ă mb > lOE la ˉ mb but: lOE l ă mbru, pl. OE h ă nd > lOE ha ˉ nd Example (16) illustrates the change before the clusters mb and nd (including inhibition in the plural because of a third consonant), but in total there are eight different clusters to which the rule described above applies: ld, mb, nd, ng / ŋ g/ , rd, rl, rn, rð. 13 Hogg (1992: § 5202) assumes that these clusters were treated as single consonants and, since in Old English the typical syllable structure of the rhyme constituent of stressed syllables was bimoric, i. e. of the type -VVC and -VCC, there was a general tendency for the short vowels in these positions to be lengthened. A different view is brought forth by Phillips (1981, 1983 b), who refers to a modern experiment by Chen (1970), which revealed that vowels were generally produced longer before the cluster liquid/ nasal plus homorganic voiced consonants than before clusters containing 12 For a different view see Bülbring (1902: §§ 91, 127, 132 - 4) and Quirk and Wrenn (1958: § 203), who argue that first fronting did not take place in the environments in which / ɑ / was restored. The main arguments of both views are provided by Hogg (1992: §§ 5.11 and 5.12). As it is not relevant for the development of the sounds to ME, I will not go into more detail. 13 The clusters with rC are irrelevant for this study, as a ˉ was inexistent in these contexts due to other phonological processes like breaking (Jones 1989: 25). 42 2. Historical Background and Previous Work <?page no="61"?> voiceless consonants. According to Phillips (1983 b: 880), this universal phonetic tendency was exaggerated in Old English: “ Since vowel length was distinctive in Old English, children learning the language interpreted these lengthened vowels as members of the group of long vowel phonemes. ” Liebl (2002, 2006, 2008) includes items with OE a ˉ due to HOL, but I have excluded them from my own research for several reasons. Unfortunately, the developments are quite irregular in the history of English. In general, lengthening was usually preserved before lC clusters; however, at the end of the 14th century, lengthening remained elsewhere only in the following environments: i+mb, o+mb, i+nd, u+nd (Mossé 1952: § 18; Smith 2007: 112). Minkova and Stockwell, who have challenged the traditional text book accounts with regard to this phenomenon, even go as far as to state that “ [h]omogranic cluster lengthening with reference to any environment other than -ld is a misnomer ” (1992: 201). The shortening before nasal clusters already started in the late 12th century (Wright and Wright 1928: § 68), so theoretically it could affect the data from LAEME. This was further complicated by the devoicing of final plosives -g, -d, and -b, which might have prevented HOL in the North and East in these environments in the first place (Brunner 1963: § 8; Jordan 1974: § 22, remark 3). For clusters with a nasal, this has the following implications for the change in Middle English: In the West Midlands one might encounter <o> spellings because of nasal influence of a short vowel or because of a ˉ > ǭ , and in the East it is not clear whether an <a> spelling might represent a short vowel or an unshifted long vowel. Clusters with lC would seem suitable at first, as the lengthening was preserved during the entire Early Middle English period; however, because vowels before lC clusters were subject to breaking in some dialects, there are already competing forms in Old English. In West Saxon the regular development was ælC > ealC, but in Anglian dialects there was retraction to alC instead (Lass 1994 a: 48 - 50): (17) West Saxon vs. Anglian eald ald heald hald beald bald Although Brunner (1963: § 10) notes that the (lengthened) diphthong e ˉ a had monophthongized to æ ˉ by the 11th century (which appears as ME e ˉ / ɛː / ), it still shows up in the LAEME CTT. A pilot study involving the clusters % ALD % and % OLD % has brought forth quite a number of <ea> spellings. This would distort the results of a study focusing on the change a ˉ > ǭ , as there would not be any data for these items from the areas where breaking took place. 2.2 OE a ˉ > ME ǭ 43 <?page no="62"?> The distinction between OE æ and a was phonemic for most of the OE period, but the two merged in / ɑ / the end of the OE period (probably in the middle of the 11th century) everywhere but in Kent. This merged vowel was most likely fronted to ME / a/ (Lass 1976; Hogg 1992: § 5215 - 216). This vowel is later affected by ‘ Middle English Open Syllable Lengthening ’ (MEOSL) (Jordan 1974: § 25, remark 4; Luick 1964: § 391; Brunner 1963: § 12; Smith 2007: 13), a change in which vowels in stressed open syllables of disyllabic words were lengthened if they were followed by unstressed -e or en, as illustrated in (18): (18) OE nama / n ɑ m ɑ / > lOE/ eME nama / nam ə / > ME name / na: m ə / OE talu / t ɑ lu/ > lOE/ eME tale / tal ə / > ME tale / ta: l ə / The change a ˉ > ǭ was still underway at the time, so theoretically words including OE ă or æ would have to be considered as well. However, Liebl (2002: 40) and Liberman (1966: 67) point out that ME a ˉ as a result of MEOSL was never affected by the change a ˉ > ǭ , very likely due to the different quality of OE a ˉ / ɑː / and the new ME a ˉ / a ː / (Liberman 1966: 68 - 69; Lass 1976: 105 - 106; Moore 1969: § 55). A short pilot study with LAEME data confirmed these findings. Therefore, items with ME a ˉ due to MEOSL will not be included in the present study. Besides the lengthening processes described above, there are also monosyllabic non-lexical items like OE hwa and swa which, in stressed position, were sometimes lengthened in OE due to what Lass (1992: 71) calls ‘ Foot-Final Lengthening ’ . Isolated OE syllables in stressed positions “ had a minimal rhyme structure of -VVC or -VCC ” (Hogg 1992: § 1.80), so especially nonlexical items of the syllable structure CV where lengthened in stressed positions to comply with this tendency (Campbell 1959: § 125; Hogg 1992: § 5198; Lass 1992: 71). As it is usually impossible to tell in a written text whether an item shows <a> because it is short or because it might be lengthened, but has not shifted yet, I do not include such items in this study. 2.2.1.5 Caveat: Late Old English shortening processes There are two shortening processes which need to be mentioned, as they fall into the Late Old English period (before 1050, Fulk 1998 a: 10) and might prevent the change a ˉ > ǭ in some instances: ‘ Shortening before Consonant Clusters ’ (SHOCC) and ‘ Trisyllabic Shortening ’ (TRISH). Both processes were critically discussed by Ritt (1994) and Minkova and Stockwell (1998). SHOCC refers to a process in the transition period to ME, according to which long vowels or diphthongs were shortened before clusters involving two 44 2. Historical Background and Previous Work <?page no="63"?> consonants, except those involved in HOL (Luick 1964: §§ 204, 352, 386; Campbell 1959: § 285; Brunner 1965: § 138; Jordan 1974: § 23): (19) OE ha ˉ tte > ME h ă tte ‘ was called ’ OE ha ˉ lgian > ME h ă lwen ‘ to hallow ’ This shortening did often not take place if the consonant cluster was regarded as being part of the following syllable, for instance in inflected forms such as ME ga ˉ stes, gen. sg. ‘ ghost ’ (Brunner 1963: § 9; Jordan 1974: § 23). This was most prominent in clusters with st or sm, and Jordan remarks that this was often also transferred to the monosyllabic form retained the long vowel, as in ME ga ˉ st. Minkova and Stockwell also note that there are “ [n]o clear examples of shortening before -ful, -hood, -less, -ship, -some, -ster ” (1998: 231). TRISH refers to the process in which, according to traditional views, long vowels or diphthongs were shortened before one consonant or a cluster involved in HOL if they precede two unstressed syllables (Campbell 1959: § 285; Jordan 1974: § 24). The most typical example is given in (20): (20) OE la ˉ ferce > ME l ă ferce ‘ lark ’ Jordan (1974: § 24) notes that a secondary-stressed element following the first syllable prevented this shortening process. Therefore, lengthening was preserved in three-syllabic compounds, as for instance in OE ha ˉ ligdæg ‘ holy-day ’ . Brunner (1963: § 9), on the other hand, lists ME h ă ligdæg as an example for shortening, but adds that both shortened and unshortened forms can be found. The existence of TRISH has been challenged by Ritt (1994) and, most prominently, by Minkova and Stockwell, who conclude that “ TRISH cannot be reconstructed for the early history of English, largely due to the paucity of words to which shortening could have applied ” (1998: 228). Minkova and Stockwell (1998: 223) note that any trisyllabic forms in Old English must have been either borrowed (in which case the first syllable would have been shortened anyway (cf. Luick 1964: § 218, Anm. 1)) or they were inflected disyllables and, unlike Luick (1898) and Lahiri, Riad, and Jacobs (1999: 348), they doubt that the shortening affected inflected forms. The lexical items with the shortening clusters as described above are nevertheless included in the data collection in order to verify whether the shortening had taken place everywhere. Also, I do not distinguish between items as simplex or in compounds in the data collection, but I discuss possible shortening processes in the first element of compounds in a separate section of chapter 5. 2.2 OE a ˉ > ME ǭ 45 <?page no="64"?> 2.2.2 OE a ˉ > ME ǭ : onset, spatial variation and completion There are various explanations as to why the change a ˉ > ǭ started in the first place. Liebl (2002: 45 - 49) presents and discusses at least nine possible explanations for the change ME a ˉ > ǭ , some of which are more convincing than others. Several scholars put forward theories which link the change ME a ˉ > ǭ to other Middle English vowels or shifts: symmetry to the upward movement of the front vowel  (Liberman 1966: 69; Prins 1972: 3.11; Labov 1994: 257), change in spelling in order to differentiate the older back vowel from the new a ˉ due to MEOSL (Gburek 1985: 143) or even, together with æ ˉ >  , as a precursor of the Great Vowel Shift (Jones 1989: 139; Lutz 2004). Many of these explanations are problematic because of the relative chronology of the changes (cf. Liebl 2002: 45 - 49); however, a satisfying answer is still wanting. The connection with the later Great Vowel Shift is very intriguing; however, neither Jones nor Lutz explains what initially triggered the shift. The raising and rounding of OE a ˉ to ME ǭ took place in all ME varieties except for the northern, and it is generally believed that the change started in the South and then spread to the North. Luick (1964: § 369) and Jordan (1974: § 44) propose the old West Saxon area and maybe, at the same time, also Kent as possible origins, but even more places of origin have been suggested more recently: The evidence suggests, I believe, that / ɑː / > / ɔː / , rather than spreading from the South to the North, might have started in Late Old English more or less simultaneously in several counties in the South as well as the East and West Midlands and radiated from there. (Liebl 2006: 30) There is no doubt that the change ME a ˉ > ǭ is apparent in the first ME written records, but there are several slightly different views as to when the change actually started. The more traditional accounts date the change to 12th century; e. g., Brunner (1963: § 11.4, note 10) remarks that <o> spellings for OE a ˉ are first attested in early 12th century manuscripts. Similarly, also Mossé (1952 § 27, Rem. II) and Lass (2006: 64) date the beginning of the change to the beginning of the 12th century. Others propose an earlier on-set for the change. Jordan (1974: § 44), following Klaeber (1902), mentions some isolated <o> spellings already in lOE and Luick (1964: § 369) and Dietz (1989: 135) also propose that a ˉ > ǭ must have started in the South earlier than 1100. Interestingly, in another publication Brunner (1960: 269) also provides the 11th century for the onset of the change. This view has recently been confirmed by Liebl (2002, 2006), who convincingly shows that the change must have started earlier than the 12th century. Liebl draws on material from Klaeber (1902: 207) and Schlemilch (1914: 16 - 46 2. Historical Background and Previous Work <?page no="65"?> 17), who present a discussion of evidence for a ˉ > ǭ from the West Saxon and Mercian written record. Liebl notes a predominance of <on> in these items for OE a ˉ n “ one ” , which could be explained by “ reduced stress with rounding before nasal or scribal confusion with the preposition ” (2006: 23). However, there are also a number of fully stressed words without a following nasal, which show <o> spellings. Hogg (1992: § 5.7, n.2) dismisses such words like <sorig> for OE sa ˉ rig ‘ sorry ’ in Oxford, Bodleian Library, Hatton 20 as hardly significant evidence and Brunner (1965: § 73, Anm. 1) explains this form as a scribal error. I agree with Liebl that there is no reason why such spellings should not be taken as early evidence for rounding and raising of a ˉ . Liebl (2002, 2006) complements his manuscript evidence by place-name evidence form the Domesday Book and summarizes the evaluation of his data as follows: [T]he highest percentages of o-spellings can be found not only in the South (Berkshire, Kent), but also the East Midlands (e. g. Northamptonshire and the East Anglian county of Norfolk) and perhaps even the North-West Midlands (Cheshire); [. . .] What is surprising, though, is the complete absence of <o> in Essex, Middlesex and Hampshire, for which no explanation can be offered. (Liebl 2002: 92) This, together with the fact that “ spelling normally lags behind changes in pronunciation ” (Lass 1992: 46), clearly supports a pre-1100 starting point for the change ME a ˉ > ǭ . The handbooks remain rather vague about the subsequent development of the change. Brunner (1963: § 11.4) notes that <o> was the usual spelling for OE a ˉ in southern manuscripts of the 13th century already, but that variation between <a> and <o> spellings lasted longer in the Midlands. Jordan (1974: § 44) remains equally vague by stating that the change was attested in most parts of the southern half of the Midlands in the second half of the 12th century and that it had spread across almost the entire Midlands in the 13th century. The later developments in the North have attracted the largest amount of scholarly attention. There is general agreement that the change had reached the North at the beginning of the 13th century (e. g. Ekwall 1938: 164; Brunner 1960: 270; Jordan 1974: § 44; Dietz 1989: 135). While this remains unchallenged, there is disagreement as to when the change was completed. Liebl concludes that this must have happened in the middle of the 13th century in the North: [T]he toponomastic material seems to be substantial enough to warrant the assumption that in Cheshire, the West Riding of Yorkshire, Derbyshire and Nottinghamshire (and perhaps even Lincolnshire - i. e. Kesteven and Holland - and Lancashire? ) a ˉ > ǭ will have been effected by 1250. (2008: 42 - 43) With this claim he contests earlier attempts by Luick (1964) or and Dietz (1989: 143), who all propose a later period for the completion of the change. 2.2 OE a ˉ > ME ǭ 47 <?page no="66"?> Furthermore, there is also disagreement as to the boundary of the a ˉ and ǭ areas after completion. The earliest scholars usually took what Dietz (1989: 135) refers to as the “ Humber-Ribble-Linie ” as the northern most boundary for the change. However, research by Moore, Meech, and Whitehall (1935) and several scholars working with place-name evidence (e. g. Ekwall 1938; Kristensson 1965; Dietz 1978) have adjusted this line, as they found <o> spellings also further to the north. As matters seem to be more complex in the west, the handbooks generally just mention the Humber as the natural northern boundary for the change (e. g. Brunner 1963: § 11.4; Jordan 1974: § 6). The latest proposal for an a ˉ / ǭ isogloss is presented by Dietz (1989: 136) based on material from LALME: Lune estuary - Claughton - Clitheroe - Colne (La) - Steeton (Aire) - Knaresborough (Nidd) - Bardsey - Bradford - Huddersfield - Morley - Aire west of Snaith - Thorne - northern tip of Nottinghamshire eLALME Maps 48 show the Middle English reflexes of OE, ON a ˉ and they imply that the boundary is more complex than Dietz suggests: <o> spellings are found as far north as Berwickshire and Northumberland (Benskin et al. 2013 - ). The maps also reveal that Late Middle English <a> spellings as the minority variant are still frequently attested as far south as southern Lincolnshire. 14 While different material will certainly yield different isoglosses it is also due to the diffusion of sound changes that boundaries are anything but stable and therefore difficult to establish. Unfortunately, it is not possible to engage in the discussion about the northern developments of the change with data from LAEME, as there is hardly any manuscript material from the north for the Early Middle English period (cf. section 3.2.2). Nevertheless, the findings of the north are important because of an aspect so far not mentioned. Both Dietz (1989: 142) and Liebl (2002, 2008) note that the spatial distribution of OE a ˉ + w/ ɣ 15 is different from a ˉ in other contexts. According to Dietz (1989: 142) the distribution is as follows: OE a ˉ w and a ˉ ɣ > ME ou (South, South Midlands)/ ME au (North, North Midlands). Spellings with <a> are found further south in these contexts than in isolation. Dietz continues that another change, ou > au, interferes in the Northwest Midlands in Late Middle English and that it is impossible to tell 14 The maps are available online (last accessed August 2013): <http: / / archive.ling.ed.ac.uk/ ihd/ elalme_scripts/ lib/ create_feature_map.php? mapid=0 480 001> (a-spellings) <http: / / archive.ling.ed.ac.uk/ ihd/ elalme_scripts/ lib/ create_feature_map.php? mapid=0 480 002> (o-spellings) 15 I follow Dietz (1989) and Liebl (2002, 2008) in the notation of w and ɣ . The latter is the IPA representation for OE intervocalic <g>. 48 2. Historical Background and Previous Work <?page no="67"?> whether spellings like awen ‘ own ’ , knaw ‘ know ’ , or saule ‘ soul ’ represent OE a ˉ or the development ou > au. Liebl (2008: 39) suggests that this change ou > au might even have been more widely spread than Dietz suggested and also Brunner (1963: § 13, note 7) assumes that ME ou (< OE a ˉ g and a ˉ w) became au not only in the West and North, but also in Kent and perhaps the South East. Lass (1992: 51), on the other hand, only states that the development of the diphthongs a ˉ w and a ˉ ɣ is parallel to the change a ˉ > ǭ in the South without mentioning this possible later developments. As I only analyze Early Middle English data, this might not interfere with my study; however, it will be essential to check whether the development of items going back to OE a ˉ w and a ˉ ɣ is different also further south. According to previous research, variation between <a> and <o> spellings should be evident in the earliest Middle English texts already. So far, most research was carried out on the basis of onomastic evidence, so the comprehensive collection of Early Middle English literary evidence in LAEME will certainly bring forth new insights even to this prominent and well-studied Middle English sound change. 2.3 Dual forms of the personal pronoun The dual number is found in many of the world ’ s languages, but it is generally regarded as an archaic feature (Wackernagel 1950: 74). Among the Indo- European languages it is attested in all the older stages of the Germanic languages, in some early Indo-Iranian languages, some Greek dialects, some Slavonic languages, Old Irish, and Tocharian; however, it is absent from Latin, Armenian, and Hittite. According to Guðmundsson (1972: 13), Sanskrit, Avestan, and Attic Greek had dual forms for nouns, pronouns and verbs; however, in the Germanic languages the dual number of nouns was lost before the first written records and the dual of verbal forms is only preserved in Gothic. All the other Germanic languages solely retain special dual forms for the personal and possessive pronouns. 16 Their use is illustrated in (21): (21) 1st person A refers to himself/ herself and B “ we two ” 2nd person A addresses both B and C “ you two ” (based on Guðmundsson 1972: 14) 16 Unless Bammesberger is correct and ċ eara in line 2733 b of Genesis A (Ne ceara incit) should be interpreted as an imperative dual verb-form (personal communication mentioned in Stiles 1996: 565). 2.3 Dual forms of the personal pronoun 49 <?page no="68"?> Dual forms for the 3rd person are not attested in any Germanic languages except for some North Frisian dialects, in which the forms are modeled after the 3rd nominative plural personal pronoun (Howe 1996: 193). While many scholars have worked on the change from a ˉ > ǭ , exactly the opposite is the case with regard to the dual forms of the personal pronouns in Middle English. Judging by the relevant entries of the Middle English grammars (Mossé 1952: § 64; Brunner 1963: § 52; Iglesias-Rábade 2003: § 75 and § 77) the feature is almost obsolete during the period and the low frequency of this feature may not have attracted much scholarly attention so far. In a paper dealing with language periodization and archaisms, Lass (2000: 38) highlights the problem as follows: There is one archaism that becomes problematical when the focus is on single texts [. . .], and this is the dual. It is perfectly possible for a dialect to have a dual, and for it not to appear in surviving texts, since the opportunities for its use are limited: in both the Orrmulum and The Owl and the Nightingale there are two-person dialogic contexts that allow it to appear; this is not the case for the Kentish sermons, though there is good reason to believe that there was no dual in their dialect. There is so much dialogue in Chaucer that there is ample opportunity for a dual to surface if it existed. Maybe because of the issue highlighted by Lass, it appears that the dual has been an understudied and neglected feature in Middle English. Most handbooks do not go beyond a (usually very short) comment that it still existed in some Middle English texts (Mossé 1952: § 64; Brunner 1963: § 52; Iglesias- Rábade 2003: § 75 and § 77). And ever since Diehn (1901), Kennedy (1915), and a somewhat obscure paper by Ladd and Radice (1951/ 52), which is mentioned in Howe (1994), there have not been any detailed studies on the dual forms of the personal pronouns in Middle English. In section 2.3.1, I highlight the development of the dual forms until the Old English period. Furthermore, I outline the occurrence of dual forms in Old English in section 2.3.2, before I introduce previous research on the dual in Middle English. On the one hand, I present the texts which are reported to still preserve dual forms and, on the other hand, I also comment on whatever information is available on the dialectal distribution of this feature. 2.3.1 The dual personal pronoun in the Germanic languages Dual forms are attested in the paradigm of the 1st and 2nd person personal pronouns in all Germanic languages. It has survived longest in the North and continental West Germanic languages (Seebold 1984: 16 - 17). The following dual forms are attested (or reconstructed) for the early Germanic languages: 50 2. Historical Background and Previous Work <?page no="69"?> Table 2.3: Dual forms of the personal pronoun in the Germanic languages according to Seebold (1984). Goth ON OE OFris OS OHG 1st Person Nom. wit vit wit *wit wit - Acc. ugkis ok(k)r unc, uncit *unk unk - Gen. *ugkara okkar uncer unkero unke ˉ r Possessives *ugkar okkarr uncer *unk unka - Dat. ugkis ok(k)r unc *unk unk - 2nd Person Nom. *jut it git *jit git *iz Acc. igqis yk(k)r inc, incit *iunk ink *ink, (*inkiz) Gen. igqara ykkar incer - *inker(o) - Possessives igqar ykkarr incer *junk inka *inker(er) Dat. igqis yk(k)r inc *iunk ink *ink It is most likely that the dental in the nominative form is a remnant of the numeral two and would directly translate into ‘ we two ’ and ‘ you two ’ ; however, the exact form that underlies the Germanic forms is doubtful: Bammesberger (1984: § 4.2.2.5) and Lass (1994 a: 140) do not even attempt to provide any reconstructed form, Prokosch (1939: 284) and Hogg and Fulk (2011: § 5.25) suggest the numeral *dw ŏ , as in L duo, Gk δύω , δύο and, finally, Schmidt (1978: 170) and Seebold (1984: 25) propose a w-less form *de, which is attested in Hittite ta- ‘ two ’ and possibly in the first element of IE*de-k ’ t ‘ ten ’ = ‘ two hands ’ . The Gothic 2nd person nominative form is not attested, but has been safely reconstructed to Goth *jut, so the first element is in line with Skr yuvám and Lith jù-du (Prokosch, 1938: 285; Seebold 1984: 26; Braune and Heidermanns 2004: § 150, Anm. 2). It is very likely that the corresponding 2nd person nominative form in the North and West Germanic languages (NWGmc *jit) is modeled after *wit by analogy (Seebold 1984: 26; Bammesberger 1984: § 4.2.2.6; Hogg and Fulk 2011: § 5.31). The forms of the oblique cases and the possessive forms are difficult to account for, but it is possible that they ultimately go back in some way or another to the 1st and 2nd person plural stems of pronouns (cf. Prokosch 1939; Seebold 1984: 37). I refrain from repeating the attempts to explain the origin of the oblique cases here, as both Prokosch ’ s and Seebold ’ s line of argumentation include several elements of the uncertain. Table 2.3 shows that the dual paradigm is only attested in its entirety in Old English (cf. section 2.3.2) and Old Norse. In the Scandinavian languages the original tri-partite number distinction was lost completely only in the 18th century. For instance, Guðmundsson (1972: 29, 87) observed a very slow demise of about 500 years in Icelandic, which started in Southern Iceland and spread to other parts of the island in the course of 2.3 Dual forms of the personal pronoun 51 <?page no="70"?> time. Generally, the dual forms have survived longer in the western Scandinavian languages (and to some extent Swedish) by adopting a plural or honorific meaning (Guðmundsson 1972: 126). This semantic shift can also be observed in Modern Yiddish (Howe 1996: 283). Although there are no attested forms for Old Frisian, we can safely assume that they must have existed, as they were preserved in Modern North Frisian until relatively recently (Howe 1996: 193 - 194). Similarly, only one single form is attested in Old High German: unke ˉ r appears once in Otfrid, and there it is even reinforced with zweio, which may indicate that the forms were unusual at the time (Howe 1996: 244; Braune and Reiffenstein 2004: § 282, Anm. 1). 17 The 2nd person dual forms are attested after the 13th century in the Bavarian and some South Westphalian varieties of Middle High German; however, they underwent a functional shift and assumed plural meaning (Kranzmayer 1954). According to Howe (1996: 257), the dual is more frequent in Old Saxon written evidence than in Old High German. However, in Old Saxon there is also evidence that the dual is declining. Manuscript C of the Heliand replaces the dual occasionally with plural forms where MS M has the dual git and, furthermore, MS C complements the dual twice with numerals, as in Otfrid above. 18 Finally, in Gothic the dual is only attested in Wulfila ’ s bible translation. As mentioned above, unlike other Germanic languages, Gothic also distinguishes dual forms in verbal inflections. The use of dual forms in the extant manuscript of Wulfila ’ s bible translation, the Codex Argenteus, is inconsistent and there are passages with dual reference in which only dual forms, only plural forms, or both alternating appear. This is treated in detail by Seppänen (1985), who critically discusses earlier studies (Cuny 1906; Meillet 1908), which had suggested that the dual in Gothic was sensitive to gender, i. e. only used when the referees are of the same sex. For almost a century this view was taken at face value, but Seppänen convincingly shows by comparison to other Germanic languages that there is free variation of dual and plural forms in contexts with dual reference but concludes that in the passages with dual reference in Wulfila ’ s bible translation, the dual is the preferred variant, “ even though the plural, accounting for well over a third of all occurrences, is far more than a merely marginal phenomenon ” (Seppänen 1985: 33). Because of the small body of surviving records in general, it is impossible to tell what the general use of the dual in Gothic might have been. 17 Kranzmayer (1954: 250) suggests that the lack of dual forms in OHG may also be attributed to the fact that a large part of OHG literature was translated from Latin, which did not know the dual. 18 C = London, British Museum, Cotton Caligula A.VII; M = Munich, Bayerische Staatsbibliothek, Cgm 25. 52 2. Historical Background and Previous Work <?page no="71"?> 2.3.2 The dual in OE All forms of the first and second person dual personal pronouns are attested in Old English, but the genitive was replaced by the relevant possessive form; i. e. the two forms are no longer distinguished as for instance in Old Norse or, to some extent, in Gothic and Old Saxon (cf. Table 2.3). The dual forms are frequent in both Early and Late West Saxon and in verse (cf. Campbell 1959: § 703; Howe 1996: 135), but elsewhere they are rarely attested: According to the standard handbooks (e. g. Campbell 1959: § 703; Howe 1996: 135), they are found only in the Mercian gloss to the Rushworth Gospels and in the inscription on the Ruthwell Cross; however, this view has to be adjusted. Stiles (1996: 557) also discusses two instances of uncet in the prose Life of St Malchus in manuscript London, British Library, Cotton Otho C.i and points out that this text has a Mercian provenance according to Sisam (1953: 210). Furthermore, Stiles (1996: 562) mentions the possessive form unce in Charter 1188 19 , which according to Bryan (1915) contains a linguistically composite text of Kentish and Anglian. Even though the dual forms are attested in all Old English dialects to some extent, they were by no means consistently used by the scribes. Seppänen (1985: 15 - 16) examined the six extant versions of the West Saxon gospels, Ælfric ’ s translation of the Old Testament and the gloss to the Rushworth Gospels and concludes that the dual is usually incoherently employed in contexts with reference to two people: There are passages in which the dual is employed consistently, passages in which only plural forms are found and, finally, passages in which dual and plural forms alternate. One could argue that this might be due to transmission errors in the process of copying these texts but, because the Mercian gloss to the Rushworth Gospels is the holograph of a priest called Farman of Harwood, Seppänen (1985: 18) concludes that this free variation of dual and plural forms was natural to speakers of Mercian and not due to corruption of the texts by later scribes. Furthermore, he argues that by analogy the same must have been the case for West Saxon. Of special interest are the accusative forms uncet, uncit and in ċ et, in ċ it, which appear to have no equivalents in any other Germanic language (Stiles 1996: 558; Hogg and Fulk 2011: 204). They are attested in Anglian texts or texts with an “ Anglian colouring ” (Stiles 1996: 557) and, generally, it is believed that the ending of the nominative dual forms wit and git were transferred to the accusative form by analogy (-et from an unaccented form) (Campbell 1959 § 703, footnote 1; Bahnick 1973: 153; Bammesberger 1984: 43; Seebold 1984: 32; Hogg and Fulk 2011: 204). Stiles (1996: 559 - 562), on the other hand, 19 The numbering of the charter corresponds to Sawyer (1968). 2.3 Dual forms of the personal pronoun 53 <?page no="72"?> attempts to explain these forms by analogy to accusative plural forms with -ik endings, as shown in (22): (22) 1st person OE (Angl) usi ċ , OS unsik, OLFrk. unsig, OHG unsih 2nd person OE (Angl) eowi ċ , OHG iuwih According to Stiles, the forms uncet, uncit and in ċ et, in ċ it are likely to show this suffix for three reasons. First, the forms are in the accusative case. Second, just like -ik the suffix -it does not cause i-mutation and, third, the forms are not attested in West Saxon and thus show the same regional distribution in Old English as the -i ċ forms of the accusative personal pronouns. While this is very convincing, the forms are more difficult to explain phonologically: The change in point of articulation that brought about its particular phonological shape, *-it, is to be explained as the result of dissimilation from the voiceless dorsal stem-final consonants of the dual pronoun forms to which it was appended. (Stiles 1996: 560) Stiles admits that “ [w]e cannot know the precise phonetic details of the process ” (1996: 560) and I think that his explanation is not very satisfactory either. Finally, a third explanation is offered by Shields (2001), who tries to link the forms with an Indo-European non-singular marker *-(e/ o)T. However, this is also problematic as it does not account for the fact that the forms are only found in Old English. While the dual forms syntactically are used like any other personal pronouns, there is one special construction in Old English (and Old Norse) consisting of the personal pronoun (mostly dual) and a personal name. Krogmann (1956: 146) discusses the constructions wit Scilling and unc Adame, which he reports to be frequent in both poetry and prose. The construction is an abbreviation for what in Modern English would have to be translated as ‘ we two, I and Scilling ’ or ‘ to us both, to me and Adam ’ . Krogmann (1956: 146 - 147) provides the following attestations for the construction: 20 (23) 1st person unc Adame GenA,B [0141(384)] wit Scilling Wid [0042(103)] wit Adam Sat [0147(405)] wyt Æþered Ch 1507 [0003(7)] uncer Grendles Beo [0559(1999)] to uncer Wulfrices ealdgemere Ch 468 [0006(6)] healf uncer Brentinges Ch 495 [0033 (33)] 2nd person git Iohannis Hell [0036(133)] 20 The constructions are referred to by their OE short title and respective citation in the Dictionary of Old English Web Corpus. 54 2. Historical Background and Previous Work <?page no="73"?> Interestingly, Krogmann (1956: 148) also lists two instances of the construction with third person pronouns, for which no dual is known: hy Osweo Bede 4 [0030 (1256.17)] and to him Arone CP [0983 (28 201.4)]. To sum up, it seems that the dual forms are attested to some extent in all OE dialects, but inconsistently. Most research was carried out on the special accusative forms of the Anglian dialects and it appears that a complete analysis of dual forms in Old English texts is still wanting. Although Campbell (1959: § 703) notes that dual forms are found frequently in Early West Saxon, Late West Saxon, and in verse, it is not evident from the handbooks on the basis of which texts this claim is made. 2.3.3 The dual in ME The current state of research is not much different for Middle English. As already mentioned above, the major Middle English handbooks only treat the dual forms very marginally and do not go further than merely mentioning that they exist: The duals wit and ȝ it still occur in the 13th century; later they are given up and replaced by the plural. (Brunner 1963: § 52) Old English still possessed special forms for the dual, for the first two persons (cp. HOE § 68 f.). A few traces of these remained up to the start of the 13th century, at least for the first person; there was the subject case wit ‘ we two ’ (5/ 14 039, 7/ 7) and the genitive (become possessive) unker (6/ 993 in unker s ı ˉ þe ‘ in our case, i. e., for both of us ’ ). (Mossé 1952: § 64) (4) The dual forms, wit ‘ we two ’ and unker, unk continued to be used in early ME until the 13th century [. . .] (9) The dual forms, ( ȝ it ‘ you two ’ ), inker, inc) continued to be used in early ME until the 13th century. (Iglesias-Rábade 2003: § 75 and § 77) While Brunner does not mention any texts, Mossé refers to a line in La ȝ amon ’ s Brut and the Ormulum in the case of the nominative and to The Owl and the Nightingale in the case of the genitive. Both Brunner and Mossé just list a few forms and the reader could get the impression that the paradigm is incomplete in Middle English. The entire paradigm is only mentioned in a more recent grammar (Iglesias-Rábade 2003); however, Iglesias-Rábade also only gives a few Middle English lines as examples, i. e. from La ȝ amon ’ s Brut, The Owl and the Nightingale and Sawles Warde. 2.3 Dual forms of the personal pronoun 55 <?page no="74"?> Table 2.4: Dual forms recorded in Diehn (1901) and Kennedy (1915: 42 - 43). 21 Text Nominative Genitive-Possessive Dative-Accusative Total 1st person 2nd person 1st person 2nd person 1st person 2nd person Form N Form N Form N Form N Form N Form N Ev L wyt ? Ev H Mt gyt ? inc ? Ev H Mc ginc ? gunc ? Ev RL hinc ? Gen. & Ex. wit ? gunker 1 unc ? gunc 1 2 Hali M. inker 2 inc 1 3 Hav. unker 1 1 Jes. unc ? Jul. inc 1 1 Kath. wit ? unc ? La ȝ . A wit ? ȝ it 9 unker ? incker 1 unc ? inc 8 26 wet ? ȝ et 4 unkere ? inckere 1 unke ? hit 1 vinkere ? unker 2 Lamb. Hom. ȝ it 1 hinc 1 2 Marh. unc ? inc 1 1 O. & N. unker ? Hunke 1 1 Orm. witt ? ȝ itt 20 ȝ unnkerr 11 unnc ? ȝ unnc 20 51 S. Warde incker 1 1 V. & V. wit ? ȝ it 4 ȝ inker 1 unc ? ȝ inc 1 10 ȝ unker 2 ȝ ing 1 ȝ ung 1 Total 38 23 38 99 A fuller account of dual forms is provided by Diehn (1901) and Kennedy (1915). The latter is a study on the pronouns of address in English literature of the 13th century and, due to its narrower scope, a more comprehensive report is only provided on the second person dual forms. According to Kennedy (1915: 42 - 43), the dual forms only survive until the first quarter of the 13th 21 Since Diehn (1901) only lists types but does not count tokens, numbers are missing for the first person forms and for those texts that contain second person forms not listed in Kennedy (1915). 56 2. Historical Background and Previous Work <?page no="75"?> century, but he mentions Genesis and Exodus and Havelok as exceptions in which dual forms appear later, though in isolation. The forms listed in Diehn (1901) and Kennedy (1915) are compiled in Table 2.4, which illustrates that the attested forms are scattered across various texts but, according to this overview, the entire paradigm is only attested in the A text of La ȝ amon ’ s Brut 22 . Kennedy was struck by the many competing forms of the dual and concluded that the oblique forms show “ a variety of forms quite out of proportion to the number of occurrences ” (1915: 43). According to Table 2.4, this is especially noteworthy for the forms in La ȝ amon ’ s Brut and Vices and Virtues. Kennedy attributes this abundance of forms to the lack of familiarity with the forms, which manifests itself especially in the following two tendencies: First, Kennedy (1915: 43) notes the occasional confusion of forms with the dual of the 1st person, as in the case of unker in the A text of La ȝ amon ’ s Brut and in Havelok. Second, Kennedy points out that the forms of the oblique cases had a tendency to add initial ȝ or g-, as in Genesis and Exodus, the Ormulum and Vices and Virtues. Howe (1996: 137 - 138) attributes this to analogy to the nominative form ȝ it or the plural forms of the personal pronoun like ȝ e ˉ , ȝ ow, ȝ ou, ȝ u ˉ r(e), ȝ owr(e), ȝ oures, ȝ ouren (cf. also DOE git). Kennedy (1915: 77 - 79) also discusses some syntactic aspects of the dual forms in Middle English and he notes that the constructions generally correspond to those of the regular plural forms. Nevertheless, his examples and comments indicate that the Middle English dual forms are very often reinforced with a form of the lexical quantifier ‘ both ’ , as in example (24). (24) to ȝ unnkerr baþre gode (Orm. 6183) þat ic mote ful ȝ in and buhsum bien ȝ ing bam (V. & V. 97, 13) This phenomenon is also discussed by Howe (1996: 141), who mentions that a form of two is also occasionally added to the dual. This supplementation of dual forms seems to be a more wide-spread phenomenon, as it is already attested in some Old English texts (cf. DOE git, 1.a.) and also known from other Germanic languages: (25) ON ykkur beggja, vit báðer, okr báða, it tueir, ykr tuá OS uuit bêðia OHG unkêr zweio (cf. Guðmundsson 1972: 95) Guðmundsson (1972: 95) explains that this phenomenon is also known from Irish, Greek, and Slavonic and notes that generally “ dual forms survived 22 Kennedy (1915: 42) states that the dual forms are replaced by plural forms in the second known version. 2.3 Dual forms of the personal pronoun 57 <?page no="76"?> longest in conjunction with words meaning two or both ” . According to Howe these quantifiers serve as “ clarifications aimed at facilitating the task of the hearer in the communicative situation ” (1996: 77). Therefore, this reinforcement of duality with lexical quantifiers suggests a lack of familiarity with the forms and indicates that these are only the last remnants of the dual system in English. Howe (1996) provides the most recent study which contains an (admittedly short) account of dual forms. He includes the following paradigm of dual forms in Middle English: Table 2.5: Dual forms in Middle English (Howe 1996: 138). 1st person 2nd person Nom. wit ȝ it Acc. unc inc, ȝ inc, ȝ unc, ((h)unk) Gen./ Poss. unker inker, ȝ inker, ȝ unker (unker) Dat. unc inc, ȝ inc, ȝ unc, ((h)unk) It is striking that in Howe ’ s table there is an abundance of different 2nd person forms, while there is only one form per case for the 1st person. The forms of the 2nd person dual pronouns correspond, in normalized form, to those listed in Kennedy (1915). This suggests that Howe ’ s main source for the dual in English, an obscure unpublished Oxford Seminar paper by Ladd and Radice (1951/ 52), must be based on Kennedy (1915). 23 This would explain why Howe does not refer to any other texts than those just mentioned. Interestingly, Howe (1996) cites Kennedy (1915) in his bibliography, but does not mention the study anywhere in the chapter on Middle English. Nevertheless, Howe ’ s (1996: 141) account of Ladd and Radice ’ s (1951/ 52) findings is still of interest for the present study, as it summarizes the findings with respect to their spatial distribution. Howe concludes that in the West Midlands and South some dual forms are still found in the early thirteenth century but must have been lost soon after as they occur inconsistently and are often replaced by the respective plural forms. Furthermore, Howe notes, probably again relying on Ladd and Radice ’ s findings, that the forms are “ best preserved, used more consistently, and survive longest in the East Midlands ” (1996: 141). This is remarkable for two reasons. First, as mentioned before, the Old English grammars report dual forms as being most frequent in West Saxon texts, i. e. in a different area, and, second, it is the South-Western 23 Unfortunately, the paper was unavailable to me and I could not verify this claim. 58 2. Historical Background and Previous Work <?page no="77"?> Middle English varieties that are generally regarded as preserving conservative forms longest. 2.4 Summary This chapter has highlighted the development of the three variables analyzed in this study. Special emphasis was given to the spatial development of the three variables which, in the case of the change OE a ˉ > ME ǭ , has already been well-studied for decades but never with a corpus of Early middle English texts as large as the LAEME CTT. The historical introduction of the three variables has shown that they offer unique and individual aspects that should be analyzed independently, but at the same time they all contribute in their own ways towards the understanding of diffusion and change in Early Middle English. As a whole, the discussion of previous work has shown that case studies like the one presented in this research project are lacking for many Middle English phenomena and up-to-date maps are not available for most variables. 2.4 Summary 59 <?page no="78"?> 3. Material and Methodology 3.1 The Linguistic Atlas of Early Middle English (LAEME) The data for this study is drawn from the LAEME Corpus of Tagged Texts (CTT). This corpus is innovative in many respects and it is certainly not exaggerated to state that it has revolutionized the field of (Early) Middle English dialectology, as for this period it is unprecedented in its size and collection of texts. The LAEME CTT includes 650,000 tagged words in 167 texts from 105 manuscripts. 1 For comparison, this is more than three times as large as the corresponding parts of the Helsinki Corpus, which only includes about 200,000 words in 54 texts from 33 files (Kytö 1996: 2 and 238 - 240). The Linguistic Atlas of Late Mediaeval English (McIntosh et al. 1986) relies on a set of questionnaires, which form the basis for the linguistic profiles that provide the source material for the maps (cf. section 1.2.3). The use of questionnaires was abandoned in the compilation of LAEME and instead it includes faithful transcriptions of manuscripts or facsimile editions (Laing and Lass 2006: 26). Lass (e. g. 1997, 2004) has repeatedly stressed the importance of consulting the original manuscripts and of using faithful transcriptions: Modern editions often alter the original text, e. g. by normalizing the spelling system, by trying to reconstruct a lost archetype or by “ correcting ” apparent errors by the scribes. Such edited texts are only partially useful for studies in language variation and change as important evidence, such as the phonological or morphological realization of a form, might be lost (cf. discussion in Vásquez and Marqués-Aguado 2012). Initially, the aim of LAEME was to include texts transcribed in their entirety but, unfortunately, this plan had to be abandoned because it would have taken too long for the project to be completed (Laing and Lass 2008 - : 3/ 2). Therefore, for some of the longer Early Middle English texts the corpus now only contains samples, e. g. the Ormulum or Ancrene Wisse. However, each of these samples contains more than 10,000 words, and it is possible that the missing parts will be transcribed at a later stage. For instance, in the entry to the Ormulum Laing expresses this wish: “ It is hoped that more will be added to this sample at a later date ” (2008 - : 162). The text files vary greatly in length, ranging 1 <http: / / www.lel.ed.ac.uk/ ihd/ laeme1/ front_page/ about_laeme.html> (accessed 16 March 2011). <?page no="79"?> from 18 words in the case of text file #130 to 32 091 words in text file #1600. 2 This has the implication that only the larger texts allow quantitative studies, but the shorter texts demand a more qualitative approach, as the occurrence of a feature might be purely coincidental. Furthermore, many of the small texts cannot be used for this study on diffusion, because they contain too little linguistic material for a safe localization of the text and, therefore, they do not appear in the maps. However, they are nevertheless included in the data collection. LAEME uses an innovative, fine-grained spatial and temporal coding system. If possible, each LAEME text file is dated to a certain quarter of a century. For this purpose, a three-part system is adopted, which includes the following elements: C12 - 14, a/ b and 1/ 2. The first part refers to a certain century, the second part to the first or second half of a century, and the final part to a certain quarter. For example, the abbreviation C13b2 refers to the last quarter of the 13th century. If a narrow dating is not possible for a certain text, it is assigned only the first or first and second part of this system. Furthermore, if a text is a copy of a pre-Conquest original, this is indicated in the corpus by an asterisk (Laing 2008 - : 1). Concerning the localization of texts, so far only larger areas like the traditional Middle English dialect areas East Midland, West Midland, Northern, Southern, and Kentish have been distinguished. This pre-categorization into certain areas was abandoned for LAEME. Instead, each localizable text file was assigned both a six-digit national grid reference and a rough location in geographic terms. For example, text file #286, which contains the output of hand A of the South English Legendary in Cambridge, Corpus Christi College 145, is localized to the national grid reference 429 195 or, in geographic terms, to NW Berkshire. Currently, only two thirds (111 texts) of all text files are localizable. 3 To sum up, the LAEME CTT includes an unprecedented collection of Early Middle English material and enables studies that have been impossible in so much detail. Although LAEME offers an innovative research tool that facilitates investigations into Early Middle English language change, there are a few methodological problems that are addressed in the next section. 2 #130 is a fragment of the lyric Stella Maris found in Oxford, Bodleian Library, Rawlinson C 510, a manuscript that otherwise contains a theological collection in Latin. #1600 includes the output of scribe A of manuscript Oxford, Bodleian Library Laud Misc 108 (Laing 2008 - : 171 and 164 - 165). 3 <http: / / www.lel.ed.ac.uk/ ihd/ laeme1/ front_page/ about_laeme.html> (accessed 16 March 2011). 3.1 The Linguistic Atlas of Early Middle English (LAEME) 61 <?page no="80"?> 3.2 Methodological considerations Many of the problems one has to face when working with LAEME, and Early Middle English material in general, were already outlined by Laing (1995) at a very early stage of the project, and then again, in more detail, in Laing (2000). The following sections take a closer look at these problems, which concern text types, coverage, lack of documentary anchor texts, and linguistically composite texts. In each section I address the implications for the data collected and analyzed in the present study. 3.2.1 Anchor texts or “ the best available witnesses ” Dees (1994: 119), who compiled a similar atlas of Old French charters (Dees 1980), states that one should rely on “ the best available witnesses ” in order to collect data for a dialect atlas and that ideally these include localizable and datable documents. The potential sources for an Early Middle English dialect atlas (cf. Laing 1993) include documentary texts; however, not all of these proved to be useful for the compilation of LAEME. While the editors of LALME (McIntosh et al. 1986) could rely on localizable Late Middle English documents to build up a network of anchor texts (cf. section 1.2.3) into which the unlocalizable texts could be “ fitted ” , this procedure was not possible for LAEME for two reasons: first, there are hardly any original English documentary texts from the thirteenth and early fourteenth centuries and, second, preliminary studies have shown that most post-Conquest documentary texts up to the 12th century are more or less faithful copies of Old English charters into the cartularies of some religious houses (Laing 1993: 3 - 4, 2000: 105; Kitson 1995: 48 - 49). Therefore, the index of sources contains only 17 texts that are listed as documentary anchor texts (Laing 2008 - ). As one has to rely on whatever material has survived from the Early Middle English period, one cannot simply dismiss material that does not fit the modern standards. Laing notes that “ [n]ot all the problems involved in making of LAEME will be solvable, but the data provided by early ME scribes are the product of intelligent, conscientious professionals and must all be treated with respect if any progress is to be made ” (1995: 5). Therefore, in the case of LAEME, the best available witnesses also include literary manuscripts, which were localized on the grounds of non-linguistic evidence. According to Laing (2000: 105) this includes, for instance, library marks or inscriptions that link a manuscript to a certain religious house. Furthermore, it is sometimes possible to localize a manuscript on the basis of content, script, or illumination. Manuscript Maidstone Museum A.13 (#66 - #68), which contains, among other content (mainly Latin, but also English), an incomplete version of Proverbs of 62 3. Material and Methodology <?page no="81"?> Alfred, has associations with the Cluniac Priory of St Andrew at Northampton. Laing explains that the English texts in the manuscript vary greatly in their language, but she localizes the Proverbs of Alfred (#66) to Northampton, as in her view the language of this text is “ consistent with an origin in the Northampton area ” (2008 - : 134 - 138). The languages of the other two text files from this manuscript differ and they have been localized independently: Text file #67 was fitted to Sussex and text file #68 was not placed at all, as it is too short and contains too little linguistic evidence for a safe localization. Any text that could not be localized on non-linguistic grounds were, if possible, fitted into the network of localizable texts by applying the fittechnique as described in section 1.2.3. In some cases the LALME locations were used to support the fittings but, if there was doubt about a certain location or if the language was too mixed, the text file was included in the LAEME corpus without being given a grid reference. Nevertheless, the location of most LAEME files are only relative to each other and the compilers of LAEME stress that “ in the sparsest areas any localisation is bound to be very approximate indeed and will always be subject to subsequent revision if more data or information becomes available ” (Laing and Lass 2008 - : 1/ 11). 3.2.2 Coverage of texts in LAEME Dees notes that “ [t]he network of geographical points should in principle be as dense as possible ” (1994: 119). Unfortunately, the Early Middle English data provides a rather patchy coverage not only spatially, but also temporally (cf. Laing 1995, 2000; Laing and Lass 2006). Curzan reminds us that “ a historical corpus can only be as thorough as the available texts ” (2009: 1098). In other words, one cannot but rely on whatever data is available from the period or, as Labov aptly put it, one cannot but make “ the best use of bad data ” (1994: 11). It is evident from any map showing all the localized LAEME texts (cf. Map 3.1, Appendix IV) 4 that there is a gap between the East and West Midlands, to which Laing (2000) refers as the ‘ East-West divide ’ and that there are hardly any surviving texts from the very North. Furthermore, the extant texts from the North are all rather late, so there are significant gaps not only in the spatial continuum, but also on a temporal axis. When Dees made the above statement on the density of atlases, he had isogloss maps in mind, in which the line between two locations relies on the 4 A map with all localized texts and a key to their respective file numbers can be downloaded from <www.lel.ed.ac.uk/ ihd/ laeme1/ pdf/ laeme1_keymap_mon.pdf> (last accessed June 2013). 3.2 Methodological considerations 63 <?page no="82"?> interpretation of the producer of the map. In this respect one can say that the denser the network, the more reliable the isoglosses (cf. section 3.5.1). Since feature maps and not isogloss maps are the intended output of both LAEME and the present study, the density of the texts is less important. 3.2.3 Text languages in LAEME As already stated in the introduction, there are very few extant holographs from the Early Middle English period (the most notable exceptions in the period being the Ormulum and Dan Michel ’ s Ayenbite of Inwyt (cf. Laing and Lass 2006: 423; Horobin and Smith 2002: 33). Therefore, when working with Early Middle English material, one has to bear in mind that in most cases one is not analyzing the original text, but copies written in different periods or locations or transcribed by scribes with different linguistic backgrounds. McIntosh (1973: 61) describes the habits of medieval scribes copying a text with a dialect different from their own as follows: Scribes hardly ever leave the text of a manuscript without any alterations (type A), more often they change it to their own dialects (type B), or they do something in between, sometimes changing a word and sometimes not (type C). Laing (1993: 10) assumes that the scribes in the Early Middle English period were trained to copy Latin texts literatim and may have transferred this practice to the copying of English. In general, there seem to be more scribes of McIntosh ’ s type A in the Early Middle English period than in later periods (Smith 1992: 583; Laing 1993: 10, 2000: 100; Laing and Lass 2008 - : 1/ 7, footnote 18). Although this means that, generally speaking, Early Middle English texts were maybe less mixed linguistically than those from later periods, all copying strategies can be shown to have been employed during the period. For this reason, it is important to make the distinction between the language of a scribe and the ‘ text language ’ as it appears in the surviving manuscript. This concept, which was developed by Fleischmann (2000), equates the output of scribes and copyists of a dead language with modern informants, i. e. the focus is on the text itself rather than on the speakers of a language. Laing and Lass (2006: 418) mention three great differences between text languages and modern informants. First, one has to rely on whatever material happens to survive, sometimes by mere chance: “ [T]he data corpus of a text language is finite; new data only become available when previously unknown documents are discovered, whether in the form of manuscripts, printed texts, tablets, etc. ” (Fleischmann 2000: 34). It is unlikely that any noteworthy amount of new material will be discovered in the future, and LAEME probably already contains most of the material that will ever be available. Although the corpus is by no means complete - at this stage some important texts have not 64 3. Material and Methodology <?page no="83"?> been transcribed and tagged yet, for instance three early 14th-century versions of The South English Legendary (Laing and Lass 2008 - : 3/ 3) - it, nevertheless, includes most known English texts from the period. Second, sociolinguistic information is scarce for this early material. The group of scribes responsible for the data included in LAEME is rather homogeneous. Laing and Lass (2006: 418) summarize it as follows: “ we are dealing with the output of literate, adult, English Catholics largely inhabiting closed institutions (monasteries, professions) where class stratification would be of little importance. ” This has the implication that other variables might have to be examined, for instance monastic order of the scribes or the scriptoria where the texts were produced (cf. Laing and Lass 2006: 419 - 420). Since many of the texts are fitted and not associated with any real scriptorium, it is not feasible to analyze the role of such factors in a study of the entire material that is collected in LAEME CTT. Therefore, I only focus on the diachronic and diatopic development of the variables in the present study. Finally, it is obvious that a historical corpus like the LAEME CTT can only include written sources. As already mentioned above, it is difficult to determine the extent to which the output of Early Middle English scribes reflects local usage and, therefore, one has to accept the texts and the kind of variation they display as linguistic witnesses in their own right: Written language, particularly in preor non-standardised varieties, such as Middle English, may show just as great a deal of variation as spoken language, on all levels of language, from orthography to discourse. For the sociolinguist, no matter whether concerned with present-day or historical data, any kind of variation will do. (Bergs 2005: 18, emphasis in original) In addition, Laing and Lass make a point that “ [t]he entire feasibility of Middle English dialectology [. . .] relies on the fact that it is nevertheless possible to isolate examples of genuine individual and local linguistic usage ” (2006: 423). However, many Early Middle English texts are far from being homogeneous. Benskin and Laing (1981) discuss the different types of scribes (McIntosh 1973) as outlined above, but add that a single scribe need not be bound to any one of these treatments over the whole course of a single text or collection of texts: although at any given point of text his treatment is describable as one and only one of these three types, a copyist may shift from one type of treatment to another, and ‘ translational drift ’ from type C to type B is in fact very common. (1981: 56, emphasis in original) Benskin and Laing refer to such a text as a “ progressively translated text ” (1981: 66). They explain that if a scribe started to copy a text with an unfamiliar language or dialect, he might have faithfully copied in the beginning; however, 3.2 Methodological considerations 65 <?page no="84"?> as he became more and more acquainted with the text, he started to convert the text into his own language or dialect. Such a procedure results in what Benskin and Laing (1981) call a ‘ Mischsprache ’ , i. e. “ linguistic output containing two or more elements that are mutually incompatible: that is, from non-continguous areas within the established dialect continuum ” (Laing and Lass 2008 - : 1/ 7). Texts containing ‘ Mischsprachen ’ cannot be assigned to a single place of origin and for this reason they are only part of LAEME if Laing and Lass were able to localize at least one part: “ It has proved possible in a number of cases to isolate within a text in mixed language one or more linguistic layers which can be taken to represent genuine regional usages ” (Laing and Lass 2008 - : 1/ 14). This means that, occasionally, one single text corresponds to different text languages, which all have separate LAEME entries. I mentioned above that there are 167 text files from 105 manuscripts. This higher number of text files is a result of the practice of allocating a separate text file to each text language. In fact, the LAEME text files correspond to several possible ways of manuscript/ scribe/ text file relation, which are illustrated in examples (26) - (29), all based on the discussion in Laing and Lass (2008 - : 3/ 3 - 4): (26) MS Scribe Text # Example (26) shows the most straightforward relationship between manuscript, scribe, and text file: The LAEME text file corresponds to the output of one single scribe in one single manuscript. For example, this is the case in #155 (Cambridge, Corpus Christi College 444), a verse version of Genesis and Exodus on 81 folios written in a single hand (Laing 2008 - : 9 - 10). (27) Scribe Text # Text # MS The second possibility, as illustrated in (27), takes into account that a single scribe in a manuscript may have produced several types of language differing considerably from each other (Laing and Lass 2008 - : 3/ 4). The example provided by Laing and Lass refers to text files #2 and #3, which contain The Owl and the Nightingale in the second part of manuscript London, British 66 3. Material and Methodology <?page no="85"?> Library, Cotton Caligula A.ix (ff. 233 r - 246 r). 5 There must have been two separate languages in an ancestor of The Owl and the Nightingale, but the scribe of the Cotton manuscript was a literatim copyist and preserved both original languages (cf. Laing 2008 - : 65). (28) MS Scribe Text # Scribe Text # In some cases more than one scribe worked on a manuscript. As a consequence the work of each scribe is then grouped in a separate text file in the LAEME CTT, even if the language of the separate scribes is homogeneous and can be located to one place. This is, for instance, the case with the output of scribes A, B, and C in manuscript London, British Library, Royal 17.A.xxvii. This manuscript appears in LAEME as text files #260, #261, and #262 (Laing and Lass 2008 - : 3/ 4). Sometimes, if scribes took turns in copying a manuscript, each passage was first given a separate file number. Some of these text files were later reassembled under a four digit number if the output proved to be homogenous after further analysis such as the respective passages of scribes A (#1200) and B (#1300) of the Trinity Homilies (Laing and Lass 2008 - : 3/ 4). (29) Text Scribe Text # Text Finally, a single scribe may have produced more than one text as shown in (29). If the output of a scribe is homogenous across several texts in the same manuscripts, they are compiled in one single text file (Laing and Lass 2008 - : 3/ 4). This is, for instance, the case in #163 (Aberdeen University Library 154), a Latin manuscript that contains a folio with a couplet and three quatrains in English, all written by the same hand (Laing 2008 - : 3). 5 Part one of the manuscript (ff. 3 r-194 v) contains version A of La ȝ amon's Brut. Laing (2008 - : 65) points out that “ earlier scholarly opinion considered the two parts of the manuscript to have been originally separate ” . She further refers to personal communication with Malcolm Parkes, who confirmed this on the grounds of paleographical evidence. 3.2 Methodological considerations 67 <?page no="86"?> This system is essential for the present study as each localizable text or text fragment is given its own national grid reference. I rely on the expertise of Laing and Lass (2008 - ) and on their decisions in the process of disentangling the various linguistically more or less homogeneous passages, as years of experience are necessary for this task. 3.2.4 Implications for the present study While the application of the fit-technique has the great advantage that texts can be localized at all, it also has the disadvantage that it is highly problematic for a study on diffusion in Early Middle English, as the texts are localized relative to each other within a continuum. As only those texts that could be localized with a high degree of confidence are given an exact location, I believe that it is nevertheless possible to use LAEME for mapping purposes as intended in the present project. However, it is important to remember two points in the discussion of the results: First, in order to avoid circular reasoning, one of the research questions of the study has to be whether the location of those texts in LAEME that are localized through the fittechnique can be confirmed on the basis of diffusion models. In this respect, such a study is essential, as the discussion of the data might confirm or put in question the relative fitting of some texts. Second, conclusions might have to be drawn on a more general level that is, for instance, in terms of East vs. West, as the exact locations of texts within a certain area are only tentative. This focus on the East and West Midlands is important for a second reason. There are hardly any texts from the South and no early texts from the North but, in the East and West Midlands, the density of texts is comparably higher and texts have survived from the entire period. In section 1.3.2 I introduced two studies on spatial diffusion in Late Middle English (Conde Silvestre and Hernández Campoy 2002; Bergs 2006) and I already addressed some of the problems of their respective approaches. Based on the assessment of the material that is available from the Early Middle English period I have to reject the methodology employed in both projects for the present study. First of all, the elaborate model developed by Conde Silvestre and Hernández Campoy (2002) cannot be applied as such to the Early Middle English period, since most innovations in Early Middle English do not originate in London (cf. chapter 2). Second, the towns which they considered in their model were not necessarily the most important ones at the beginning of the Early Middle English period. Donkin notes that “ [i]n England, as in other parts of Europe, the twelfth and thirteenth centuries were characterized by a considerable urban development. Existing centers of trade increased in size; villages grew to urban rank; and entirely new towns were founded ” 68 3. Material and Methodology <?page no="87"?> (1973: 123). Similarly, Astill (2000: 46) describes a “ rapid urban growth ” in the Early Middle English period, which is evident for instance in taxation records. Donkin (1973: 134) presents an overview of the ranking of provincial towns based on the feudal aids collected under Henry II (1154 - 89) and on the Lay Subsidy Rolls (1334), and it is evident from this comparison that there are great differences within the Early Middle English period. For instance, Exeter ranks as the 6th town in the 12th century data, but only as 26th in the data from the Lay Subsidy Rolls. Hereford, on the other hand, ranks as 30th in the data from the time of Henry II, but as 14th in the later data. Finally, most locations of LAEME are only tentative and often it is not possible to associate their text languages with real towns or scriptoria. Bergs ’ s (2006) approach cannot be transferred to the Early Middle English period because a dense structure of different locations is essential for the successful application of this method. Since the network of localizable texts included in LAEME does not comply with this prerequisite, it is not possible to divide the entire area in such smaller units. The only area where it might be possible to proceed in a similar way is the West Midlands but, as I already mentioned in section 1.3.2, the methodology is somewhat problematic anyway, because the size of the honeycomb structure or the exact location of the grid on the map might influence the results. Furthermore, the methodology does not take into account temporal differences but treats the entire data as synchronic. As the Early Middle English period is one of great change and variation, it would be a pity not to make use of the fine-grained system of the 25-year periods as presented in LAEME. Since I cannot rely on the methodologies developed by Conde Silvestre and Hernández Campoy (2002) and Bergs (2006) I decided to display the data in feature maps. The representation of spatial diffusion on maps can be described as follows: [D]iffusion describes a dynamic process across time and space in which an increasing number of speakers add a linguistic feature to their repertoire. The distribution of a feature on a linguistic map is the frozen geographic reflex of a shift in linguistic behavior - a change in language. (Haas 2010: 649) In order to reveal diffusion patterns in Early Middle English, it is important that the maps include not only the spatial dimension but also a time and frequency dimension. The idea of diffusion as the “ frozen geographic reflex ” (Haas 2010: 649) on a map is applied to represent the Early Middle English data from LAEME in the present study. 3.2 Methodological considerations 69 <?page no="88"?> 3.3 The system of lexico-grammatical tags The linguistic variation that is evident in the LAEME texts renders it necessary to devise a system that allows the retrieval of different variants of the same linguistic form. This section looks into the system of lexico-grammatical tagging adopted for LAEME. 6 Each lexical item in every text of the LAEME corpus is tagged; a typical item has the following format: (30) $be/ vps11_AM The beginning of a tag is always indicated by $. The dollar sign is followed by the lexical part of the tag, called ‘ lexel ’ . The lexel is separated from the grammatical part of the tag, called ‘ grammel ’ 7 , by a slash. Finally, the lexicogrammatical part of the tag is separated by an underscore from the actual form of the word, as it is found in the manuscript. This system allows the identification of the lexico-semantic and morpho-syntactic properties of each item and, as a consequence, the retrieval of data through well-defined search queries. In order to keep the format as simple as possible, only ASCII characters were used for tagging and transcribing (Laing and Lass 2008 - : 3/ 11). A lexel is defined as “ an atheoretical taxonomic convenience, merely a name given to an item ” (Laing and Lass 2008 - : 4/ 3). Such identifiers are necessary because of the heterogeneous nature of Early Middle English texts. Only because of the lexels it is possible to find different spelling variants and different grammatical forms of the same word, as illustrated with WORD 8 in example (31), which is taken from the LAEME tag dictionary: (31) $word/ n *WORD +WORD +wORD WORD WORD+ WORDE WOURD wORD wORDE wORT wORd w\ORD In most cases, these lexels are Modern English “ descendants or semantic equivalents ” of the Middle English item (Laing and Lass 2008 - : 4/ 4). However, if this was not possible - because the word has not survived in any equivalent forms - the Middle English, or even West Saxon or Old Norse 6 A detailed description and discussion of the tagging system in LAEME can be found in chapter 4 of Laing and Lass (2008 - ). The grammel of example (30) is explained below. 7 Laing and Lass explain that “ [t]he nonce-terms lexel and grammel have been invented deliberately, to avoid the potential theoretical baggage (as well as definitional specificity) carried by terms like 'lexeme', 'lexical morpheme' and 'grammatical morpheme' “ (2008 - : 4/ 2). A more detailed explanation of this particular grammel is given below. 8 I follow the LAEME practice of printing the lexels in small capitals (cf. Laing and Lass 2008 - : 2/ 5, footnote 9). 70 3. Material and Methodology <?page no="89"?> form of the word was used. Some of the lexels contain a one-letter code within curly brackets, if, for instance, a lexical item is semantically ambiguous. Laing and Lass (2008 - : 4/ 5) provide the example of $before{p}/ and $before{t}/ , in which {p} refers to the spatial ({p}=place) and {t} to the temporal dimension of the same word ({t}=time). The grammel contains a morpho-syntactic description of a particular word form. In example (30) above, the grammel part reads “ vps11 ” , identifying the form of the copula as a verb (v) conjugated for the present tense (ps), singular number (1) and first person (1). Laing and Lass state that the type of grammatical tags used “ are essentially those of ‘ traditional ’ (Latinate) grammar ” (2008 - : 4/ 5), in other words, all items in LAEME are part-of-speech tagged including additional reference to subcategories like case, person, number, tense, etc. Furthermore, in addition to grammatical reference, a grammel may also contain supplementary tags, such as “ {rh} ” for an item in rhyming position or “ -k ” for the second element of a compound. Finally, as mentioned above, an underscore marks the end of a grammel and separates it from the actual word form that is tagged, as it appears in the manuscript. This part of the tag contains a system of upper case for the regular letters of the Latin alphabet and lower case letters for special characters like thorn, edh, ash, yogh, wynn, and insular < ᵹ >. Furthermore, anything else that needs special treatment in a diplomatic transcription, e. g. special symbols, signs of abbreviation, superscripts, diacritics, punctuation, deletions, insertions, line ends, folio references, etc. are given a special ASCII character code in order not to lose any valuable information in the transcription process. However, it would exceed the limits of this section to go into details, especially as all conventions are fully explained in Laing and Lass (2008 - : 3/ 13 - 28). In the present study, whenever a passage from LAEME is cited, the forms are transferred into Unicode characters, as it is more reader friendly than the LAEME format. In this process any graphetic information is retained that is encoded in the LAEME forms, such as capitalization (marked by a preceding *), nasal strokes (indicated through a lower case <n>), or other abbreviations. The signs of abbreviation as used in LAEME are described in Laing and Lass (2008 - : 3/ 14 - 17). All the word forms which are tagged in the LAEME corpus or any extralinguistic references or comments, are separated by a line break. Example (32) illustrates a typical passage from a LAEME text file: (32) $too/ av_*TO $long/ av_LONGE $/ P11N+H_ICH $have/ vps11_HABBE 3.3 The system of lexico-grammatical tags 71 <?page no="90"?> $child/ n_CHILD $be/ vSpp_I+BE+N $ge-/ xp-vpp_I+ $/ vSpp[V]_+N $in{m}/ pr_A $word/ n<pr_wORDE $&/ cj_& $in{m}/ pr_A $deed/ n<pr{rh}_DADE {.} {\} (#4) ‘ Too long I have been a child in word and deed. ’ This passage contains the line To longe ich habbe child iben a ƿ orde ⁊ a dade (line 3) from the version of the Poema Morale in manuscript Cambridge, Trinity College B.14.52 (335). Each tagged word form is listed on a separate line, including the tags {.} and {\}, which refer to a ‘ punctus ’ and a line break, respectively. Furthermore, the passage illustrates that in addition to individual words also some affixes (prefixes, derivational and inflectional suffixes) were given separate tags; e. g., the form iben in the sixth line of example (32) contains additional tags for the prefix and the inflectional ending of the past participle. The same practice was adopted for the individual parts of compounds. Finally, the passage also shows that not all items in LAEME necessarily contain a lexeme but, for some word forms, e. g., pronouns, determiners, and inflectional endings, the grammel is sufficient to identify a tag (Laing and Lass 2008 - : 4/ 1). In the example above, this is the case for the first person singular nominative of the personal pronoun, which is tagged as $/ P11N, or again, in the case of the tag for the past participle ending $/ vSpp[V]. Occasionally, there are comments by the editors of the texts on certain items. They follow the tagged items within curly brackets and the equal sign, as illustrated in (33): (33) $come/ vSpt13_COM {=C written over an erasure.=} (#286) Such comments include information on deletions and insertions in the manuscript, on different hands, on difficult readings, or on the shape of certain letters (Laing and Lass 2008 - : 3/ 28). Sometimes they contain more general observations and are not directly concerned with the tag preceding them but, as they often contain useful information on a certain word form, it is essential that these are consulted for the discussion of the variables in this study. 72 3. Material and Methodology <?page no="91"?> 3.4 Data retrieval 3.4.1 Reduction of the number of grade vowels of strong verbs The tag ‘ vSpt ’ identifies all verbs (v) with non-weak (S) past forms (pt), which allows a complete extraction of all possible strong past forms from the LAEME CTT. In LAEME this tag is linked to inflectional endings and, if one carries out a simple search for the grammel ‘ vSpt ’ through the ‘ Tagged Texts ’ option in the ‘ LAEME Task ’ section of LAEME, the result page only displays the plural inflectional endings per text files. The actual forms can only be accessed if the options ‘ lexis ’ and ‘ search by TAG ’ are selected for the query. In the field for lexel, one needs to add a space and, in the field for grammel, the tag mentioned above (in both cases the option ‘ % ’ in the drop-down list has to be selected, which includes any character(s) following or preceding the string). This type of search not only lists the actual forms, but also all inflectional endings and because they are not needed for the present study, I opted for data retrieval through an external script. First, I downloaded all the corpus files in the format.tag from LAEME. The files can also be accessed through the ‘ Tagged Texts ’ option by first selecting ‘ search by county ’ and then by choosing ‘ all ’ . 9 These files were then processed through a PHP-script, 10 which extracted all non-weak past forms from the LAEME CTT and transferred the raw data into a spread sheet in the following format: Table 3.1: Sample entry of processed LAEME data through a PHP-script. Text file Tag Form Number Person Comments 0277 $cweYan/ vSpt13_I+QUEd $ge-/ xpvpt13_I+ iqueð sg. 3 - This script extracted the form, number, and person as it is recorded in the LAEME tag and converted the ASCII-text of the tagged form into the corresponding Unicode characters. Furthermore, it also scanned the follow- 9 Because the data retrieval for this variable was carried out in the course of 2008, the data for the chapter on strong verbs is based on LAEME 1.1, which was superseded by LAEME 2.1 in December 2008. According to a note on versions on the entry site of LAEME “ this [update] follows addition of punctuation to a number of tagged texts with consequent updating of information in the Index of Sources ” (<http: / / www.lel.ed.ac.uk/ ihd/ laeme1/ front_page/ laeme_versions.html>, accessed April 2009). This means that with regard to the morphosyntactic tagging of the text files there should not be any changes to the version used here. 10 All the PHP-scripts that are used in this study are written by Dieter Studer-Joho. 3.4 Data retrieval 73 <?page no="92"?> ing line for comments and added them to the spread sheet, which would have been more time consuming through the integrated search option of LAEME. As the focus of the study on past grade reduction lies on individual classes, each verb was then manually allotted to one of the seven classes of strong verbs or one of their subclasses (cf. section 2.1.2). At the same time as assigning the verb forms to the classes of strong verbs, I tagged the grade vowels. This was fairly straightforward in most cases and usually the first vowel grapheme in a word stem represents the grade vowel. Nevertheless, due to Middle English scribal practices, a few decisions had to be taken in cases in which the grade vowel was not immediately evident. First of all, if the grapheme <u> clearly denotes a voiced, labio-dental fricative [v] or labio-velar approximant [w], these where not regarded as being part of the grade vowel as illustrated in (34). (34) Verb form Grade vowel OE drifan > ME driuen, 3pl i (#121) OE swican > ME suyken, 3pl y (#149) Furthermore, if a palatalized consonant was followed by i + a/ e, only the vowels following the i were regarded as stem vowels, because the only Middle English diphthong with / i/ as the first element is / iu/ (Brunner 1963: § 6.C; Mossé 1952: § 30 - 31): (35) Verb form Grade vowel ME gialde, 2sg a (#296) ME ȝ ief, 3sg e (#147) In addition to this, it was also necessary to pay special attention to those verbs in which the phoneme structure -VgVappeared in the past forms either primarily or through Verner ’ s Law (and this includes the contracted verbs where g may surface again). In verbs with this structure, vocalization of Old English palatal g / j/ takes place after front vowels and the new diphthongs ai and ei emerge (Brunner 1963: § 31.1). Vocalization of velar g / ɣ / occurs intervocalically after back vowels, which leads to the new diphthongs au and ou (Brunner 1963: § 31.3). The same diphthongs also emerge because of the development of epenthetic vowels before the phonemes / x/ and / ç/ (usually spelled <gh> or <h>): au and ou before / x/ and ei before / ç/ (Brunner 1963: § 31.2). As it is always transparent how the diphthongs developed, it was still possible to determine the first element of these diphthongs as the underlying grade vowel: 74 3. Material and Methodology <?page no="93"?> (36) Verb form Grade vowel ME drei (OE dre ˉ ogan), 3sg e (#1700) ME sau ȝ (OE se ˉ on), 3sg a (#158) ME steih (OE st ı ˉ gan), 3sg e (#64) ME astauh (OE ast ı ˉ gan), 3sg a (#173) Furthermore, there are some verbs for which no grade vowel could be determined because they were abbreviated or because the grade vowel was missing or not evident: (37) Verb form ꝙ , 3sg (OE cweþan) (#118) S ƿ ngke, 1pl (OE swincan) (#8) Although it is very likely that in the latter case the grade vowel would be / u/ following wynn, this cannot be determined with any certainty, therefore these abbreviations were not assigned a grade vowel. Finally, there are a few individual verb forms among the data that have caused difficulties and had to be excluded: Table 3.2: ‘ vSpt ’ forms not included in the present studies because of anomalies. Text file Tag Form Number Person Comments 0118 $ingo/ vSpt13_In\-zONG ı ˉ \ ȝ ong sg. 3 yes 11 0137 $gitan? / vSpt13_GETTE gette sg. 3 yes 12 1400 $forspae: can? / vSpt11_FOR-SWAT for swat sg. 1 - 2000 $abe: odan? / vSpt13_O+BED $a-/ xp-v_O+ obed sg. 3 yes 13 Although it would be possible to associate gette (#137) and obed (#2000) with strong verbs, I have decided not to include them because Laing (2008 - ) raises doubts about the validity of the verb form in the comment following the LAEME entry (cf. footnote 12 and 13). Furthermore, it is not clear whether ı ˉ \ ȝ ong (#118) is a verb form at all (cf. footnote 11). If so, it would be associated with OE ga ˉ n, which has suppletive past forms and is therefore not included in 11 “ {or the word 'had' or 'took' is missing and InzONG is a noun. Against that is that A and C also have verb missing and N too though with further corruption. All the texts can be read with good sense if INzONG is taken as a verb} ” (#118). 12 “ {=CB reads GELTE which is certainly wrong. It could be read GEITE, but the sense is still somewhat obscure. The form (in either spelling) would suggest a form of the word ge: atan 'grant' or 'concede' which the context does not support. It could possibly be from ge: atan in the sense 'confess'=} ” (#137). 13 “ {B an odd shape perhaps a mistake for CwED} ” (#2000). 3.4 Data retrieval 75 <?page no="94"?> this study. Finally, for-swat (#1400) is tagged with the lexel FORSPAE : CAN . Although this fits the context of the line in which the form appears, and <p> in an earlier copy of the manuscript could very well have been mistaken for wynn < ƿ > and was therefore transcribed as <w>, the verb could not be allocated and, consequently, I exclude this item. Table 3.3: Strong verbs in Middle English based on Brunner (1963: § 69). Present 1st Past 2nd Past Past. Part. I ME ı ˉ N/ eME a ˉ , S ǭ N a ˉ , S i i II ME e ˉ ̣ / u ˉ ę ̄ u/ o ǭ IIIa ME N + C ı ˉ ā / ǭ ou/ u ː / ou/ u ː / IIIb ME l + C e a u o IIIc ME r/ h + C e a u o IVa ME regular ę ̄ a (a ˉ ) ę ̄ , e ˉ ̣ ǭ IVb ME niman/ cuman i/ u a/ o ˉ o ˉ u V ME ę ̄ a (a ˉ ) ę ̄ , e ˉ ̣ ę ̄ VI ME a ˉ o ˉ ̣ o ˉ ̣ a ˉ VII ME V e ˉ ̣ e ˉ ̣ V Unfortunately, not all verb classes are suitable for an analysis of past grade vowel mergers. Table 3.3 shows that classes IVb, VI, and VII already have the same grade in their past singular and past plural stem and are therefore excluded from this study. Moreover, class IIIa distinguishes the past forms by a ˉ / o ˉ and u, which does not seem to be problematic at first sight. However, there is great variation with regard to a ˉ and o ˉ before nasals, especially before consonant clusters (cf. Mossé 1952: § 25). This is further complicated by the Middle English convention of writing <o> for / u/ , again especially before nasals (Brunner 1963: § 6). This problem is also highlighted by Lass: Verbs of this type are problematic, since <-oN-> spellings (bigon, bigonnen) are difficult to interpret. In a western text, <bigon> could be (phonologically) a continuation of Old English pret 1 bigann/ -gonn, with West Saxon pre-nasal raising and rounding. On the other hand, given the convention of writing <o> for <u> in minim environments, it could be a “ Western Preterite ” shift to / u/ in pret 1 . (1994 b: 92) For these reasons, it is impossible to determine the exact phonological quality of the grade vowels in class IIIa and it is, therefore, excluded as well. 76 3. Material and Methodology <?page no="95"?> All the other classes are suitable for an analysis of past vowel mergers and, therefore, the verb forms of the remaining classes were processed by another computer script. This script compiled all the variants of grade vowels per class and text in spread sheets (one each for 1st/ 3rd singular, 2nd singular, and the plural forms) that allow individual discussion of each class. While the 1st and 3rd singular forms, on the one hand, and the plural forms, on the other hand, were grouped together in one file each, the 2nd singular forms were separated from the plural forms as a different development might be responsible for the grade vowel in these forms. However, this distinction is only used for the discussion of the individual classes; in the compilation of the maps the 2nd singular and the plural forms are combined. A discussion of previous work with regard to the past grade mergers in section 2.1 has revealed that sometimes the past participle vowels are also involved in the mergers. Table 3.3 shows that a clear distinction of the past plural and past participle vowel is only evident in classes IVa, VI, and VII. In class V the distinction is based on / e ː / and / ɛː / , a distinction which is not always made explicit in the graphematic representation of the two sounds (Brunner 1963: § 6). Furthermore, in classes IIIb and IIIc the difference is o vs. u and, as already mentioned above, it is difficult to determine the sound that underlies a certain grapheme, as short u was spelled as <u> and, especially before <n, m, v, w>, alternatively as <o> (Brunner 1963: § 6). Finally, some plural forms of class II show o, which in writing may be the same as the past participle vowel ǭ . The distinction between the past plural and past participle vowel is thus difficult but, for the purpose of this study, I accept all possible spelling variants of the plural vowel in the singular without considering the influence of the past participle any further. 3.4.2 OE a ˉ > ME ǭ For the second variable I had to take a different approach in order to retrieve all forms that include a Middle English reflex of OE a ˉ , as there is not just one tag that can be employed. For the collection of lexels that could possibly be included in this study, the option ‘ Tag Dictionary ’ proved to be very helpful. This tool, which is part of the section ‘ LAEME tasks ’ , compiles a list of all tags found in the LAEME text files in alphanumeric order. The first step was to collect a list of all lexels which include <a: > because, if a lexel has not survived to Modern English, it contains the grapheme <a: > to represent a ˉ (e. g. BLA : C , GA : L or SCENDLA : C ). In a second step, I had to go through the entire ‘ Tag Dictionary ’ section of LAEME manually in order to check whether any of the listed forms included variation of <a> or <o>. Then I cross-checked the lexel with the OED and/ or MED in order to ensure that it had developed from OE a ˉ . 3.4 Data retrieval 77 <?page no="96"?> This procedure yielded a list of almost 600 lexical items, but not all items were suitable for the present study. First, the list does not only include words that developed from OE a ˉ , but also those of Latin or French origin that entered English with a ˉ , such as POLE L pa ˉ lus > OE pa ˉ l (OED pole, n. 1 ) or TA : MEN L atta ˉ mina ˉ re > OF atamer > tamer (OED attame, v., tame, v. 2 ). While the originally Old English Latin loans were included in the present study because they have developed the same way as OE a ˉ (cf. section 2.2.1.3), the French loans were excluded because the grapheme <a: > in their lexel represents / a ː / , which was not affected by the change a ˉ > ǭ (Strang 1970: § 134; Liebl 2002: 44). Furthermore, since low stress level is known to have already been responsible for variation between <a> and <o> in Old English (Hogg 1992: § 5.7; cf. also section 2.2.1.1), I decided to limit this case study to nouns, adjectives, and verbs and, therefore, I did not include suffixes, numerals, adverbs, and interjections. In many cases it is not possible to tell whether the change had not yet taken place, or whether the <a> spelling should be attributed to low stress level and subsequent shortening of the vowel. The list of all lexels is included in Appendix II. In a next step I had to conduct manual queries for every single lexel with the ‘ Tagged Texts ’ option and selecting ‘ Lexis ’ and ‘ Search by TAG ’ . The data for each lexical item was then transferred to individual spread sheets as shown in Table 3.4: Table 3.4: LAEME Results for BOAT . Text file Lexel Grammel Form <o> <a> ‘ other ’ 0064 BOAT n<pr bote 1 0 0 0173 BOAT n bot 1 0 0 0261 BOAT n<pr bat 0 2 0 0272 BOAT n bat 0 1 0 As strong verbs show different grade vowels in the past and past participle, I only collected the forms with the grammels “ vi ” (= infinitive) and “ vps ” (= present tense) for non-weak verbs. If a lexel included words from different word classes, they were collected in different spread sheets. In total I collected more than 13,000 individual tokens, and, therefore, I believe that my case study on a ˉ > ǭ is based on firm ground. Finally, there were a few tokens among the data with forms that I excluded at a very early stage, as they do not go back to OE a ˉ at all. First of all, Germ. *slahan developed into West Saxon sle ˉ an, and only Old English northern varieties retained a ˉ in the infinitive sla ˉ n. According to the OED (slay, v.1), “ [a]ll parts of the verb exhibit a great variety of Old English and Middle English forms, partly 78 3. Material and Methodology <?page no="97"?> through natural phonetic development, and partly by assimilation to each other. ” Therefore, I only take into account those forms that either show <a> or <o>; all others are excluded. Furthermore, there is SLOWTH , which in Middle English shows both forms from OE slæ ˉ wth and forms that are derived from the adjective ME slaw/ slo ˉ w (OED sloth, n.1). Only the latter are included in the present study. Moreover, I excluded tokens, which show i-mutation to mark grammatical contrast, as in GO and GOAT . The former may show i-mutation in the 2nd and 3rd singular present forms (Brunner 1965: § 430). Furthermore, OE ga ˉ t had i-mutated dative singular and nominative plural forms such as gæ ˉ t (OED goat, n.) and these plural forms are occasionally also found in Middle English texts (Mossé 1952: § 58). Again, these forms do not regularly follow the change a ˉ > ǫ and were not taken into consideration. In addition to these forms, there are also some forms of A : NLE : PIG , which already had i-mutated forms in the first syllable in Old English (OED onlepy, adj.; Campbell 1959: § 204.2, n.3). Some of these forms survived to the Middle English period and had to be excluded from the data. Also the present paradigm of HA : TAN { C } and HA : TAN { N } sometimes contain forms that do not go back to OE a ˉ . In Middle English the verb occasionally has present forms that are derived from the past form ME he ˉ t, especially in the meaning “ to be named ” (Brunner 1963: § 69, n.25). These forms are specially marked in LAEME by the tag ‘ ptform ’ , and I have excluded them all. Moreover, the paradigm of BOTH / aj also contains some inflected forms that do not contain a ˉ but go back to OE be ˉ gen (nom./ acc. masc.) and these were also left away. Finally, only those forms of LA : T / n that are derived from ON lát (OED let v.1) were included, as the native forms go back to OE æ ˉ and are not relevant for the change to ME ǭ . In each spread sheet I counted the occurrences of <a> and <o> forms in each file and added a separate group for any other vowel that might arise (cf. the last three columns of Table 3.4). 14 In this process I occasionally faced the same problems with diphthongs, for instance before h or g, as discussed with regard to the strong verbs, and I applied the same principles as described above in section 3.4.1. Nevertheless, there were the following two special cases that I have not mentioned yet. First, some of the northern texts display forms with <ai> graphemes that look like diphthongs, for instance waik, a form of WEAK in text file #297. Both Brunner (1963: § 5, note 5) and Jordan (1974: § 19) note that the original diphthong ai was regularly monophthongized to a ˉ in the North and that the <i> was reanalyzed as a length marker and also appeared in lexical items in which, historically, it was not to be expected. Since this process 14 Long vowels are also represented by the digraphs <aa> and <oo> but, for the sake of simplicity, I do not always name both <a>/ <aa> and <o>/ <oo>. I simply refer to <a> and <o> spellings, even if a certain text file shows the digraph. 3.4 Data retrieval 79 <?page no="98"?> started in the middle of the 14th century, it is possible that the later northern texts show this <i> as a length marker. However, because all northern forms with diphthong can be identified as Scandinavian loan words, I assume that also this form of WEAK shows the diphthong of ON veik r (OED weak, adj.). Any form of Scandinavian origin with such a diphthong is not included. Furthermore, I occasionally encountered the digraph <oa>. Lass offers the following explanation for this type of spelling: “ The spelling <ea, oa> (vs. <ee, oo>) seem to be diacritic in origin: the <a> means ‘ opener version of the first category ’“ (Lass 1997: 64, footnote 26). Also Moore (1969: 76) and Jordan (1974: 73) take <oa> to represent / ɔː / . However, as this digraph also appears in some texts in addition to both <a> and <o> spellings, as for instance in LAEME file #124 in the case of WOE , I decided to mark these spellings as ‘ other ’ and to discuss these spellings separately. The spread sheets were then further processed. First, I had to determine for each lexel how many <a>, <o> and ‘ other ’ spellings there were in total for each text file. These files then served as a basis for another PHP-script, which counted the total number of <a>, <o > and ‘ other ’ spellings for each LAEME text file across all lexels. On the basis of these figures I eventually compiled the maps. 3.4.3 Dual forms of the personal pronoun While the extraction of the past vowels of strong verbs and of items reflecting OE a ˉ need more detailed explanations, this is not necessary for the dual forms of the personal pronoun, as the extraction of dual forms from the LAEME CTT is a relatively straightforward task. The ‘ Concordancing ’ option among the various ‘ LAEME TASKS ’ allows the retrieval of forms embedded in their context. This is necessary to provide information of potential reinforcement of dual forms by quantifiers (cf. section 2.3.3). For the compilation of the concordance I chose ‘ Grammatical Words ’ and entered P2 for a plural personal pronoun as ‘ initial character of the tag ’ . This opened up a new input screen, where I chose the maximum of 20 words each to precede and follow a form, in order to have as much context as possible. Then, I chose all tags which included a supplement D for ‘ dual ’ ; for each dual form found in LAEME this produced the following type of output: (38) BO yIN EzE *wEyer IS BET+erE OF TwERE TwOM *yAT MON BO BLIyE OyER GROM *SO BO HIT EUer IN / P21GD - UNKer SIyE *yAT yU BO SORI & ICH BLIyE *zUTyUAISH+EISTwI ICH NE FARE *IN-TO Oyer LONDE & SINGE || cotowlat.tag | 2 | C13b2 80 3. Material and Methodology <?page no="99"?> Transcription: ‘ bo þin e ȝ e Ƿ eþer is betere of t ƿ ere t ƿ om Þat mon bo bliþe oþer grom So bo hit euer in unker siþe Þat þu bo sori & ich bliþe Ȝ ut þuaisheist ƿ i ich ne fare In to oþer londe ⁊ singe ’ This output includes both information on the form and the context in which it is embedded, plus a reference to the file (cotowlat.tag), its number (2), and dating (C13b2). 3.5 Linguistic maps Maps are a wonderful way to illustrate the areal distribution of linguistic features and, as Wikle nicely put it, “ they can be used to assist us in the identification of unique spatial characteristics or distributions on landscapes ” (1997: 417). There are several ways to map linguistic features. Depending on the nature of the data or on the objective of the map, a different type is more useful, and it is therefore important that an adequate map type is chosen (cf. Girnth 2010: 101). In this section, I first look into different ways to map linguistic features in order to find a suitable one for the present data and in the second section I explain how the maps that display the results of this study were created. 3.5.1 Types of linguistic maps A general distinction is made between ‘ display ’ and ‘ interpretive maps ’ , which are defined as follows: Display maps simply transfer the tabulated responses for a particular item onto a map, putting the tabulation into a geographical perspective. Interpretive maps attempt to make a more general statement, by showing the distribution of predominant variants from region to region. (Chambers and Trudgill 1998: 25) The first type, which Kretzschmar (2003) also refers to as ‘ descriptive maps ’ , simply displays the answer of a particular informant for a linguistic item at a certain location, while in an interpretive map the individual answer is not necessarily distinguishable and might be grouped together with others in a larger entity. Chambers and Trudgill (1998: 28 - 29) point out that both types of maps have advantages and disadvantages and that neither is really satisfying, as the descriptive map shows too much detail and the interpretive map does not include all the pieces of information that a researcher might be interested in. 3.5 Linguistic maps 81 <?page no="100"?> Glauser (1994: 40) notes a continuum between both types and explains that both types can be combined. The view that no map is perfect is widespread. For instance, Kretzschmar, who is working on Southern English, a variety of English spoken in the United States, notes the following: No map of Southern English is the ‘ best ’ one, because each has arisen from different assumptions and principles. However, maps can be considered as products of the decisions of their makers, and in those terms their relative success can be assessed. (Kretzschmar 2003: 130) In the first part of his essay, Kretzschmar discusses maps by Hempl (1896), Kurath (1949), and of Labov ’ s Telsur Project (as published in Labov, Ash, and Boberg 2006), which all include dialect boundaries or isoglosses. There is no doubt that isoglosses are useful to highlight general tendencies; however, Kretzschmar ’ s evaluation of some of Kurath ’ s (1949) maps on Southern English illustrates nicely how, in each case, we have to trust the researcher. For instance, Kurath, who had extensively collected data through fieldwork in the area in question, is certainly well-informed with regard to the features that he mapped. Nevertheless, the exact progression of each isogloss and, as a consequence, also of the major isogloss bundles, necessarily relies on the researcher ’ s perception and are an interpretation of the collected data. Furthermore, since linguistic change is gradual, which implies that in most cases there is a transition zone between different realizations of the same variable, it is inadequate to display this transition zone by a seemingly fixed boundary on the map (cf. Wikle 1997: 418; Girnth 2010: 112 - 113). The early publications on Middle English dialects all included isogloss maps (most prominently Oakden 1930 and Moore, Meech, and Whitehall 1935). Dialects were seen as solid entities and were delineated on the basis of the isoglosses of several features that divided England into five distinct dialect areas (sometimes adding London as a sixth). Since variation in Middle English is more diverse than just five distinct areas, I think that it is essential to avoid isogloss maps in this study. Kretzschmar (2003) also evaluates descriptive maps, such as the Linguistic Atlas of the Gulf States (LAGS) (Pederson et al. 1986 - 92). He states that “ [t]he advantage of these maps is that they show what they mean [. . .] The disadvantage of their descriptive maps is also that they show what they mean ” (Kretzschmar 2003: 138). In other words, it is an advantage of the maps that they plot the responses of the individual informants within predefined survey areas, but it is a disadvantage that they do not show any general tendencies or define larger dialect areas. With regard to the Early Middle English period it is not useful to show general tendencies again; this 82 3. Material and Methodology <?page no="101"?> has already been done for more than a century (cf. section 1.2). Therefore, I follow the tradition of LALME and LAEME and display my data in descriptive maps. I naturally offer a discussion and interpretation of the data represented in the maps in chapters 4 to 6 in order to make general observations, but I think it is important that the maps themselves also leave room for the individual interpretation of any reader (cf. Girnth 2010: 101). Descriptive maps are further distinguished between ‘ quantitative ’ and ‘ qualitative maps ’ . Qualitative maps simply display the location of a linguistic feature as point text/ symbol maps or area text/ symbol maps (Girnth 2010: 108). Girnth (2010: 118) suggests that point text maps “ offer the most exact representation of linguistic reality and with it also the avoidance of abstractions ” , which would be necessary in a symbol map. LALME (McIntosh, Samuels, and Benskin 1986), for instance, maps some features as point text maps, where all the possible answers for a feature are listed in each location. However, this is not applicable to the Early Middle English period, as some of the texts display such a vast array of spelling variants that the map would be unreadable. Quantitative maps, on the other hand, also map the frequency of a feature in a certain place. In this study it is important that frequency is taken into consideration as it is prerequisite to establish transition zones and the diffusion of a feature. Wikle (1997) distinguishes three types of quantitative maps: areal frequency mapping, surface mapping, and point mapping. The first type lumps together the data from a certain area, usually pre-existing entities like counties, towns, or neighborhoods. The most popular type of such maps are choroplethic maps, where the “ [s]hading or color patterns inside boundaries reflect each area ’ s average value for the linguistic feature represented ” (Wikle 1997: 419). Because of the thin coverage of texts in LAEME such maps are not suitable. In addition, surface maps require data which is regularly spread out, as they map the frequency of a feature by means of isopleths or a threedimensional model of the surface where an isopleth or “ height ” in the threedimensional model represents frequency (Wikle 1997: 430 - 432). While such a map would be perfect to display the diffusion of a feature, it is not possible to produce a representative map due to the scarcity of texts from the Early Middle English period. Therefore, the LAEME material should be displayed in point maps, which are used to display not only the statistical means of an area, but also discrete locations, where each dot represents one respondent. Furthermore, according to Wikle they are “ particularly well suited ” (1997: 429) to map data that is discontinuous in nature, such as the LAEME data. Wikle (1997: 427 - 430) discusses different ways to display frequency with a single dot or symbol: Pie maps display the frequency of a feature in a certain location by means of a segmented circle; graduated point maps use the size of a dot or 3.5 Linguistic maps 83 <?page no="102"?> symbol to represent the frequency in a location. Occasionally it is useful to combine the two, for instance in order to represent the number of respondents. Furthermore, it is even possible to include further dimensions, meaning a certain social variable might be encoded through the coloring or shading of a dot or symbol. Both LALME (McIntosh, Samuels, and Benskin 1986) and also the maps that are already available for LAEME display the results in dot maps of different sizes. While this practice highlights the areas of highest density, it is not possible to easily discern the different sized dots at first glance. Furthermore, all dots have the same color and the editors left away the possibility to add another dimension by additional information encoded in the color of the dots. I believe that more information can be gathered from maps if one observes what Girnth (2010: 102) refers to as ‘ visual economy ’ : A map presents its object with visual economy when the recipient obtains maximum information for a minimum of cognitive effort. To this end, a map should combine clarity and illustration and make it possible for the recipient to grasp the areal circumstances quickly and at one glance, not only after a prolonged examination of the map or the supplementary comments. For this reason, I decided to use segmented symbols to map the grade vowel mergers and the change a ˉ > ǭ , as they immediately reveal whether a feature is the main variant in a certain place or not. This is not necessary for the dual forms, as frequency does not play an important role there. The objective of mapping the dual forms is simply to highlight the locations where they have survived in the Middle English period. 3.5.2 The present maps The data is presented in descriptive maps that display the dimensions space, time and frequency for the reduction of strong past grades of strong verbs and for the distribution of <a> and <o>. In the maps of the dual, only the dimensions space and time are included. The spread sheets containing the data were processed into maps by means of another PHP-script. This script synchronizes the spread sheets that contain the frequency per location with a spread sheet that includes all the necessary information on location and date of a text in order to plot the relevant, pre-defined symbols on a base map similar to the one of the Computer Developed Linguistic Atlas of England (CLAE) (Viereck and Ramisch 1991 - 1997). For the grade vowel mergers of the strong verbs, I displayed the occurrence of the other past grade vowel in a given location with three different symbols: 84 3. Material and Methodology <?page no="103"?> (39) majority variant = above 66.7 % mixed = between 33.3 % and 66.7 % minority variant = below 33.3 % A text file is mapped as ‘ mixed ’ if the percentage lies between 33.3 % and 66.7 %. The majority or minority variants are defined by any quotient that lies outside this spectrum. For these calculations I disregarded any abbreviated forms such as ꝙ (#118) or forms where the grade vowel is missing as in S ƿ ngke (#8). Because Girnth suggests that visual economy is also gained if “ an iconic relationship between types of symbols and linguistic phenomena is established ” (2010: 112), I chose the following symbols, which immediately reveal the areas of highest <a> or <o> density: (40) only <a> = <a> higher than 95 % majority variant <a> = <a> between 66.7 % and 95 % mixed = <a>/ <o> below 66.7 % majority variant <o> = <o> between 66.7 % and 95 % only <o> = <o> higher than 95 % If the percentage of either <o> or <a> spellings is above 95 %, I assigned the text file to the category of either <a> or <o> spellings. If the percentage of the dominant variant is between 95 % and 66.7 %, I assigned the text file to the categories ‘ more a ’ or ‘ more o ’ , respectively, but if it falls below 66.7 %, I regarded a text file as ‘ mixed ’ . I had to include this 5 % margin at either end of the scale because in some manuscripts the distribution is <o> 3 x, <a> 249 x and ‘ other ’ 1 x (#297) and this should definitively not be grouped together with a text that has a distribution of <o> 3 x, <a> 26 x and ‘ other ’ 2 x (#304). If the majority variant is ‘ other ’ , the text file is tagged as such, but it does not appear on the maps. The texts with any other vowel as the dominant form are discussed separately in section 5.4. A finer-grained system of symbols was necessary to map the variation between <a> and <o>, as the frequencies are generally higher and as some texts show only one of either vowels. Since the system of symbols varies from variable to variable but always includes a circle, it is essential that each map includes a key in order to avoid confusion. In addition to the symbols just described, I also adopt a color scheme to add the dimension of time, which for a study on diffusion is essential. My maps are 3.5 Linguistic maps 85 <?page no="104"?> innovative in this respect, as this additional dimension has not been included in Middle English dialect maps so far. The following system is based on the LAEME coding as described in section 3.1 (the exact colours are evident in the linguistic maps of Appendix IV): (41) < 1200 12 b 1200 - 1225 13a1 1225 - 1250 13a2 1250 - 1275 13b1 1275 - 1300 13b2 1300 < 14 a I chose to combine 12b1 and 12b2 because there are very few texts from the third quarter of the 12th century and, similarly, all post-1300 texts are collected in one group as there are only two texts in LAEME from after 1325 (Laing 2008 - ). Some of the texts do not have a dating within a quarter of a century; in this case I took the earliest possible dating. Furthermore, the dots on the maps do not reveal whether a text files goes back to a pre-Conquest original, but these vital pieces of information are included in the discussion of the results in chapters 4 to 6. Finally, if there is more than one text file in one location, I have slightly spread out the dots in an overlapping row, but in a manner that all the mapped properties of a location are still visible. This is for instance necessary for the four text files of Vices and Virtues in manuscript London, British Library, Stowe 34, which is divided into files #64, #65, #302 and #304. The basic map with all text files is shown in Map 3.1 (cf. Appendix IV). The map also includes those text files that could not be located, but it displays the color code of the dating next to the text file number. I decided against adding the numbers of the text files to the locations in the maps, as they would hardly be readable anymore and it would not comply with the concept of visual economy. This choice is in line with the maps provided in LALME (McIntosh et al. 1986) and LAEME (2008 - ). In the discussion of the results, I refer to individual text files and, because the maps also include the historical county boundaries, it is possible to refer to individual files in geographic terms. 86 3. Material and Methodology <?page no="105"?> 3.6 Summary LAEME offers an innovative tool that allows studies which, so far, were not possible on such a large scale. The corpus of texts compiled in LAEME outshines any similar collection of texts from the Early Middle English period, not only with regard to its size, but also with regard to the finer-grained system of localization and dating that was employed. Furthermore, the taggingsystem with lexel and grammel, which was devised especially for LAEME, removes the methodological problem of spelling variants that has been one of the problems of historical corpus linguistics. The discussion of the surviving material that is included in LAEME has shown that the methodology of similar studies on Late Middle English (Conde Silvestre and Hernández Campoy 2002; Bergs 2005), cannot be applied to the Early Middle English data. For this reason, the data is presented in quantitative point maps, which are capable of revealing patterns of spatial diffusion by means of also representing the dating of a text and the relative frequency of a variable in a location. Some of the texts are only tentatively placed and Laing and Lass remind anyone working with LAEME that “ the display of linguistic data in map form at all is a convenient but highly generalized abstraction ” (2008 - : 1/ 11). It is therefore essential that the maps are complemented by a qualitative discussion of the results, which addresses the limitations of the Early Middle English data as described in this chapter with regard to text type, coverage, and localization. 3.6 Summary 87 <?page no="106"?> 4. Reduction of the Number of Grades in Strong Verbs The basic search query for the tag ‘ vSpt ’ through the PHP-script yields a total number of 10,510 tokens in 127 texts files. As explained in Section 3.3.2, only those verb classes that have distinct grade vowels in the first and second past stem are included. Taking only classes I, II, IIIb, IIIc, IVa, V into account, the data is reduced to 4,354 tokens in 101 text files. Apart from the verb classes that are excluded in the first place, there are a few verbs or verb forms that have to be treated separately. These are the 210 tokens of verbs that are of Old Norse origin and a short list of 18 tokens that historically belong to the group of weak verbs but which show past forms without dental suffix. In the following presentation and discussion of the results I address the leveling of the past grade vowels under the singular grade vowel in section 4.1, followed by a discussion of the opposite process, i. e. leveling under the plural vowel, in section 4.2. Furthermore, section 4.3 looks into verbs of Scandinavian origin and I address any tendencies concerning the past grade vowel mergers with regard to these items. Finally, this chapter ends with a short discussion of those verbs that are historically weak in section 4.4. 4.1 Singular forms Among the text files assembled for LAEME there is a total of 3,356 1st and 3rd singular past forms for the classes discussed in this study. Table 4.1 combines the figures and frequencies, and a more detailed table with the frequencies for each text file for each verb class can be found in Appendix I. Table 4.1 shows that the second past grade vowel is attested in singular forms for any class, except class IIIb. All the other classes contain at least one singular form with the plural grade vowel and the frequencies range from 0.2 % in class I to 29.1 % in class V. Furthermore, it is evident that class V is by far the largest class with regard to singular past forms in the LAEME CTT. There are 2,091 tokens for class V but only 437 for class I, the next largest class. In some classes the category ‘ other ’ is rather frequent and there are several possible explanations for these deviating grade vowels, such as class mixing, abbreviations, or scribal errors. With regard to the first group, there are, for instance, 41 forms of STI : GAN with <e> or <ea>. This is a verb which originally belonged to class I but which shows forms with the 1st past grade vowel of <?page no="107"?> class II (cf. Brunner 1963: § 69, note 2). Abbreviations are most frequent in class V, in which CWE Y AN 1 is abbreviated 102 times. Table 4.1: Frequency of each grade vowel per class in the 1st and 3rd past singular. Sg. Pl. Sg. vowel Pl. vowel ‘ other ’ Total Texts n= % n= % n= % n= n= I a ˉ / o ˉ i 361 82.6 1 0.2 75 17.2 437 57 II e ˉ u/ o 231 82.2 17 6.0 33 11.7 281 58 IIIb a u 49 77.8 0 0 14 22.2 63 28 IIIc a u 158 65.0 38 15.6 47 19.3 243 46 IVa a (a ˉ ) e ˉ 189 79.7 43 18.1 9 3.8 241 48 V a (a ˉ ) e ˉ 1323 63.3 609 29.1 159 7.6 2091 78 Since the variables are quite unevenly distributed across the manuscripts in which they actually occur, it is not possible to draw any further conclusions from Table 4.1 and, although it provides a first impression about the general frequency of the occurrence of the plural grade vowel in the singular, it is essential that each class is treated individually. 4.1.1 Class V Table 4.1 shows that with regard to the plural vowel in the singular this class is the most advanced in the data from the LAEME CTT. The total number of 2,091 1st and 3rd singular forms is distributed among 78 texts and the forms with the plural vowel are found in 58 of these. The following passage from the Ancrene Riwle (#118) contains three verbs of class V, which serve to illustrate both possible grade vowels: (42) Eue beheld o þe forbodene appel & seh hire \ fair & feng to deliten iþe bihal\dinge & toc hire lust þer to ƿ ard \ & nom & et þer of & ȝ af hire la\uerd (#118) ‘ Eve saw the forbidden apple and it appeared fair to her and she began to take delight in its visual appearance and took her delight towards it and took and ate thereof and gave some to her lord. ’ In example (42), the second past grade vowel <e> is found in the 3rd singular form seh SEE and et EAT , but ȝ af GIVE displays the expected singular vowel <a>. 1 According to the LAEME conventions a capital <Y> in a lexel represents a thorn <þ>. 4.1 Singular forms 89 <?page no="108"?> In total, the plural vowel e ˉ is attested to be the majority variant of the singular grade in 27 texts, as the minority variant in 20 texts and in 11 texts the distribution of a and e ˉ is roughly even. The discussion of the relevant handbooks and grammars in section 2.1.2.3 has shown that the forms with the plural grade vowel in the singular have to be expected in texts from the South-West and the southern part of the West Midlands. Map 4.1 2 confirms that the feature is most prominent in the West Midlands. With three exceptions (#158 NE Gloucestershire, #171 Worcester and #249 SE Herefordshire) all text files that contain 1st and 3rd past singular forms of class V contain verbs with <e> as past singular grade vowel at least as the minority variant. However, especially as a minority variant, the phenomenon is attested in all areas. At first, it seems that the data confirms the general view that the mergers in the South-West Midlands operate from the past plural to the singular and it seems that class V is very advanced in the change. However, there might be an alternative explanation for this apparent merger. Old English had e instead of æ (< Gmc a) in the West Midlands and in Kent and this vowel only developed to a in the 13th century in the West Midlands and a century later in Kent (cf. DeCamp 1958: 234; Brunner 1963: § 11, note 2; Brunner 1965: § 19.3; Jordan 1974: § 32). In this respect, it is very doubtful whether the numerous forms with e as grade vowel in the singular in the West Midlands can actually be attributed to the plural vowel. In order to verify this claim I compared my data with Map 107 3 in the ‘ MAPPING ’ section of LAEME (cf. Map 4.4). This map contains all forms of AFTER with <e>, which serves to illustrate the attestations of <e> for æ in Middle English. With the exception of #7, all text files from the southern part of the West Midlands that are dated to the first half of the 13th century or earlier contain ME efter either as the predominant or as the only form. Furthermore, the two texts from Kent that have the plural vowel <e> in the singular show ME efter (#8 and #291). In the later text files from the West Midlands the 2nd past grade vowel in the singular is only a minority variant or competes with the singular form. Similarly, it is evident from LAEME Map 105 4 (cf. Map 4.5) that after is the majority form in these later text files. The change only took place in the 14th century in Kent (Brunner 1963: § 11, note 2; Jordan 1974: § 32), 2 All maps are printed in Appendix IV. 3 <http: / / archive.ling.ed.ac.uk/ ihd/ laeme1_scripts/ create_map3.php? id=FMW/ fmw_after- EFT.pts> (last accessed August 2013). 4 <http: / / archive.ling.ed.ac.uk/ ihd/ laeme1_scripts/ create_map3.php? id=FMW/ fmw_after- AFT.pts> (last accessed August 2013) 90 4. Reduction of the Number of Grades in Strong Verbs <?page no="109"?> which explains why e is still found in the Ayenbyet of Inwyt dated to 1340. These findings confirm my earlier claim that the apparent merger has phonological explanations and that e as grade vowel is not the plural vowel but a regular variant of OE æ in the variety in question. Map 4.4: LAEME Map 107 showing all forms of AFTER with <e> (reproduced by permission of Ordnance Survey and LAEME). 4.1 Singular forms 91 <?page no="110"?> Map 4.5: LAEME Map 105 showing forms of AFTER with <a> (reproduced by permission of Ordnance Survey and LAEME). While this explains the forms with <e> in the West Midlands and Kent, it does not explain the similar forms in the East Midlands and elsewhere, where the sound change did not take place. It appears that the forms with <e> were so dominant in the West Midlands and Kent that they spread to other areas. Map 4.1 suggests that the forms with e in the singular diffused northward through the East Midlands. In the LAEME data, e appears in 10 texts from the East Midlands as the singular grade vowel in competition with the regular grade vowel or as the minority variant as illustrated in Table 4.2: 92 4. Reduction of the Number of Grades in Strong Verbs <?page no="111"?> Table 4.2: The distribution of a and e as singular grade vowel of class V in MSS from the East Midlands. Text file Location a e 1300 W Suffolk 39 31 1200 NW Essex 7 7 137 E Cambridgeshire 3 2 4 W Essex 1 1 282 Isle of Ely, Cambridgeshire 19 3 Minority variant Mixed 285 W Norfolk 108 3 64 SW Essex 26 2 65 SW Essex 5 1 155 W Norfolk 111 1 1400 W Norfolk 9 1 The forms with e and a mixed in an area around North-West Essex, East Cambridgeshire, and West Suffolk in the 12th century. In the 13th century the forms are found as the minority variant in Essex, on the one hand, but also further in the north. The forms with e in the 1st and 3rd singular must have spread from South to North because they fade out the further north the forms are located (cf. Map 4.1). With regard to the temporal distribution of the feature it is remarkable that there are very early texts (that is pre-1200) in which there are past singular forms with the plural vowel. This is earlier than mentioned in the handbooks (cf. section 2.3). In fact the feature is present in six out of the ten earliest texts either as the majority variant or as a competing form. Concerning the later temporal development there is evidence that the phenomenon began disappearing again towards the end of the Early Middle English period, as in most of the later texts in which we find the plural grade vowel for past singular it is only a minority variant. It seems that there was some change around the turn of the 13th century that had an impact on the feature. This is actually parallel to the change OE e > ME a in the 13th century in the West Midlands and supports the claim that the cluster of forms around Worcestershire has to be explained by the variation of OE æ ~ e rather than by extension of the past vowel to the singular. However, it seems that the change of West Midland OE e > a did not fully replace the e in the past singular with the regular singular grade vowel, as Long (1944) still mentions the occurrence of e in the singular of verbs of class V in her data from Late Middle English. 4.1 Singular forms 93 <?page no="112"?> 4.1.2 Class IVa Verbs with 1st and 3rd singular forms that belong to class IVa are found in 48 text files (30 less than for class V). The plural grade vowel is the predominant form in 15 texts, in 9 it is the minority variant, and the two grade vowels are mixed in 4 further text files. Examples (43) and (44) illustrate the two possible grade vowels with regard to class IV: (43) I suge þat icham ouercome \ Þoru hire þat bar þat holi sone (#2002) ‘ I say that I am overcome by her who bore the Holy Son. ’ (44) Þe lioun ber þe kni ȝ t child wiþ him \ Awey he wende wroþ and grim (#2002) ‘ The lion carried the boy with him. Away he went angry and grim. ’ Both examples contain a 3rd singular form of BEAR from text file #2002. Example (43) shows a form with the singular grade vowel <a> and example (44) a form with the plural grade vowel <e>. Map 4.4 illustrates the temporal and spatial distribution of verbs of class IVa with the plural grade vowel in the singular. The map shows a similar picture as for class V. Unfortunately, the data from class IVa includes much fewer tokens and, therefore, some of the texts that provided data for class V do not contain verbs belonging to class IVa. Again, the texts with the plural grade for the singular cluster around the same area in the South-West Midlands, and, furthermore, there are two texts from Kent that contain e as the singular past grade vowel. This is further support for the claim that the realization of OE æ is still e in the West Midlands and Kent in the Early Middle English period. Just as in class V the forms with e decline in the second half of the 13th century or compete with a, which corresponds to the appearance of a for OE æ even in these areas (Brunner 1963: § 11, note 2; Jordan 1974: § 32). Unlike class V, there are hardly any forms of class IV with e in the singular outside the South-West Midlands and Kent. There are plenty of text files from the East Midlands that are large enough to contain 1st and 3rd singular forms of class IVa, but we find the plural grade in the singular only as the minority variant (one form in each manuscript) in texts #64 and #155. Both texts also contain plural grades for the singular in class V. In the case of #155, rhyme might have been responsible for the only singular form with the plural past grade vowel. (45) And zelfa t ƿ o sunes him ber · \ Lia calde is · Gad · and asser · (#155) ‘ And Zelfa bore him two sons, Lia called these Gad and Asser. ’ 94 4. Reduction of the Number of Grades in Strong Verbs <?page no="113"?> In this example ber BEAR rhymes with the name Asser. While the rhyme must have influenced this particular form, it nevertheless indicates that it was in the repertoire of the scribe of Genesis and Exodus or one of its possible ancestors. The data from Long (1944) suggests that the forms with e in the entire past of class IVa spread in the course of the later Middle English period, and she reports the intrusion of the plural vowel in the singular for several texts from the East, like the works of Lydgate, the Paston Letters, and the Shillingford Letters. Although it is found in rhyming position, it might be an early attestation of the plural vowel for the singular in this text from the East Midlands. The lack of texts with e in class IVa in texts from the East Midlands allows for the conclusion that the verbs of class V were leading the change in the East. It is at least probable that the high frequency of class V verbs was influential. The data from class IVa confirms the claim that there are phonological reasons for the apparent occurrence of the past plural vowel in the singular. Already Bülbring (1889: 116) concluded that the extension of the past vowel to the singular grade vowel did not really take place; however, he did not discuss the same development but referred to the lengthening of the singular vowel in some verbs of class IVa and V, which might have accounted for the merger. But apart from a few isolated forms - eet (#273), of seeh (#280), and seet (#2002) - there is no evidence from class IVa or V in the West Midlands which would suggest any such lengthening process of the singular past grade vowel. 4.1.3 Class IIIc The absolute numbers and frequencies in Table 4.1 suggest that class IIIc behaves very similarly to class IVa. In this class the past vowels are quite distinct, so a merger of the two grade vowels for phonological reasons is less likely. However, if we look at the data for this class, it is immediately evident that the picture that emerges from Table 4.1 is misleading: there are 38 verb forms with the u from the plural in class IIIc; however, 34 out of these 38 forms are attested in one text only. In total, there are only five manuscripts that contain 1st and 3rd singular forms with the plural grade vowel of class IIIc, as Table 4.3 illustrates: 4.1 Singular forms 95 <?page no="114"?> Table 4.3: The temporal and spatial distribution of the plural grade vowel in the singular for class IIIc. Text file Location Date u 63 W Berkshire 1175 - 1200 1 150 W Norfolk 1275 - 1325 1 155 W Norfolk 1300 - 1325 34 273 Leominster, N Herefordshire 1225 - 1250 1 278 NW Worcestershire 1250 - 1275 1 It seems that, except in #150 and #155, one cannot recognize any temporal or spatial connection whatsoever. Therefore, there have to be other reasons for the appearance of u in the singular past of class IIIc in these manuscripts. However, I would like to add here that text #155 shows the plural grade for the singular as in other classes and in this case very predominantly. A closer look at the actual forms in question reveals that there might be some phonotactic explanations. All the forms with u belong to verbs of the structure wVrC, i. e. the grade vowel is preceded by a labio-velar approximant and followed by a consonant cluster with r (OE weorþan and weorpan). It is well possible that win the syllable onset triggered the plural grade vowel in these cases. However, all the other manuscripts that contain forms of OE weorþan and weorpan in the 1st and 3rd past singular show the traditional past singular vowel, so this might be an idiosyncrasy of the scribe of #155. The claim that the appearance of the plural vowel in the singular is not regular in class IIIc is supported by Long (1944: 59), who only mentions the intrusion of o into the singular in two texts. 5 4.1.4 Class IIIb Class IIIb is very small compared to the others and only contains 63 1st or 3rd plural forms that are recorded in 28 text files. As there are no plural grade vowels in the past singular of class IIIb, this class seems to be stable with respect to the singular grade vowel. However, on the basis of this data I hesitate to conclude that the phenomenon is not present at all in class IIIb in the Early Middle English period. Since the class is attested so infrequently, it is quite possible that the textual evidence is not representative. 5 In Long's (1944) study classes IIIb and IIIc are treated as one. 96 4. Reduction of the Number of Grades in Strong Verbs <?page no="115"?> 4.1.5 Class II The 1st and 3rd singular past forms of class II are quite widely distributed among the text files (282 forms in 58 texts), but forms with the plural grade vowel only appear in 13 texts: in 3 as the majority variant, in 8 as the minority variant, and in 2 text files the plural grade vowel is competing with the singular grade. The following two examples from St Katherine (#260) illustrate the two grade vowels: (46) þes maxence ƿ es ouercumen ant fleh into ali\xaundre (#260) ‘ This Maxence was defeated and fled to Alexandria. ’ (47) humah\te he drehen ꝥ he droh and deien so derfliche (#260) ‘ How could he endure what he endured and die so miserably? ’ The form fleh FLEE in example (46) is a 3rd singular form with the original grade vowel, which in Middle English generally was <e>, and droh DRE : OGAN in example (47) shows a form with the plural vowel <o>. I counted both forms with <u> and <o> spellings for the plural grade vowel because both can be graphic representations for / u/ . Brunner (1963: § 69, note 4) argues that not all <o> spellings in the past plural can be explained as spelling variants for <u> and that some have been transferred from / o ː / in the past participle. Long (1944) and Brook (1963) also mention that the past participle vowel spreads to the past forms in the southern parts of the West Midlands and in the South. However, in the LAEME data the <u> spelling is prevalent in the South-West Midlands and the <o> spellings are more common in the texts from the North and the East Midlands, so it could well be that it is more of a regional variant. In any case, there are no indications in my data that would suggest that <o> is not the plural vowel. Most earlier attestations of class II with the plural vowel in the singular are localized in the West Midlands (cf. Map 4.5). It is possible that classes IVa and V were influential and that in class II the plural vowel appears by analogy in some instances. There are also a few singular forms with the plural vowel outside the West Midlands (in text files #1200, #285 and #155). Interestingly, these are all texts that already contained some forms with the plural vowel in class V. For instance, with the exception of IIIb, text file #155 contained plural grade vowels in all the classes discussed so far. However, as the phenomenon is so rare, it might be useful to take a closer look at the singular forms that are attested with the plural grade vowel. Forms with <u> are attested in lu ȝ e LE : OGAN (#10), schute SCE : OTAN (#120, #260, #1000), and lute, alute ( A ) LU : TAN (#234). With the exception of lu ȝ e (#10), the verbs end in -ute, which suggests that the -t might have had some influence. 4.1 Singular forms 97 <?page no="116"?> It seems that these are not strong past forms at all, but that these are in fact weak past forms, in which t was reanalyzed as the dental element of the weak past forms in analogy to weak verbs with the same syllable structure where contraction occurred (cf. Brunner 1965: § 405 and § 407.3) and a note in Brunner (1963: § 69, note 6) even mentions weak forms for schote(n). However, the case of SCE : OTAN all infinitive and present forms in the LAEME CTT are attested with <e>, <eo>, <o>, or <y>, which would also be expected in corresponding weak forms, but there is without doubt a vowel change in schute. This confirms that it is a strong form and that the plural vowel appears in this form. Example (49) above also illustrates a further problem, which I have not mentioned so far, namely that the <o> spellings in the past singular of DRE : OGAN might not be an example of the plural vowel for the singular but of class mixing (cf. section 2.1.2.2). DRE : OGAN is very similar to DRAW (< OE dragan), a strong verb of class VI. The 1st or 3rd singular past form of DRAW can be droh (and is actually attested as such in text file #260, from which the example above was taken) and a certain influence is not deniable. The other <o> forms in the singular are attested for TE : ON , SCU : FAN , and FORBID . TE : ON is a verb that shifted to class II from class I (cf. section 2.1.2.2), where <o> would be a possible spelling for the singular vowel under the assumption that the shift a ˉ > ǭ was already completed. The form tog is found in text files #155 and #246 and in both text files <o> is the dominant variant for OE a ˉ (cf. chapter 5). So the past form tog in these verbs might be a relict form that has survived from class I. FORBID (< OE forbe ˉ odan), on the other hand, has close proximities with BI : DAN , also a verb from class I. The form forbod e is attested in #1200, which also has dominantly <o> for OE a ˉ (cf. chapter 5). This means that shof SCU : FAN (#285) might be the only form in which <o> is extended from the plural grade vowel to the singular. In class II there are not many attestations of the plural vowel in a singular form. However, there are a few forms which cannot be explained only on phonological grounds or by means of class mixing. This allows for the conclusion that, again, the many verbs from classes V and IVa that show the same grade vowel in the two past forms plus the few verbs that show the same grade vowel due to class mixing must have been influential and triggered the merger in a few other instances of class II. 4.1.6 Class I Among the 436 1st and 3rd past singular forms of class I (in 57 manuscripts) there is just one single form with the plural grade vowel in the version of Ancrene Riwle found in text file #118: smit SMITE . Since there is only one form with the plural vowel in this class, I conclude that class I is very stable with 98 4. Reduction of the Number of Grades in Strong Verbs <?page no="117"?> regard to the mergers of the past singular under the vowel of the past plural. Long (1944: ch1) does not mention any form with the plural vowel in the singular either. I argue that this stability in class I can be explained by the fact that the class has only three separate grade vowels, two of which are distinguished by quantity alone. Therefore, the constraint mentioned by Lass (2006: 77) of “ no more than three grades per verb ” is already complied with without any vowel mergers. 4.2 2nd singular and plural forms In the classes that were analyzed for this study there are 999 past plural and 2nd past singular verb forms. Out of these only 141 are 2nd singular forms (85 of which belong to class V), the vast majority consists of 3rd plural forms. The figures concerning the distribution of the singular and plural grade vowels in the past plural forms are compiled in Table 4.4: Table 4.4: Frequency of each grade vowel per class in the past plural. Sg. Pl. Sg. vowel Pl. vowel ‘ other ’ Total Texts n= % n= % n= % n= n= I a ˉ / o ˉ i 12 11.1 87 80.6 9 8.3 108 47 II e ˉ u/ o 19 15.0 105 82.7 3 2.4 127 43 IIIb a u 3 16.7 13 72.2 2 11.1 18 12 IIIc a u 6 8.7 53 76.8 10 14.5 69 25 IVa a (a ˉ ) e ˉ 33 23.2 108 76.1 1 0.7 142 48 V a (a ˉ ) e ˉ 111 20.7 410 76.6 14 2.6 535 65 Classes IVa and V also show the highest percentage of the 1st past grade vowel in the 2nd singular and plural forms, but the difference between the lowest frequency (8.7 % in class IIIc) and the highest (23.2 % in class IVa) is not as drastic as it was for the plural grade vowel in the singular (cf. Table 4.1). Again, in this section I discuss the developments of each class individually. I would like to recall that this phenomenon is reported to have started in the North before the appearance of the first known texts from the area (cf. section 2.1.2.3). Moreover, it is said to have spread southwards through the Midlands in the 13th and 14th century, while the South remained the most conservative area (e. g. Brunner 1963: § 69). 4.2 2nd singular and plural forms 99 <?page no="118"?> 4.2.1 Class V A total of 440 past plural forms and 85 2nd singular forms is found in 65 text files. Forms with the singular grade vowel instead of the plural grade vowel appear in 10 texts as the majority variant, in 5 texts as the minority variant and in 3 texts the forms are about equally distributed. The following example from Havelok illustrates the phenomenon of grade vowel mergers under the plural grade vowel in class V: (48) HE geten children hem bi twene \ Sones and douthres rith fiuetene (#285) ‘ The two of them conceived children, sons and daughters, exactly fifteen. ’ (49) And gaten mani children samen \ And liueden ay in blisse and game (#285) ‘ And together [they] conceived many children and lived forever in happiness and joy. ’ Both examples show a plural form of GITAN ; in example (48) the form shows the original plural grade vowel <e>, but in example (49) the same form is spelled with the singular grade vowel <a>. Map 4.6 illustrates the data from the 2nd singular and the plural forms combined. With few exceptions the forms all occur in texts from the East Midlands and in the longer and later texts from the North. Long (1944: 135) observed the same diatopic variation in later Middle English. It is very remarkable that the phenomenon is already attested in the earliest texts, namely in the second continuation of the Peterborough Chronicle (#149) and also in another very early text, the Ormulum (#301). The Peterborough Chronicle was written “ in or shortly after 1154 ” (Laing 2008 - : 169) and the Ormulum is dated to the end of the 12th century (Laing 2008 - : 161). In other words, the feature is already present in the North-East Midlands in the 12th century, which allows the conclusion that the merger started earlier than the 13th century, which was suggested for instance by Lass (2006: 77). Furthermore, the two text files with the output of the two main scribes of Vices and Virtues (#64 and #65) also show several plural forms with the singular vowel. This manuscript from the early 13th century is located in South-West Essex (Laing 2008 - : 116). This implies, therefore, that the change does not evenly diffuse southwards in the 13th century as the handbooks imply (cf. Markus 1990; Brunner 1963) because the mergers are already present in the southern parts of the East Midlands in the beginning of the 13th century. I have already stressed the necessity to treat the 2nd singular and the plural forms individually because, for instance, Brunner (1963: § 69.1) states that the 2nd singular was first to adopt the singular vowel in the Midlands. Unfortu- 100 4. Reduction of the Number of Grades in Strong Verbs <?page no="119"?> nately, there is no 2nd singular past for of any class V verb in the text file of the Peterborough Chronicle (#149), so the full dot in Map 4.6 for the singular vowel in the 2nd past grade is based entirely on the plural forms 2 x drapen DREPAN , 2 x iafen GIVE , aiauen AGIVE , in opposition to undergæton UNDERGITAN . It is difficult to tell whether <æ> in the last form represents the singular or plural grade vowel, as in West Saxon varieties the two grade vowels were distinguished in quantity only (æ vs. æ ˉ ) (Brunner 1965: § 391). In non-West Saxon varieties it can only be the singular vowel but, as a continuation of the Anglo-Saxon chronicle, text file #149 might show some West Saxon influence - although generally the language is described as vernacular East Midland Middle English (Clark 1958: xl). In the tagged sample of the Ormulum (#301), badd BIDDAN is the only 2nd singular past form of class Vand <a> represents the singular vowel in this form. The other two forms, bæden n BIDDAN and spæcen n SPEAK , are plural forms that contain <æ> as their grade vowel. Again, the grapheme could represent both singular and plural vowel; however, in the case of the Ormulum - thanks to Orm ’ s eccentric spelling system - it is possible to distinguish vowel length. Both forms appear to contain the plural vowel because Orm generally indicates a short vowel by the doubling of the following consonant (cf. Burchfield 1956; Anderson and Britton 1997, 1999). Therefore, in the Ormulum the singular grade vowel is only found in the 2nd singular past, which permits the conclusion that the 2nd singular took the lead. In Vices and Virtues all 2nd singular past tense forms contain the singular grade (with one exception), but the plural forms are mixed in the case of scribe A (#64) and only contain e in the case of scribe B (#65). The one exception mentioned with regard to the 2nd singular needs some explanation: The form was transliterated as ‘ gE>A>UE ’ (#65) in the LAEME text file, which means that the <a> was inserted into the word (Laing and Lass 2008 - : 3/ 25) and is actual evidence for the deliberate insertion of the singular grade vowel into the 2nd singular. But again, it is difficult to tell whether a phonological development may have enhanced the merger, especially in the more southern texts. On the one hand, Brunner points out that the occurrence of a ˉ in the second grade vowel of classes IVa and V was possible both in the North and South and that it might be “ a transfer of the vowel of the sg. lengthened in open syllables ” (1963: § 69). However, in another paragraph he explains that OE æ ˉ became a ˉ in the East Saxon area, which he defines as “ Essex, Hertfordshire, Bedfordshire and probably also parts of Cambridge and Middlesex ” (Brunner 1963: § 11.2). It is therefore possible that the early occurrence of the singular grade vowel in the 2nd past grade vowel coincides with ME a ˉ for OE æ ˉ in the southern parts of the East Midlands. Under the assumption that a ˉ in the plural past grade has a phonological explanation in Essex, one might argue that the merger under the 4.2 2nd singular and plural forms 101 <?page no="120"?> singular vowel nevertheless gradually diffuses southwards as Brunner (1963) and Markus (1990) observed. There is admittedly not much evidence, but one can deduce from Map 4.6 that the frequency of the forms declines towards the south, which allows this tentative conclusion. Map 4.6 also shows three manuscripts in the South and South-West Midlands that contain the singular vowel as the minority variants (text files #278, #286 and #1600). Plural forms are involved in all three cases, and the forms with the singular grade vowel in forms of the second past grade are documented only very scarcely: 1 out of 32 (#278), 3 out of 78 (#286), and 2 out of 64 (#1600). With regard to the 2nd singular past vowel there is little useful information in two of the texts: The grade vowel of the only 2nd singular past form in text file #278 is <æ> and, as mentioned before, it is difficult to tell whether this is the singular or plural grade vowel. Unfortunately, text file #1600 does not provide any kind of information on the 2nd singular past vowel in class V, as no such form is found in the tagged sample. However, there are several 2nd singular past forms in text file #286 and all these show the original 2nd past grade vowel. At this stage it is not clear how such isolated forms outside the East Midlands can be explained. It is possible that they are simply scribal errors; however, if such forms are also attested in the other classes, other explanations might have to be found. 4.2.2 Class IVa In class IVa there are a total of 142 forms that regularly show the past plural grade in 48 text files. Out of these only 24 involve 2nd singular past forms. The following example from two versions of the Poema Morale illustrates the different past grade vowels of class IVa: (50) Al sal þar ben þanne cuð þat men lu ᵹ en her & halen Al sal þar ben þanne un ƿ rien þat men her hudden & stalen (#4) ‘ All shall there be known then what men lied and concealed here. All shall there be uncovered then what men hid and stole here. ’ 6 (51) Al scal þer bon þenne cud þer me ˉ n \ lu ᵹ en her ent stelen Al scal þer bon þa ˉ ne \ un ƿ ron  ꝥ men ƿ ru ᵹ en her & helen · (#5) ‘ All shall there be known then what men lied and stole here. All shall there be uncovered then what men covered and concealed here. ’ In text file #4, a version from Essex, the rhyme halen/ stalen shows the singular vowel whereas text file #5, a version from North-West Worcestershire, shows 6 Translation by Carla M. Thomas in Treharne (2010: 345). 102 4. Reduction of the Number of Grades in Strong Verbs <?page no="121"?> the original plural grade vowel. In total, there are 13 text files in which the singular past vowel is found (9 as the majority variant, 2 as mixed and 2 as the minority variant). Among the 9 text files in which the singular grade vowel is dominant, there are five later texts from the North, which generally show the singular vowel in the plural also in the Late Middle English period (Long 1944: 113). Map 4.7 shows that, again, there are very early attestations of the singular past tense vowel in 2nd singular and plural forms. It is found both in the Peterborough Chronicle (#149) and in one of the scribal texts of Vices and Virtues (#64), two manuscripts that already showed the phenomenon for class V. Furthermore, there are two additional early text files in which plural forms with the singular vowel are documented. In text file #4, which contains a version of the Poema Morale (dated to the fourth quarter of the 12th century), there are three plural forms with the singular vowel, and in text file #138, a London manuscript from the beginning of the 13th century, there is evidence for a 2nd singular past form with the singular vowel. Again, it is possible that in text files #4 and #64 this might be evidence of the change OE æ ˉ > ME a ˉ in Essex (Brunner 1963: § 11.2). However, if this were the case, the apparent merger must subsequently have spread to the London text, as Brunner explicitly excludes London from the area where the change OE æ ˉ > ME a ˉ had taken place. So the data for class IVa confirms the findings of class V, namely that the merger is earlier than the handbooks suggest and that it does not spread southwards in the 13th century, but is already present in Essex at the end of the 12th century. Unfortunately, there is too little evidence from class IVa to make any observations on whether the merger is earlier in the 2nd singular forms. The only text from the West Midlands with the singular vowel in the plural is text file #278, which contains those parts of La ȝ amon ’ s Brut that are written by scribe A. This text also shows a few forms with the singular vowel in the 2nd past tense in verbs of class V. Again, the feature is not very prominent in this text. In class IVa there is only the form braken in opposition to five other forms of BREAK or TOBREAK that show <e>. 4.2.3 Class IIIc Class IIIc contains even less 2nd singular and plural past forms than classes IVa and V. The 69 forms are found in 25 text files, but forms with the singular vowel are found in only five of these. The following two forms from the version of St Katherine in text file #123 illustrate the two possibilities. 4.2 2nd singular and plural forms 103 <?page no="122"?> (52) And he as ha het him hef ꝥ hatele s ƿ ord up & s ƿ ipte hire of þat heaued I þat ilke stede anan i ƿ urðen \ t ƿ a ƿ undres (#123) ‘ And he, as she had commanded him, lifted that violent sword and chopped off her head. In that very same place immediately two wonders happened. ’ (53) Porph\ire & Auguste i ƿ arðen of þes \ ƿ ordes se s ƿ iðe ƿ el c ƿ eme & se hardi for þi ꝥ ha hefden isehen \ sihðen of heouene (#123) By these words Porphire and Auguste became so very gracious and so fearless because they had seen the face of heaven. ’ Example (52) includes the form i ƿ urðen WEOR Y AN with the historical plural vowel and example (53) has exactly the same form with <a>, which equals the 1st past grade vowel. In the LAEME CTT, the forms with singular grade appear as the majority variant only three times and once each as the minority variant or coexisting with the plural grade. Two of these text files (#123 and #262) are localized to the West Midlands, one to the North-West Midlands (#122) and, finally, two are found in relatively late texts from the North (#295 and #297). In text files #122 and #262 the singular past vowel appears in the 2nd singular. In #122 there is only the 2nd singular form faht FIGHT twice, but no plural form. In #262 both the 2nd singular form faht FIGHT and the plural form bursten BURST are attested. Only the 2nd singular form shows the singular grade vowel, which is further support for the claim that the merger starts in the 2nd singular form. However, text file #123 does not contain any 2nd past singular forms at all and, among the four plural forms, one shows the singular vowel as illustrated in example (53). Again, there is very little evidence to observe a possible start of the merger in the 2nd past singular form. Unfortunately, there are very few texts from the East Midlands that actually contain verbs of class IIIc in the 2nd singular or plural past. All three early texts that contain such forms - the Peterborough Chronicle (#149), the Poema Morale (#4), and the Ormulum (#301) - show the historical 2nd past grade vowel <u>. These are all plural forms so there is no data available on 2nd singular forms through LAEME with regard to these texts. According to Long (1944), the original plural vowel of her class IIIB (which corresponds to the classes IIIb and IIIc of this study) was completely gone in Late Middle English. She reports the intrusion of the past singular vowel in some texts, but she also mentions several forms that contain the past participle vowel. However, it is not clear how she distinguishes the past plural and the past participle vowel in this class as the same graphemes are sometimes used for / u/ and / o/ (cf. section 3.3). In the Early Middle English period the merger under the singular vowel is not yet evident anywhere in the East Midlands in this class. 104 4. Reduction of the Number of Grades in Strong Verbs <?page no="123"?> 4.2.4 Class IIIb While there are no singular forms with the plural vowel for class IIIb at all, there are a few plural forms that show the singular vowel. However, the percentage of 16.7 % is somewhat misleading: In the LAEME CTT there are only 18 2nd singular and plural forms attested for class IIIb. Furthermore, there are only 4 text files where the singular vowel is present (each time as the only option) and, with one exception, only plural forms are involved. The only 2nd singular form with the 1st past grade vowel is found in gialde GIELDAN in text file #296, which is a late text from the North. With the exception of text file #245 (West Worcestershire), the files that show evidence for a singular grade vowel in a plural form contain the later Northern text already mentioned in the discussion of the other classes. The only text file from the East Midlands that contains any plural form at all in class IIIb is #155; however, this text shows the regular plural grade vowel. It seems that class IIIb is, in general, very stable with regard to the singular vowel in the 2nd singular and plural forms of the past. 4.2.5 Class II In class II there are a total of 15 2nd past singular and 112 past plural forms. In 12 text files there are forms with the singular grade: 6 times it is the majority variant, 4 times the minority variant, and twice the frequency of forms with singular and plural grade are about equal. The following examples from the part of Cambridge, Trinity College B.14.39 (323) written by scribe B illustrate the phenomenon with regard to class II: (54) sone muchel a hire boden · a ˉ t more ahire bi hete (#247) ‘ straight away they offered her much and more they promised her. ’ (55) godes word ful wel þov cnewe ful mildeliche þer to þov bewe (#247) ‘ you know God ’ s word very well, you bow thereto very humbly. ’ The first example shows one of the possible plural vowels in the form boden BE : ODAN but, in the second example, the 2nd singular form bewe BU : GAN shows the vowel of the 1st past grade. In class II the singular vowel in the plural is again present in the later Northern texts, but also in a few West Midland texts and in one text file from West Norfolk, namely text file #285 containing Havelok (cf. Map 4.8). The six plural forms with the singular vowel in this text file are the only attested forms in the LAEME CTT for class II in the entire East Midlands. All the other text 4.2 2nd singular and plural forms 105 <?page no="124"?> files from the East Midlands that contain 2nd singular and plural forms of the past (including the earliest text discussed above) contain only the regular plural past vowel. As already mentioned, the feature is also present in some South-West Midlands texts and these are all dated to the first quarter of the 13th century (text files #247, #248 and #278). In text files #247 and #248 the forms involved are all 2nd singular past forms. Again this is further support that the merger starts in the 2nd singular form. In the third text, i. e. text file #278, there are also some plural forms with the singular grade vowel. This is the same text file in which plural forms with the singular grade are also recorded in classes IVa and V. Text files #247 and #248, which both contain texts from manuscript Cambridge, Trinity College B.14.39 (323), have some affiliation with the East Midlands, where the singular vowel in the 2nd singular and in the plural forms of the past is prominent in classes IVa and V. According to Laing (2008 - : #247 and 248) there is a Latin Epitaph of Bishop Robert Grossetest of Lincoln in this manuscript, but she reckons that the language is not East Midland in character at all. Moreover, with regard to class II, the phenomenon is very rare in the East Midlands, so eastern influence is very doubtful here. In addition to these three texts from the southern part of the West Midlands there is also text file #122, which is located in the North-West Midlands and contains a 2nd singular past form with the older diphthong ea in accordance with the 1st and 3rd past singular. This text already contains the singular vowel in the 2nd singular in class IIIc; as the text is located to the North-West Midlands (NE Cheshire), it is possible that this is evidence for the diffusion of the singular vowel from the North. The merger under the singular vowel is very infrequent in all areas and there are only a few isolated forms that are recorded in the Early Middle English period. It seems that the frequency of the singular grade vowel in the 2nd singular and plural forms of class II did not increase any further in the rest of the Middle English period, since Long (1944) does not mention such forms from her Late Middle English data. 4.2.6 Class I Class I contains 108 verbs in 47 manuscripts, but only five forms are 2nd singular past forms. Table 4.4 shows that only 11 forms contain the singular vowel and these are distributed among 7 manuscripts only (always as the majority variant except in text file #155, where the feature is a minority variant only). With the exception of text file #271, the merger under the singular vowel is only present in 14th century texts (including the larger Northern texts). I argue, therefore, that the 2nd singular and plural grade vowels of class I 106 4. Reduction of the Number of Grades in Strong Verbs <?page no="125"?> are also quite stable, at least in the Early Middle English period. It seems, however, that the mergers in class I under the singular vowel become more frequent, as Long (1944: 2 - 3) concludes that the extension of the past singular vowel into the past plural is dominant in the North but also present in several more southern texts in the late Middle English period. 4.3 Verbs of Old Norse origin As mentioned above, there are 210 forms, which are found in 35 different text files. However, the great majority of these forms - that is 197 tokens - belong to ON taka > ME take(n). The following verbs of Old Norse origin are found in the LAEME CTT: (56) Class I: 1 x NI : TA (ON níta), 2 x RI : FA , 1x rive, 1x TORIVE (ON (to)rífa), 2 x THRIVE (ON þrífa-sk) Class IIIa: 1 x SLING (ON slyngva) Class VI: 4 x BETAKE , 3x OFTAKE , 3x OVERTAKE , 163x TAKE , 21x TAKE { G }, 3x UNDERTAKE (ON -taka) Weak: 4 x CAST (ON kasta), 1 x SY : TA (ON sýta) Among the verbs of class I there are four 1st or 3rd singular forms, one 2nd singular form, and two plural forms. The singular forms all show the historical singular vowel. Among the plural forms there is raf (#298), which also shows the vowel of the singular. The other two plural forms are þriue (#282) and nit (#296). The latter form differs from similar native verb forms, as in text file #296; otherwise, only forms with the singular grade vowel are found in the plural of class I. Classes IIIa and VI are not part of the present study, so it is not possible to make any comparisons. It appears that the only form of Old Norse origin in class IIIa shows the singular vowel in the plural form slongen. However, as explained in section 3.3.2, it is difficult to tell whether <o> is the singular vowel or whether it represents / u/ (cf. Lass 1994 b: 92). Class VI has the same past vowel in the singular and plural anyway. Only three forms out of the 196 forms of -take do not show <o> as the past grade vowel: taken (#214), taken (#285), and oueR teoc (#261). Finally, there are two verbs that were weak but show past forms with vowel change in LAEME. The forms of CAST cannot be associated with any particular class. There are two explanations for the form sat SY : TA in text file #297, which contains the version of the Cursor Mundi found in the manuscript from the Royal College of Physicians in Edinburgh: On the one hand, the verb could have been reanalyzed as belonging to class I. The OED (site, v 1 ) also 4.3 Verbs of Old Norse origin 107 <?page no="126"?> cites the 1st singular present form site from the Vespasian version of the Cursor Mundi and the spelling <i> in the present rather than <y> supports the assumption that this verb was regarded as a strong verb from class I. However, there is an alternative explanation that would yield the same vowel, and this is the similarity with sit of class V, which also regularly has a in the singular past grade. In conclusion, there is very little evidence for past grade mergers in verbs of Old Norse origin as most verbs belong to classes where the two grades are the same or similar anyway. Those forms with the opposite past grade vowel show the same pattern as the native forms from the same class. 4.4 Historically weak verbs There are 18 verb forms in 9 different text files which were given the tag ‘ vSpt ’ , although historically they belong to one of the weak classes (cf. Table 4.5). In most of these verbs the influence of a similar strong verb is evident. The forms fande (#296) and faand (#298) seem to be influenced by the past forms of FIND , which in the two text files are attested as fande, fa ˉ de, fa ˉ d (sg.)/ fa ˉ d (pl.) (#296) and fand, fande (sg.)/ fand, Fand (pl.) (#298). The second singular form in (#296) shows the merger under the singular vowel. CLEOFIAN , on the other hand, could easily be confused with OE cle ˉ ofan, a strong verb of class II. The form claue (#296) might reflect the plural vowel of class II through the change ME ou > au (Brunner 1963: § 13, note 7). However, it is a 3rd singular form and the plural vowel in the singular is only the minority variant in this text, so it is unlikely. It is possible though that this form appears for the sake of the rhyme with d a ve (#296) in the previous line. This form shows the regular grade vowel for DRIVE . A verb of class IIIb might be responsible for the form fullen (#278) with <u> spellings in the plural grade: The smoothing of the diphthong in OE fe ˉ olan (< *feolhan) (cf. Brunner 1965: § 387, note 2) would have rendered the two verbs indistinguishable in Middle English. In text file #278 the grade vowel <u> is also found in the plural forms of FALL , which goes back to OE feallan, a verb of class VII. A verb of class IVb, NIMAN , might have influenced neme (#280). The grade vowel of this verb would regularly be <o> in Middle English, but the 3rd plural form nemen for NIMAN is attested twice in text file #280. Finally, there are six forms of BELAE : FAN , all found in Genesis and Exodus (#155). The infinitive is attested in this text file as bi leuen and it is possible that the scribe misinterpreted the verb as being part of class V, as the syllable structure of the stem is also eC [-R] . If this were the case, then the singular shows the grade vowel of the plural. But this seems to be very unlikely, as there is only one out of 108 4. Reduction of the Number of Grades in Strong Verbs <?page no="127"?> a total of 112 verb forms of class V in text file #155 that shows the plural grade vowel in the singular past form. Another possible explanation would be the influence of verbs of class VII with the grade vowel <e> in their past forms. A possible candidate is LAE : TAN , which in text file #155 is attested as leten (#155) in the infinitive and as let, leet (#155) in the 3rd person singular. It seems that this particular verb of the same syllable structure (leC-) has maybe influenced the past tense forms of BELAE : FAN in Genesis and Exodus (#155). Table 4.5: Verb forms with the tag ‘ vSpt ’ that historically belong to the weak verbs. Text file Tag Form Person Number OE inf. 0296 $fandian/ vSpt12_FANDE fande sg. 2 fandian 0298 $fandian/ vSpt13_FAAND faand sg. 3 fandian 0296 $cleofian/ vSpt13{rh}_CLAUE claue sg. 3 cleofian 0286 $fast/ vSpt23_VAST+E vaste pl. 3 fæsten 2000 $fast/ vSpt23_FEST\+EN fest\en pl. 3 fæsten 0278 $fell/ vSpt23_FULL+EN fullen pl. 3 fællan 0155 $belae: fan/ vSpt13_BI+LEF bilef sg. 3 belæ ˉ fan 0155 $belae: fan/ vSpt13_BI-LEF bi lef sg. 3 belæ ˉ fan 0155 $belae: fan/ vSpt13_BI-LEF bi lef sg. 3 belæ ˉ fan 0155 $belae: fan/ vSpt13_BI-LEF bi lef sg. 3 belæ ˉ fan 0155 $belae: fan/ vSpt13_BI-LEF bi lef sg. 3 belæ ˉ fan 0155 $belae: fan/ vSpt13_BI-LEAF bi leaf sg. 3 belæ ˉ fan 0280 $name/ vSpt23_NEM+E neme pl. 3 namian 0285 $set/ vSpt13+P23Od{rh}_SET+ES setes sg. 3 settan 1600 $spit/ vSpt13_SPATTE spatte sg. 3 spæ ˉ tan 1600 $spit/ vSpt13_SPATTE spatte sg. 3 spæ ˉ tan 1600 $Yyncan/ vSpt13_yOUzH+TE þou ȝ hte sg. 3 þyncan 0278 $bethink/ vSpt13_BI-yOHTE bi þohte sg. 3 beþencan There are three verbs in the list of Table 4.5 with stem final -t: FAST , SET , and SPIT . The dental element of the past forms of these verbs had already conflated with the stem final -t in Old English (cf. Brunner 1965: § 405 and § 407.3). But again, a similar strong verb accounts for the vowel change in the forms spatte (#1600) rather than spitte as in text files #118, #272 and #273: OE spæ ˉ tan is synonymous with OE sp ı ˉ wan, a strong verb of class I, in which <a> is the regular grade vowel of the singular. With regard to the other two verbs, I am not sure why they were tagged with ‘ vSpt ’ . Elsewhere, the singular past forms of FAST are simply tagged with ‘ vpt ’ : 2 x faste (#65), faste (#220) and feste (#2000). 4.4 Historically weak verbs 109 <?page no="128"?> Furthermore, the past tense form set or sette is also otherwise always tagged as ‘ VPT ’ . It seems to me that the specific reference to a non-weak verb is not necessary in these forms. Finally, I believe that also the last two verbs on the list do not qualify for the tag ‘ vSpt ’ . The vowel change in the past of Y YNCAN and BETHINK can be explained by the non-occurrence of i-mutation in the past tense forms of these verbs (Brunner 1965: § 407). Elsewhere in the LAEME CTT these forms are always tagged as ‘ vpt ’ , even in text files #278 and #1600. So it seems to me that the tag ‘ vSpt ’ found in the grammel of the forms þou ȝ hte (#1600) and bi þohte (#278) could be attributed to an oversight in the tagging process. These verb forms allow for the conclusion that the number of weak verbs that shifted to a strong class is relatively small in the Early Middle English period. However, a survey of all verbs with the tag ‘ vpt ’ would be necessary in order to check whether they also include past forms with vowel change rather than dental suffix to mark the past tense forms. I have shown that some of the verbs that were tagged as ‘ vSpt ’ should maybe simply receive the tag ‘ vpt ’ and not be given a tag for a non-weak past form. Finally, if a verb shows a vowel change in the past form, it always corresponds to a similar strong verb. In a few instances, the vowel merger under the singular vowel was evident. 4.5 Summary Classes IVa and V are clearly the most advanced with regard to the past vowel mergers in both directions. I have shown that the apparent merger under the plural vowel, which is most evident in the West Midlands and Kent, can be attributed to a phonological variant in these varieties, which is represented by the same graphemes. Nevertheless, this similarity of grade vowels, which were only distinguished by length, must have been influential and the e then spread northwards in the East Midlands. With regard to the merger under the singular vowel, the opposite has become evident. The data from the LAEME CTT confirms a gradual southward diffusion in the East Midlands but only in class V. There are two factors that might have been influential in this class: On the one hand, the high frequency of tokens in class V might have played a role. The large number of weak verbs in general was also given as a possible reason for the shift from strong verbs to the weak class (cf. Brook 1958; Williams 1975), so it is possible that the high number of verbs in class V might have exerted an influence over the other classes in Late Middle English, where the mergers are better attested (cf. Long 1944). On the other hand, the similarity of the grade vowels in the past forms of classes IVa and V in Old English and the Middle English varieties of the West Midlands and Kent, but 110 4. Reduction of the Number of Grades in Strong Verbs <?page no="129"?> also of classes VI and VII, in which this was the case anywhere, might have played a major role in the transmission of the change. Class I, on the other hand, is clearly the most stable in the LAEME data. This can be explained by the fact that in Old English there were not more than three grades in this class and two of these were only distinguished by quantity, as mentioned in section 4.1.6. It seems that this is a general tendency, since also Michelau (1910: 79) notes the same with regard to the shift to weak, i. e. that class I is the most stable. The shift in the past forms of class I starts more than a century later: in the LAEME data the merger under the singular vowel in class I is most prominent in the 14th century texts from the Northern and East Midlands. Also the findings of Long (1944) confirm that the singular vowel must have gained more ground in the later Middle English period. The picture that emerges from the discussion of each individual class in the singular and the plural is rather patchy. Not all text files in LAEME contain past forms of strong verbs and, moreover, if past forms are recorded, they are not necessarily evenly distributed among the classes, persons, and numbers. Nevertheless, however patchy the data is, it provides some important insights concerning the early developments of the past form mergers. Furthermore, there is some evidence that the 2nd singular past forms are maybe somewhat earlier in adopting the singular vowel (cf. Berndt 1956: 267; Brunner 1963: § 69.1). Unfortunately, there are too few 2nd singular past forms in the LAEME CTT to put these tentative claims on firmer grounds. With higher levels of confidence I can conclude that the dating of the mergers is generally earlier than noted in the handbooks (e. g. Lass 1994 b: 88; Brunner 1963: § 69.1). In classes IVa and V there is evidence from 12th century texts for the mergers under the singular vowel, and for class V there is also evidence for the merger under the plural vowel in the East Midland, which cannot be explained phonologically, but should probably be attributed to northward diffusion from Kent. 4.5 Summary 111 <?page no="130"?> 5. OE a ˉ > ME ǭ 5.1 General results The data collection described in section 3.3.3 resulted in a list of 377 lexical items (213 nouns, 84 adjectives, and 80 verbs) (cf. Appendix II). The input for the results and maps in this chapter is based on the 13,053 tokens (7,716 types) found through the LAEME queries for each of these individual items. Out of the 167 text files of LAEME there are 147 text files which contain at least one noun, adjective, or verb form that is relevant for the change ME a ˉ > ǭ and Table 5.1 summarizes the results overall and per word class. Table 5.1: General distribution of <a> and <o> among the LAEME text files in total and per word class. <a> More <a> Mixed More <o> <o> Other Texts n= % n= % n= % n= % n= % n= % n= Nouns 29 22.1 14 10.7 9 6.9 33 25.2 43 32.8 3 2.3 131 Adjectives 31 25.8 7 5.8 9 7.5 16 13.3 53 44.2 4 3.3 120 Verbs 35 31.8 7 6.4 5 4.5 32 29.1 27 24.5 4 3.6 110 Overall 30 20.4 16 10.9 9 6.1 43 29.3 44 29.9 5 3.4 147 In almost 60 % of all Early Middle English texts <o> is the majority or only variant, whereas only 30 % of all texts contain <a> as the majority or only variant. This distribution is more or less similar across all word classes, but such an overview is very general and, at best, it shows that the change ME a ˉ > ǭ has already gained momentum in the Early Middle English period across all word classes. Such a generalization is not very diagnostic for two reasons: First, it displays the results detached from any embedding in the spatial and temporal continuum and, second, although the results are very similar across all word classes, the total number of texts shows that not all text files provided results for all three word classes and, as a consequence, it is difficult to compare the figures. In this section I discuss the results from all three categories and in section 5.3 I take a closer look at the three individual classes, only considering those texts that yield results for all three classes. The spatial and temporal distribution of <a> and <o> in the LAEME CTT is visualized in Map 5.1 (cf. Appendix IV) and, as can be gathered from the small black circles on the map, there are only eight locations without any data for this <?page no="131"?> variable. An additional set of 36 text files do not appear on the map, although they provided data. The reason for their absence is that their text files are not localized, but I discuss the data from these texts in section 5.1.4 in order to analyze whether it is possible to embed them in the dense network of textual witnesses for this variable. In the following section I briefly discuss the results for the South and the North before focusing on the developments in the West and East Midlands in sections 5.1.2 and 5.1.3. 5.1.1 The South and the North In the South, <a> is preserved as the only or majority variant in three very early text files localized to Hampshire and Surrey (#143, #184 and #304) and in one text file from Somerset (#157). The fact that, in three of these text files, the frequency of <a> spellings is below 95 % indicates that the change is already well on its way everywhere. The evidence from #8 and #67 in the South East and #63 in western Berkshire shows that <o> is the dominant variant in very early southern texts. In the South East, <o> is the majority variant for the entire Early Middle English period; in the South West it is almost categorical for the more northern parts and, with the exception of the already mentioned text file #157, it is also the dominant variant for the rest of the South West. With regard to the North, Map 5.1 nicely shows that the <a> forms prevail in the North until the end of the Early Middle English period north of the line running west from the Humber. Since the texts are too scattered, it is not possible to draw a more exact demarcation line. All the data from the first part of the 13th century show <a> spellings as the only or majority variant. The only early text which does not have more than 95 % of <a> spellings is text file #231. The reason for this are not <o> spellings but two forms of HOLY which have <e>: +eli and hely. These are examples of the very few attestations of HOLY in LAEME that go back to Northumbrian hæ ˉ lig (cf. OED holy, adj. and n.). The more southern text files from Cheshire (#124 and #136) and Lincolnshire (#159 and #169) still show mixed <a> and <o> forms in the second half of the 13th century or at the beginning of the 14th century. There is only one northern text that has more than 95 % of <o> forms at the end of the 13th century. Text file #151 contains all the English material written by hand D from a macaronic sermon in Latin and English (Laing 2008 - : 143). The text has been fitted and Laing refers to Fletcher, who argues that “ scribe D produces a variety of written Middle English locatable in the north Lancashire or west Yorkshire region ” (1994: 221). Since all the other texts from the same period or later still show rather high frequencies of <a> spellings, the text is possibly not as northern as Fletcher concludes or maybe linguistically mixed. For instance, the text also attests fram FROM , which according to LAEME map 207 is rarely 5.1 General results 113 <?page no="132"?> attested that far north. On the other hand, the text file contains the 3rd singular form es BE , which according to LAEME map 130 is confined to the North. Further research is necessary into the language of this text file. In sum, the areas of early <o> spellings in the South correspond exactly to those proposed by Luick (1964: § 369) and Jordan (1974: § 44) as possible origin of the change (cf. section 2.2.2). But the early <o> spellings also in the central parts of the South confirm Liebl ’ s (2002, 2006) claim that the change had probably started in the South already before the appearance of the first Middle English texts. In the North, the <a> spellings clearly prevail for the entire period. This means that manuscript evidence does not support Liebl ’ s (2008: 42 - 43) findings based on onomastic evidence, which suggest that the change ME a ˉ > ǭ is completed in the North by 1250 (cf. section 2.2.2). 5.1.2 The East Midlands In the East Midlands <a> is still the main variant in the earliest texts in the northern parts, i. e. in The Peterborough Chronicle (#149) and The Ormulum (#301) and in the southern parts of Essex, i. e. in the various text languages of Vices and Virtues (#64, #65, #302) and in a version of The Creed (#183). Although <a> is the dominant form in the different text files associated with Vices and Virtues, they show a great deal of variation with regard to <a> and <o>, which suggests that the change had definitively started (cf. Lass 2006: 64): 470 x <a>, 97 x <o>, 3 x ‘ other ’ (#64), 105 x <a>, 9 x <o>, 1 x ‘ other ’ (#65), and 5 x <a>, 2 x <o> (#302). Between these peripheral texts with dominant <a> forms, there are four very early text files that show a majority of <o> forms (#4, #1200 and #1300) or only <o> (#266). The first three are part of the same manuscript, Cambridge, Trinity College B.14.52, and contain a version of the Poema Morale (#4) and two different scribal outputs of the Trinity Homilies (#1200 and #1300) and they are localized to north-western Essex and western Suffolk, respectively. Text file #266, on the other hand, is localized to Ramsey in Huntingdonshire. It is actually a Latin manuscript with very little English content; in total there are only 118 words (Laing 2008 - : 39). The manuscript is roughly dated to the 13th century and, according to the practice adopted in this study, it was assigned the color code for its earliest possible dating, thus the first quarter of the 13th century. This text is a literary anchor text, so its location is not fitted. Therefore, on the basis of the categorical <o> spellings in this short text, it seems that the text should maybe be dated more towards the end of the 13th century. Categorical <o> spellings in this location at the beginning of the 13th century are very unlikely, as all the nearby older texts show more than 95 % of <a> spellings and the later texts in the area either show <a> spellings or less than 95 % of <o> spellings. 114 5. OE a ˉ > ME ǭ <?page no="133"?> Unfortunately, there are hardly any other texts in the East until the end of the 13th or until the early 14th century in the more eastern parts of the area. With a few exceptions these texts all show a majority of <o> forms, but in most of the later texts the change is still ongoing. The only later texts which contain <o> spellings throughout are the lyrics found in London, British Library, Royal 12 E.i (#269 and #270), a manuscript that is localized to King ’ s Lynn. There are four late text files that contain only <a> (#131, #133 and #135) or mostly <a> (#185). All four text files go back to pre-Conquest originals and it is possible that the <a> spellings in these texts reflect an older language layer. Text files #133, #135, and #186 are either from the same location or close to the location of the literary anchor text #266, which has only <o> spellings. A comparison to the data from this text supports the view that the <a> spellings are probably not contemporary. Both #133 and #135 include content from manuscript London, Kew, The National Archives, E 164/ 28 and, according to Laing, there is reason to believe that the language of both hands is local as “ [t]he Old English of the original has been converted (either at this stage or some earlier stage) to Middle English ” (2008 - : 124 - 125). While this generally might be the case, I argue that, with regard to OE a ˉ , these texts preserve the old variant rather than convert it to contemporary Middle English because the <a> spellings do not correspond to the evidence from #266. Text file #131, on the other hand, contains short passages in English that are preserved in the Cartulary from the Abbey of St. Benet of Holme; Laing (2008 - : 89) notes that “ [t]he text is very short so there is not much to go on, but it is plausibly local language. ” I believe, however, that the five <a> spellings found in this text file also reflect an older language layer, as all the other contemporary text files from north-western East Anglia show a majority of <o> spellings. In general, the change is already evident in the East Midlands in the 12th century, earliest in north-western Essex, which supports Liebl ’ s findings that it “ might have started in Late Old English more or less simultaneously in several counties in the South as well as the East and West Midlands ” (2006: 30). Since the texts in the East Midlands are scattered across a large area and because there is no evidence for the middle of the 13th century, it is difficult to tell how the change then spread northwards. It is certain, however, that at the end of the Early Middle English period the change had reached the area around the Wash and generally shows very high levels of <o> spellings anywhere except the copied pre-Conquest documentary texts. However, it was definitively not completed yet, as both early 14th century texts from northern Norfolk still show <a> spellings in 13.4 % (Havelok #285) and in 8.5 % (Genesis and Exodus #155) of all tokens. This conclusion is also supported by eLALME Map 48 (Benskin et al. 2013 - ), which shows that, in the Late Middle English period, a few <a> spellings are still attested in southern Lincolnshire. This provides 5.1 General results 115 <?page no="134"?> further evidence for the finding that in the northern parts of the East Midlands the change is not complete at the end of the Early Middle English period. 5.1.3 The West Midlands In the West Midlands, especially in the South-West, there is much more textual evidence from the Early Middle English period and the text files provide evidence from all periods. The diffusion of the change OE a ˉ > ME ǭ can therefore be more thoroughly studied in this area. It is evident from Map 5.1 that the more northern text files still mostly show <a> spellings, while the more southern text files mainly have <o> spellings. In the area in-between, the text files show both <a> and <o> spellings, which is indicative of a transition zone between the southern and northern parts. Since the network of texts that contain the variable is much denser in this area and, as a consequence, the map is difficult to read, I decided to provide separate maps for each sub-period for this area (cf. Map 5.2). The earliest records show that the change had begun in the South-West Midlands in the earliest Middle English period already. The earliest text in north-west Oxfordshire (#232) already shows a majority of <o> spellings and the only form to preserve <a> is ihal ᵹ ed, which involves a consonant cluster (cf. section 5.2.2). The only other 12th century text, a literary anchor text from Worcester (#170), only has <o> once as opposed to twelve attestations of <a>. In the next period (1200 - 1225) the text files from Worcester (#171, #172, and #173) already show <o> as the main variant. These text files include three different text languages that were written by the well-known Worcester Tremulous Hand (Laing 2008 - ). All three text files go back to pre-Conquest texts, but <o> is nevertheless the dominant variant, which is different from the pre-Conquest documents from the East. Occasionally, the same lexical items shows both <a> and <o> spellings: BONE 1 x ban, 4 x bon(es), LA : C 1 x lac, 1 x loc, which indicates that the change is highly variable at the time in this place. The two forms of LA : C are even found in adjoining lines. In addition, the other more southern text (#66) mainly shows <o> forms, which is congruent with the earlier data from #232. In the more northern parts of the South-West Midlands, the <a> forms are still preserved at the beginning of the 13th century in the following text files: in MS A of the Ancrene Wisse (#272), in the texts of the Katherine group from manuscript London, British Library, Royal 17 A xxvii (#260, #261, #262), in the Poema Morale (#5), and the Lambeth Homilies (#2000 and #2001) that are part of manuscript London, Lambeth Palace Library 487. All these texts show very high levels of <a> spellings (more than 96 %), but #261 and #262 are the only text files from the period in the South-West Midlands in which the change is not yet attested. In addition to these text files with <a> as the 116 5. OE a ˉ > ME ǭ <?page no="135"?> dominant form, text file #189 is mapped as ‘ mixed ’ . This text file contains a passage from the same manuscript as #5, #2000, and #2001, but it contains the output of scribe B. Laing (2008 - : 131) notes that the language was “ a little later than Hand A ” , and this is also reflected in the variation of <a> and <o> spellings - <a> 13 x, <o> 15 x, <ao> 1 x - which allows the assumption that the change was already at a more advanced stage. In the period between 1225 - 1250 there are still three more northern text files (#273, #275, and #1000) that contain high levels of <a> spellings (in all three cases at least 94 %). In several respects these texts are similar to those from the earlier period. Text file #273 contains another version of Ancrene Wisse ( ‘ MS C ’ ) and #275 collects the corrections by scribe B to the same manuscript. Text file #1000, on the other hand, contains the text from the Katherine Group in manuscript Oxford, Bodleian Library, Bodley 34. This is the manuscript whose language, together with the one of MS A of Ancrene Wisse, is referred to as AB language. Smith noted that the spellings of the text associated with AB language are conservative in general: [A] conscious effort has been made to retain forms which must have been archaic at the time of writing. This is indicated by occasional scribal “ errors ” , such as o for Old English / ɑː / (usually spelt a in “ AB-language ” ); such occasional slips indicate that the scribes, left to their own devices, would doubtless have introduced the advancing form o. (Smith 1992: 583) Elsewhere, Smith has argued that AB language is “ simply a particular parochial usage belonging to a particular locality in the South-West Midlands ” (2000: 131). The evidence of the variable OE a ˉ > ME ǭ from the West Midlands supports Smith ’ s more recent view: There is no reason to believe that the texts were particularly archaic with regard to high levels of <a> spellings and I argue that that the occasional <o> spellings should rather be explained in the light of an early stage in the diffusion of the change in this area. Liebl reaches the same conclusion through comparison with onomastic material: “ In some cases, however, onomastic data proved helpful in confirming intuitions about the phonology of textual witnesses: the almost exclusive retention of <a> in the ‘ AB language ’ , for example, might thus actually have been a local dialect trait ” (2002: 193). In addition to these <a> spellings in the northern parts, there is also a more southern version of the Poema Morale (#6), which contains 94 % of <a> forms. This is a remarkably high number as all the other texts that are localized close by (i. e. text files #7, #234, #245, and #1800) have <o> as the dominant variable at this period. A close look at the data reveals that the high number of <a> spellings is based only on MORE / aj, which has <a> 6 x and <o> 2 x. The high frequency of this item might have caused the longer preservation of <a> 5.1 General results 117 <?page no="136"?> spellings. Finally, the evidence of this period also indicates that the change must have reached as far north as southern Staffordshire, as #146 already shows a majority of <o> forms. The diffusion of the change is further observable in the next period. Manuscript ‘ G ’ of Ancrene Riwle (#276) and the A version of La ȝ amon ’ s Brut (#277 and #278) in north-western Worcestershire still have <a> as the dominant form, but the two text files from La ȝ amon ’ s Brut show a very mixed type of language: <a> 120 x, <o> 20 x, ‘ other ’ 19 x (#277) and <a> 123 x, <o> 15 x, ‘ other ’ 31 x. The overall findings confirm Lass ’ s (2006: 64) observation that was based on three lexical items only, as in these spellings the diffusion of the change is indeed evident. The data from this and the earlier period allows the conclusion that the diffusion of the sound change reached the northern parts of the South-West Midlands sometime in the middle of the 13th century. The more southern texts show <o> as the dominant form in text files #246, #248, and #249 or, with very few exceptions, in the forms of text files #158, #247, and #271. The text files #246 - #249 contain different scribal outputs of MS Cambridge, Trinity College B.14.39 (323). Laing (2008 - : 21 - 29) remarks that the manuscript contains an epitaph of Bishop Robert Grosseteste of Lincoln but that the language is not “ East Midland in character ” . She explains that, although the manuscript only received one localization in LALME (LP 7721), they treated the different scribal outputs separately. Based on the evidence of the variable a ˉ > ǭ this turns out to be the right decision, as the percentages of <o> forms vary greatly: 71.0 % (#248), 86.4 % (#249), 92.3 % (#246), and 98.2 % (#247). The location of text file #247 more to the south-west of the other text files - and thus closer to the other text files of the period with almost categorical <o> spellings - is justified on the basis of the evidence of this case study. In the final quarter of the 13th century, <o> is the dominant if not the only form, even in northern Worcestershire. The two northern-most texts that are dated to this period correspond to the two text languages found in the Cotton manuscript of The Owl and the Nightingale (#2 and #3) and the two languages are also reflected in the state of the change OE a ˉ > ME ǭ in these text files: <a> 3 x, <o> 86 x (#2) and <a> 21 x, <o> 55 x, ‘ other ’ 3 x (#3). The first text language shows 96.6 % of <o> forms, which is very high compared to the 69.6 % of <o> spellings in the parts corresponding to the second text language. This suggests that, with regard to this change, the text language of #3 appears to be more conservative. The more southern texts all show at least 88 % of <o> forms (#10, #161, #1100, #1600, and #2002) and two texts even contain only <o> forms (#229 and #264). From the last sub-period there is only one text file which provides any data. It is a copy of an originally Old English writ of King Edward in the register of 118 5. OE a ˉ > ME ǭ <?page no="137"?> Richard Swinfield of Hereford Cathedral (Laing 2008 - : 47). The text is very short (only 116 words) and there is, in fact, only one single form homesokene HA : MSO : CN that provides any data. Therefore, it is impossible to tell whether the change was generally completed in the South-West Midlands at the beginning of the 14th century. In sum, the data from the West Midlands allows a more comprehensive discussion of the spatial diffusion of the change a ˉ > ǭ . The dominant <a> forms which are found at the beginning of the period to the north-west of Worcester are decreasing in the course of the period, and <o> forms are diffusing northwards from the south-eastern parts. The highest levels of <o> spellings are found towards the end of the 13th century in the southern parts of the South-West Midlands. 5.1.4 Unlocalized texts There is a long list of texts providing results for the change OE a ˉ > ME ǭ that could not be given an exact localization in LAEME. In some cases these texts were still given a general location in geographic terms. A discussion of these texts is essential, as the data from the change OE a ˉ > ME ǭ might confirm some of these tentative locations through the localized LAEME texts as presented in Map 5.1. The 36 texts that provide data are listed in Table 5.2 together with information on their general location, as presented in Laing (2008 - ). There are four text files listed in Table 5.2 from the first quarter of the 13th century. The first in the list, text file #14, was only vaguely dated to the 13th century and is thus listed in the group of its earliest possible dating. Laing (2008 - : 152) refers to Liebl (2005: 71), who noticed that the language of this text file is very similar to #292 (both contain a version of Candet Nudatum Pectus). Text file #292 is dated to the last quarter of the 13th century, so maybe the dating of #14 should be narrowed down to a later period in the 13th century too. Moreover, text files #265 and #267 are only given a very general dating and, unfortunately, there is very little early material from their potential areas of origin that could be used for comparison. However, it seems to me that <o> as the dominant variant in both text files either points to a more southern type of language or that the dating of the texts should be narrowed down to a later part of the 13th century only. Finally, the last text in this group #268 also has a general “ C13 ” dating. The only form found is <a> in halidey and because HOLYDAY only has <a> spellings except in two text files dated to the early 14th century, it is not possible to narrow down the dating (cf. section 5.2.3). In any case, it does not contradict a possible origin in Worcester. 5.1 General results 119 <?page no="138"?> Table 5.2: List of those LAEME text files which provide data for the change a ˉ > ǭ but which could not be given a more exact location (ordered by dating). In the period between 1225 and 1250 there are seven text files that were not given a more exact location. Four of these text files - #119, #120, #121, and #123 - are part of MS London, British Library, Cotton Titus D xviii and 120 5. OE a ˉ > ME ǭ <?page no="139"?> contain texts associated with the Katherine Group. The language of these text files is a mixture of North-West Midland and “ something more southerly ” (Laing 2008 - : 97). In this area <a> is the dominant variant at the period as far south as northern Worcestershire and Herefordshire, so it is not possible to tell on the basis of this variable how much more southern influence there is. Text files #235 and #236 might have an origin of South-West Worcestershire or North Gloucesterhire and this is supported well by the <o> spellings in these text files. Text file #17 also has <o> spellings throughout and Thomson (1935: 103) suggests an origin of the text in the Cambridge and Bury St Edmunds area. This seems to be based on paleographic evidence, but Laing notes that “ it is not clear on what basis this statement was made ” (2008: 16). The text is dated to a period between 1225 and 1300. Again, there is very little comparable evidence, but since the later text files from the area all show <o> as the dominant but by far not the only form, an early dating to the second quarter of the 13th century seems unlikely to me. In the next period there are two text files, #176 and #228. The former is not placed, but the language of the second might be from the South-West Midlands. This text file has a general dating to the second half of the 13th century but in both sub-periods a more southern origin is more likely, as the more northern texts still show some <a> spellings at that period. Finally, the great bulk of unlocalized texts is dated to the last quarter of the 13th century; about half of them could not be placed at all: #174, #238, #239, #241, #242, #243, #244, #259, #292, #294. The first six texts in this list (#238 - #244) are all found in the second part of MS London, British Library, Cotton Caligula A ix, i. e. the same manuscript that contains the Caligula version of The Owl and the Nightingale. However, Laing (2008 - : 68 - 79) notes that the language of each of these texts is different from the two text languages found in The Owl and the Nightingale (#2 and #3). As <o> is the dominant form in all texts from the Midlands and South at the end of the 13th century, it is not possible to reach any conclusions about possible locations of the majority of these text files on the basis of this variable only. There is only text file #179 that mainly has <a> spellings. These attestations of <a> for this variable support the assumption that the language is northern (Laing 2008 - : 149). Because there are only three lexical items that provide any data for this text, the conclusion is not based on very firm grounds. In general, this variable provides very little evidence that would provide any support for the general locations given in Laing (2008 - ). It is especially difficult in the East and North, where evidence from the early period is missing. However, in some cases I was able to make suggestions on the dating of the text or to confirm some of the more general locations. 5.1 General results 121 <?page no="140"?> 5.2 Linguistic environment The development of both OE a ˉ and a is closely connected to the linguistic environment of the variables in question (cf. section 2.2.1). With regard to the change OE a ˉ > ME ǭ both Dietz (1989: 142) and Liebl (2002, 2008) noticed that it behaved differently before w or ɣ in the North. Furthermore, I also pointed out in section 2.2.1.5 that some late Old English shortening processes (SHOCC and TRISH) might have prevented the change in some items in the first place. In the following sections I take a closer look at the developments in these contexts. 5.2.1 OE a ˉ + w/ ɣ Both Dietz (1989: 142) and Liebl (2002, 2008) noted that spellings with <a> before w or ɣ are found further to the south than <a> in other linguistic environments. In order to compare my data to their findings, I compiled separate maps for these surroundings with the following lexical items as input: (57) OE a ˉ + w: BLOW (OE bla ˉ wan), CROW (OE cra ˉ we), HLA : W (OE hla ˉ w), KNOW / BEKNOW (OE cna ˉ wan), SLOTH (eME sla ˉ wðe), SLOW (OE sla ˉ w), SNOW / SNOWWHITE (OE sna ˉ w), SOUL (OE sa ˉ wol), THROE / THROW (OE ðra ˉ wan) OE a ˉ + ɣ : A : GAN / A : GANNOT (OE a ˉ gan), A : GNIAN (OE gea ˉ gnian), FA : H (OE fa ˉ g), LOW (eME la ˉ g- < ON lágr), MA : GE (OE ma ˉ ge), OWN / OWNWILL (OE a ˉ gen), WA : G (OE wa ˉ g), Y RA : G (OE þra ˉ g) The distribution of <a> and <o> in these environments is displayed in Map 5.3 (a) and (b). The data from LAEME confirms the observation that, in these phonetic contexts, the <a> spellings are found further south than in other environments, which corresponds to Dietz ’ and Liebl ’ s findings. The map reveals that <a> is found in both contexts in more than 95 % of all forms everywhere north of a line between Cheshire and the Wash. A comparison with Map 5.1 shows that this is different from the overall results. Dietz (1989: 142) explained this difference by the change ou > au, which is said to operate in Late Middle English especially in the North-West Midlands. It is difficult to tell whether this change is already evident in the Early Middle English data or whether the change to ME ǭ is just slowed down. For instance, text file #136, which is localized to Stanlaw Abbey near Chester, has <o> as the dominant spelling in other linguistic environments, but it is uncertain whether the <a> spelling in the form fa ƿ e FA : H is already evidence for the change ou > au. Before both w and ɣ the level of <a> spellings is also higher in the southern parts of Shropshire and northern Herefordshire and Worcestershire. These texts are much earlier and the high levels of <a> in general show that the 122 5. OE a ˉ > ME ǭ <?page no="141"?> change a ˉ > ǭ was definitively not completed at the time and, therefore, there is no reason to believe that ou > au had already operated into the other direction again. It rather seems that a following w and ɣ slowed down the diffusion of the change in the northern parts of the South-West Midlands. A closer look at the South-West Midlands shows that the development of OE a ˉ + w is different from OE a ˉ + ɣ in the later period, as <a> before w is found much further to the south and until much later. This difference could be attributed to the emerging shift of ou to au because, according to Brunner (1963: § 13, note 7), the change did not take place when the diphthong was derived from ǭɣ . Several text files show <a> as the dominant form or are mapped as ‘ mixed ’ . However, it turns out that one lexical item alone is responsible for the high level of <a> spellings in the South-West Midlands: SOUL clearly retains OE a ˉ the longest or, alternatively, shows the earliest evidence of the change ou > au in the South-West Midlands. Brunner (1963: § 13, note 7) observes that ME ou also became au in Kent and also possibly in the South East. Map 5.3 (a) shows that this is indeed already evident in the Early Middle English period. OE a ˉ + w only has <a> spellings in The Kentish Sermons (#142) and The Ayenbite of Inwyt (#291). The evidence from the latter is especially strong as <a> before w is attested in 53 different forms. On the other hand, both texts have <o> for the context OE a ˉ + ɣ and, again, there is strong evidence from #291, which has a total of 33 <o> spellings but no <a> spellings in this context. Furthermore, the Kentish version of the Poema Morale (#8) also has high levels of <a> before w (5 x <a>, 4 x <o>), but la ᵹ en LOW is the only attested form with <a> in the context OE a ˉ + ɣ . It is not quite clear what Brunner (1963: § 13, note 7) meant exactly with the South East, but LAEME also provides evidence from Essex for the change ME ou > au. Two texts from the last quarter of the 13th century from northern Essex show <a> as the dominant spelling before w but not before ɣ (#160 and #162). Just as in the West Midlands it is OE a ˉ + w that shows higher levels of <a> and, in this case, it is clear that this is early evidence for the change ME ou > au, as the earliest Kentish texts have only <o> or mostly <o> before w. Again, these findings confirm Brunner ’ s (1963: § 13, note 7) observation that ME ou > au did not affect diphthongs that historically involved ɣ . The a ˉ w and a ˉ ɣ clusters are still graphetically distinct in the Kentish texts: w is spelled < ƿ > (#8), <u> (#142), or <w>/ <u> (#291), but ɣ is spelled < ᵹ > (#8), <gh> (#142), or < ȝ > (#291). However, in the two Essex texts the velar fricative has already disappeared and is represented by <u> or <w>, as evident, for instance, in the following forms of A : GAN : howest 1 x, oute 2 x, aute 1 x (#160), and owet (#162). However, even in these texts the difference between the two contexts is retained: ME ou > au did not operate if historically the diphthong included ɣ . 5.2 Linguistic environment 123 <?page no="142"?> 5.2.2 Shortening before consonant clusters (SHOCC) Many of the variants that show <a> in those texts that otherwise have <o> as the dominant variant include a consonant cluster that caused the shortening of OE a ˉ already before the Middle English period (Luick 1964: §§ 204, 352, 386; Campbell 1959: § 285; Brunner 1965: § 138; Jordan 1974: § 23). Words affected by shortening are nevertheless included in the present study (cf. section 2.2.1.5), in order to find out whether the shortening was categorical, or whether there are a few items among the Middle English data which were still affected by the change to ME ǭ . Table 5.3 lists the total of <a> and <o> spellings for those items that historically had a non-lengthening consonant cluster. That means that items with homorganic clusters (cf. section 2.2.1.5) or clusters caused through inflectional endings, suffixes, or the individual items of compounds are not included. Table 5.3: Distribution of <a> and <o> of lexical items with OE a ˉ plus a non-lengthening consonant cluster. <a> <o> other OE forms A : DL n 1 0 0 a ˉ dl A : DLIAN v 2 0 0 a ˉ dlian A : DLIG aj 1 0 0 a ˉ dlig A : GNIAN v 3 3 1 a ˉ gnian A : HT aj 13 5 3 a ˉ ht AMA : NSIAN v 16 0 0 ama ˉ nsian AMA : NSUMIAN v 1 0 0 ama ˉ nsumian GHOST n 256 201 2 ga ˉ st GHOSTLY aj 96 28 0 ga ˉ stl ı ˉ c GNA : ST n 2 0 0 gna ˉ st HA : LGA n 87 0 1 ha ˉ lga HA : LGIAN v 39 1 0 ha ˉ lgian HA : LSIAN v 16 0 0 ha ˉ lsian HA : LWENDE aj/ n 18 0 0 ha ˉ lwende LA : TTE : OW n 6 0 0 la ˉ tte ˉ ow, la ˉ dto ˉ w MA : D M ( HUS ) n 15 0 0 ma ˉ ðm SPA : TL n 1 4 0 spa ˉ tl 124 5. OE a ˉ > ME ǭ <?page no="143"?> The great majority of these lexical items consistently display <a> in texts from all areas, which indicates that the vowel must have shortened before the change a ˉ > ǭ started. To these items I also add HA : LGIAN with one <o> form and HA : LGA with one form in the category ‘ other ’ . The only <o> form of HA : LGIAN is found in the Lord ’ s Prayer (#258): (58) Hure wader þat is in euene þyn oli name beyn olid. (#258) ‘ Our father who is in heaven, your holy name be hallowed. ’ It is very probable that olid HA : LGIAN is influence by oli HOLY in the same line. The form without <a> in HA : LGA is Hlhen (#272). In the text file it is followed by the comment “ {=sic. Two-line coloured initial *H with the first element ascending} ” (#272) and it seems that the scribe simply forgot to write the vowel after this more elaborate initial. The two other forms of HA : LGA in text file #272 both have <a>. OE ha ˉ tte > ME h ă tte ‘ was called ’ , the passive form of HA : TAN { N }, was also affected by this shortening process (cf. section 2.2.1.5). The form hatte appears 48 times in the LAEME CTT and it is always spelled exactly in this form with <a>. It is likely that the vowel was shortened, but it is also possible that the form fossilized and simply preserved the Old English form including the vowel. In addition, there are also some contracted third singular forms of the verbs BEHA : TAN , HA : TAN { C }, and HA : TAN { N } (tagged with ‘ -ct ’ ), which generally only show <a>: (59) BEHA : TAN : behat 3 x, bihait 1 x, bihat 19 x, bi hat 6 x, bi hath 1 x, bihot 1 x HA : TAN { C }: hat 38 x, hate 1 x, hot 5 x, hóót 2 x HA : TAN { N }: hat 6 x, hate 2 x In Middle English these contracted forms do not show a cluster anymore but at some stage, the inflectional ending contracted with stem final t. With very few exceptions - HA : TAN { N }: hot (#2), hot (#8), hot, hóót (#142), hot (#1100); BEHA : TAN : bihot (#10) - these contracted forms only have <a>. Both #8 and #142 are localized to Kent. Interestingly, also #2 and #1100 have a South- Eastern connection: Both text files contain versions of The Owl and the Nightingale and the common ancestor is thought to be of South-Eastern origin. The more recent publications on the subject discussed the evidence for Guildford in Surrey (Fletcher 1999) and a possible Kentish origin of the poem (Cartlidge 1998; Cartlidge 2001: XV). Cartlidge ’ s assessment is based on linguistic criteria, but the feature presented above provides evidence not yet discussed. There are five lexical items with a consonant cluster that have rather high levels of <o> too: A : GNIAN , A : HT , GHOST , GHOSTLY , and SPA : TL . Since the items 5.2 Linguistic environment 125 <?page no="144"?> GHOST and GHOSTLY are so frequent, I discuss them separately below. The following <o> forms are found for the other three items: (60) A : GNIAN : ionnie (#173), hoþ ȝ endede (#280), o ȝ ninge (#291) A : HT : oht/ ohte (#278), hoþte (#280) SPA : TL : spot (#264), spotel (#274), 2 x spotle (#1600) It is possible that the velar fricatives (/ x/ and / ɣ / ) that were involved in the first two clusters, and which eventually caused the diphthongization of the preceding vowel (Brunner 1963: § 13), prevented the shortening. With regard to SPA : TL it is possible that the epenthetic vowel, which is visible in spotel (#274), emerged before the shortening took place. Furthermore, the variant spa ˉ ld already existed side by side with spa ˉ lt in Old English (OED spold, n., spattle, n. 1 ), which retained its length in any case and which might have further influenced the forms in the three text files. However, I should also add that both #264 and #1600 are localized to eastern Gloucestershire and western Oxfordshire, respectively, so it might well be a local variant (#274 has not been placed). Unfortunately, there are only five forms in total attested in LAEME, so further evidence is missing for the Early Middle English period. There are the many forms of GHOST and GHOSTLY that show high levels of <o> spellings. Both Brunner (1963: § 9) and Jordan (1974: § 23) mention the cluster st as the exception that did not cause the shortening. The absolute numbers of <a> and <o> spellings in Table 5.3 already show that, in the case of GHOST , almost 50 % of all forms show the change to ǭ . Map 5.4 shows that the diatopic and diachronic variation of <a> and <o> of the item GHOST generally corresponds to the overall picture presented in Map 5.1. 1 The distribution in the West Midlands only shows <a> spellings in the early texts and in the more northern text. However, the picture is somewhat different in the East. In Map 5.1 there are mainly <o> spellings in East Anglia; however, <a> is the more dominant in forms of GHOST in text files #137, #150, #155, and #300. This suggests that if the shortening before st-clusters ever took place, it is only noticeable in very few texts of the East Midlands. Both Brunner (1963: § 9) and Jordan (1974: § 23) mention that the cluster st did not cause shortening if an inflectional syllable followed and the cluster was regarded as a part of this second syllable. However, there is no evidence in the Early Middle English material that would support this claim. If it were true one would expect some texts in which <o> is found only in (historically) inflected forms, but <a> elsewhere. The only text to which this applies is the 1 The map of GHOSTLY would look the same, but with fewer text files that contain the item and, therefore, I do not add it here. 126 5. OE a ˉ > ME ǭ <?page no="145"?> output of scribe B of Vices and Virtues (#65): gast 2 x, goste 1 x, but it is difficult to believe that this claim should be based on one text only, especially as, before LAEME, the different scribal outputs were not necessarily distinguished in the discussion of forms and scribe A has <o> also in uninflected forms. On the basis of the LAEME data I conclude that if OE a ˉ was ever shortened in GHOST , it took place in the East, but not in the West Midlands, and regardless of the morphological environment. Shortening before non-lengthening clusters is generally reflected in the LAEME data for OE a ˉ , as with very few exceptions none of the vowels changed in this position. The data also confirms the longstanding claim that the shortening did not take place before st clusters, but the change to ǭ is apparent in all morphological contexts and is less prominent in the East Midlands. Finally, consonant clusters that emerged through contraction of the third singular forms of BEHA : TAN , HA : TAN { C }, and HA : TAN { N } did not cause this type of shortening in Kent. 5.2.3 Trisyllabic shortening (TRISH) Traditionally, vowels were thought to have been shortened in the antepenultimate syllable of trisyllabic words if two unstressed syllables followed (Campbell 1959: § 285; Jordan 1974: § 24). Minkova and Stockwell (1998) have argued that inflected disyllables are not affected by this supposed shortening process. According to Jordan (1974: § 24), the first element of compounds also retains its length. In texts that have <a> as the dominant form, it is difficult to establish whether TRISH prevented the change to ME ǭ because the <a> spelling might simply reflect the unchanged vowel quality in any case. However, a closer look at the spellings with <a> in texts with <o> as the dominant form may reveal whether inflected disyllables or first elements of compounds are frequent among those items that still preserve <a> longer. In the list of lexels that have served as input for the present study (cf. Appendix II) there is not one single lexical item which is truly trisyllabic, i. e. which has three syllables not counting derivational or inflectional affixes or parts of compounds. Minkova and Stockwell have already come to the conclusion that TRISH (cf. section 2.2.1.5) cannot be established “ due to the paucity of words to which shortening could have applied ” (1998: 228). The lack of any trisyllabic items for OE a ˉ in the narrow sense in my data confirms their conclusion. With regard to inflected forms there are just two forms that are trisyllabic through inflection and which show <a> in texts that otherwise show a majority of <o> forms. The short list presented in (61) only includes those items which were not already discussed in sections 5.2.1 or 5.2.2 above, where the 5.2 Linguistic environment 127 <?page no="146"?> phonological environment is responsible for the preservation of <a> in trisyllabic forms like hal ᵹ ede (#147). (61) hatere (#4 and #7) ƿ anunge (#173) The discussion of the two forms shows that they must be regarded as exceptions: first, hatere is a comparative form of HOT , which elsewhere is spelled with the geminate < tt - >: hattre (#5, #6), hatter (#10), hattore (#214), hatture (#1100), 2 x hattere (#1300). In the entire LAEME CTT there is only one comparative form with <o> in text file #8, which corresponds to the exceptions mentioned in section 5.2.2 above with regard to other clusters. In general, the comparative forms of HOT show shortening before consonant clusters. Second, a comparison of the second form with other texts also shows that it does not reveal anything about shortening in inflected disyllabics. The distribution of <a> and <o> in trisyllabic forms of the verbal noun of WA : NIAN in the LAEME CTT includes 9 x <o> and 7 x <a>, which shows that shortening did not take place. The findings with regard to OE a ˉ confirm Minkova and Stockwell ’ s claim that inflected disyllabics are not affected by shortening. The first elements of compounds, on the other hand, show occasional <a> spellings in texts that otherwise have <o> as the dominant form for OE a ˉ . The following list includes all compound nouns that show <a> in these cases: (62) GOATHEORD : 1 x <a> (#245) GOATTICCEN 1 x <a> (#245) HA : LEWAE : GE 1 x <a> (#150), 1 x <a>, 2 x <ea> (#245) HA : MSO : CN 3 x <a> (#1400) HOLYDAY 1 x <o>, 1 x <a> (#140), 1 x <a> (#227), 1 x <a> (#244), 2 x <a>, 1 x <e> (#245), 1 x <a> (#246), 1 x <a> (#1600), 1 x <a> (#1700) HOLIDOM 1 x <a> (#1400) HOLIHOOD 1 x <a> (#155) HOLYGHOST 2 x <a> (#155), 1 x <o>, 1 x <a> (#300) HOLYWATER 1 x <a> (#245), 1 x <a> (#1700) LORD 1 x <o>, 3 x <a> (#3), 1 x <o>, 1 x <a> (#7), 1 x <o>, 1 x <a> (#137), 4 x <o>, 1 x <a> (#248), 49 x <o>, 4 x <a> (#1200), 72 x <o>, 2 x <a> (#1300), 25 x <o>, 4 x <a> (#1400) LORESPEL 2 x <o>, 2 x <a> (#1300) LORE Y E : OW 10 x <o>, 1 x <a> (#1300) While it is certainly not possible to draw any conclusions from those coumpounds which are only attested once, there is a tendency that <a> is the majority option in the first elements of compounds in texts that otherwise show <o> as the dominant form. There is one great exception: <o> is generally 128 5. OE a ˉ > ME ǭ <?page no="147"?> more dominant in the forms of LORD . This might be explained by the fact that the forms of OE hla ˉ fweard > hla ˉ ford, which in Early Middle English most commonly appear as variants of lauerd or louerd, are not transparent as a compound in the Early Middle English period anymore. It is difficult to tell whether the vowel is actually shortened in the first element. Since most of these compounds also show forms with <o> in the LAEME CTT, it appears that OE a ˉ in the first element of compounds is maybe just somewhat later in undergoing the change and that Jordan (1974: § 24) is right when he states that the first elements of compounds usually retain their length. However, there is still the possibility that the compounds later adopted the new vowel through analogy to the simplex forms. More research with other vowels would be necessary to confirm or dismiss Jordan ’ s claim. 5.3 Word class and frequency Only nouns, adjectives, and verbs are included in this study in order to eliminate the factor of stress (cf. section 3.4.2). Figure 5.1 shows the overall figures and percentages in total and per word class. As already mentioned, not all text files contain data from all three classes and it is therefore difficult to compare all the data. For this reason, Figure 5.1 only displays the data from those text files that provide data for all three word classes. This was the case in a total of 91 text files and the figure shows how many of these text files belong to the categories ‘ <o> ’ , ‘ <a> ’ , ‘ more <o> ’ , ‘ more <a> ’ , ‘ mixed ’ , or ‘ other ’ . Figure 5.1: Distribution of <a> and <o> in those 91 LAEME text files that provide data for all three word classes. 5.3 Word class and frequency 129 <?page no="148"?> With regard to dominant <a> or <o> forms in general, the adjectives and nouns show a very similar picture. There are more or less the same number of texts that generally have <a> or <o> as the dominant form, and the only difference is that adjectives generally have higher <o> frequencies, i. e. more texts with more than 95 % of <o> forms. Figure 5.1 also shows that there are slightly more texts which show dominant <a> forms in verbs, but the difference is almost negligible. In order to fully understand whether there are indeed differences among the word classes, it is necessary to have a closer look at individual forms for each lexical item. The higher number of <a> forms among the verbs can be explained by only a few single verbs that are responsible for this difference: (63) HA : TAN { N } hat (#159) A : NCENNED { B } ancenneden (#171) BEHA : TAN bihastest (#189) BEHA : TAN bi hat (#189) HA : LGIAN ihal ᵹ ed (#232) HA : LGIAN Ihalseed* (#265) *{or from ha: lsian? But sense requires ‘ hallowed ’ not ‘ greeted ’ } These are all lexical items that show a consonant cluster that caused shortening and show <a> in general (cf. section 5.2.2). It is also necessary to discuss the high level of <o> spellings among the adjectives, to find out whether there is a reason for the higher number of <a> spellings among the nouns and verbs and as a consequence a smaller percentage of text files that show more than 95 % for <o>. It becomes apparent that a few lexical items among the nouns and verbs are responsible for the differences observed above. In (64) I list those items that showed <a> in more than three text files that otherwise have <o> as the majority variant: (64) A : GAN (#10, #160, #286, #1200, #2000) BEHA : TAN only contracted forms (#4, #8, #66, #245, #249, #291, #1800) GHOST (#137, #182, #246, #282, #1200) GO (#160, #161, #220, #291) HA : LGA (#142, #245, #249, #286, #291, #1200, #1600, #1800) HA : LGIAN (#182, #232, #245, #249, #291, #1200,#1800) HA : LSIAN (#245, #1700, #1800) HA : TAN { C } only contracted forms (#4, #245, #246, #291, #1600) HA : TAN { N } only in hatte (passive form) (#158, #280, #291, #1200) HOLYDAY (#245, #246, #1600, #1700) LORD (#137, #241, #1200) SOUL (#8, #10, #142, #160, #161, #162, #241, #249, #291, #1700) 130 5. OE a ˉ > ME ǭ <?page no="149"?> Again, these mostly include those items that were already discussed in the previous sections with regard to their linguistic environment. Therefore, it is safe to conclude that the three word classes that are included in the present study do not differ with regard to OE a ˉ > ME ǭ and none is promoting the change. The differences that are noticeable can be explained by variation due to the phonological environment of the stem vowel. Frequency is often mentioned as a key factor in the implementation of a sound change (cf. discussion in Phillips 1983 a) and, therefore, I divided my data into four groups according to token frequency. In order to make the data comparable, it was again necessary to include only those text files that contain data from all four ranges. With regard to frequency, I distinguished the following categories, which are adapted from Phillips (1983 a: 489): 1 - 11, 11 - 50, 51 - 100, and 101+. Figure 5.2 summarizes the results of those texts that include data from all four categories: Figure 5.2: Distribution of <a> and <o> categories in those 71 LAEME text files that provide data for all four frequency ranges. The columns on the left illustrate the percentage of those text files in which <a> is still the dominant form. The columns on the right, on the other hand, provide the percentages for those text files in which <o> is the dominant form. First of all, Figure 5.2 indicates that the class ‘ mixed ’ is the smaller the more frequent the lexical items are. Furthermore, the figure also shows that the more frequent an item is, the more likely the text files are to show <o> as their dominant form. In other words, the overall frequency of texts that still show a majority of <a> forms remains more or less the same, but competing <a> and <o> forms are resolved towards <o> spellings the more frequent an item is and the more frequent items appear to be leading the change OE a ˉ > ǭ . 5.3 Word class and frequency 131 <?page no="150"?> While word class does not seem to be an influencing factor in the implementation of the change, there is a slight tendency in the group of the rare tokens that a text file shows more mixed <a> and <o> spellings rather than <o> as the dominant form. In other words, the more frequent the tokens are, the more the category ‘ mixed ’ is resolved towards <o> spellings; however, the overall differences are very small. 5.4 <ao> Spellings In my data there are five text files that were not mapped because the majority of forms belongs to the category ‘ other ’ . These are text files #11, #12, #13, #139, and #175. In text file #13 the only evidence comes from bleyc BLA : C , which shows a form that derives from the loanword ON bleikr (OED blake, adj.). In text file #139 only some forms of SORE , OWN , and GO are attested and, while OWN and GO both show <o> forms, they are overruled by the four forms of SORE with <e> in stanza II, line 13 in The Proverbs of Hending. Both text files are too short to provide enough linguistic evidence for a localization in LAEME. Table 5.4: Distribution of <a> and <o> in text files that contain <oa> at least once. Text file Date Localization <a> <o> <oa> Dominant form 11 13b1 SW Essex 1 2 5 <oa> 12 13b1 not placed 0 1 6 <oa> 64 13a1 SW Essex 470 97 3 <a> 65 13a1 SW Essex 105 9 1 <a> 124 13b1 Chester 4 2 3 <a> 155 14a1 W Norfolk 13 130 4 <o> 172 13 a Worcester 4 69 2 <o> 173 13 a Worcester 62 335 3 <o> 175 13b2 NW Norfolk 0 0 2 <oa> 277 13b1 NW Worcs 120 20 1 <a> 1800 13a2 W Worcs 9 128 1 <o> 2000 13a1 NW Worcs 482 6 1 <a> 132 5. OE a ˉ > ME ǭ <?page no="151"?> The other three texts all have <oa> spellings as their dominant form but, in text file #12, the total number of forms also includes one <o> form and text file #11 also has <a> once as well as two <o> spellings. The lyric of text file #175 attests boan in a series of rhymes that include leman, iohan, man, wan, and can. Since this text file is localized to the East Midlands where the <a> in these items would not be rounded and raised, it is likely that <oa> represents the unchanged vowel. Traditionally, the digraph <oa> is thought to represent / ɔː / (Umpfenbach 1935: 102; Moore 1969: 76; Jordan 1974: 73; Lass 1997: 64, footnote 26). The examples just mentioned indicate that this was probably not always the case in Early Middle English. Table 5.4 lists the distribution of <a> and <o> in all text files that have at least one <oa> spelling in the word classes noun, verb, and adjective. It shows that there are five text files with <a> as the dominant form as opposed to four text files that contain more <o> spellings. With the exception of #12 and #175, the forms always appear in texts that contain both <a> and <o> spellings. From the data presented in Table 5.4, three major areas emerge in which <oa> spellings are attested: in the West Midlands in an area between Worcester and Chester, 2 in Essex, and in Norfolk. According to Map 5.1, <a> is clearly the dominant form in the northern parts of the South-West Midlands and Chester until the third quarter of the 13th century and in the more southern parts of Essex until the first quarter of the 13th century. Therefore, it is difficult to say which phonetic value <oa> actually represented and it might just indicate that the scribes struggled in deciding how to put the underlying sound into writing. It is possible that it represented any vowel in-between <a> and <o>, for instance, a rounded version of <a> / ɑː / . Lass assumes that the <a> in the digraph marks the first vowel as having a more open pronunciation: The spelling <ea, oa> (vs. <ee, oo>) seem to be diacritic in origin: the <a> means ‘ opener version of the first category ’ . The twelfth-century Icelandic First Grammarian (Haugen 1950) uses a subscript hook (< ę , ǫ >) to mark openness, and says explicitly that it represents part of the letter <a>. (1997: 64, footnote 26) Lass compares the possible meaning of <a> in the digraph to the subscript hook used by the Icelandic First Grammarian to mark openness. According to Haugen (1950: 31), the same grapheme could also mean “ less open than a ” and it is possible that the same type of variation was distinguishable in Early Middle English. Based on the findings from LAEME there is no reason to believe that <oa> always represented / ɔː / but, in some attestations, it might have been an opener version of / ɑː / or / ɒː / . 2 There are further attestations of <oa> spellings in text files localized to Cheshire for lexical items from other word classes. 5.4 <ao> Spellings 133 <?page no="152"?> 5.5 Summary The analysis of lexical items that include OE a ˉ has shown that at the beginning of the Early Middle English period the change to ME ǭ had already started in the South West, South East, South-West Midlands, and southern parts of the East Midlands. Liebl has reached similar conclusions based on his research on onomastic evidence: The evidence suggests, I believe, that / ɑː / > / ɔː / , rather than spreading from the South to the North, might have started in Late Old English more or less simultaneously in several counties in the South as well as the East and West Midlands and radiated from there. (2006: 30) The evidence from LAEME confirms Liebl ’ s conclusion that the change must have started in the Old English period, as many of the very early texts in the areas mentioned already show <o> as the dominant form. Based on the Early Middle English evidence it is difficult to tell where exactly the change started. I find it difficult to believe that a change could have started simultaneously in several places; maybe Liebl ’ s findings have to be interpreted that the change leaped from one area to others at a very early stage; the area which took the lead will probably remain unsolved. The northwards diffusion of OE a ˉ > ME ǭ can be traced in the West Midlands, where the surviving texts from the Early Middle English period provide a dense enough network both spatially and diachronically to track the diffusion of the change. Unfortunately, there are very few early texts in the East Midlands in general and, because all the later texts (except the copies of pre-Conquest documents) show <o> as the dominant form, one cannot deduce anything about the diffusion of the change in the stages in-between. This case study confirms Dietz (1989: 142) and Liebl (2002, 2008), who argue that <a> spellings are attested further south before w and ɣ , due to the later change ou > au. Before w, <a> spellings are also more frequent in the southern parts of the West Midlands and in Essex and Kent. Furthermore, the data for this study confirms the traditional view that non-lengthening consonant clusters had shortened the vowel before the change took place. The items GHOST and GHOSTLY are the prominent exceptions in my data but, unlike Brunner (1963: § 9) and Jordan (1974: § 23), I did not observe higher frequencies in inflected forms. However, the data from LAEME indicate that the shortening or preservation of length in these cluster is spatially conditioned, as the texts from East Anglia generally show shortening even in the items just mentioned. Finally, there is also regional variation with regard to original <-tt-> clusters in the contracted forms of BEHA : TAN , HA : TAN { C } and 134 5. OE a ˉ > ME ǭ <?page no="153"?> HA : TAN { N } and the comparative form of HOT . While all text files generally show <a> in these forms, there are <o> forms in the text files from Kent and in The Owl and the Nightingale (#1100), which is thought to have Kentish origins. The discussion has shown that with regard to the three major word classes analyzed in this case study, there are no striking differences with regard to the distribution of <a> and <o>. However, frequency appears to play a role with regard to the category ‘ mixed ’ : the more frequent a lexical items is, the more likely it is to display dominant <o> forms. The frequency of <a> spellings remains roughly the same across the entire frequency range. Finally, I have also shown that <ao> spellings, which were thought to stand for / ɔː / (e. g. Moore 1969: 76; Jordan 1974: 73), might not represent exactly this vowel quality in Early Middle English. The LAEME data suggests that, in some texts, the represented vowel might be closer to / ɒː / than / ɔː / . 5.5 Summary 135 <?page no="154"?> 6. The Decline of the Dual in Middle English In the LAEME CTT there are 114 dual forms in 21 different text files. This equals 17.5 times per 100 ’ 000 words, which is a relatively small number. This is more than 300 times less than the total number of first and second personal pronouns in the LAEME CTT 1 and this suggests that the system of the dual forms is not very active in Middle English. Table 6.1 summarizes the total number of forms found per LAEME text file: Table 6.1: Total number of dual forms per LAEME text file. Text file Manuscript 1st Person 2nd Person Total 2 London, BL, Cotton Caligula A ix, Part II 3 0 3 3 London, BL, Cotton Caligula A ix, Part II 5 0 5 64 London, BL, Stowe 34 6 10 16 120 London, BL, Cotton Titus D xviii 0 1 1 121 London, BL, Cotton Titus D xviii 0 3 3 155 Cambridge, Corpus Christi College 444 3 3 6 172 Worcester Cathedral, Chapter Library F 174 10 0 10 173 Worcester Cathedral, Chapter Library F 174 1 0 1 185 Cambridge, University Library, Add. 3020 0 6 6 249 Cambridge, Trinity College B.14.39 (323) 1 0 1 260 London, BL, Royal 17 A xxvii 2 0 2 261 London, BL, Royal 17 A xxvii 0 2 2 262 London, BL, Royal 17 A xxvii 0 1 1 277 London, BL, Cotton Caligula A ix, part I 0 1 1 278 London, BL, Cotton Caligula A ix, part I 1 14 15 285 Oxford, Bodleian Library, Laud Misc 108 1 0 1 301 Oxford, Bodleian Library, Junius 1 8 0 8 1000 Oxford, Bodleian Library, Bodley 34 0 3 3 1100 Oxford, Jesus College 29, part II 8 0 8 1400 Cambridge University Library Ff.II.33 18 0 18 2000 London, Lambeth Palace Library 487 1 2 3 Total 21 text files/ 14 different manuscripts 68 46 114 1 The sum of all tokens retrieved through queries for the tags ‘ P01 % ’ , ‘ P02 % ’ , ‘ P11 % ’ , ‘ P12 % ’ , ‘ P21 % ’ , and ‘ P22 % ’ is 36,899. <?page no="155"?> The 114 forms consist of 68 1st person dual forms and 46 2nd person dual forms. The total number of 1st person dual forms is higher also in terms of the total number of text files in which they are attested: 14 text files contain 1st person forms and only 11 text files contain 2nd person forms. In total, there are only 4 text files in LAEME that contain both 1st and 2nd person dual forms. The special construction described in section 2.3.2 - unc Adame “ to us both, to me and Adam ” , which is attested in several Old English texts - is not recorded anywhere in the Middle English material. Table 6.1 illustrates that the number of 21 text files is a slightly misleading figure concerning the general distribution of the forms because the forms are in fact only found in 14 different manuscripts. The design of the LAEME CTT, which distinguishes text languages rather than manuscripts or literary pieces, accounts for differing counts. Moreover, not only manuscripts, but also some of the content is repeated. Text files #2 and #3 combined and #1100 contain the two extant versions of The Owl and the Nightingale. Moreover, there are various texts that are traditionally grouped together in the Katherine Group. In Table 6.1 these are the following texts and text files: Sawles Warde (#120), Hali Meiðhad (#121), St Katherine (#260), Sawles Warde and St Juliana (#261), St Margaret (#262), and Hali Meiðhad and Sawles Warde (#1000). 2 Among these, Sawles Warde is represented three times and Hali Meiðhad twice. Finally, files #277 and #278 combined contain the output of the two scribes working on the A version of La ȝ amon ’ s Brut. To sum up, the 21 text files only contain 17 different texts. Five of the text files that are listed in Table 6.1 provide forms which were not mentioned in any overview of the dual in Middle English such as Diehn (1902) or Kennedy (1915): #249 is a version of the Proverbs of Alfred, #172 and #173 contain copies of Old English texts by the Tremulous Worcester Hand, and #185 and #1400 are copies of pre-Conquest documents, which were judged to contain contemporary language and were therefore included in the Middle English corpus. But only four of the texts that provide additional attestations of the dual forms go back to Old English originals. Although the name is misleading, the Proverbs of Alfred are of Middle English origin and their date of composition is at the beginning of the second half of the 12th century (Arngart 1978: 5) or shortly after the turn of the 12th century (Skeat 1901: xxxix). The poem exists in various versions but, according to Skeat, the dual form is only found in exactly this passage preserved in manuscript Cambridge, Trinity College B.14.39 (323). 2 If a text file contains several texts, I only mention the text of the passage that contains the dual form(s). 6. The Decline of the Dual in Middle English 137 <?page no="156"?> This first synopsis of the total number of dual forms as found in the LAEME CTT suggests that it is rare in Middle English both in terms of total attestations and also in terms of text files that contain it, which implies that the dual is almost non-existent in the Middle English period. In the next section, I take a closer look at the contexts in which the dual survives longer. This is followed by a discussion of the dual forms that are attested in the LAEME CTT and a comparison with the results of previous studies on the dual (cf. section 2.3). Finally, in section 6.4 I put the findings into perspective in order to examine whether Howe ’ s (1996) claim that the dual forms have survived better in the East Midlands can be verified. 6.1 Contexts of duals As already explained in section 2.3, the specific context that is needed for the use of a dual form in the first place is one of the reasons for the scarcity of its occurrence. The dual is restricted to contexts in which person A refers to himself/ herself and B (1st person forms) or addresses both B and C but not anyone else (2nd person forms) (cf. Guðmundsson 1972: 14). Biber et al. draw on Modern English corpus data and conclude that “ first and second person pronouns, referring to the speaker and addressee, are naturally very common in conversation ” (2000: 333). This in return implies that the written data compiled in the LAEME CTT does not necessarily reflect the proportional frequency in spoken Early Middle English. Furthermore, 1st and 2nd person plural forms of the personal pronoun are very rare in general: 46 of the 167 LAEME files do not even contain 1st or 2nd person plural forms, which, however, is a prerequisite for the occurrence of dual forms. Biber et al. also add that a context in which 1st and 2nd person pronoun forms are potentially used in written sources is given in “ speech situations depicted in fictional dialogue ” (2000: 333). And indeed, most dual forms found in the LAEME CTT appear in passages in which dialogue is embedded in a narrative frame. This is, for instance, the case in Sawles Warde (#120 and #260) 3 in a passage in which the personified cardinal virtue Temperance addresses two messengers. Examples (65) and (66) show this passage as it appears in the two text files: 3 Text file #1000, which also contains a version of Sawles Warde, breaks off before this passage that contains a dual form in other versions. 138 6. The Decline of the Dual in Middle English <?page no="157"?> (65) Eiðer of o ƿ qð meað haueð his stude to spekene ne nis inker no ƿ ðeres tale to schunien ı ˉ his time (#120) (66) Eiðer of o ƿ haueð his stunde to speokene ne nis ı ˉ ckeR noðres tale to schunien in his time (#261) ‘ Either of you (quoth Temperance) has his turn to speak, nor is the tale of neither of you two to be rejected during his time. ’ The utterance by Temperance is marked as direct speech in the passage of text file #120 by the phrase qð meað ‘ quoth Temperance ’ . In #261, on the other hand, Temperance is not mentioned as the speaker, but the entire passage contains quoð several times to mark the turn-taking of the speakers. Apart from this attestation, the dual is also found in the fictional dialogues of the texts of the Katherine Group (#121, #260, #261, and #262), in Havelok (#285), Genesis and Exodus (#155), in La ȝ amon ’ s Brut (#277 and #278), two of the Lambeth Homilies (#2000), and some parts of The Ormulum (#301). Obviously the content of the dialogues varies, but often the dual forms refer to biblical couples, such as Adam and Eve (#155), Zachary and Elizabeth (#301), and Ananias and Sapphira (#2000) and frequently these and several other passages are concerned with marriage or sexual intercourse. These are traditionally associated with duality, which might have been emphasized by the use of the dual forms. In addition to these texts with just some dialogic passages, the dual forms are also preserved in texts with longer debates between two individuals. The most prominent example is the debate between the owl and the nightingale in text files #2/ #3 and #1000, respectively. Example (67) illustrates a passage at the very end of the poem in which the nightingale even makes a reference to the entire debate by mentioning unker speche: (67) Do ƿ e þe ni ȝ tegale seide Ac ƿ a schal unker speche rede An telle to uore unker deme (#3) ‘ Let ’ s do that, the nightingale said, but who will read our speech and relate it before our judges. ’ This dialogue is embedded in a quotative frame, as þe ni ȝ tegale seide in the first line of example (67) shows, and the same is true for the dual forms of the Worcester Fragments, which are embedded in a passage containing a debate, namely The Soul ’ s Address to the Body (#172). In addition, all but two dual forms of Vices and Virtues (#64) are recorded in a longer passage concerned with the body and soul. 6.1 Contexts of duals 139 <?page no="158"?> There are only a few attestations in which the dual forms appear outside of dialogue. First, there are a few texts in which someone in particular is addressed at the beginning of the text, such as Orm ’ s brother Walter in The Ormulum (#301), the son of the speaker in the passage of the Proverbs of Alfred (#249) or King Cnut and Emma in the Will of Mantat (#185). If only one individual is addressed, there are only 1st person dual forms. Second, there is the passage of Ælfric ’ s Grammar and Glossary (#173), in which Latin uter ‘ which of you two ’ is translated as vnker oþer (#173). Finally, the dual also appears in wills that are copied into the Sacrist ’ s Register of Bury St Edmund; most forms are found in what seem to be formulaic expressions: vnker bother/ bothre day ‘ the day of both of us ’ or vnker bother(e) soule ‘ the soul of both of us ’ (#1400). The discussion of contexts in this section has shown that the dual is more often attested in texts that include dialogue. Not all compositions in the Early Middle English period contain such passages and it is possible that the forms were more frequent in spoken Early Middle English. The following two sections discuss the attested duals forms, which might provide further insight as to whether the scribes were familiar with the forms or not. 6.2 1st person dual forms The 68 attestations of the 1st person dual forms are divided into 21 nominative forms, 15 accusative/ dative forms, and 32 genitive/ possessive forms. The genitive/ possessive forms seem to have survived somewhat better, as they not only show the highest number of attestations, but also are preserved in the highest number of LAEME text files: genitive/ possessive forms are found in 10 different text files closely followed by the accusative/ dative forms that are found in 9 and the nominative forms that are found in 7 text files. Map 6.1 (cf. Appendix IV) shows the spatial and temporal distribution of the 1st person dual forms. Most forms are preserved in texts from the South- West Midlands, but there is another cluster around the Wash, which includes #155, #301, #285, and #1400 and, finally, there is #64 in isolation in SW Essex. There are four text files which preserve the entire set for 1st person dual forms in the LAEME CTT, namely #64, #172, #301, and #1400, which is remarkable for several reasons when compared to Table 2.4, which collects the findings from the early studies by Diehn (1902) and Kennedy (1915): On the one hand, the findings by Diehn and Kennedy do not include text files #172 and #1400 and Table 2.4 does not list genitive or possessive forms for Vices and Virtues (#64) and The Ormulum (#301). On the other hand, one can gather from Diehn and Kennedy that the A text of La ȝ amon ’ s Brut contains all 1st person dual 140 6. The Decline of the Dual in Middle English <?page no="159"?> forms; however, the LAEME CTT only provides unker (#278), but no nominative or accusative/ dative form. Table 6.2: 1st person dual forms in the LAEME CTT (P21% + D). Text file Nom. Acc./ Dat. Gen./ Poss. Total n= Forms n= Forms n= Forms n= 2 3 unker 3 3 1 Hunke 4 unker 5 64 3 Ꝩ it, ꝩ it (2) 2 unc 1 unker 6 155 2 ƿ e it, ꝩ it 1 unc 3 172 4 ꝩ it 3 unc 3 unker, vnker, [ ]nker 10 173 1 vnker 1 249 1 unc 1 260 1 ꝩ it 1 unc 2 278 1 unker 1 285 1 unker 1 301 4 Ꝩ itt, ꝩ itt, Ꝩ it t , ꝩ it t 3 unnc, un n c (2) 1 unnkerr 8 1100 1 hunke 7 vnker (5), vnker (2) 8 1400 6 wit 2 vnc 10 vnker 18 2000 1 ꝩ it 1 Total 21 15 32 68 The nominative forms are all quite regular, apart from variation in the spelling of <w> and < ƿ > and Orm ’ s convention of marking short vowels by a following double consonant (#301). There is just one form which deviates from the others, i. e. ƿ e it (#155): (68) ðo q~at laban frend sule ƿ it ben And tre ƿ ðe ƿ ligt* nu unc bi t ƿ en And make ƿ e it her an hil of ston *{=sic - error for PLIGT=} (#155) ‘ Then Laban said: we two shall be friends and make a promise between the two of us and let the two of us make a hill of stone here. ’ In the first line of this passage from Genesis and Exodus, the scribe used the expected form ƿ it and in the following line there is also the usual oblique form unc. However, in the third line of example (68) it is evident that the scribe was not entirely familiar with the dual forms, as he appears to confuse the dual form 6.2 1st person dual forms 141 <?page no="160"?> with the 1st plural personal pronoun we plus the 3rd singular neuter personal pronoun it, which Arngart (1968: 31) explains by obsolescence of the original form when the surviving exemplar was copied. Such an interpretation by the scribe does not seem entirely unfounded from a semantic perspective. Moreover, the accusative/ dative does not display a great range of forms, apart from some orthographic variation. The forms in the Ormulum (#301) again show the idiosyncratic doubling of the consonant described above and the scribe of the Sacrist ’ s Register of Bury St Edmunds frequently spells initial u with the graphemic variant <v>, which is also evident in the form vnc (#1400). Nevertheless, there is again one form, which needs discussion, namely Hunke (#3)/ hunke (#1100) in The Owl and the Nightingale: (69) Hunke* schal itide harm & scho ˉ de Ȝ ef ȝ edoþ griþ bruche on his londe · (#3) *{=Error for zUNKE? “ you two ” ? =} ‘ Harm and disgrace shall happen to both of you if you breach the peace in this land. ’ hunke schal ityde harm & schonde If we doþ gryþbruche on his lond e · (#1100) ‘ Harm and disgrace shall happen to both of you if we breach the peace in this land. ’ This form is recorded in both manuscripts that contain the poem, which suggests that it was already present in a common ancestor of these manuscripts. While unetymological initial <h>-insertion is attested for the entire South in Middle English (McIntosh et al. 1986: 548, dot map 1172; Lass and Laing 2010), the ending -e is more difficult to explain. Cartlidge (2001: xlv) points out that -e frequently serves as a marker for the singular oblique case in The Owl and the Nightingale and this could be an explanation for this form here, although the dual of the oblique cases was never inflected and rhythmically it is mute in both examples. The fact that neither of the two scribes corrected this form suggests that they were not familiar with this 1st person accusative/ dative dual form (Cartlidge 2001: 140, note to line 1733). This view is further supported by the fact that the scribe of #3 continues with the 2nd person pronoun ȝ e in the following line (in the manuscript there is no space between ȝ e and the doþ ‘ do ’ ), where the scribe of #1100 has the respective 1st person pronoun. The scribe of #2 and #3 is known for literatim copying, while the scribe of #1100 freely adapted and generally corrected the language of the exemplar from which he was copying (Laing 2007: 445). There are two possible explanations for the occurrence of we in #1100: Either the scribe of #1100 interpreted hunke as a 1st 142 6. The Decline of the Dual in Middle English <?page no="161"?> person dual form and consciously corrected this error to what he thought was correct here, or hunke is indeed a 1st person dual form and the literatim scribe of #3 simply confused < ȝ > and < ƿ >, which seemed to have been very similar in the common ancestor of #2/ #3 and #1100, as editorial comments in LAEME like the following suggest: (70) $wo: h/ ajplOd{rh}_wOz+E $/ plajOd{rh}_+E (#2) {=wO written over an erasure, wynn replacing what looks like yogh. This} {may be another indication of the equivalence in the exemplar of the} {figurae of thorn/ wynn and yogh=} The confusion of < ȝ > and < ƿ > is discussed in more detail by Laing (2007: 465 - 467), who argues that this similarity was already present in the common ancestor of the two manuscripts. If the second interpretation of the confusion of ȝ e (#3) and we (#1100) is accepted then it is further support for this view. The comment to the form in the text file of #3, however, raises the conjecture that the form could at some stage have been the 2nd person dual form * ȝ unke, and Diehn (1902: 48) and Kennedy (1915: 43) list hunke as a 2nd person form. Moreover, the context of the line in which the wren addresses both the nightingale and the owl also points towards a 2nd person pronoun. While there is little doubt that the scribe of the common ancestor already introduced or copied hunke, it is beyond determination whether he continued with a 1st or 2nd person plural pronoun in this line. In any case, the confusion presented in this salient form appears to be older than the two extant versions of The Owl and the Nightingale. The genitive/ possessive forms do not seem to demand any further discussion, as they simply display the kind of variation that was already described for the other cases plus some forms with abbreviations for -er that are tagged as follows in LAEME (cf. Laing and Lass 2008 - : 3/ 14 - 17): $/ P21GD_UNKer (#2) and $/ P21GD_VNK~ (#1100). But appearances are often deceptive: although the forms displayed in Table 6.2 show a rather homogeneous picture, there is at least one form that Kennedy (1915: 42) takes as actually representing a 2nd person form (cf. section 2.3.3), namely unker (#285), the only dual form in Havelok: (71) Roberd willam hware ar ye Gripeth eþer unker a god tre And late we nouth þise doges fle (#285) ‘ Robert, William, where are you? Each of us two must take hold of a good stick and we should not let these dogs flee. ’ 6.2 1st person dual forms 143 <?page no="162"?> In LAEME this form has received the tag $/ P21GD_UNKER; however, in this passage Hugh Raven clearly addresses his brothers Robert and William, so I think that Kennedy (1915) is right in identifying this form as a 2nd person dual form and that the tag in LAEME should probably be P22GD, maybe followed by a comment explaining that the form appears to be a 1st person dual form. The confusion of the form with the 1st person form could possibly have been caused by the other forms of the 2nd oblique dual cases showing <u> in the East, as discussed in the next section. Kennedy has remarked “ a variety of forms quite out of proportion to the number of occurrences ” (1915: 43) in his discussion of the 2nd person dual forms (cf. section 2.3.3), but this is definitively not the case for the 1st person dual forms. The 1st person dual forms that are attested in Table 6.2 do not display the vast array of forms mentioned by Kennedy and the discussion in this section has shown that there are only a few forms that are different from the majority. 6.3 2nd person dual forms The 2nd person dual forms are slightly less frequent than the 1st person dual forms. The 46 forms amount to a total of 15 nominative, 19 accusative/ dative, and 12 genitive/ possessive forms that are recorded in 11 LAEME text files. All but two text files with 2nd person dual forms preserve accusative/ dative forms, 7 text files contain genitive or possessive forms, and only 5 text files show at least one attestation of a nominative form. The forms as attested in the LAEME CTT are displayed in Table 6.3. The locations of all texts which contain at least one 2nd person dual personal pronoun are shown in Map 6.2 and the picture is very similar to Map 6.1. There is a cluster of texts in the South-West Midlands, a few texts around the Wash (#155, #185 and #301), plus file #64 in isolation in SW Essex. Table 6.3 suggests that there are only two texts that preserve the entire set of 2nd person dual forms. Yet, if one also takes into account the forms mentioned by Diehn (1902) and Kennedy (1915), there are two additional texts that contain all 2nd person dual forms, namely #155 and #301. It is a very unfortunate circumstance that most 2nd person dual forms are attested in passages of the Ormulum, which are not transcribed for the LAEME CTT. Kennedy (1915) lists a total of 51 forms in this manuscript: 20 x ȝ itt, 11 x ȝ unnkerr, and 20 x ȝ unnc. A simple search for these two forms in the entry for the Ormulum in the Corpus of Middle English Prose and Verse (accessed 14 Oct 2011) confirms the 11 attestations of ȝ unker but yields even 21 attestations for ȝ unnc. This indicates that if one solely relies on LAEME for such a context- 144 6. The Decline of the Dual in Middle English <?page no="163"?> specific low-frequency variable as the dual, the emerging picture might be highly distorted. Table 6.3: 2nd person dual forms in the LAEME CTT (P22% + D). Text file Nom. Acc./ Dat. Gen./ Poss. Total n= Forms n= Forms n= Forms n= 64 4 ᵹ it 3 ᵹ ing, ᵹ ung, ᵹ inc 3 ᵹ inker, ᵹ u ˉ ker (2) 10 120 1 inker 1 121 1 inc 2 inker 3 155 2 ge it 1 gunker 3 185 1 yt 5 inc 6 261 1 inc 1 ı ˉ ckeR 2 262 1 inc 1 277 1 inc seluen 1 278 7 ȝ it (6), ȝ et 5 inc (4), hit 2 incker, inckere 14 1000 1 inc 2 inker 3 2000 1 ᵹ it 1 h ı ˉ c 2 Total 15 19 12 46 The five text files which contain nominative forms of the 2nd person dual display an unusually high number of five different forms. The Middle English forms that correspond exactly to the Old English forms are ᵹ it (#64 and #2000) or ȝ it (#278). But the one attestation of ᵹ it from those parts of the Lambeth Homilies that are transcribed in #2000 shows that what appears to be a quite regular form might have also caused some trouble to the copying scribe: (72) Ð c ƿ ~ Petr ꝰ · h ƿ i i ƿ earð h ı ˉ c s ƿ a  ᵹ it* dursten fondian godes  (#2000) *{=MS gIFT with F subpuncted=} ‘ Then Peter quoth, why is it so agreeable to the two of you that you dare to put God on trial? ’ In this passage, in which St Peter addresses Ananias and his wife Sapphira, the scribe first wrote gift but then realized his error and consciously corrected it to the dual personal pronoun. This would, however, indicate that he actually understood the dual form. Laing (2008 - : 127) assumes that the main scribe himself is responsible for the corrections. 6.3 2nd person dual forms 145 <?page no="164"?> In Genesis and Exodus (#155), on the other hand, the scribe again appears to confuse the underlying dual form with a 2nd person plural personal pronoun plus it: (73) Sone ge it ðor of hauen eten · Al ge it sullen ƿ itent and nogt forgeten · (#155) ‘ as soon as the two of you have eaten thereof the two of you will know everything and never forget (it). ’ This is the same construction that I have already discussed above with regard to the first person dual form we it in example (68). Similarly, there is confusion with the third person singular neuter pronoun in another text from the same area: (74) Gehealde inc here on liue heofan engle kinge · and ᵹ e lede inc on his lihte mid him þer yt wiþ uten sorh ᵹ e euere wunian · amen. (#185) ‘ The king of angels may preserve you two here in the life of heaven and may lead you two into his light where you two should live with him ever without sorrow. ’ Mantat addresses King Cnut and Emma in his will and this passage contains the object dual form inc twice. However, in the last subordinate clause the 2nd dual nominative personal pronoun is expected as subject, but the recorded form is yt, which has the appearance of a 3rd person singular neuter pronoun. The form yt is rarely recorded in Early Middle English, but all three attestations (text files #169, #188, and #282) are found in texts from the North or North-East Midlands, like the Will of Mantat (#185). The last form of the 2nd person nominative dual that needs some discussion is ȝ et in #278. The majority variant of the 2nd person nominative form is ȝ it in La ȝ amon ’ s Brut but, in the passage reproduced in example (75), in which King Belin addresses two wise men which he captured, it is replaced with ȝ et once. However, only a few lines further down, the 2nd person nominative is embedded in a similar structure again but, in this instance, in the majority variant ȝ it: (75) Ic inc habbe beiene an inne mine bende ˉ ȝ if ȝ et wulled af ter ræde faren ich hit* wulle freoi ȝ en Mid gode ræue bihon And makien inc riche faren 146 6. The Decline of the Dual in Middle English <?page no="165"?> mid me suluen & ich inckere freond wurðe ȝ if ȝ it me wulleð laden (#278) *{sic - error for INC? } ‘ I have both of you two only in my bonds; if you two want to proceed according to my advice, I will set you two free, dress you with good garments and make you two rich. You should go with me and I become the friend of you both if you two will lead me. ’ In text file #278 the form ȝ et appears another 6 times as the adverbial YET . It is possible that the scribe mistook the dual form for YET , as it would fit the context semantically, although a subject would be missing for wulled. Another form in this passage implies that the scribe was not entirely familiar with the dual: In the third line he writes hit, where an accusative form of the 2nd person dual is expected, which is noted in the LAEME comment to this form. Kennedy (1915: 43) attributes this to a scribal error “ due to a confusion of hinc and nom. ȝ it ” . But, again, there could be confusion with the 3rd singular nominative neuter form, which in #278 is spelled with initial <h-> everywhere but once. A similar type of confusion could have caused the form h ı ˉ c in text file #2000, in which the nasal is only represented through a nasal stroke. I already mentioned above that the scribe must have had some troubles with the dual, as the <f> of the form gift is subpuncted to correct it to the appropriate dual form ᵹ it and in the manuscript this is just in the line following h ı ˉ c. For this reason one could argue that the scribe maybe wrote hit first, before correcting it to hinc through a nasal stroke. After all, the shape of the graphemes for <c> and <t> in insular scripts looks very similar. However, an examination of the microfiche facsimile of manuscript London, Lambeth Palace, MS 487 (Doane and Hussey 2000) has revealed that the present scribe undoubtedly wrote <c> in the first place, so a correction in the present version is unlikely. However, this does not rule out the possibility that the form was hit in the exemplar from which the scribe was copying. The three accusative/ dative forms that are attested in Vices and Virtues (#64) show another type of apparent confusion. The forms ᵹ ing, ᵹ ung, and ᵹ inc all have word-initial g - , which very likely represents a voiced palatal approximant, and the forms variably have <i> or <u> as their medial vowel and <c> and <g> as their final consonant. The latter probably reflects the voicing of plosives, which started in Old English in Kent and neighboring areas exactly in nc clusters (Jordan 1974: § 178, remark 2). The variation of <i> and <u> needs further discussion. This variation, together with an initial consonant spelled < ᵹ -> or <g->, is also found in the genitive/ possessive forms of #64 and #155. As mentioned above, both Diehn (1902) and Kennedy (1915) also list the forms ȝ unnc and ȝ unnker for the Ormulum. Furthermore, Kennedy (1915: 43) 6.3 2nd person dual forms 147 <?page no="166"?> mentions one occurrence of gunc in Genesis and Exodus, which is in a line that is not included in #155, and Diehn (1902: 48) provides the forms ginc and gunc for the Gospel of St Marc in manuscript Oxford, Bodleian Library, Hatton 38, which is not part of LAEME. Diehn (1902) does not comment on these forms and Kennedy (1915: 43) merely mentions their “ tendency to adopt an initial consonant ” . Howe (1996: 137 - 138) explains the forms by analogy to the nominative 2nd person dual form git or to the 2nd plural pronouns with initial / j/ . Such forms are also found in the object cases of the 2nd person plural personal pronouns such as ȝ ou or ȝ ow, from which ModE you developed. Mossé (1952: 55) explains these forms either by accent shift in éow > eów or as a blend of the subject ȝ e ˉ with the object form ou. Howe (1996: 137 - 138) states that the dual forms with initial / j/ are more likely to have developed by analogy to the 2nd person dual nominative forms with initial / j/ as accent shift cannot have played a role. However, I do not think that the 2nd personal plural personal pronouns should be disregarded completely, as analogy to these forms is just as likely. Furthermore, Howe explains the appearance of <u> in these forms by analogy with the 1st person dual oblique forms unk and unker, which means that the initial consonant and the vowel <u> have to be attributed to the influence of two different sets of forms. Analogy with the 2nd person plural forms, on the other hand, accounts not only for the appearance of / j/ but also for the variation of <i> and <u>, as forms like ȝ uw or ȝ ou would have been pronounced as / ju ː / . 4 All dual forms with initial / j/ are attested in text files from the East Midlands so, if analogy to the oblique forms of the 2nd person plural personal pronoun is the accurate explanation, then the 2nd person plural forms with / j/ should theoretically also be attested mainly in this area. Map 6.3 displays the occurrence of the oblique cases of the plural with initial <g>, < ȝ >, and < ᵹ > to represent / j/ in the LAEME text files. This map is based on a LAEME query for the 2nd plural personal pronoun tag P22 % with g% (not case sensitive) and z% entered in the field for form and subsequent exclusion of all nominative and also dual forms. The map shows that forms with initial / j/ in the oblique forms of the 2nd person plural forms are first attested in the earliest texts from the East and then spread to the South-West Midlands towards the end of the 13th century. 4 Variation of <u> and <i> could possibly also be attributed to Scandinavian influence. The respective forms are ON yk(k)r (acc./ dat.) and ykkar (gen.) (Nedoma 2001: 26). Phonologically speaking, / y/ was most likely a sound between / i/ and / u/ , sharing frontness with / i/ and rounding with / u/ , which might account for <i> and <u> spellings. However, there is only one possible dual form with yin LAEME (cf. example (74)) and, moreover, there are no dual forms whatsoever in Old Danish sources (Howe 1994: 311), so it is unlikely that language contact played any role here. 148 6. The Decline of the Dual in Middle English <?page no="167"?> Two of the western texts (#278 and #249) that contain dual forms of the oblique cases without initial <g>, < ȝ >, or < ᵹ > show oblique plural forms with initial / j/ . This allows the conclusion that at some stage the confusion of forms with the oblique plural pronouns must have occurred somewhere in the East and that subsequently the / j-/ forms spread before they disappeared completely. Support for this claim is found in the two forms that are not even recorded in the LAEME CTT. Above I already mentioned that ginc and gunc are also attested in the Gospel of St Marc in manuscript Oxford, Bodleian Library, Hatton 38. The former is listed as the only variant reading with initial gfor inc in Mark 11: 2 among the eight collated manuscripts of the West-Saxon Gospels in Liuzza (1994: 65). Both Diehn (1902: 48) and the DOE cite Mark 1: 17 as the only attestation of gunc in the same manuscript, but Liuzza does not list this variant reading. According to Skeat ’ s edition (1970 [1871]), it is the only variant with initial g-. Laing (1993: 133) included the Hatton manuscript of the West-Saxon Gospels in her catalogue for possible LAEME sources; it is, however, not part of the final corpus, possibly because it is ultimately a copy of the 11th century version in Oxford, Bodleian Library, Bodley 441. According to Laing, the manuscript has associations with Canterbury, Kent; she refers to Ker, who notes that “ [t]he linguistic forms are Kentish ” (1957: 387). The scribe of this manuscript is known to have ‘ modernized ’ the lexicon of the West-Saxon Gospels in many places (cf. Fischer 1997), and he appears to have ‘ updated ’ the Old English forms to the distinct Eastern forms to which he might have been more accustomed. These attestations of gforms in Kent also provide support for the assumption that the underlying form in The Owl and the Nightingale for Hunke (#3) and hunke (#1000) might have been ȝ unke as suggested in the comment following the form in #3 and as shown in (69) above. Several linguistic features point to an original Kentish or generally South- Eastern provenance of the poem (Cartlidge 2001: XV) and the form ȝ unke can be confirmed by the evidence from other eastern texts. On the basis of the attestation of the forms with initial / j/ , for which there is no parallel in Old English according to the DOE, one may conclude that these forms diffused across a larger area in the East, which must have extended as far as the Wash in the North and to Kent in the South. I think it is unlikely that exactly the same kind of confusion happened in several places independently, but the scarcity of material containing dual forms only permits a tentative conclusion. It remains impossible to tell with certainty to what degree the lack of familiarity with dual forms in general might have played a role, but all the evidence suggests that the forms with / j/ must have spread across the East Midlands and Kent before the dual was lost completely. 6.3 2nd person dual forms 149 <?page no="168"?> 6.4 The survival of the dual in ME Howe concluded that the dual forms are “ best preserved, used more consistently, and survive longest in the East Midlands ” (1996: 141), probably relying on the findings by Ladd and Radice (1951/ 52) (cf. section 2.3.3). However, the data from LAEME, supplemented by the findings of Diehn (1902) and Kennedy (1915) as presented in Table 2.4, presents a picture that is not as clear-cut as Howe ’ s summary suggests with regard to preservation, consistency, and duration. 6.4.1 Preservation Map 6.4 displays the locations of all 21 LAEME text files that contain at least one dual form. Map 6.1 and Map 6.2 have already shown that the forms are attested on either side of the Early Middle English East-West divide with two relic areas on either side of the gap plus text file #64 in isolation in Essex. It is certainly not exaggerated to state that the cluster of text files with at least one dual form is greatest in the South-West Midlands. Furthermore, the discussion of individual forms in sections 6.2 and 6.3 indicates that the forms, when compared to their Old English equivalents, are better preserved in the South-West Midlands, as only a few deviate from the paradigm presented in Table 2.4. However, one should not disregard the fact that the forms might have changed towards Middle English and, therefore, the forms of the East Midlands have to be taken into account as well. In several East Midland texts there is a vast array of different forms, such as the 2nd person object and genitive forms with initial ȝ -, as displayed in the forms ȝ ing, ȝ ung, ȝ inc, ȝ inker, ȝ u ˉ ker in #64 and similarly in #155 and #301. Together with #285 these are the texts from the East Midlands that are not direct copies from pre-Conquest documents. I have already mentioned above that it is difficult to tell whether these forms show lack of familiarity of the scribes and point therefore towards the demise of the dual system, or whether they should be regarded as an eastern innovation which managed to diffuse across a larger area. Moreover, the fact that the western forms represent more or less exactly the Old English system could also imply that they are not used anymore and are therefore preserved in a fossilized manner. Having said that, there is no denying that the variation between forms with initial i-, ȝ i-, and ȝ uor the forms Ƿ e it and ge it in #155, and the many forms found in #64 point towards a general instability of the system in the East Midlands. Among those texts which cluster in the West-Midlands, such a large amount of variation is only found in the A version of La ȝ amon ’ s Brut (#278) from the northernmost part of the South-West Midlands. 150 6. The Decline of the Dual in Middle English <?page no="169"?> In sections 6.1 and 6.3 I stated that there are only very few Middle English texts that preserve either the 1st or 2nd person set of dual personal pronouns. In total, there are only three texts that contain the entire dual personal pronoun paradigm. These three texts mark a triangle between the two relic areas in the South-West Midlands and just south of the Wash and the single location in SW Essex. In other words, the three text files that preserve all forms are scattered across a large area and include both the East and the West. The attested forms definitively do not allow for the conclusion that the dual is better preserved in the East Midlands. If Howe (1996: 141) actually refers to the preservation of the Old English forms or to the total amount of attested forms, then the dual forms are better preserved in the South-West Midlands. Furthermore, the distribution of the few texts that contain all forms is found in all relic areas. There is no evidence in the Early Middle English data that supports Howe ’ s conclusion with regard to this aspect. 6.4.2 Duration Howe (1996: 141) also claims that the dual has survived longest in the East Midlands. It is indeed the case that three of the manuscripts in the East Midlands with dual forms are dated to the 14th century, which is later than anywhere else (cf. Map 6.4). However, one of these three texts is a copy of pre- Conquest documents. The Will of Mantat (#185) in the Red Book of Thorney Abbey is preserved in a cartulary, which also contains English boundary clauses to Latin writs about which Laing (2008 - : 35) notes that they “ are in only slightly modified Old English ” . The language of Mantat ’ s will, however, was regarded as a suitable documentary anchor text by the compilers of LAEME. Nevertheless, while the language in general might represent more recent or even contemporary Middle English, it is likely that the dual forms were part of the original already. The other two post-1300 texts are Havelok (#285) and Genesis and Exodus (#155) and, as discussed above in sections 6.2 and 6.3, both scribes seem to have problems with the dual forms, which suggests that they might not have used it actively anymore. Genesis and Exodus is unique in this version (cf. Laing 2008 - : 10), but both books of Moses were, of course, well known texts in the Middle Ages and it is possible that the scribe had some older version of the text at hand when he composed his own version. Furthermore, the original of Havelok was possibly written already in the second half of the 13th century. Holthausen (1928: XII) notes that Robert Manning quotes four verses from Havelok in his Handling Sinne, which Holthausen dates to 1303. Furthermore, the fourth text in the area, the Sacrist ’ s Register of Bury St Edmunds (#1400), is only slightly older but, like the Will of Mantat, this is a copy of an Old English text. According to Laing (2008 - : 36), the Register 6.4 The survival of the dual in ME 151 <?page no="170"?> contains “ copies of 47 pre-Conquest documents relating to the Abbey of Bury St Edmunds consisting of royal grants (including vernacular writs and privileges) and wills ” . Laing continues that, although these were transcribed from Old English originals, the language is actually adapted and modernized. But again I believe that the dual forms represent an older layer of language and should maybe not be taken as evidence for the contemporary East Midlands variety. This is further supported by the fact that 8 out of the 18 dual forms attested in #1400 are found in the expression vnker bother/ bothre day (6 x) and vnker bother(e) soule (2 x). The DOE web corpus provides additional attestations of the second construction: (76) uncer begea sawle Ch 1496 [0001(1)] uncere begra sawle Ch 1498 [0003(10)] uncer begra saule Ch 1503 [0005(9)] uncer begea sawla Ch 1509 [0001(1)] uncer begea sawla Ch 1533 [0002(2)] uncer begea sawla Ch 1533 [0012(27)] This suggests that these are formulaic expressions that were already used in the Old English period and that preserved the dual in text file #1400. In the South-West Midlands the dual is also preserved in a group of text files that contain copies of Old English texts: #172 and #173 refer to manuscript Worcester Cathedral, Chapter Library F. 174, which was written by the famous Worcester Tremulous Hand (Laing 2008 - : 189 - 190). The section of #172 which contains the dual forms is a fragment of The Soul ’ s Address to the Body; text file #173, on the other hand, contains a copy of Ælfric ’ s Grammar and Glossary, where Latin uter ‘ which of you two ’ is translated as vnker oþer. Laing (2008 - : 187) states that “ [t]he language of this text has been influenced by being copied from an Old English original. ” In this case it is possible to consult an older version of the text in order to verify whether the dual form was already present in an ancestor of the text file included in LAEME: Manuscript Oxford, St. John ’ s College 154 contains an early 11th century version of Ælfric ’ s Grammar and Glossary and the dual is already attested in the same passage there (folio 60 r). 5 Furthermore, the ‘ Body and Soul ’ theme is already known in Old English literature, so again, it is most probable that the dual forms belong to an older language layer. This might also be the case in some of the other attestations of the dual in the South-West. For instance, some of the Lambeth Homilies (#2000) are modernized versions, others free adaptations of Old English homilies (cf. Stadlmann 1921: 1). 5 Available as a digital facsimile at <http: / / image.ox.ac.uk/ show? collection=stj&manuscript=ms154> (accessed 18 October 2011). 152 6. The Decline of the Dual in Middle English <?page no="171"?> Since many of these texts go back to earlier versions, it is difficult to tell how long the dual really survived into Middle English. Both original Middle English compositions that are recorded in two early 14th century manuscripts from the East Midlands show great confusion of forms. I assume that the scribes of the two texts were not very familiar with the dual anymore and, if one accepts the date of composition of these texts in the second half of the 13th century and disregards the contemporary copies of the two pre-Conquest documents, one cannot conclude that the forms survived longer in the East Midlands. In other words, while it is true that the forms were preserved in later manuscripts of the East, there is no evidence that they also survived this long in the spoken register. 6.4.3 Consistency Howe ’ s (1994: 141) third point concerns the consistency with which the forms survived. In the LAEME data there are several examples like (77) and (78) in which dual forms occur side by side with a regular plural form in the same utterance, such as the 1st person nominative plural forms in these examples: (77) Ac ar ƿ e to unker dome fare · (#2) ‘ But before we (pl.) proceed to our (dual) judgment. ” (78) & ƿ e su\len unc do delen (#249) ‘ And we (pl.) shall separate us (dual). ’ Furthermore, many of the LAEME text files that contain dual forms also have at least one normal plural form that is supplemented by what Howe refers to as “ lexical dual quantifiers ” (1996: 141), i. e. a lexical item which makes explicit that exactly two individuals are meant. Example (79) lists some of these pairs: (79) hure eiþer (#2) us ba (#261) bo þe ƿ e (#3) us bi\tuei ᵹ en (#277) ᵹ eu baðe (#64) ȝ e ba (#278) o ƿ baðe (#120) us boþe (#285) bituhhen o ƿ (#121) vr eyþer (#1100) gu to (#155) vre boþere time (#1400) o ƿ ba (#260) ure efrec oðer (#2000) This short list shows that dual forms are not always consistently used in texts from either side of the East-West divide. It appears that Howe ’ s claim is based solely on the fact that The Ormulum “ has duals throughout ” (1994: 141) and, as far as I can tell, this is indeed the case. Nevertheless, I find it problematic to draw conclusions for an entire area on the basis of one single text. All other 6.4 The survival of the dual in ME 153 <?page no="172"?> East Midland texts that are not direct copies from pre-Conquest documents are not consistent in the use of the dual. On the other hand, I also do not want to claim that the forms are more consistently used in the South-West Midlands. One should bear in mind that already in Old English and also in other Germanic languages the dual forms are not attested consistently (cf. section 2.3); therefore, I believe that in general, consistency does not reveal anything about the survival of the dual in Middle English and should not be given too much emphasis. I pointed out in section 2.3.3 that in many languages dual forms are reinforced by lexical quantifiers before they are lost completely. A few examples are listed in (80): (80) unc t ƿ a (#260) unc bi t ƿ en (#155) ƿ it baðe (#64) ȝ it . . . beien (#278) inker eiðer (#1000) ȝ u ˉ ker noðer (#64) vnker oþer (#173) Such reinforcements are present in 12 out of the 21 text files that contain any dual form of the personal pronoun. Forms of BOTH are found 21 times, NEITHER 4 times, BETWEEN and EITHER twice each and both OTHER and the cardinal number 2 only once. The picture is slightly distorted as 9 reinforcements with BOTH are found in the formulaic expressions vnker bother/ bothre day (6 x) and vnker bother(e) soule (2 x) of text file #1400, which I already mentioned above. However, even if those formulaic expressions are disregarded, the reinforcement of the dual with BOTH is clearly favored. Table 6.4 compiles the figures for all LAEME text files concerning the reinforcement of the dual forms. Two thirds of all texts in the East Midlands show this reinforcement of the dual but only about half of all texts from the West. Most of the nine text files that do not show this phenomenon at all are restricted to the southern parts of the South-West Midlands: #2, #3, #172, #249, #277, #1100 and #2000. Map 6.5 shows that this phenomenon is mostly found in the East Midlands and northern part of the South-West Midlands. The only exception here is the copy of Ælfric ’ s Grammar and Glossary, which has vnker oþer (#173) and is located slightly more to the south than the other text files, but otherwise this reinforcement is not found in the South-West Midlands. Guðmundsson states that “ dual forms survived longest in conjunction with words meaning two or both ” (1972: 95) and it is likely that the occurrence of the dual forms with such lexical items indicates that these are the last remnants of the dual forms in these 154 6. The Decline of the Dual in Middle English <?page no="173"?> areas and that it is exactly these constructions that preserved it longer. Thus, this provides further support for the claim that the dual forms may have survived better in the South-West Midlands, where these pairs are not found. In other words, the withdrawal of the dual forms seems to go along with this reinforcement through lexical quantifiers, which already started in the Old English period (cf. DOE, git 1.a.), but which is not attested in a relic area in the South-West Midlands. Table 6.4: Reinforcement of the dual through lexical quantifiers. 1st Person 2nd Person Combined Total Lexical Quantifiers Total Lexical Quantifiers Dual Lexical Quantifiers n= n= % n= n= % n= n= % 2 3 0 0.0 0 0 0.0 3 0 0.0 3 5 0 0.0 0 0 0.0 5 0 0.0 64 6 3 50.0 10 4 40.0 16 7 43.8 120 0 0 0.0 1 1 100.0 1 1 100.0 121 0 0 0.0 3 1 33.3 3 1 33.3 155 3 1 33.3 3 0 0.0 6 1 16.7 172 10 0 0.0 0 0 0.0 10 0 0.0 173 1 1 100.0 0 0 0.0 1 1 100.0 185 0 0 0.0 6 0 0.0 6 0 0.0 249 1 0 0.0 0 0 0.0 1 0 0.0 260 2 1 50.0 0 0 0.0 2 1 50.0 261 0 0 0.0 2 2 100.0 2 2 100.0 262 0 0 0.0 1 1 100.0 1 1 100.0 277 0 0 0.0 1 0 0.0 1 0 0.0 278 1 0 0.0 14 3 21.4 15 3 20.0 285 1 0 0.0 0 0 0.0 1 0 0.0 301 8 3 37.5 0 0 0.0 8 3 37.5 1000 0 0 0.0 3 2 66.7 3 2 66.7 1100 8 0 0.0 0 0 0.0 8 0 0.0 1400 18 9 50.0 0 0 0.0 18 9 50.0 2000 1 0 0.0 2 0 0.0 3 0 0.0 Total 68 18 26.5 46 14 30.4 114 32 28.1 6.4 The survival of the dual in ME 155 <?page no="174"?> 6.5 Summary Dual forms in Middle English are very rare, which might explain but not justify the fact that the dual is treated only marginally in the Middle English handbooks and grammars. However, in some texts the entire dual paradigm is preserved and the dual forms are not only attested in copies from Old English texts but also in original Middle English compositions or translations, such as the A text of La ȝ amon ’ s Brut, The Owl and the Nightingale or The Ormulum. This case study has shown that even a low-frequency variable may provide valuable insights into diffusion and language change in Early Middle English. First of all, the data from the East Midlands with initial / j/ indicates that also a variable that is almost gone from the written records may show diatopic variation and that this new variant must have diffused across a relatively large area before the forms disappeared completely. Forms with initial / j/ are not attested in Old English and, in the Early Middle English period, they are not attested anywhere in the South-West, which suggests that these forms have to be regarded as regional variants rather than evidence for a lack of familiarity with the forms, as has been previously suggested. The data from LAEME has brought to light five texts that have not been mentioned in any study on the dual in Middle English so far. Four of these are copies from Old English originals, so the forms do not shed new light on the dual in Middle English, but the text known as Proverbs of Alfred is an original Middle English composition, which also contains a dual form. Furthermore, the LAEME CTT provided forms that were hitherto not included in larger scale studies, but the retrieved data also lacked some of the forms that are listed in Diehn (1902) and Kennedy (1915). The missing dual forms are those that are attested in passages not transcribed for the LAEME CTT. This was most notably the case for the data from The Ormulum (#301). In the LAEME text file there are no 2nd person dual forms, but a total of 55 forms were found by Kennedy (1915). This means that for such a low-frequency variable, it is necessary to check the entire text, as the passage transcribed in LAEME might miss some important data. However, this does not devalue the benefit of LAEME. A search query in LAEME provides the user with a list of possible forms, which then facilitates the search within an unannotated electronic version of the text. Finally, I was able to show that the claim that the dual forms are better preserved in the East Midlands is a misconception of Howe (1996). In the LAEME CTT there are clearly more texts from the West Midlands that contain any dual forms and, with regard to preservation, consistency, and duration, there is no evidence in the LAEME data that would favor the East over the West. 156 6. The Decline of the Dual in Middle English <?page no="175"?> 7. Spatial Diffusion in Early Middle English The previous chapters concentrated on the results and analyses of the individual case studies. All three variables have provided new insight into change and diffusion in Early Middle English. In this chapter I summarize the most important findings concerning the spatial diffusion of the three features in order to make more general observations on diffusion patterns in Early Middle English. Most importantly, I address the theoretical and methodological implications of the findings from chapters 4 - 6 and the applicability of modern diffusion models to Early Middle English data. 7.1 Spatial diffusion in the three case studies 7.1.1 Reduction of the number of grades in strong verbs Chapter 4 has shown that the leveling of the past grade vowels is not as prominent in the period as expected (cf. section 2.1.2.3). In fact, apart from classes IVa and V there are only a few texts that contain forms that show the unetymological past grade vowel. There is strong evidence that the seeming extension of the plural to the singular past grade vowel in classes V and IVa in the West Midlands and Kent has to be attributed to a phonological variant. In the Kentish and Mercian varieties of Old English the past grade vowels were already very similar: The first past grade vowel OE æ (< Gmc a) appears as e in these varieties (cf. Brunner 1963: § 11, note 2; Brunner 1965: § 19.3; Jordan 1974: § 32) and is consequently only distinguished in length from the second grade vowel e ˉ . Since in some texts from the East Midlands verbs of class V also show <e> in the singular, where the above-mentioned phonological feature is not attested in Old English, there is reason to believe that these forms have spread northwards from Kent. The so called ‘ northern leveling ’ under the singular vowel is also attested in classes V and IVa in the East Midlands. In the other classes it is not prominent at all in the Early Middle English data apart from the northernmost early 14th century texts. This phenomenon is reported to have spread southwards in the course of the Middle English period (Brunner 1963: § 69.1; Markus 1990: 152; Lass 1994 b: 88). The few attestations of the feature in the northern parts of East Anglia support this observation as they generally show higher frequencies of the singular past grade vowel in plural forms. The few early attestations of the merger in classes V and IVa in Essex is most likely due to another <?page no="176"?> phonological variant: OE æ ˉ became a ˉ in the South-East Midlands (Brunner 1963: § 11.2). This vowel represents the 2nd past grade vowel of classes V and IVa, which means that the two past grade vowels are only distinguished in quality. In writing, this difference in length is not distinguished at all and both appear as <a>. This means that in the East Midlands the two diffusion processes operate from two different directions both with regard to geographical distribution and with regard to grade vowel. In Map 7.1 (a) and (b) (cf. Appendix IV) I juxtapose the contrasting developments in two excerpts that show the verbs of class V in the (South) East. Map 7.1 (a) shows the northwards diffusion of <e> spellings in the forms that historically would have the 1st past grade <a>. Apart from Vices and Virtues (#64 and #65), in which <a> is the dominant past grade vowel both in the singular and plural, there is a decrease in frequency from south to north, i. e. <e> as the majority variant in the Kentish text files (#8 and #291), mixed in the text files from Western Suffolk (#1300) and Eastern Cambridgeshire (#137), and <e> as the minority variant in the four northernmost text files where the feature is attested (#155, #282, #285, #1400). Map 7.1 (b) on the other hand, shows the southward diffusion of the leveling under the singular vowel. The singular grade vowel as the majority variant in the text files from Essex has to be disregarded here, as these text files show <a> due to the phonological process described above. Apart from these exceptions, the northern most text files have <a> as the majority variant (#149, #182, #150). In text files #282 and #285 the two grade vowels are roughly equally frequent and in #137 - that is the southern most of these text files - <a> is only the minority variant in the forms where the 2nd past grade vowel would be expected. In text files #137, #282, and #285 both processes are attested, but the evidence nicely illustrates how the one process pushes from the south (majority variant in #137) and the other process from the north (majority variant in #282 and #285). All three text languages have been fitted (Laing 2008 - : 56, 165 and 167) and - under the assumption that both processes spread gradually - the evidence from class V supports the relative localization of these three text files. Although it is apparent that past grade vowel mergers in the Early Middle English period are in most cases a phonological phenomenon, I was nevertheless able to show the diffusion of the mergers in the East Midlands, where both types of merger are operating and competing. Unfortunately, the network of texts in the East Midlands is not very dense and, from a temporal perspective, also somewhat patchy; for this reason the conclusions are not based on very firm ground. However, the evidence from the available data allows the tentative conclusion that the merger of the past grade vowels of class V and, with respect to the merger under the singular vowel also of class IVa, was indeed spreading in the East Midlands. 158 7. Spatial Diffusion in Early Middle English <?page no="177"?> 7.1.2 OE a ˉ > ME ǭ The diffusion of the change OE a ˉ > ME ǭ is most prominent in the West Midlands. Map 7.2 shows a detailed view of this area, where a transition zone between the dominant <a> spellings in the North and the dominant <o> spellings in the southern parts is evident. The <a> spellings in the north are receding as time progresses and the <o> spellings are pushing northwards from the south. However, it is difficult to tell how much of this transition zone has to be attributed to the localization of the text files through the fit-technique, in which texts are only localized relative to each other. In any case, the diffusion patterns as shown in Map 7.2 do not contradict the placement of the text files in the South-West Midlands and, on a very general level, the diffusion is certainly evident even through this relative localization of the texts; however, more detailed descriptions are not possible. For instance, Map 7.2 implies that the Severn was influential in the diffusion of the change, as there are increasing frequencies of <o> spellings along the river in northern Worcestershire and possibly also in the rather high frequency of <o> spellings in southern Staffordshire (#146). This would correspond to Conde Silvestre and Hernández Campoy ’ s (2002) model, according to which the waterways played an influential role in the diffusion of innovations in Late Middle English. However, as the exact location of the texts might be further away from the river, one cannot make any safe conclusions about the role of the Severn in the diffusion of the change. While it is impossible to clarify what role waterways might have played in the general diffusion of the change a ˉ > ǭ in this area with the densest network of texts in LAEME, they might help to explain the occurrence of the change ou > au in Kent, as discussed in section 5.2.1. According to Brunner, this change is attested in Kent and in some areas of the West and North (1963: § 13, note 7) and, in my data, it was also evident in some later Essex texts (#160 and #162). 1 The change must have spread from elsewhere, as the earliest texts in these areas have <o> as the dominant form in text file #67 or at least in direct competition with <a> in the case of text file #8. Since <o> is the dominant spelling in the more eastern texts from the West Midlands and in the texts from East Anglia, it is possible that trade from the North or West Midlands along the coast to London was responsible for the sudden appearance of this feature in the South-East towards the end of the 13th century and that the change radiated from there, which would explain the <au> spellings for OE a ˉ + w both in Kent and Essex. These trade routes correspond to those that were 1 The two main text files from Vices and Virtues (#64 and 65) also shows <a> spellings in this environment but, in these text files, <a> is the dominant spelling in all contexts. 7.1 Spatial diffusion in the three case studies 159 <?page no="178"?> established by Conde Silvestre and Hernández Campoy (2002: 171) for the Late Middle English period (cf. Map 1.3). Further support for potential diffusion around the coast comes from text file #131, which contains some English passages from the Cartulary of the Abbey of St Benet of Holme. This text is a documentary anchor text, whose location on the coast is not established in relation to other texts (Laing 2008 - : 88). In this text, there is only evidence from three lexical items ( HA : MSOCN , OWN , SOUL ), which all have <a>. Two of these items show exactly this change, but there is very little comparative evidence from other linguistic environments. Furthermore, the text is a copy of a pre-Conquest document and, therefore, it is difficult to tell whether the <a> spellings might simply reflect archaisms. However, Laing notes that “ it is plausibly local language ” (2008 - : 89), so there is no reason to dismiss the evidence completely, but one has to bear in mind that the conclusion remains very tentative. 7.1.3 Dual forms The dual forms have receded to two distinct relict areas in the South-West Midlands and the North-East Midlands and possibly a third area in Essex, as attested in the output of scribe A of Vices and Virtues (#64). However, since the dual is only attested in text file #64 but not in the output of scribe B in #65, it is possible that the language of scribe A is more archaic than that of scribe B and the relict area cannot be established as firmly as in the South-West and North- East Midlands. The archaic nature of the dual forms was one of the reasons why I chose to include this variable. In section 2.3 I claimed that this feature might be indicative of older language layers in certain texts. However, the fact that the attestation of the dual is limited to relatively distinct relict areas suggests that the feature may not have been as archaic in all areas as first believed. Furthermore, a comparison with <o> spellings for the change a ˉ > ǭ in the West Midlands reveals that the more southern texts that contain some dual forms regularly display the change a ˉ > ǭ . Even the texts attributed to the Worcester Tremulous Hand (#172 and #173), in which the dual forms are likely to belong to an older language layer, show an increase of <o> spellings. Furthermore, with the exception of the very early Ormulum #301 and The Will of Mantat #185 (<o> 25 %), all the texts from the East Midlands that contain some dual forms have <o> as the dominant variant. In other words, the occurrence of archaic features like dual forms are not necessarily indicative of archaic language in general and the same text languages might even be innovative with regard to other features. 160 7. Spatial Diffusion in Early Middle English <?page no="179"?> The most important finding concerning the spatial distribution of the dual forms is the insight that even a variable which has almost disappeared from the written sources may show spatial variation. It appears that in the East Midlands the oblique forms adopted initial / j/ by analogy to the oblique forms of the plural personal pronoun, as evident in forms like ȝ ing, ȝ ung, ȝ inc, ȝ inker, ȝ u ˉ ker (#64), gunker (#155), or ȝ unnc, ȝ unnker in passages of the Ormulum that are not included in LAEME. Although there are also oblique plural forms with initial / j/ in the later western texts, this variant is absent in the oblique forms of the dual in the South-West Midlands. Furthermore, such variants are not attested in any Old English written records. This corroborates the conclusion that before the dual forms were lost completely the forms with initial / j/ had diffused across the East Midlands, which in my opinion was only possible if the dual forms were more frequently used than the Middle English written records suggest. 7.1.4 General observations In the introduction I noted that very general claims about the diffusion of changes are common. Lass, for instance, notes that “ [a]lmost everything new begins in the north and percolates down through the east midlands ” (2006: 71). He continues that the South-West Midlands as well as the South-East “ remain the most conservative ” . Pilch ’ s comment is very similar: “ It has been noted for a long time that innovations in late Old English and Early Middle English spread from North to South ” (Pilch 1997: 450). Judging by the three case studies of this research project, I believe that such claims are not generally true for the Early Middle English period. On the one hand, there is just not enough early data from the North to observe the earliest appearance of any innovations. Futhermore, the network of texts in the East Midlands is simply too patchy to observe exactly how the innovations spread southwards through this area in the earliest written records. The observations made above on the southwards diffusion of the past vowel merger under the singular vowel were based on a few texts only. The change OE a ˉ > ME ǭ , on the other hand, is reported to spread from the South to the North (e. g. Brunner 1963: § 11.4; Luick 1964: § 369; Dietz 1989: 135). Judging from statements like the ones mentioned above, one gets the impression that it is exceptional if a change spreads northwards in the Middle English period. However, this change is a prominent example to illustrate that general statements on the conservative nature of the South-West do by no means apply to all features. With regard to the dual forms, the second relict area lies in the North-East Midlands, which is otherwise generally regarded as being more innovative. Furthermore, the South-West Midlands and Kent 7.1 Spatial diffusion in the three case studies 161 <?page no="180"?> were innovative in that the change ou > au for the context of OE a ˉ + w took place in these areas, but not elsewhere. Moreover, both in this process and even more so in the past grade vowel merger under the plural, a northwards diffusion is noticeable on the maps in the more southern parts of the East Midlands. The relict area of dual forms in the North-East Midlands also allows for the tentative conclusion that the complete replacement of dual forms by plural pronouns might have pushed northwards from the South; however, further research is necessary to fully support this claim. This evidence shows that for the Early Middle English period the origin and diffusion of innovations is highly variable-specific and that statements like the one by Pilch (1997: 450), which postulate that it is generally known that Early Middle English innovations spread from North to South, should be phrased more carefully. Both the studies by Conde Silvestre and Hernández Campoy (2002) and Bergs (2006) stress the main communication routes along which innovations diffuse in Late Middle English. As repeatedly mentioned, the network of texts is too loosely-knit to assess in detail, what role the roads and rivers might have played in Early Middle English. I have discussed the West Midlands, where the exact role that the Severn might have played cannot be established. Moreover, in the East Midlands I was not able to discern any distinct trade routes in the diffusion patterns. Bergs was able to track the diffusion of <th - > forms of the third plural personal pronoun in East Anglia and Kent and concluded that “ [t]here appear to be distinctive lines running to and from London northwards, southwards (the coast), and in a north-easterly direction, roughly where Norwich must be. ” (2006: 17). Both the maps of the past grade vowel mergers and of the change OE a ˉ > ME ǭ display such ‘ distinctive lines ’ from London northwards and towards Kent. However, a map that simply displays the locations of all LAEME text files shows these exact lines as well: There are simply no other text files on either side in the Early Middle English period and consequently it is impossible to tell whether the diffusion of the innovations was faster along these routes (cf. Map 3.1). On a more general level the trade route around the South coast might explain the close similarity between the South-West Midlands, on the one hand, and Kent and Essex, on the other hand, which was evident several times. The evidence from the Early Middle English period suggests that the trade route between London and Bristol might have played a role in the transmission of certain linguistic innovations, as they cannot be observed in the area inbetween. In the case study on the reduction of the past grade vowels of strong verbs, there were two texts which showed strong affinities to texts from other varieties in more than one class. Section 4.1 has revealed that Genesis and 162 7. Spatial Diffusion in Early Middle English <?page no="181"?> Exodus (#155) occasionally shows the western practice of leveling the past vowels under the vowel of the plural. At an early stage of the project, David Britain (p. c.) suggested that the close proximity of the medieval port in Lynn might have played a role in the diffusion of the western forms to this East Midland text. Furthermore, the data for the change OE a ˉ > ME ǭ shows that at the end of the 13th century the number of <o> spellings were generally higher around Lynn and the Wash than elsewhere in the East at the time and thus similar to the contemporary western data (cf. Map 5.1). Finally, the second relict area of dual forms lies exactly in this area too, and it is possible that the contact with the South-West Midlands, where the dual had apparently survived better, might have been influential here (cf. Map 6.5). To sum up, the concept of the trade routes is helpful for research on diffusion patterns in the Early Middle English period, but only on a very general level. The limited amount of written records that have survived and the tentative placement of some of the text files do not allow a more detailed discussion of the results. Nevertheless, these models from the later Middle English period give us some ideas about the possible routes along which the changes might have been transmitted. 7.2 Theoretical implications In the introduction I pointed out that one has to be careful about conclusions on diffusion based on Early Middle English data, as the fit-technique, which was applied in the localization of the texts, works under the assumption that innovations diffuse in a wave-like fashion and that this results in a dialect continuum into which the texts can be placed. I suggested that the question has to be turned around and that one should instead ask whether the diffusion as displayed on maps based on LAEME data can confirm the location of texts. On the basis of the three case-studies, this question can be answered affirmatively with one exception. The evidence from the change OE a ˉ > ME ǭ suggests that text file #151 has more southern traits than comparable text files from the same area and period. Furthermore, there are a few text files that are only very vaguely dated and that could be dated to more confined periods. For instance, based on their <o> spellings, I argue that text files #266 and possibly some of the unlocalized texts (cf. section 5.1.4) should be dated closer to the end of the 13th century. The other two variables of this study provided too little evidence for such claims; a temporal and spatial embedding is not possible if the network of the texts that provide data is too small. Nevertheless, although the exact location of most LAEME texts is not firmly established, it was possible to make, admittedly very general, claims 7.2 Theoretical implications 163 <?page no="182"?> about diffusion in Early Middle English. Most of the changes described diffuse along wave-like patterns or leap from one area to another along the trade routes described above and then show a wave-like pattern in the new area, as in the case of the past tense vowel mergers under the second grade in class V in the southern parts of the East Midlands, or the change ou > au in Kent and Essex. While the wave-like diffusion is noticeable on the maps presented in this study, it is not possible to observe diffusion patterns that involve the size or importance of towns. For instance, models like the ‘ urban hierarchical ’ (Trudgill 1974, 1983, 1986; Callary 1975, Gerritsen and Jansen 1980; Bailey et al. 1993; Wikle and Bailey 1997; Hernández Campoy 2003; Britain 2005 a) or ‘ contra-hierarchical ’ (Trudgill 1986; Bailey et al. 1993; Wikle and Bailey 1997) diffusion models (cf. section 1.3.1) are not suitable for the Early Middle English period because the exact origin of most text languages is not known, which has two implications: First, a distinction between rural and urban texts cannot be made and, second, the size of the town where a text was written cannot be discerned. Astill comments on research in the urban history of Britain in the period between 600 and 1300 that “ [f]or most of the period under review the data allow a qualitative, rather than a quantitative, approach to towns and, therefore, important issues such as the relative size of towns can be addressed only in an oblique fashion ” (2000: 27). This means that even if the exact location of a LAEME text is known, this does not necessarily apply to the size of the town from which the text originates. Furthermore, it is possible that the size, importance, and order of a monastery is more influential, but again, very often we do not have access to these pieces of information (cf. discussion below). The concept of ‘ cultural hearth ’ diffusion, on the other hand, seems more applicable to the Early Middle English context. This model describes the diffusion of an innovation across one particular region before it appears in another, disregarding the size or importance of a town (Horvath and Horvath 1997, 2001, 2002). Furthermore, the exact location of the texts is not important if one applies this model to the Early Middle English period and it disregards the gap of texts between the West and East as described by Laing (2000). This type of spatial diffusion is observable in those innovations which first diffused through the South-West Midlands and then appeared in the South-East, but not in-between, as for instance with regard to ou > au as described above. Furthermore, this model can also account for the diffusion of / j/ initial oblique dual forms in the East, which only diffused through this area but disappeared before such forms could spread to other areas. Since spatial diffusion can only be observed and described on a more general level, it is important that a description of the diffusion pattern in Early Middle English is enhanced by a discussion of questions like those proposed by 164 7. Spatial Diffusion in Early Middle English <?page no="183"?> Britain (2005 b: 997). Britain ’ s revised model redresses the concerns about purely spatial models by removing the focus from the belief that innovations simply replace the traditional form and by taking a closer look at what exactly happens to the variables through dialect contact: The success of innovations appears to be contingent on the social, interactive, and structural compatibility of the new form with the old one. If that compatibility is not present, diffusion may be slowed down, succeed only partially, or in some cases, be rejected entirely. (Britain 2005 b: 996) In this section I take Britain ’ s model as a basis for the discussion as to whether such questions as Britain proposes are also useful for historical data. Britain ’ s first point addresses the issue of whether an innovation is “ linguistically identical in its new resting place as it was where it began, with the same social and stylistic status ” (2005 b: 996). With regard to the decline of the dual forms it is possible that their stylistic status enabled them to survive longer in some texts. In section 6.1 I note that dual forms feature prominently in dialogues in which exactly two individuals argue, and it is also attested in several passages that deal with marriage or sexual intercourse where the dual nature of the cause is semantically charged. Therefore, there is reason to believe that the dual forms survived longer as a stylistic device to emphasise the duality in these contexts. Social status, on the other hand, cannot be included in a study on Early Middle English, as very little is known about the scribes of the Early Middle English texts. Laing and Lass note that “ we are dealing with the output of literate, adult, English Catholics largely inhabiting closed institutions (monasteries, professions) where class stratification would be of little importance ” (2006: 419). They add, however, that there might be other variables than could be studied instead of the social variables that are typically included in modern sociolinguistic studies. Laing and Lass suggest that, for instance, text type, monastic order, or a certain scriptorium might play a role in the transmission of a change. 2 2 Lenker (2000), for instance, has successfully applied network theory to the monasteries of the Benedictine Reform and the school of Æthelwold. 7.2 Theoretical implications 165 <?page no="184"?> Figure 7.1: Diachronic distribution of prose and verse texts that contain OE a ˉ . In the present study I did not analyze the variables text type or monastic order for the following reasons. First of all, as the LAEME CTT only contains 167 texts, it is not possible to get a balanced sample of any text type across any subperiod or area, which would be essential. Figure 7.1 includes the 147 texts that contain the variable OE a ˉ and illustrates the distribution of prose and verse texts over the course of the Early Middle English period. The figure shows that the number of verse texts that contain the variable increases substantially over times. This has the implication that, if one simply took into consideration prose vs. verse texts, one would get the result that the verse texts are more advanced in the promotion of the change OE a ˉ > ME ǭ , whereas in fact this is simply a matter of the diachronic distribution of the two text types. With regard to genre there is the same problem. In her diachronic study on word formation processes Gardner (forthcoming) distinguishes several genres, but she is only able to establish “ religious texts ” as a suitable category for both a diachronic and diatopic comparison in the Early Middle English period (based on LAEME) and even there she observes the same problems concerning the unbalanced distribution of the texts. Moreover, it is not useful to include monastic order or a certain scriptorium into the equation. To illustrate this I take again the example of those texts that provide evidence for OE a ˉ > ME ǭ . Among the 147 text files, there are only 29 that probably have an affiliation with a religious house. The following orders are represented among these texts (cf. Laing 2008 - ): Arroaisian (maybe #301), Augustinian (e.g. #146, #291), Benedictines (e. g., #131, #143, #184), Carthu- 166 7. Spatial Diffusion in Early Middle English <?page no="185"?> sians (#163), Cistercians (e. g. #129, #304), Franciscans (maybe #137, #178 - #181). This leaves only a few texts per order, in most cases only one each per place and period. Those text files that are comparable either spatially or temporally show a very patchy picture. For instance, the five text files that might have a Franciscan affiliation are all dated to the last quarter of the 13th century and text files #178 - #181 all appear in a manuscript that might have been produced by the Oxford Franciscans (Laing 2008 - : 148 - 150); however, their language is very different: Text files #137 and #178 contain <o> as the dominant form, #181 has <o> as the only form, #180 shows a mixture of <o> and <a> spellings, and #179, a northern type of text, has <a> as the dominant form. So for this set of comparable data the variable religious house does not seem to be influential. The other three questions which Britain proposes are more useful for the Early Middle English period, as they can be discussed on the basis of language internal evidence. His second question looks into the linguistic constraints that play a role in the distribution of the variants of a variable (Britain 2005 b: 997). In my case studies this was most prominent in the change OE a ˉ > ME ǭ , which was inhibited in the North-West Midlands in diphthongs with voiced labio-velar approximants. The same phenomenon is noticeable in Kent, but it seems that the change OE a ˉ > ME ǭ was categorical at first and that it was later reversed again in this environment. It is difficult to tell whether there were also constraints in the past vowel mergers; there is just too little evidence in the Early Middle English period. In any case, the evidence suggests that some environments favor the mergers, which leads to the third question; i. e., whether the traditional form is compatible with the innovation (Britain 2005 b: 997). In his own study on the diffusion of four consonantal and five vocalic variables Britain concludes “ [i]f the dialect receiving the innovation does not have a similar phonetic form to the innovative form in the donor dialect, it is less likely to accept the change or will do so more slowly ” (2005 b: 1018). Since the past grade vowels of class V and IVa were already very similar in Old English and only distinguished by vowel quantity, it is no surprise that the mergers are more prominent in these classes. This confirms the findings by Branchaw, who, in a quantitative study on the shift of strong verbs to the weak declension, concluded that “ the single most important factor in preserving the strong inflection of a verb was the perceptual ease of distinguishing the vowels of the principal parts ” (2010: 101). Branchaw noticed that those verbs that had less distinct grade vowels were more likely to shift, and the evidence from LAEME suggests that the same is true for the reduction of past grade vowels. Finally, Britain proposes that one should also address the outcome of the change: “ What are the linguistic consequences of the contest between the 7.2 Theoretical implications 167 <?page no="186"?> innovation and the traditional form? Victory for the innovation? Survival of the traditional form? Or a structural compromise between the two? ” (2005 b: 997). With regard to the change OE a ˉ > ME ǭ the innovation eventually replaced the traditional form in the South; however there was a clear ‘ victory ’ for the traditional form in the North, where the diffusion of the change was stopped (e. g. Ekwall 1938; Brunner 1963: § 11.4; Kristensson 1965; Jordan 1974: § 6; Dietz 1978). With regard to the past grade vowel mergers, it seems that a structural compromise was finally reached. Long (1944: 263) concluded, in her study on Late Middle English data, that the merger under the 1st past grade vowel was more likely to take place in classes I and III, whereas in classes II and IV the past participle vowel is generally expanded to the past tense forms. And while the dual forms were clearly replaced by the plural forms of the personal and possessive pronoun, it seems that a structural compromise took place for this variable too. Before the forms were lost completely, they were enhanced with a lexical quantifier like ‘ both ’ or ‘ other ’ in order to make the dual reference explicit. This seems to be a wide-spread phenomenon, as it did not only take place in English but in other languages that knew dual forms, too (Guðmundsson 1972: 95). This practice is still employed today to refer to exactly two individuals. I believe that such issues as raised by Britain are essential in the discussion of diffusion and change in varieties which only offer a limited amount of material useful for research. They allow a more detailed discussion of the diffusion of the innovations and do not simply address the spatial dimension, which in many texts is highly uncertain in the Early Middle English period. Britain ’ s questions that are concerned with language internal factors are more suitable for the Early Middle English period, as information on external factors associated with the text files is often missing. I propose one addition to Britain ’ s model for the Early Middle English context. Several of the forms that were discussed in chapters 4 to 6 are attested in pre-Conquest documentary texts. Although these are only included in LAEME if the language is transferred to a contemporary variety of English, they might preserve some traditional forms that have already been replaced in contemporary texts. The same applies to texts that are copied from earlier Middle English versions. Therefore, the questions proposed by Britain (2005 b) should be supplemented by the following in any study of spatial diffusion concerned with the Early Middle English period: Is the traditional form in any area mostly attested in texts that go back to earlier versions of the texts or even of pre-Conquest documents? Is the linguistic variable the same in different versions of the same text? 168 7. Spatial Diffusion in Early Middle English <?page no="187"?> For instance, there is evidence that several of the dual forms discussed in chapter 6 are already attested in earlier versions of the texts. This piece of information is essential for the interpretation of the results and this additional issue should be addressed in any discussion of LAEME material. 7.3 Methodological implications Because of the limitations of the Early Middle English material as described in section 3.2.4 I chose to display the results in symbol maps. The addition of the dimension of time proved to be very effective, as it allows the observation of changes not only diatopically but also diachronically. However, only by adding the dimension of the relative frequency of a feature as well was it possible to show the fading out of the diffusion of a feature, as shown in Map 7.1 (a) and Map 7.1 (b) displaying the past grade vowel mergers of class V or to highlight transition zones between two variants of a variable as evident in the change OE a ˉ > ME ǭ in the South-West Midlands in Map 7.2. A disadvantage of including all these dimensions is that it is difficult to read the maps in areas that contain a denser network of texts, as is the case in the South-West Midlands. The amount of information that is encoded in a map like Map 7.2 may be overwhelming and other ways of display appear to be more suitable for this area. For this reason I decided to display the results from this area also with a set of six period specific maps, which are much easier to read than a map that just displays all evidence at once. Among the three case-studies, the variable OE a ˉ > ME ǭ has clearly yielded more evidence than the other two variables, as it is independent of particular contexts such as terms of direct address, in the case of dual, or past tense, in the case of the past tense mergers. Nevertheless, I do not believe that one should only focus on high-frequency variables, as all the three case studies based on the LAEME CTT have shown spatial variation. Nevertheless, there is one aspect that has to be borne in mind with regard to low-frequency variables. The case study on the dual has shown that the samples of the longer texts that are included in LAEME are maybe not sufficient if such a feature is analyzed. If a variable is context specific like the dual forms, it is necessary to include the entire texts as valuable data might be lost. For instance, there are no 2nd person dual forms in the LAEME sample of the Ormulum (#301), but Kennedy (1915: 42 - 43) lists a total of 51 forms. Nevertheless, even for such a lowfrequency variable it is useful to access the LAEME CTT first because its innovative tagging system allows the retrieval of variant forms for any variable, which can then be used to search the electronic version of an untagged corpus such as the Corpus of Middle English Prose and Verse. 7.3 Methodological implications 169 <?page no="188"?> 8. Conclusion and Summary In this study I have shown that, although the data retrieved from LAEME is limited with regard to temporal and spatial coverage, it is possible to discern specific patterns in the diffusion of variants. In this final chapter, I present a summary of the most important results of the individual case studies and highlight those findings that were only gained through a study of new data from LAEME. 8.1 Summary of the individual findings from the three case studies Spatial diffusion of innovations was apparent for all three variables. The findings with regard to spatial diffusion are summarized and discussed in detail in chapter 7. I have shown that, with regard to the past grade vowel merger of strong verbs this was most evident in the East Midlands, where leveling under the singular vowel diffused from north to south and the opposite process, namely leveling under the plural vowel, diffused from Kent northwards (cf. chapter 4). In the change from OE a ˉ to ME ǭ a general diffusion from south to north was evident both in the West and East Midlands, but the data from the West Midlands clearly allow the most detailed discussion (cf. chapter 5). Finally, the study of the dual forms of the personal pronoun revealed some forms that must have spread in the East Midlands and Kent. The diffusion of the oblique forms of the 2nd person dual forms with initial / j/ must have fallen into the transition period between Old and Middle English, as they are not attested in any pre-Conquest material and the origin of the change cannot be deduced from the Middle English data. 8.1.1 Reduction of the number of grades in strong verbs In this study I analyzed the past grade vowels of verbs from classes I, II, IIIb, IIIc, IVa, and V, which have the most distinct past grade vowels in the Middle English period. Apart from classes II, IVa, and V the grade system of strong verbs is still relatively stable in the Early Middle English period. For many classes the text files show the expected past grade vowels in the 1st and 3rd singular form, on the one hand, and in the 2nd singular and plural forms, on the other hand (cf. Appendix I). <?page no="189"?> Both Brunner (1963: § 69.1) and Brook (1963: 72) mention the leveling of the past grade vowels under the plural vowel, which means that 1st and 3rd singular forms of the past tense show the same grade vowel as the 2nd singular and the plural forms. In chapter 4, I showed that the apparent analogical extension of the plural vowel to the singular has to be attributed to a phonological variant of the 1st past grade in the West Midlands and Kent. Both classes IVa and V show a majority of <e> spellings in this grade vowel, and the spatial distribution of these forms correspond to the areal distribution of OE e (< Gmc *a) (cf. Brunner 1963: § 11, note 2; Brunner 1965: § 19.3; Jordan 1974: § 32). This had the implication that the past grade vowels in classes IVa and V were represented by the same grapheme. This similarity in vowel length plus the large number of verbs of class V must have also triggered the merger under the plural vowel in class II, which cannot be explained through phonological changes. Furthermore, as discussed in detail in section 7.1.1, in class V the extension of the plural past vowel to the 1st and 3rd singular is also evident in verbs from the East. Forms with this grade vowel must have spread northwards from Kent, where OE e (< Gmc *a) is generally attested. That verbs of classes IV and V take the lead in a change is not a new phenomenon in the Early Middle English period. The verbs from these classes were the most unstable of the first five classes already at an earlier stage in the history of strong verbs; in the first recorded Germanic languages they already show morphologically unexpected grade vowels in the 2nd past and past participle grade of class V and the 2nd past grade of class IV (cf. section 2.1.1). The data from LAEME confirms the leveling of the past grade vowels under the singular past vowel, which started in the North (Long 1944: 263; Brunner 1963: § 69.1; Brook 1963: 72; Nielsen 1985: 45; Iglesias-Rábade 2003: § 197). The phenomenon is present in all 14th century texts from the North that contain strong verbs, and it is attested in all classes which provide any data from this area (text files #169, #188, #295 - #298, cf. Appendix I). Both Berndt (1956: 267) and Brunner (1963: § 69.1) note that the 2nd singular past forms are earlier in adopting the singular grade vowel. In section 4.2 I have pointed out that in the LAEME data, there are vague tendencies that this was indeed the case but, as there are too few 2nd singular past forms in the classes I analyzed, I cannot draw any firmer conclusions. In areas other than the North, the leveling is only evident in some classes of strong verbs. In the East Midlands, the merger spreads southwards only in classes IVa and V and is evident in the more northern text files of the East. There it is already attested in some 12th century text files, which is earlier than expected (cf. Lass 1994 b: 88; Brunner 1963: § 69.1). The merger is also apparent in the Southern parts of the East Midlands at the beginning of the Early Middle English period, but most likely this has to be attributed to a 8.1 Summary of the individual findings from the three case studies 171 <?page no="190"?> phonological change too: OE æ ˉ , which is the grade vowel of the plural in classes IVa and V in Old English, became a ˉ in these areas (Brunner 1963: § 11.2), so the two grade vowels were both represented by <a>. The merger under the singular vowel is also attested in class II in some texts from the third quarter of the 13th century; however, unlike in class IVa and V, these texts are not localized in the East, but in the West Midlands. These texts do not correspond to those that attest the phenomenon with regard to leveling under the plural vowel and it seems that there is just a general tendency in this area that the four grades in class II were reduced to three, but with no regular pattern. Finally, in sections 4.3 and 4.4 I discussed verbs of Scandinavian origin and verbs that historically belong to the weak conjugation. The Scandinavian verbs only yielded very little evidence from the classes with distinct past grade vowels, but those that are attested correspond to the native verbs of the same class. The number of weak verbs that shifted to strong is very small and most of the verbs that were annotated with the tag for a non-weak past form in LAEME turned out to be weak after all. These are verbs whose stem ends in -t, which had conflated with the dental suffix of the weak past inflection. In some verb forms, a vowel change was evident in one of the past grades and these always corresponded to a strong verb of a similar syllable structure. 8.1.2 OE a ˉ > ME ǭ The change OE a ˉ > ME ǭ is evident in the South from the earliest Middle English written records onwards and is also attested in the more southern texts from the West and East Midlands. This distribution of early <o> spellings across a relatively large area suggests that the change was already well under way at the beginning of the Early Middle English period, which confirms Liebl ’ s (2002, 2006) conclusion that the change must have started in the Late Old English period. In general, the data indicates that the factors ‘ word class ’ and ‘ frequency ’ do not play a major role in the diffusion of the change; however, there are a few linguistic environments which inhibit or even reverse it. The present study has shown that in the North and West Midlands the <a> spellings are found further to the south before w and ɣ , which supports the findings of Dietz (1989: 142) and Liebl (2002, 2008). Before w the higher frequencies of <a> spellings are also evident in Essex and Kent, where the diphthong ou in the later texts regularly changed to au (cf. Brunner 1963: § 13, note 7). In general, shortening took place before non-lengthening consonant clusters (cf. section 2.2.1.5). However, the data revealed that, in Kent, this 172 8. Conclusion and Summary <?page no="191"?> shortening apparently did not take place before the cluster <-tt->, as lexical items which historically contained these clusters, such as contracted forms of ( BE ) HATAN or comparative forms of HOT regularly changed to <o>. The same is apparent in text files #2 and #1100, which contain two separate versions of The Owl and the Nightingale. This evidence provides further support for the South- Eastern origin of the poem. In general, the shortening did not take place in GHOST and GHOSTLY either, and these items show the regular <a> and <o> distribution in most areas. On the one hand, both Brunner (1963: § 9) and Jordan (1974: § 23) note a difference between uninflected and inflected forms of items with -st clusters, but this was not apparent in the LAEME data. On the other hand, the data from LAEME reveals that the shortening did take place in East Anglia, as the two items show higher frequencies of <a> spellings in these areas. In section 5.4 I was able to show that, based on the Early Middle English evidence, the <oa> spellings might not only represent / ɔː / , as previously suggested (cf. Umpfenbach 1935: 102; Moore 1969: 76; Jordan 1974: 73; Lass 1997: 64, footnote 26). The data from LAEME permits the conclusion that the grapheme might represent a vowel closer to / ɑː / or / ɒː / in some texts. This case-study has confirmed some of the locations of texts. For instance, text files #246 - #249 contain different text languages of MS Cambridge, Trinity College B.14.39 (323). The manuscript only received one localization in LALME (LP 7721) (cf. Laing 2008 - : 21 - 29), but the evidence from the variable a ˉ > ǭ indicates that the location of text file #247 more to the south-west of the other text files is the right decision of the compilers of LAEME and can be confirmed by respective frequencies of <o> spellings. There is only one text for which there is strong evidence that the location should be reconsidered. Text file #151, which is located to eastern Lancashire, shows <o> forms in all nine forms that include OE a ˉ . Since categorical <o> spellings in a text of this period and location is very unusual, I assume that the text is more southern than the present LAEME location. Some other text files like #14 or #265 - #267 have only received a vague dating in LAEME and the evidence from this change suggests a dating to more narrow periods. Finally, the data from this case study supports Liebl ’ s (2002: 193) claim that AB language is not as archaic as believed with regard to <a> spellings in contexts of OE a ˉ (cf. Smith 1992: 583). The diffusion of the change as evident in the West Midlands shows that the change had not reached the area to which these texts are localized at the time of their writing. 8.1 Summary of the individual findings from the three case studies 173 <?page no="192"?> 8.1.3 The decline of the dual Dual forms are found only in 21 LAEME text files that are located in two relict areas in the South-West Midlands, in the North-East Midlands and in one location in Essex. In section 2.3.3 I stated that the dual forms are only briefly mentioned in the Middle English handbooks and grammars and that the entries sometimes even suggest that not all forms are attested (e. g. Mossé 1953; Brunner 1963). However, I was able to show that there are three texts which preserve the entire paradigm: the part of hand B of the A version of La ȝ amon ’ s Brut (#278), the Ormulum (#301) and the part of hand A of Vices and Virtues (#64). Howe (1996) suggests that the forms have survived better in the East Midlands, but the data from LAEME clearly does not support such a claim and suggests that the forms are generally better attested in the South-West Midlands. So far, the oblique forms of the 2nd person dual personal pronoun with initial / j/ have been attributed to the scribes ’ lack of familiarity with the forms. The fact that the forms are only found in the East Midlands allows for the conclusion that they should be regarded as a regional variant. These forms can either be explained by analogy to the nominative of the 2nd person dual git or by analogy to the 2nd person plural forms (cf. DOE git; Howe 1996: 137 - 138). In my opinion it is most likely that analogy to the oblique forms of the 2nd person plural pronouns played the more important role, as this explanation accounts both for initial / j/ and the appearance of / u/ in forms like ȝ ung (#64) or gunker (#155). The form gunc is also attested in MS Oxford, Bodleian Library, Hatton 38, which contains a Kentish version of the West-Saxon Gospels that is not part of LAEME, although it is dated to the Middle English period (cf. Liuzza 1994: xxxiii - xxxvi). An editorial comment to the form Hunke (#2) in the Owl and the Nightingale speculates whether this form might be a scribal error for * ȝ unke. As already mentioned above, the poem has a possible origin in the South-East, which corresponds exactly to the spatial distribution of such forms and supports this comment. 8.2 Spatial diffusion in Middle English The linguistic maps presented in this study have mostly revealed diffusion in a wave-like fashion. In the three case studies this was best observable in the diffusion of past grade vowel mergers of class Vand in the northwards diffusion of ME ǫ in the South-West Midlands. The discussion of the results has shown that the descriptions have to remain on a very general level due to the limited extent of the surviving material from the period. Some of the LAEME texts are 174 8. Conclusion and Summary <?page no="193"?> only localized relative to each other and, therefore, a more detailed analysis of exact patterns is not possible. Nevertheless, I was able to show that trade along the coast might have played a major role in the diffusion of innovations from one area to others, as several Middle English innovations are attested in Kent and in the West Midlands but do not appear elsewhere. Since it is not possible to detect more distinct patterns, I have argued in section 7.2 that any study of diffusion in the Early Middle English period should be complemented by questions such as those proposed by Britain (2005 b: 997) because they not only focus on the spatial aspect but also address the linguistic consequences of the diffusion of innovations. Britain grounds his discussion of these issues on a quantitative analysis of several linguistic contexts and several social factors. If LAEME data is included in a study, the emphasis should lie on a detailed qualitative analysis of the results, as many smaller texts do not provide enough data to distinguish several linguistic contexts. If a quantitative study of several linguistic contexts is envisaged, it might be wise to include only the more substantial text files from LAEME. Most LAEME text files were written by anonymous scribes; Dan Michel (#291) and Orm (#301), who identify themselves as the authors of the work they present, are the exceptions. For many of the texts in LAEME, we do not have any access to extra-linguistic factors and even style or genre is possibly problematic, as the corpus of surviving texts is very unbalanced in this respect (cf. section 7.2). Nevalainen and Raumolin-Brunberg (2003) stress the importance of including period specific social variables but, for the Early Middle English period, these are simply not accessible in most cases. Some of the texts can be linked with religious institutions, but the entire corpus is also unbalanced with regard to this aspect. For this reason, when discussing diffusion and language change in Early Middle English, the focus has to rely on questions that include language internal factors only. Nevertheless, there is one period specific factor that has to be considered in any study of diffusion in Early Middle English. Many of the extant medieval texts have survived as copies of older texts and, for this reason, I believe that it is necessary to augment Britain ’ s model with questions that take this into account. A linguistic variable that is attested in such a text should not be studied in its own right, but in the light of earlier versions of the same text and thus its possible archaic nature (cf. section 7.2). The discussion of such issues is essential as it helps to detect older language layers, which might influence the discussion of the findings from a certain text. Furthermore, if a text exists in more than one copy, the absence or inclusion of a variable in one text might reveal the origin of a variable. 8.2 Spatial diffusion in Middle English 175 <?page no="194"?> 8.3 Outlook The three case-studies have revealed that LAEME sheds new light on already well-studied variables. For instance, all of the following are remarkable findings that were only discovered through the analysis of LAEME data: First, leveling under the plural vowel in class V is evident in text files from the southern parts of the East Midlands. Second, shortening before consonant clusters did not take place before <-tt-> clusters in Kent. Third, oblique forms of the 2nd person dual pronoun with initial / j/ are typical variants from the East Midlands and Kent. It is assumed that future case studies on any features will yield similarly unexpected, new insights, which increases the demand for revised and refined entries in Middle English handbooks and grammars. In general, only the findings from the case-study on the change OE a ˉ > ME ǭ are substantial enough to confirm or question the location or dating of texts. In most cases the findings supported the localizations of LAEME, with the exception of text file #151 mentioned above. More research into the language of this text is necessary to evaluate the exact nature of the language of this text file. Moreover, the dating of a few texts like text file #14 or #265 - #267 could be narrowed down to a more confined period based on comparative evidence from the same area. Again, further research from other variables is necessary to base these claims on firmer ground. The maps as presented in this study have proved to be a useful tool in the visual representation of spatial diffusion. I was able to show that, even with the mentioned limitations of Early Middle English data, it is possible to visualize linguistic change across time and space if the dating of a text and the frequency of the variants are included in the linguistic maps. At the time of research, it was necessary to produce such maps by means of a PHP-script, as the mapping section of LAEME only included maps of pre-selected features that do not display the time dimension. A “ Create/ Edit a Feature Map ” tool is now available in the mapping section of LAEME 3.2. This certainly facilitates the creation of user-defined maps and, consequently, the comparison of a wider range of Early Middle English linguistic features. In the present study I offer a set of plausible hypotheses on spatial diffusion patterns in Early Middle English, which hopefully spark an interest to explore this fascinating topic further, as a very important question remains: How far are the findings of this study generalizable? Future research is needed to assess whether similar patterns emerge when other variables are studied and whether the findings also hint at the importance of waterways around the coast in the transmission of a change. It is therefore hoped that more research on LAEME data is conducted in a similar manner in order to support or challenge the findings of the present exploratory study. 176 8. Conclusion and Summary <?page no="195"?> 9. Bibliography Anderson, John and Derek Britton. 1997. “ Double Trouble: Geminate Versus Simplex Graphs in the Ormulum. ” In Jacek Fisiak (ed.). Studies in Middle English Linguistics. (Trends in Linguistics: Studies and Monographs 103). Berlin, New York: Mouton de Gruyter, 23 - 58. Anderson, John and Derek Britton. 1999. “ The Orthography and Phonology of the Ormulum. ” English Language and Linguistics 3: 299 - 334. Anderwald, Lieselotte. 2009. The Morphology of English Dialects: Verb Formation in Non- Standard English. Cambridge: University Press. Arngart, Olof. 1949. “ Middle English Dialects. ” Studier i modern språkvetenskap 17: 17 - 29. Arngart, Olof. 1968. The Middle English Genesis and Exodus. Re-edited from MS. C. C. C. C. 444 with Introduction, Notes and Glossary. (Lund Studies in English 36). Lund: Gleerup. Arngart, Olof. 1978. The Proverbs of Alfred: An Emended Text. Lund: Gleerup. Astill, Grenville. 2000. “ General Survey 600 - 1300. ” In D. M. Palliser (ed.). The Cambridge Urban History of Britain. Volume 1, 600 - 1540. Cambridge: University Press, 27 - 49. Bahnick, Karen R. 1973, The Determination of Stages in the Historical Development of the Germanic Languages by Morphological Criteria: An Evaluation. The Hague, Paris: Mouton de Gruyter. Bailey, Guy and Tom Wikle, Jan Tillery and Lori Sand. 1993. “ Some Patterns of Linguistic Change. ” Language Variation and Change 5: 359 - 390. Bammesberger, Alfred. 1984. A Sketch of Diachronic English Morphology. (Eichstätter Materialien. Abteilung Sprache und Literatur 7). Regensburg: Pusted. Bammesberger, Alfred. 1986. Der Aufbau des germanischen Verbalsystems. (Untersuchungen zur vergleichenden Grammatik der germanischen Sprachen 1). Heidelberg: Winter. Bammesberger, Alfred. 1996. “ The Preterite of Germanic Strong Verbs in Classes Four and Five. ” NOWELE 27: 33 - 43. Bammesberger, Alfred. 2006. „ Das Futhark und seine Weiterentwicklung in der anglofriesischen Überlieferung. ” In Alfred Bammesberger and Gaby Waxenberger (eds.). Das Futhark und seine einzelsprachlichen Weiterentwicklungen: Akten der Tagung in Eichstätt vom 20. bis 24. Juli 2003. (Ergänzungsbände zum Reallexikon der germanischen Altertumskunde 51). Berlin: de Gruyter. Bauer, Gero. 1986. “ Medieval English Scribal Practice: Some Questions and Some Assumptions. ” In Dieter Kastovsky and Aleksander Szwedek (eds.). Linguistics Across Historical and Geographical Boundaries: In Honour of Jacek Fisiak on the Occasion of His Fiftieth Birthday. Vol. 1: Linguistic Theory and Historical Linguistics. (Trends in Linguistics: Studies and Monographs 32). Berlin, New York: Mouton de Gruyter, 199 - 210. Baugh, Albert C. and Thomas Cable. 1993. A History of the English Language. 4th ed. London: Routledge. Bech, Gunnar. 1969. Das germanische reduplizierte Präteritum. (Historisk-filosofiske meddelelser 44: 1). Kopenhagen: Komm. Munksgaard. Benskin, Michael and Margaret Laing. 1981. “ Translations and Mischsprachen in Middle English Manuscripts. ” In Michael Benskin and M. L. Samuels (eds.). So Meny People <?page no="196"?> Longages and Tonges: Philological Essays in Scots and Mediaeval English Presented to Angus McIntosh. Edinburgh: Benskin and Samuels, 55 - 106. Benskin, Michael, Margaret Laing, Vasilis Karaiskos and Keith Williamson. 2013 - . An Electronic Version of A Linguistic Atlas of Late Mediaeval English. The University of Edinburgh. <http: / / www.lel.ed.ac.uk/ ihd/ elalme/ elalme.html> Benskin, Michael. 1981. “ The Middle English Dialect Atlas. ” In Michael Benskin and M. L. Samuels (eds.). So Meny People Longages and Tonges: Philological Essays in Scots and Mediaeval English Presented to Angus McIntosh. Edinburgh: Benskin and Samuels, xxvii - xli. Benskin, Michael. 1991. “ The ‘ Fit ’ -technique Explained. ” In Felicity Riddy (ed.) Regionalism in Late Medieval Manuscripts and Texts. Cambridge: D. S. Brewer, 9 - 26. Bergs, Alexander. 2005. Social Networks and Historical Sociolinguistics: Studies in Morphosyntactic Variation in the Paston Letters (1421 - 1503). (Topics in English Linguistics 51). Berlin, New York: Mouton de Gruyter. Bergs, Alexander. 2006. “ Spreading the Word: Patterns of Diffusion in Historical Dialectology ” . In Markku Filppula, Juhani Klemola, Marjatta Palander, Esa Penttilä (eds). Topics in Dialectal Variation. Joensuu: Joensuu University Press, 5 - 30. Berndt, Rolf. 1956. Form und Funktion des Verbums im nördlichen Spätaltenglischen: Eine Untersuchung der grammatischen Formen und ihrer syntaktischen Beziehungsbedeutungen in der grossen sprachlichen Umbruchsperiode. Halle: Max Niemeyer. Biber, Douglas, Stig Johansson, Geoffrey Leech, Susan Conrad and Edward Finegan. 2000. Longman Grammar of Spoken and Written English. Harlow: Longman/ Pearson Education Björkman, Erik. 1900 - 1902. Scandinavian Loan-Words in Middle English. 2 vols. Halle: Max Niemeyer. Blake, Norman. 1994. “ Premises and Periods in a History of English. ” In Francisco Fernández, Miguel Duster and Joan José Calvo (eds.). English Historical Linguistics 1992: Papers from the 7th International Conference on English Historical Linguistics. (Current Issues in Linguistic Theory 113). Amsterdam, Philadelphia: John Benjamins, 37 - 46. Branchaw, Sherrylyn. 2010. “ Survival of the Strongest: Strong Verb Inflection from Old English to Modern. ” In Robert A. Cloutier, Anne Marie Hamilton-Brehm, William A. Kretzschmar (eds.). Studies in the History of the English Language V: Variation and Change in English Grammar and Lexicon: Contemporary Approaches. Berlin, New York: Mouton de Gruyter, 87 - 104. Braune, Wilhelm and Frank Heidermanns. 2004. Gotische Grammatik. 20th, rev. ed. Tübingen: Max Niemeyer Verlag. Braune, Wilhelm and Ingo Reiffenstein. 2004. Althochdeutsche Grammatik I: Laut- und Formenlehre. 15th ed. Tübingen: Max Niemeyer Verlag. Brinton, Laurel J. and Leslie K. Arnovick. 2011. The English Language: A Linguistic History. 2nd ed. Oxford: University Press. Britain, David. 2002. “ Space and Spatial diffusion. ” In J. K. Chambers, Peter Trudgill and Nathalie Schilling-Estes (eds.). Handbook of Variation and Change. Oxford: Blackwell, 603 - 637. Britain, David. 2005 a. “ Geolinguistics - Diffusion of Language. ” In Ulrich Ammon, Norbert Dittmar, Klaus J. Mattheier and Peter Trudgill (eds.). Sociolinguistics: An International Handbook of the Science of Language and Society. Berlin: Mouton de Gruyter, 34 - 48. 178 9. Bibliography <?page no="197"?> Britain, David. 2005 b. “ Innovation Diffusion: ‘ Estuary English ’ and Local Dialect Differentiation: The Survival of Fenland English. ” Linguistics 43: 995 - 1022. Britain, David. 2012. “ Innovation Diffusion in Sociohistorical Linguistics. ” In Juan Manuel Hernández-Campoy and Juan Camilo Conde-Silvestre (eds.). The Handbook of Historical Sociolinguistics. Chichester: Wiley-Blackwell, 451 - 464. Brook, G. L. 1958. A History of the English Language. London: Andre Deutsch Limited. Brook, G. L. 1963. English Dialects. London: Andre Deutsch Limited. Brunner, Karl. 1960 - 1962. Die englische Sprache: Ihre geschichtliche Entwicklung. 2 vols. 2nd rev. ed. Tübingen: Niemeyer. Brunner, Karl. 1963. An Outline of Middle English Grammar. 5th ed. translated by Grahame Johnston. Cambridge, MA: Harvard University Press. Brunner, Karl. 1965. Altenglische Grammatik: Nach der angelsächsischen Grammatik von Eduard Sievers. 3rd, rev. ed. Tübingen: Niemeyer. Bryan, William Frank. 1915. Studies in the Dialects of the Kentish Charters of the Old English Period. Menasha, WI: George Banta Publishing: Bülbring, Karl. 1889. Geschichte des Ablauts der starken Zeitwörter innerhalb des Südenglischen. Strassburg: Verlag Karl J. Trübner. Bülbring, Karl. 1902. Altenglisches Elementarbuch. Heidelberg: Carl Winter. Burchfield, R. W. 1956. “ The Language and Orthography of the Ormulum MS. ” Transactions of the Philological Society 55: 56 - 87. Bybee, Joan L. and Dan I. Slobin. 1982. “ Rules and Schemas in the Development and Use of the English Past Tense. ” Language 58: 265 - 289. Callary, Robert. 1975. “ Phonological Change and the Development of an Urban Dialect in Illinois. ” Language in Society 4: 155 - 170. Campbell, Alistair. 1959. Old English Grammar. Oxford: Clarendon Press. Campbell, Bruce M. S., James A. Galloway, Derek Keene and Margaret Murphy. 1993. A Medieval Capital and its Grain Supply: Agrarian Production and Distribution in the London Region c. 1300. (Historical Geography Research Series 30). Cheltenham: Historical Geography Research Group. Cartlidge, Neil. 1998. “ The Linguistic Evidence for the Provenance of The Owl and the Nightingale. ” Neuphilologische Mitteilungen 99: 249 - 68. Cartlidge, Neil. 2001. The Owl and the Nightingale: Text and Translation. Exeter: University of Exeter Press. Chambers, J. K. and Peter Trudgill. 1998. Dialectology. 2nd ed. Cambridge: University Press. Chaytor, H. J. 1966. From Script to Print. London: Sidgwick & Jackson. Chen, Matthew. 1970. “ Vowel Length Variation as a Function of the Voicing of the Consonant Environment. ” Phonetica 22: 129 - 159. Cheshire, Jenny. 1994. “ Standardization and the English Irregular Verbs. ” In Dieter Stein and Ingrid Tieken-Boon van Ostade (eds.). Towards a Standard English 1600 - 1800. Berlin, New York: Mouton de Gruyter, 115 - 133. Clark, Cecily. 1958. The Peterborough Chronicle 1070 - 1154: Edited from MS. Bodley Laud Misc. 636 with Introduction, Commentary, and an Appendix Containing the Interpolations. Oxford: University Press. Conde Silvestre, Juan Camilo and Juan Manuel Hernández Campoy. 2002. “ Modern Geolinguistic Tenets and the Diffusion of Linguistic Innovations in Late Middle English. ” Studia Anglica Posnaniensia 38: 147 - 177. 9. Bibliography 179 <?page no="198"?> Corpus of Middle English Verse and Prose. Last update February 22, 2006. Frances McSparran et al. (eds.). University of Michigan. <http: / / quod.lib.umich.edu/ c/ cme/ > Crowley, Joseph Patrick. 1986. “ The Study of Old English Dialects. ” English Studies 67: 97 - 112. Cuny, Albert L. M. 1906. Le nombre duel en grec. Paris: Librairie C. Klincksieck. Curzan, Anne. 2009. “ Historical Corpus Linguistics and Evidence of Language Change. ” In Anke Lüdelin and Merja Kytö (eds.). Corpus Linguistics: An International Handbook. Part 2. Berlin, New York: Mouton de Gruyter, 1091 - 1109. DeCamp, David. 1958. “ The Genesis of the Old English Dialects: A New Hypothesis. ” Language 34: 232 - 244. Dees, Anthonji. 1980. Atlas des formes et des constructions des chartes françaises du 13 e siècle. (Beihefte zur Zeitschrift für romanische Philologie 178). Tübingen: Niemeyer. Dees, Anthonji. 1994. “ Historical Dialectology and Literary Text Traditions. ” In Margaret Laing and Keith Williamson (eds.). Speaking in our Tongues: Proceedings of a Colloquium on Medieval Dialectology and Related Disciplines. Cambridge: D. S. Brewer, 117 - 125. Diehn, Otto. 1901. Die Pronomina im Frühmittelenglischen: Laut- und Flexionslehre. Heidelberg: Carl Winter ’ s Universitätsbuchhandlung. Dietz, Klaus. 1978. “ Zur mittelenglischen a ˉ / ǭ -Grenze in Lincolnshire. ” Beiträge zur Namenforschung 13: 185 - 193. Dietz, Klaus. 1989. “ Die historische Schichtung phonologischer Isoglossen in den englischen Dialekten: II. Mittelenglische Isoglossen. ” In Andreas Fischer (ed.). The History and the Dialects of English: Festschrift für Eduard Kolb. (Anglistische Forschungen 203). Heidelberg: Winter, 133 - 175. Doane, Alger N. and Matthew T. Hussey (eds.). 2000. Wulfstan Texts and Other Homiletic Materials. (Anglo-Saxon Manuscripts in Microfiche Facsimile Vol. 8). Tempe, AZ: Arizona Center for Medieval and Renaissance Studies. DOE = Dictionary of Old English. Antonette Di Paolo Healy et al. (eds.). Last updated April 04, 2014. University of Toronto. <http: / / www.doe.utoronto.ca/ index.html> DOE = Dictionary of Old English. Antonette Di Paolo Healy et al. (eds.). Last updated April 04, 2014. University of Toronto. http: / / www.doe.utoronto.ca/ index.html> DOE Web Corpus = Dictionary of Old English Web Corpus. Antonette Di Paolo Healey with John Price Wilkin, Xin Xiang (eds.). Last updated November 24, 2009. University of Toronto. <http: / / tapor.library.utoronto.ca/ doecorpus/ > Donkin, Robin Arthur. 1973. “ Changes in the Early Middle Ages. ” In Henry Clifford Darby (ed.). A New Historical Geography of England Before 1600. Cambridge: University Press, 75 - 135. Eichner, Heiner. 1990. “ Die Ausprägung der linguistischen Physiognomie des Englischen anno 400 bis anno 600 n. Chr. ” In Alfred Bammesberger and Alfred Wollmann (eds.). Britain 400 - 600: Language and History. Heidelberg: Carl Winter, 307 - 333. Ekwall, Eilert. 1913. “ Ortsnamenforschung ein Hilfsmittel für das engl. Sprachgeschichte-Studium. ” Germanisch-romanischen Monatsschrift 5: 592 - 608. Ekwall, Eilert. 1938. “ The Middle English a ˉ / o ˉ -boundary. ” English Studies 20, 147 - 168. Fischer, Andreas. 1997. “ The Hatton MS of the West Saxon Gospels: The Preservation and Transmission of Old English. ” In Paul E. Szarmach and Joel T. Rosenthal (eds.). The Preservation and Transmission of Anglo-Saxon Culture: Selected papers from the 1991 180 9. Bibliography <?page no="199"?> Meeting of the International Society of Anglo-Saxonists. (Studies in Medieval Culture XL). Kalamazoo: Medieval Institute Publications. Fisiak, Jacek. 1968. A Short Grammar of Middle English. Part I: Graphemics, Phonemics and Morphemics. Warsaw: Pa ń stwowe Wydawnictwo Naukowe. Fisiak, Jacek. 1994. “ Linguistic Reality of Middle English. ” In Francisco Fernández, Miguel Duster and Joan José Calvo (eds.). English Historical Linguistics 1992: Papers from the 7th International Conference on English Historical Linguistics. (Current Issues in Linguistic Theory 113). Amsterdam, Philadelphia: John Benjamins, 47 - 61. Fleischman, Suzanne. 2000. “ Methodologies and Ideologies in Historical Working with Older Languages. ” In Susan C. Herring, Pieter van Reenen and Lene Schøsler (eds.). Textual Parameters in Older Languages. Amsterdam/ Philadelphia: John Benjamins, 33 - 58. Fletcher, Alan J. 1994. “‘ Benedictus qui venit in nomine Domini ’ : A Thirteenth-Century Sermon for Advent and the Macaronic Style in England. ” Mediaeval Studies 56: 217 - 245. Fletcher, Alan J. 1999. “ The Genesis of The Owl and the Nightingale. ” The Chaucer Review 34: 1 - 17 Fulk, Robert D. 1987. “ Reduplicating Verbs and Their Development in Northwest Germanic. ” Beiträge zur Geschichte der deutschen Sprache und Literatur 109: 159 - 178. Fulk, Robert D. 1998 a. “ The Role of Syllable Structure in Old English Quantitative Sound Changes. ” NOWELE 33: 3 - 35. Fulk, Robert D. 1998 b. “ The Chronology of Anglo-Frisian Sound Changes. ” In Rolf H. Bremmer, Jr., Thomas S. B. Johnston, and Oebele Vries (eds.). Approaches to Old Frisian Philology. (Amsterdamer Beiträge zur Älteren Germanistik 49). Amsterdam: Rodopi, 139 - 54 Gardner, Anne. forthcoming. Derivation in Middle English: Regional and Text Type Variation. (Mémoires de la Société Néophilologique 92). Helsinki: Société Néophilologique. Gburek, Hubert. 1985. “ The vowel / a ː / in English. ” In Jacek Fisiak (ed.). Papers from the 6th International Conference on Historical Linguistics. (Current Issues in Linguistic Theory 34). Amsterdam: John Benjamins, 139 - 148. Gburek, Hubert. 1986. “ Changes in the Structure of the English Verb System - Evidence from Scots. ” In Dietrich Strauss and Horst W. Dressler (ed.). Scottish Language and Literature, Medieval and Renaissance. Frankfurt: Peter Lang, 115 - 124. Gerritsen, Marinel and Frank Jansen. 1980. “ The Interplay of Dialectology and Historical Linguistics: Some Refinements of Trudgill ’ s Formula. ” In Peter Maher (ed.). Proceedings of the Third International Congress of Historical Linguistics. Amsterdam: Benjamins, 11 - 38. Girnth, Heiko. 2010. “ Mapping Language Data. ” In Alfred Lameli, Roland Kehrein and Stefan Rabanus (eds.). An International Handbook of Linguistic Variation. Vol. 2: Language Mapping. (Handbooks of Linguistics and Communication Science 30.2). Berlin, New York: Mouton De Gruyter, 98 - 121. Glauser, Beat. 1994. “ Dialect Maps: Depicting, Constructing or Distorting Linguistic Reality. ” In Gunnel Melchers and Nils-Lennart Johannesson (eds.) Nonstandard Varieties of Language: Papers from the Stockholm Symposium, 11 - 13 April, 1991. (Stockholm Studies in English 84). Stockholm: Almqvist and Wiksell, 35 - 52. Görlach, Manfred. 1996. “ Morphological Standardization: The Strong Verbs in Scots. ” In Derek Britton (ed.). English Historical Linguistics 1994: Papers from the 8th 9. Bibliography 181 <?page no="200"?> International Conference on English Historical Linguistics (8. ICEHL, Edinburgh, 19 - 23 September 1994). Amsterdam and Philadelphia: John Benjamins, 161 - 181. Gregory, Derek. 1985. “ Suspended Animation: The Stasis of Diffusion Theory. ” In Derek Gregory and John Urry (eds.). Social Relations and Spatial Structures. Basingstoke, London: Macmillan, 296 - 336. Griffiths, Ralph A. 1996. “ The Later Middle Ages (1290 - 1485). ” In Kenneth O. Morgan (ed.). The Oxford Illustrated History of Britain. Oxford: University Press, 166 - 222. Grimm, Jacob. 1819. Deutsche Grammatik. Göttingen: Dieterich. Guðmundsson, Helgi. 1972. The Pronominal Dual in Icelandic. (University of Iceland. Publications in Linguistics 2). Reykjavík: Institute of Nordic Linguistics. Haas, Walter. 2010. “ A Study on Areal Diffusion. ” In Alfred Lameli, Roland Kehrein and Stefan Rabanus (eds.). An International Handbook of Linguistic Variation. Vol. 2: Language Mapping. (Handbooks of Linguistics and Communication Science 30.2). Berlin, New York: Mouton De Gruyter, 649 - 667. Haugen, Einar. 1950. The First Grammatical Treatise. The Earliest Germanic Phonology: An Edition, Translation and Commentary. (Language Monographs 25). Baltimore: Waverly Press. Hempl, George. 1896. “ Grease and Greasy. ” Dialect Notes 1: 438 - 444. Hernández Campoy, Juan Manuel and Juan Camilo Conde Silvestre. 2005. “ Sociolinguistics and Geolinguistics Approaches to the Historical Diffusion of Linguistic Innovations: Incipient Standardisation in Late Middle English. ” International Journal of English Studies 5: 101 - 134. Hernández Campoy, Juan Manuel. 1999. “ Geolinguistic Models of Analysis of the Spatial Diffusion of Linguistic Innovations. ” Studia Anglica Posnaniensia 34: 7 - 42. Hernández Campoy, Juan Manuel. 2003. “ Exposure to Contact and the Geographical Adoption of Standard Features: Two Complementary Approaches. ” Language in Society 32: 227 - 255. Higham, Nicholas (ed.). 2007. Britons in Anglo-Saxon England. Woodbridge: Boydell and Brewer. Hinde, Andrew. 2003. England ’ s Population: A History since the Domesday Survey. London: Oxford University Press. Hogg, Richard M. 1982. “ Was There Ever an / ɔ / -Phoneme in Old English? ” Neuphilologische Mitteilungen 83: 225 - 229. Hogg, Richard M. 1992. A Grammar of Old English. Vol. 1: Phonology. Oxford: Blackwell. Hogg, Richard M. and Robert D. Fulk. 2011. A Grammar of Old English. Vol. 2: Morphology. Oxford: Wiley-Blackwell. Holthausen, Ferdinand. 1928. Havelok mit Einleitung, Glossar und Anmerkungen. 3rd rev. ed. Heidelberg: Carl Winter. Horobin, Simon and Jeremy J. Smith. 2002. An Introduction to Middle English. Edinburgh: University Press. Horvath, Barbara and Ron Horvath. 1997. “ The Geolinguistics of a Sound Change in Progress: / l/ Vocalisation in Australia. ” In Charles Boberg, Miriam Meyerhoff and Stephanie Strassel (eds.). A Selection of Papers from NWAVE 25. (Special Issue of University of Pennsylvania Working Papers in Linguistics 4). Philadelphia: Penn Linguistics Club, 109 - 124. Horvath, Barbara and Ron Horvath. 2001. “ A Multilocality Study of a Sound Change in Progress: The Case of / l/ Vocalisation in New Zealand and Australian English. ” Language Variation and Change 13: 37 - 58. 182 9. Bibliography <?page no="201"?> Horvath, Barbara and Ron Horvath. 2002. “ The Geolinguistics of / l/ Vocalisation in Australia and New Zealand. ” Journal of Sociolinguistics 6: 319 - 346. Hotta, Ryuichi. 2009. The Development of the Nominal Plural Forms in Early Middle English. Tokyo: Hituzi Syobo Publishing. Howe, Stephen. 1994. The Personal Pronouns in the Germanic Languages: A Study of Personal Pronoun Morphology and Change in the Germanic Languages from the First Records to the Present Day. (Studia Linguistica Germanica 43). Berlin: de Gruyter. Hulbert, J. R. 1921. “ The West-Midland of the Romances. ” Modern Philology 19: 1 - 16. Iglesias-Rábade, Luis. 2003. Handbook of Middle English: Grammar and Texts. München: LINCOM GmbH. Jones, Charles. 1989. A History of English Phonology. London: Longman. Jordan, Richard. 1974 [1934]. Handbook of Middle English Grammar: Phonology. Translated and revised by Eugene Joseph Crook. (Janua Linguarum. Series Practica 218). The Hague & Paris: Mouton de Gruyter. Kaluza, Max. 1901. Historische Grammatik der englischen Sprache. 2nd ed. Berlin: Felber. Keene, Derek. 2000. “ Metropolitan Values: Migration, Mobility and Cultural Norms, London 1100 - 1700. ” In Laura Wright (ed.). The Development of Standard English, 1300 - 1800: Theories, Descriptions, Conflicts. Cambridge: University Press, 93 - 114. Kennedy, Arthur G. 1915. The Pronouns of Address in English Literature of the Thirteenth Century. Stanford: University Publications. Ker, N. R. 1957. Catalogue of Manuscripts Containing Anglo-Saxon. Oxford: Clarendon Press. Kitson, Peter R. 1992. “ On the Chronological and Geographical Spread of Old English Combinative u-mutation. ” Studia Neophilologica 64: 3 - 23. Kitson, Peter R. 1995. “ The Nature of Old English Dialect Distributions, Mainly as Exhibited in Charter Boundaries. ” In Jacek Fisiak (ed.). Medieval Dialectology. (Trends in Linguistics: Studies and Monographs 79). Berlin and New York: Mouton de Gruyter, 43 - 135. Klaeber, Friedrich. 1902. “ Zur altenglischen Bedaübersetzung. ” Anglia 25: 257 - 315. Kortlandt, Frederik. 1991. “ The Germanic Seventh Class of Strong Verbs. ” NOWELE 18: 97 - 100. Kortlandt, Frederik. 1992. “ The Germanic Fifth Class of Strong Verbs. ” NOWELE 19: 101 - 107. Kranzmayer, E. 1954. “ Der pluralische Gebrauch des alten Duals ‘ eß ’ and ‘ enk ’ im Bairischen. ” In Festschrift für Dietrich Kralik. Horn: Berger, 249 - 259. Kretzschmar, William A. 2003. “ Mapping Southern English. ” American Speech 78: 130 - 149. Kristensson, Gillis. 1965. “ Another Approach to Middle English Dialectology. ” English Studies 46: 138 - 156. Kristensson, Gillis. 1967. A Survey of Middle English Dialects 1290 - 1350: The Six Northern Counties and Lincolnshire. (Lund Studies in English 35). Lund: C. W. K Gleerup. Kristensson, Gillis. 1987. A Survey of Middle English Dialects 1290 - 1350: The West Midland Counties. (Skrifter Utgivna av Vetenskaps societeten i Lund 78). Lund: University Press. Kristensson, Gillis. 1995. A Survey of Middle English Dialects 1290 - 1350: The East Midland Counties. (Skrifter Utgivna av Vetenskaps societeten i Lund 88). Lund: University Press. 9. Bibliography 183 <?page no="202"?> Kristensson, Gillis. 2001. A Survey of Middle English Dialects 1290 - 1350: The Southern Counties. Vol. I. Vowels (except Diphthongs). (Skrifter Utgivna av Vetenskaps societeten i Lund 93). Lund: University Press. Kristensson, Gillis. 2002. A Survey of Middle English Dialects 1290 - 1350: The Southern Counties. Vol. II. Diphthongs and Consonants. (Skrifter Utgivna av Vetenskaps societeten i Lund 94). Lund: University Press. Krogmann, Willy. 1956. “ As. *unk Adama? ” Zeitschrift für Vergleichende Sprachforschung 74: 146 - 161. Krygier, Marcin. 1994 a. The Disintegration of the English Strong Verb System. (Bamberger Beiträge zur Englischen Sprachwissenschaft 34). Frankfurt am Main: Peter Lang. Krygier, Marcin. 1994 b. “ The Strong Verb System in the Peterborough Chronicle. ” Studia Anglica Posnaniensia 28: 49 - 69. Krygier, Marcin. 2001. “ Reconsidering the History of the English Verbal System. ” Studia Anglica Posnaniensia 36: 51 - 59. Kuhn, Sherman M. 1961. “ On the Syllabic Phonemes in Old English. ” Language 37: 522 - 538. Kurath, Hans. 1949. A Word Geography of the Eastern United States. Ann Arbor: University of Michigan Press. Kytö, Merja. 1996. Manual to the Diachronic Part of the Helsinki Corpus of English Texts: Coding Conventions and Lists of Source Texts. 3rd ed. Helsinki: Helsinki University Printing House. Labov, William, Sharon Ash and Charles Boberg. 2006. Atlas of North American English: Phonetics, Phonology and Sound Change. Berlin: Mouton de Gruyter. Labov, William. 1972. Sociolinguistic Patterns. Philadelphia: University of Philadelphia Press. Labov, William. 1994. Principles of Linguistic Change. Vol. 1: Internal Factors. Oxford: Blackwell. Ladd, C. A. and I. Radice. 1951/ 52. “ The Survival of the Dual - Middle English. ” Unpublished Seminar Paper, University of Oxford. LAEME = A Linguistic Atlas of Early Middle English, 1150 - 1325. Version 2.1. Compiled by Margaret Laing. Electronic text corpus with accompanying software, index of sources and theoretical introduction (with Roger Lass). Edinburgh: © 2008 - The University of Edinburgh. <http: / / www.lel.ed.ac.uk/ ihd/ laeme1/ laeme1.html> LAEME 3.2 = A Linguistic Atlas of Early Middle English, 1150 - 1325. Version 3.2. Compiled by Margaret Laing. Edinburgh: © 2013 - The University of Edinburgh. <http: / / www.lel.ed.ac.uk/ ihd/ laeme2/ laeme2.html> Lahiri, Aditi, Tomas Riad and Haike Jacobs. 1999. “ Diachrony. ” In Harry van der Hulst (ed.). Word Prosodic Systems in the Languages of Europe. Berlin, New York: Mouton de Gruyter, 335 - 424. Laing, Margaret and Roger Lass. 2006. “ Early Middle English Dialectology: Problems and Prospects. ” In Ans van Kemenade and Bettelou Los (eds.). The Handbook of the History of English. Malden, MA: Blackwell, 417 - 451. Laing, Margaret and Roger Lass. 2008 - . “ Introduction. ” A Linguistic Atlas of Early Middle English 1150 - 1325. Edinburgh: University of Edinburgh. <http: / / www.lel.ed.ac.uk/ ihd/ laeme1/ laeme1_intro.html> Laing, Margaret. 1993. Catalogue of Sources for a Linguistic Atlas of Early Medieval English. Cambridge: D. S. Brewer. 184 9. Bibliography <?page no="203"?> Laing, Margaret. 1995. “ A Linguistic Atlas of Early Middle English. ” Medieval English Studies Newsletter 33: 1 - 6. Laing, Margaret. 2000. “ Never the twain shall meet: Early Middle English - The East-West Divide. ” In Irma Taavitsainen et al (eds.). Placing Middle English in Context. (Topics in English Linguistics 35). Berlin, New York: Mouton de Gruyter, 97 - 124. Laing, Margaret. 2006. “ The Early Middle English Scribe: Sprach er wie er schrieb? ” In Marina Dossena, Richard Dury and Maurizio Gotti (eds.). English Historical Linguistics 2006. Vol. III: Geo-Historical Variation in English. Selected Papers from the Fourteenth International Conference on English Historical Linguistics (ICEHL 14), Bergamo, 21 - 25 August 2006. Amsterdam: John Benjamins, 1 - 44. Laing, Margaret. 2007. “ The Owl and the Nightingale: Five New Readings and Further Notes. ” Neuphilologische Mitteilungen 108: 445 - 477. Laing, Margaret. 2008 - . “ Index of Sources. ” A Linguistic Atlas of Early Middle English 1150 - 1325. Edinburgh: University of Edinburgh. Last accessed October 2011. <http: / / www.lel.ed.ac.uk/ ihd/ laeme1/ pdf/ LAEME_Index_of_Sources.pdf> Lass, Roger and Margaret Laing. 2005. “ Are Front Rounded Vowels Retained in West Midland Middle English? ” In Nikolaus Ritt and Herbert Schendl (eds.). Rethinking Middle English: Linguistic and Literary Approaches. (Studies in English Medieval Language and Literature 10). Frankfurt am Main: Peter Lang, 280 - 290. Lass, Roger and Margaret Laing. 2010. “ In Celebration of Early Middle English ‘ h ’ . ” Neuphilologische Mitteilungen 111: 345 - 354. Lass, Roger. 1976. English Phonology and Phonological Theory. Cambridge: University Press. Lass, Roger. 1992. “ Phonology and Morphology. ” In Norman Blake (ed.). The Cambridge History of the English Language. Vol. II: 1066 - 1476. Cambridge: University Press, 23 - 155. Lass, Roger. 1994 a. Old English: A Historical Linguistic Companion. Cambridge: University Press. Lass, Roger. 1994 b. “ Proliferation and Option-cutting: The Strong Verb in the Fifteenth to Eighteenth Centuries. ” In Dieter Stein and Ingrid Tieken-Boon van Ostade (eds.). Towards a Standard English 1600 - 1800. Berlin, New York: Mouton de Gruyter, 81 - 113. Lass, Roger. 1997. Historical Linguistics and Language Change. (Cambridge Studies in Linguistics 81). Cambridge: University Press. Lass, Roger. 2000. “ Language Periodization and the Concept ‘ Middle ’ . ” In Irma Taavitsainen, Terttu Nevalainen, Päivi Pahta and Matti Rissanen (eds.). Placing Middle English in Context. Berlin, New York: Mouton de Gruyter, 7 - 41. Lass, Roger. 2004. “ Ut custodiant litteras: Editions, Corpora and Witnesshood. ” In Marina Dossena and Roger Lass (eds.). Methods and Data in English Historical Dialectology. (Linguistic Insights 16). Bern: Peter Lang, 21 - 48. Lass, Roger. 2006. “ Phonology and Morphology. ” In Richard Hogg and David Denison. A History of the English Language. Cambridge: University Press, 43 - 108. Lenker, Ursula. 2000. “ The Monasteries of the Benedictine Reform and the ‘ Winchester School ’ : Model Cases of Social Networks in Anglo-Saxon England? ” European Journal of English Studies 4, 225 - 238. Liberman, A. S. 1966. “ On the History of Middle English a ˉ and a. ” Neuphilologische Mitteilungen 67: 66 - 71. 9. Bibliography 185 <?page no="204"?> Lieberman, Erez, Jean-Baptiste Michel, Joe Jackson, Tina Tang and Martin A. Nowak. 2007. “ Quantifying the Evolutionary Dynamics of Language. ” Nature 449: 713 - 716. Liebl, Christian. 2002. “ The Origins and Geographical Diffusion of a Medieval English Sound-Change: A Study of a ˉ > ǭ , Based on Onomastic and Textual Evidence. ” Unpublished M. A. Thesis, University of Vienna. Liebl, Christian. 2005. “ Two Unnoticed Early ME Versions of Candet Nudatum Pectus. ” In Niklaus Ritt and Herbert Schendl (eds.). Rethinking Middle English: Linguistic and Literary Approaches. (Studies in English Medieval Language and Literature 10). Frankfurt am Main: Peter Lang, 67 - 74. Liebl, Christian. 2006. “ The A and O of a Medieval English Sound-change: Prolegomena to a Study of the Origins and Early Geographical Diffusion of / ɑː / > / ɔː / . ” In Niklaus Ritt, Herbert Schendel, Christiane Dalton-Puffer and Dieter Kastovsky (eds.). Medieval English and its Heritage: Structure, Meaning and Mechanisms of Change. Frankfurt am Main: Peter Lang, 19 - 35. Liebl, Christian. 2008. “ Nothing New in the North? The Reflexes of OE a ˉ in Early Middle English Place-Names. ” Vienna English Working Papers 17: 28 - 62. Liuzza, R. M. 1994. The Old English Version of the Gospels. Vol. 1: Text and Introduction. (Early English Text Society 304). Oxford: University Press. Long, Mary McDonald. 1944. The English Strong Verb From Chaucer to Caxton. Menasha, WI: George Banta Publishing. Luick, Karl. 1898. “ Die quantitätsveränderungen im laufe der englischen sprachentwicklung. ” Anglia 20: 335 - 362. Luick, Karl. 1964 [1914 - 21]. Historische Grammatik der Englischen Sprache. Stuttgart: Tauchnitz. Lutz, Angelika. 2004. “ The First Push: A Prelude to the Great Vowel Shift. ” Anglia 122: 209 - 224. Mailhammer, Robert. 2007. The Germanic Strong Verbs: Foundations and Development of a New System. Berlin: Mouton de Gruyter. Markus, Manfred. 1990. Mittelenglisches Studienbuch. Tübingen: Francke Verlag. McIntosh, Angus, M. L. Samuels and Michael Benskin. 1986. A Linguistic Atlas of Late Mediaeval English. Aberdeen: University Press. McIntosh, Angus, M. L. Samuels, Michael Benskin, Margaret Laing and Keith Williamson. 1987. Guide to A Linguistic Atlas of Late Mediaeval English. Aberdeen: University Press. McIntosh, Angus. 1963. “ A New Approach to Middle English Dialectology. ” English Studies 44: 1 - 11. McIntosh, Angus. 1973. “ Word Geography in the Lexicography of Mediaeval English. ” Annals of the New York Academy of Sciences 211: 55 - 66. McWhorter, John H. 2011. Linguistic Simplicity and Complexity. Boston/ Berlin: Walter de Gruyter. MED = Middle English Dictionary. Frances McSparran et al. (eds.). Last updated 18 December 2001. University of Michigan. <http: / / quod.lib.umich.edu/ m/ med/ >. Meid, Wolfgang. 1971. Das Germanische Präteritum. Indogermanische Grundlagen und Ausbreitung im Germanischen. (Innsbrucker Beiträge zur Sprachwissenschaft 3). Innsbruck: Institut für Vergleichende Sprachwissenschaft der Universität Innsbruck. Meillet, Antoine. 1908. “ Notes sur quelques faits gotiques. ” Mémoires de la Société de Linguistique de Paris 15: 73 - 103. 186 9. Bibliography <?page no="205"?> Michelau, Erich. 1910. Der Übertritt starker Verba in die schwache Coniugation im Englischen. Königsberg: Karg und Manneck. Milroy, James. 1992. “ Middle English Dialectology. ” In Norman Blake (ed.). The Cambridge History of the English Language. Vol. II. 1066 - 1476. Cambridge: University Press, 156 - 206. Minkova, Donka and Robert Stockwell. 1992. “ Homorganic Clusters as Moric Busters in the History of English: The Case of -ld, -nd, -mb. ” In Matti Rissanen, Ossi Ihalainen, Terttu Nevalainen and Irma Taavitsainen (eds.). History of Englishes: New Methods and Interpretations in Historical Linguistics. Berlin, New York: Mouton de Gruyter, 191 - 206. Minkova, Donka and Robert Stockwell. 1998. “ The Origins of Long-short Allomorphy in English. ” In Jacek Fisiak and Marcin Krygier (eds.). Advances in English Historical Linguistics. Berlin: Mouton de Gruyter, 211 - 39 Mitzka, Walther. 1952. Handbuch zum Deutschen Sprachatlas. Marburg: Elwert. Moore, Samuel, Sanford Brown Meech and Harold Whitehall. 1935. Middle English Dialect Characteristics and Dialect Boundaries: Preliminary Report of an Investigation Based Exclusively on Localized Texts and Documents. Ann Arbor: University of Michigan Press. Moore, Samuel. 1969. Historical Outlines of English Sounds. rev. by Albert H. Marckwardt. Ann Arbor: Wahr. Mossé, Fernand. 1952. A Handbook of Middle English. Trans. James A. Walker. Baltimore: The Johns Hopkins Press. Mottausch, Karl-Heinz. 1998. “ Die reduplizierenden Verben im Nord- und Westgermanischen: Versuch eines Raum-Zeitmodells. ” NOWELE 33: 43 - 91. Mottausch, Karl-Heinz. 2000. “ Das Präteritum der 4. und 5. starken Verbklassen im Germanischen. ” NOWELE 36: 45 - 58. Nedoma, Robert. 2001. Kleine Grammatik des Altisländischen. Heidelberg: Carl Winter. Nerbonne, John and Wilbert Heeringa. 2007. “ Geographic Distributions of Linguistic Variation Reflect Dynamics of Differentiation. ” In Sam Featherston and Wolfgang Sternefeld (eds.). Roots: Linguistics in Search of its Evidential Base. Berlin: Mouton de Gruyter, 267 - 297. Nevalainen, Terttu and Helena Raumolin-Brunberg. 2003. Historical Sociolinguistics: Language Change in Tudor and Stuart England. London: Longman. Nevanlinna, Saara, Päivi Pahta, Kirsti Peitsara and Irma Taavitsainen. 1993. “ Middle English. ” In Matti Rissanen, Merja Kytö, Minna Palander-Collin. Early English in the Computer Age: Explorations Through the Helsinki Corpus. Berlin, New York: Mouton de Gruyter, 33 - 51. Nielsen, Hans F. 1985. “ Tendencies in the Evolution of the Modern English Irregular Verbs. ” Journal of English Linguistics 18: 41 - 53. Oakden, James P. 1930. Alliterative Poetry in Middle English. Vol. 1: The Dialectal and Metrical Survey. Manchester: University Press. OED = Oxford English Dictionary. 2011. Oxford: University Press. <www.oed.com> Orton, Harold et al. (ed.). 1962 - 71. Survey of English Dialects. Leeds: Arnold. Pederson, Lee A., Susan Leas McDaniel, Guy Bailey, Marvin H. Bassett, Carol M. Adams, Caisheng Liao and Michael Montgomery (eds.). 1986 - 92. Linguistic Atlas of the Gulf States. 7 vols. Athens: University of Georgia Press. Peters, Hans. 1988. “ On the State and Possible Aims of Middle English Word Geography. ” In Jacek Fisiak (ed.). Historical Dialectology: Regional and Social. The Hague: Mouton de Gruyter, 397 - 416. 9. Bibliography 187 <?page no="206"?> Phillips, Betty S. 1981. “ The Phonetic Basis of a Late Old English Sound Change. ” In Wolfgang U. Dressler, Oskar E. Pfeiffer, and John R. Rennison (eds.). Phonologica 1980. (Innsbrucker Beiträge zur Sprachwissenschaft 36). Innsbruck: Institut für Sprachwissenschaft der Universität Innsbruck, 337 - 341. Phillips, Betty S. 1983 a. “ Lexical Diffusion and Function Words. ” Linguistics 21: 487 - 499. Phillips, Betty S. 1983 b. “ Constraints on Syllables and Quantitative Changes in Early English. ” Linguistics 21: 879 - 895. Pilch, Herbert. 1997. “ Middle English Phonetics: A Systematic Survey Including Notes on Irish and Welsh Loanwords. ” In Jacek Fisiak (ed.). Studies in Middle English Linguistics. (Trends in Linguistics: Studies and Monographs 103). Berlin and New York: Mouton de Gruyter, 439 - 464. Pinsker, Hans E. 1963. Historische englische Grammatik. München: Max Hueber. Price, Hereward Thimbleby. 1910. A History of Ablaut in the Strong Verbs from Caxton to the End of the Elizabethan Period. Bonn: Hanstein. Prins, Anton A. 1972. A History of English Phonemes: From Indo-European to Present-Day English. 2nd ed. Leiden: Leiden University Press. Prokosch, Eduard, 1939. A Comparative Germanic Grammar. Philadelphia: Linguistic Society of America. Quirk, Randolph and Charles Leslie Wrenn. 1958. An Old English Grammar. 2nd edition. London: Methuen. Richter, Michael. 1985. “ Towards a Methodology of Historical Sociolinguistics. ” Folia Linguistica Historica 6: 41 - 61. Ritt, Nikolaus. 1994. Quantity Adjustment: Vowel Lengthening and Shortening in Early Middle English. Cambridge: University Press. Rothwell, Harry. 1975. English Historical Documents 1189 - 1327. Volume 3. London: Eyre & Spottiswoode. Russell, John C. 1949. British Medieval Population. Albuquerque: The University of New Mexico Press. Sawyer, Peter H. 1968. Anglo-Saxon Charters: An Annotated List and Bibliography. London: Office of the Royal Historical Society. Schlemilch, Willy. 1914. Beiträge zur Sprache und Orthographie spätaltengl. Sprachdenkmäler der Übergangszeit (1000 - 1150). (Studien zur englischen Philologie 34). Halle a. S.: Niemeyer. Schmidt, Gernot. 1978. Stammbildung und Flexion der Indogermanischen Personalpronomina. Wiesbaden: Harrassowitz. Schmidt, Johannes. 1872. Die Verwandtschaftsverhältnisse der Indogermanischen Sprachen. Weimar: Böhlau. Schrijver, Peter. 2009. “ Celtic Influence on Old English: Phonological and Phonetic Evidence. ” English Language and Linguistics 13(2): 193 - 211. Scragg, Donald G. 1974. A History of English Spelling. Manchester: Manchester University Press. Scragg, Donald G. 1992. “ Spelling Variations in Eleventh-Century English. ” In Carola Hicks (ed.). England in the Eleventh Century: Proceedings of the 1990 Harlaxton Symposium. Stamford: Paul Watkins, 347 - 354. Seebold, Elmar. 1984. Das System der Personalpronomina in den frühgermanischen Sprachen: Sein Aufbau und seine Herkunft. Göttingen: Vandenhoeck and Ruprecht. 188 9. Bibliography <?page no="207"?> Seppänen, Aimo. 1985. “ On the Use of the Dual in Gothic. ” Zeitschrift für deutsches Altertum und deutsche Literatur 114, 1 - 41. Serjeantson, Mary S. 1922. “ The Dialectal Distribution of Certain Phonological Features in Middle English. ” English Studies 4: 93 - 109, 191 - 198, 223 - 233. Shields, Kenneth. 2001. “ On the Origin of Old English uncet and in ċ it. ” Neuphilologische Mitteilungen 102: 211 - 216. Sisam, Kenneth. 1953. Studies in the History of Old English Literature. Oxford: Clarendon Press. Skeat, Walter W. 1901. The Proverbs of Alfred: Re-Edited from the Manuscripts. Oxford: Clarendon Press. Skeat, Walter W. 1970 [1881 - 1887]. The Gospel According to Saint Matthew and According to Saint Mark. Unrevised reprint. Darmstadt: Wissenschaftliche Buchgesellschaft. Smith, Jeremy J. 1992. “ A Linguistic Atlas of Early Middle English: Tradition and Typology. ” In Matti Rissanen et al. (eds.). History of Englishes: New Methods and Interpretations in Historical Linguistics. (Topics in English Linguistics 10). Berlin and New York: Mouton de Gruyter, 582 - 591. Smith, Jeremy J. 2000. “ Standard Language in Early Middle English? ” In Irma Taavitsainen et al. (eds.). Placing Middle English in Context. (Topics in English Linguistics 35). Berlin and New York: Mouton de Gruyter, 125 - 139. Smith, Jeremy J. 2007. Sound Change and the History of English. Oxford: University Press. Stadlmann, Alois. 1921. Die Sprache der Mittelenglischen Predigtsammlung in der Handschrift Lambeth 487. Wien und Leipzig: Wilhelm Braunmüller. Stenroos, Merja. 2009. “ MEG: Project Summary. ” The Middle English Grammar Project. University of Stavanger. <http: / / www.uis.no/ research/ culture/ the_middle_english_grammar_project/ project_summary/ > Stiles, Patrick V. 1995. “ Remarks on the ‘ Anglo-Frisian ’ Hypothesis. ” In Volkert F. Faltings, Alastair G. H. Walker and Ommo Wilts (eds.). Friesische Studien II: Beiträge des Föhrer Symposiums zur Friesischen Philologie vom 7. - 8. April 1994. (NOW- ELE Supplement 12). Odense: Odense Universitetsforlag, 177 - 220. Stiles, Patrick V. 1996. “ Old English uncet and in ċ it. ” In Adeline Petersen and Hans F. Nielsen (eds.). A Frisian and Germanic Miscellany Published in Honour of Nils Århammer on his Sixty-Fifth Birthday, 7 August 1996. Odense, University Press, 557 - 568. Strang, Barbara. 1970. A History of English. London: Methuen. Taylor, Ann. 1994. “ Variation in Past Tense Formation in the History of English. ” Penn Working Papers in Linguistics 1: 143 - 158. Thomson, Samuel Harrison. 1935. “ The Date of the Early English Translation of the Candet Nudatum Pectus. ” Medium Ævum 4: 100 - 105. Treharne, Elaine (ed.). 2010. Old and Middle English c.890 - c.1450: An Anthology. 3rd ed. Oxford, Malden: Wiley-Blackwell. Trudgill, Peter. 1974. “ Linguistic Change and Diffusion: Description and Explanation in Sociolinguistic Dialect Geography. ” Language in Society 3: 215 - 246. Trudgill, Peter. 1983. On Dialect: Social and Geographical Perspectives. Oxford: Blackwell. Trudgill, Peter. 1986. Dialects in Contact. Oxford: Blackwell. Umpfenbach, Heinz. 1935. Die oa-Schreibung im Englischen: Ein Beitrag zur Geschichte der Englischen Orthographie. (Palaestra 201). Leipzig: Mayer & Müller. 9. Bibliography 189 <?page no="208"?> van Coetsem, Frans. 1990. Ablaut and Reduplication in the Germanic Verb. (Untersuchungen zur vergleichenden Grammatik der germanischen Sprachen 3). Heidelberg: Winter. Vásquez, Nila and Teresa Marqués-Aguado. 2012. “ Editing the Medieval Manuscript in its Social Context ” . In Juan Manuel Hernández-Campoy and Juan Camilo Conde- Silvestre (eds.). The Handbook of Historical Sociolinguistics. Chichester: Wiley-Blackwell, 123 - 139. Viereck, Wolfgang and Heinrich Ramisch. 1991 - 1997. The Computer Developed Linguistic Atlas of England (CLAE). 2 vols. Tübingen: Niemeyer. Wackernagel, Jacob. 1950. Vorlesungen über Syntax mit besonderer Berücksichtigung von Griechisch, Lateinisch und Deutsch. Basel: Verlag Birkhäuser. Wells, John E (ed.). 1907. The Owl and the Nightingale. Boston, London: D. C. Heath and Co. We ł na, Jerzy. 1991. “ The Strong to Weak Shift in English Verbs: A Reassessment. ” Kalbotyra 42: 129 - 139. Wikle, Tom and Guy Bailey. 1997. “ The Spatial Diffusion of Linguistic Features in Oklahoma. ” Proceedings of the Oklahoma Academy of Science 77: 1 - 15. Wikle, Tom. 1997. “ Quantitative Mapping Techniques for Displaying Language Variation and Change, ” In Cynthia Bernstein, Thomas Nunnally and Robin Sabino (eds.). Language Variety in the South Revisited. Tuscaloosa, AL: University of Alabama Press, 417 - 433. Williams, Joseph M. 1975. Origins of the English Language: A Social and Linguistic History. New York: Free Press. Williamson, Keith. 2008 - A Linguistic Atlas of Older Scots, Phase 1: 1380 - 1500. University of Edinburgh. <http: / / www.lel.ed.ac.uk/ ihd/ laos1/ laos1.html>. Wright, Joseph and Elizabeth Mary Wright. 1928. An Elementary Middle English Grammar. Oxford: University Press. Wyld, Henry C. 1913. “ The Treatment of OE. ȳ ̆ in the Dialects of the Midland and South Eastern Counties in ME. ” English Studies 47: 1 - 58. Wyld, Henry C. 1914. “ OE. ȳ ̆ in the Dialects of the South and South Western Counties in ME. ” English Studies 47: 145 - 166. Wyld, Henry C. 1921. “ South-eastern and South-east Midland Dialects in Middle English. ” Essays and Studies by Members of the English Association 6: 112 - 145. 190 9. Bibliography <?page no="209"?> Appendix I - Distribution of Grade Vowel per Text File and Verb Class Class I Text file Singular Forms Plural Forms 1st Past vowel 2nd Past vowel ‘ other ’ 1st past vowel 2nd past vowel ‘ other ’ n= % n= % n= % n= % n= % n= % 2 2 100.0 0 0.0 0 0.0 - - - - - - 3 1 100.0 0 0.0 0 0.0 - - - - - - 5 - - - - - - 0 0.0 2 66.7 1 33.3 6 - - - - - - 0 0.0 2 100.0 0 0.0 7 - - - - - - 0 0.0 0 0.0 1 100.0 64 7 87.5 0 0.0 1 12.5 0 0.0 1 100.0 0 0.0 65 0 0.0 0 0.0 1 100.0 - - - - - - 118 2 50.0 1 25.0 1 25.0 0 0.0 2 100.0 0 0.0 119 1 100.0 0 0.0 0 0.0 - - - - - - 121 - - - - - - 0 0.0 1 100.0 0 0.0 123 9 69.2 0 0.0 4 30.8 0 0.0 1 100.0 0 0.0 124 1 100.0 0 0.0 0 0.0 - - - - - - 128 1 100.0 0 0.0 0 0.0 - - - - - - 137 4 100.0 0 0.0 0 0.0 - - - - - - 140 1 50.0 0 0.0 1 50.0 - - - - - - 142 3 100.0 0 0.0 0 0.0 - - - - - - 149 - - - - - - 1 33.3 2 66.7 0 0.0 150 2 66.7 0 0.0 1 33.3 - - - - - - 155 19 82.6 0 0.0 4 17.4 1 20.0 4 80.0 0 0.0 158 2 66.7 0 0.0 1 33.3 0 0.0 0 0.0 1 100.0 169 1 50.0 0 0.0 1 50.0 1 100.0 0 0.0 0 0.0 171 1 33.3 0 0.0 2 66.7 - - - - - - 172 2 66.7 0 0.0 1 33.3 0 0.0 1 100.0 0 0.0 173 6 100.0 0 0.0 0 0.0 0 0.0 1 100.0 0 0.0 175 - - - - - - 0 0.0 1 100.0 0 0.0 182 1 100.0 0 0.0 0 0.0 0 0.0 1 100.0 0 0.0 183 1 50.0 0 0.0 1 50.0 - - - - - - 188 - - - - - - 0 0.0 1 100.0 0 0.0 189 - - - - - - 0 0.0 2 100.0 0 0.0 <?page no="210"?> Text file Singular Forms Plural Forms 1st Past vowel 2nd Past vowel ‘ other ’ 1st past vowel 2nd past vowel ‘ other ’ n= % n= % n= % n= % n= % n= % 214 1 100.0 0 0.0 0 0.0 0 0.0 1 100.0 0 0.0 229 - - - - - - 0 0.0 0 0.0 1 100.0 231 2 66.7 0 0.0 1 33.3 - - - - - - 234 1 100.0 0 0.0 0 0.0 0 0.0 1 100.0 0 0.0 245 2 40.0 0 0.0 3 60.0 0 0.0 1 50.0 1 50.0 246 3 50.0 0 0.0 3 50.0 0 0.0 2 100.0 0 0.0 247 5 100.0 0 0.0 0 0.0 - - - - - - 260 8 66.7 0 0.0 4 33.3 0 0.0 1 100.0 0 0.0 261 3 100.0 0 0.0 0 0.0 0 0.0 4 100.0 0 0.0 262 7 87.5 0 0.0 1 12.5 0 0.0 5 100.0 0 0.0 265 1 100.0 0 0.0 0 0.0 - - - - - - 269 1 100.0 0 0.0 0 0.0 - - - - - - 271 - - - - - - 1 100.0 0 0.0 0 0.0 272 3 75.0 0 0.0 1 25.0 - - - - - - 273 2 66.7 0 0.0 1 33.3 0 0.0 1 100.0 0 0.0 275 0 0.0 0 0.0 1 100.0 - - - - - - 276 - - - - - - 0 0.0 1 100.0 0 0.0 277 22 78.6 0 0.0 6 21.4 0 0.0 4 100.0 0 0.0 278 8 80.0 0 0.0 2 20.0 0 0.0 7 100.0 0 0.0 280 20 87.0 0 0.0 3 13.0 0 0.0 3 75.0 1 25.0 282 2 100.0 0 0.0 0 0.0 - - - - - - 285 24 96.0 0 0.0 1 4.0 0 0.0 2 100.0 0 0.0 286 34 100.0 0 0.0 0 0.0 0 0.0 7 100.0 0 0.0 291 5 71.4 0 0.0 2 28.6 0 0.0 1 100.0 0 0.0 295 7 70.0 0 0.0 3 30.0 1 100.0 0 0.0 0 0.0 296 12 92.3 0 0.0 1 7.7 4 100.0 0 0.0 0 0.0 297 7 77.8 0 0.0 2 22.2 1 100.0 0 0.0 0 0.0 298 10 100.0 0 0.0 0 0.0 2 100.0 0 0.0 0 0.0 300 0 0.0 0 0.0 1 100.0 - - - - - - 301 10 100.0 0 0.0 0 0.0 - - - - - - 1000 - - - - - - 0 0.0 1 100.0 0 0.0 1100 4 100.0 0 0.0 0 0.0 0 0.0 2 66.7 1 33.3 1200 7 63.6 0 0.0 4 36.4 0 0.0 1 100.0 0 0.0 1300 13 54.2 0 0.0 11 45.8 0 0.0 2 100.0 0 0.0 192 Appendix I - Distribution of Grade Vowel per Text File and Verb Class <?page no="211"?> Text file Singular Forms Plural Forms 1st Past vowel 2nd Past vowel ‘ other ’ 1st past vowel 2nd past vowel ‘ other ’ n= % n= % n= % n= % n= % n= % 1600 34 100.0 0 0.0 0 0.0 0 0.0 10 83.3 2 16.7 1700 1 100.0 0 0.0 0 0.0 - - - - - - 1800 1 50.0 0 0.0 1 50.0 0 0.0 2 100.0 0 0.0 2000 17 81.0 0 0.0 4 19.0 0 0.0 2 100.0 0 0.0 2001 7 100.0 0 0.0 0 0.0 0 0.0 2 100.0 0 0.0 2002 10 100.0 0 0.0 0 0.0 0 0.0 2 100.0 0 0.0 Total: 361 82.6 1 0.2 75 17.2 12 11.1 87 80.6 9 8.3 Class II Text file Singular Forms Plural Forms 1st Past vowel 2nd Past vowel ‘ other ’ 1st past vowel 2nd past vowel ‘ other ’ n= % n= % n= % n= % n= % n= % 2 3 100.0 0 0.0 0 0.0 - - - - - - 4 - - - - - - 0 0.0 1 100.0 0 0.0 5 - - - - - - 0 0.0 1 100.0 0 0.0 6 - - - - - - 0 0.0 1 100.0 0 0.0 7 - - - - - - 0 0.0 1 100.0 0 0.0 8 - - - - - - 0 0.0 1 100.0 0 0.0 10 0 0.0 1 100.0 0 0.0 - - - - - - 64 3 100.0 0 0.0 0 0.0 0 0.0 1 100.0 0 0.0 65 2 100.0 0 0.0 0 0.0 - - - - - - 118 2 66.7 1 33.3 0 0.0 0 0.0 2 100.0 0 0.0 119 2 100.0 0 0.0 0 0.0 0 0.0 1 100.0 0 0.0 120 0 0.0 1 100.0 0 0.0 - - - - - - 121 3 100.0 0 0.0 0 0.0 - - - - - - 122 - - - - - - 1 100.0 0 0.0 0 0.0 123 3 75.0 1 25.0 0 0.0 0 0.0 4 100.0 0 0.0 137 4 100.0 0 0.0 0 0.0 - - - - - - 143 0 0.0 0 0.0 1 100.0 - - - - - - 149 4 100.0 0 0.0 0 0.0 0 0.0 5 100.0 0 0.0 150 1 100.0 0 0.0 0 0.0 - - - - - - Class II 193 <?page no="212"?> Text file Singular Forms Plural Forms 1st Past vowel 2nd Past vowel ‘ other ’ 1st past vowel 2nd past vowel ‘ other ’ n= % n= % n= % n= % n= % n= % 155 39 95.1 2 4.9 0 0.0 0 0.0 7 100.0 0 0.0 158 1 100.0 0 0.0 0 0.0 1 100.0 0 0.0 0 0.0 169 1 100.0 0 0.0 0 0.0 - - - - - - 170 1 100.0 0 0.0 0 0.0 0 0.0 1 100.0 0 0.0 172 2 100.0 0 0.0 0 0.0 0 0.0 2 100.0 0 0.0 173 4 100.0 0 0.0 0 0.0 - - - - - - 182 1 100.0 0 0.0 0 0.0 - - - - - - 214 1 100.0 0 0.0 0 0.0 0 0.0 1 100.0 0 0.0 220 2 100.0 0 0.0 0 0.0 - - - - - - 229 1 100.0 0 0.0 0 0.0 - - - - - - 231 0 0.0 0 0.0 1 100.0 - - - - - - 234 1 33.3 2 66.7 0 0.0 - - - - - - 244 1 100.0 0 0.0 0 0.0 - - - - - - 245 2 66.7 0 0.0 1 33.3 0 0.0 2 100.0 0 0.0 246 2 40.0 1 20.0 2 40.0 - - - - - - 247 2 100.0 0 0.0 0 0.0 1 50.0 1 50.0 0 0.0 248 1 100.0 0 0.0 0 0.0 1 100.0 0 0.0 0 0.0 260 3 60.0 2 40.0 0 0.0 0 0.0 2 100.0 0 0.0 261 9 90.0 1 10.0 0 0.0 0 0.0 3 100.0 0 0.0 262 3 100.0 0 0.0 0 0.0 0 0.0 3 100.0 0 0.0 272 2 66.7 0 0.0 1 33.3 0 0.0 2 100.0 0 0.0 273 2 66.7 0 0.0 1 33.3 0 0.0 2 100.0 0 0.0 276 2 100.0 0 0.0 0 0.0 0 0.0 3 100.0 0 0.0 277 11 64.7 0 0.0 6 35.3 1 6.7 14 93.3 0 0.0 278 8 50.0 0 0.0 8 50.0 1 10.0 8 80.0 1 10.0 280 14 100.0 0 0.0 0 0.0 0 0.0 10 100.0 0 0.0 282 3 75.0 0 0.0 1 25.0 0 0.0 1 100.0 0 0.0 285 6 75.0 2 25.0 0 0.0 6 60.0 4 40.0 0 0.0 286 8 88.9 0 0.0 1 11.1 2 28.6 5 71.4 0 0.0 291 2 66.7 0 0.0 1 33.3 0 0.0 1 100.0 0 0.0 295 4 80.0 0 0.0 1 20.0 1 100.0 0 0.0 0 0.0 296 4 100.0 0 0.0 0 0.0 1 100.0 0 0.0 0 0.0 297 1 50.0 1 50.0 0 0.0 2 100.0 0 0.0 0 0.0 298 6 100.0 0 0.0 0 0.0 1 100.0 0 0.0 0 0.0 194 Appendix I - Distribution of Grade Vowel per Text File and Verb Class <?page no="213"?> Text file Singular Forms Plural Forms 1st Past vowel 2nd Past vowel ‘ other ’ 1st past vowel 2nd past vowel ‘ other ’ n= % n= % n= % n= % n= % n= % 301 0 0.0 0 0.0 6 100.0 0 0.0 2 100.0 0 0.0 304 - - - - - - 0 0.0 1 100.0 0 0.0 1000 3 75.0 1 25.0 0 0.0 - - - - - - 1100 4 100.0 0 0.0 0 0.0 0 0.0 1 100.0 0 0.0 1200 3 75.0 1 25.0 0 0.0 - - - - - - 1300 10 90.9 0 0.0 1 9.1 - - - - - - 1400 1 100.0 0 0.0 0 0.0 0 0.0 1 100.0 0 0.0 1600 9 90.0 0 0.0 1 10.0 0 0.0 3 60.0 2 40.0 1700 15 100.0 0 0.0 0 0.0 - - - - - - 2000 4 100.0 0 0.0 0 0.0 0 0.0 6 100.0 0 0.0 2001 2 100.0 0 0.0 0 0.0 - - - - - - 2002 3 100.0 0 0.0 0 0.0 0 0.0 1 100.0 0 0.0 Total: 231 82.2 17 6.0 33 11.7 19 15.0 1- 05 82.7 3 2.4 Class IIIb Text file Singular Forms Plural Forms 1st Past vowel 2nd Past vowel ‘ other ’ 1st past vowel 2nd past vowel ‘ other ’ n= % n= % n= % n= % n= % n= % 2 2 100.0 0 0.0 0 0.0 - - - - - - 64 3 100.0 0 0.0 0 0.0 - - - - - - 65 1 100.0 0 0.0 0 0.0 - - - - - - 118 2 66.7 0 0.0 1 33.3 - - - - - - 123 1 50.0 0 0.0 1 50.0 0 0.0 1 1.0 0 0.0 124 - - - - - - 0 0.0 1 1.0 0 0.0 155 4 80.0 0 0.0 1 20.0 0 0.0 1 1.0 0 0.0 169 1 100.0 0 0.0 0 0.0 - - - - - - 181 1 100.0 0 0.0 0 0.0 - - - - - - 245 0 0.0 0 0.0 2 100.0 0 0.0 0 0.0 1 1.0 247 1 33.3 0 0.0 2 66.7 - - - - - - 260 0 0.0 0 0.0 1 100.0 0 0.0 1 1.0 0 0.0 Class IIIb 195 <?page no="214"?> Text file Singular Forms Plural Forms 1st Past vowel 2nd Past vowel ‘ other ’ 1st past vowel 2nd past vowel ‘ other ’ n= % n= % n= % n= % n= % n= % 272 2 100.0 0 0.0 0 0.0 - - - - - - 273 2 100.0 0 0.0 0 0.0 - - - - - - 276 - - - - - - 0 0.0 1 1.0 0 0.0 277 1 100.0 0 0.0 0 0.0 - - - - - - 278 - - - - - - 0 0.0 2 1.0 0 0.0 280 1 100.0 0 0.0 0 0.0 - - - - - - 286 2 66.7 0 0.0 1 33.3 0 0.0 4 1.0 0 0.0 291 1 100.0 0 0.0 0 0.0 - - - - - - 295 3 75.0 0 0.0 1 25.0 - - - - - - 296 2 100.0 0 0.0 0 0.0 2 100.0 0 0.0 0 0.0 297 0 0.0 0 0.0 1 100.0 0 0.0 0 0.0 1 1.0 298 3 75.0 0 0.0 1 25.0 1 100.0 0 0.0 0 0.0 301 2 100.0 0 0.0 0 0.0 - - - - - - 1100 2 100.0 0 0.0 0 0.0 - - - - - - 1300 2 100.0 0 0.0 0 0.0 - - - - - - 1600 7 100.0 0 0.0 0 0.0 0 0.0 2 1.0 0 0.0 2000 1 100.0 0 0.0 0 0.0 - - - - - - 2001 0 0.0 0 0.0 2 100.0 - - - - - - 2002 2 100.0 0 0.0 0 0.0 - - - - - - Total: 49 77.8 0 0.0 14 22.2 3 16.7 13 0.7 2 0.1 Class IIIc Text file Singular Forms Plural Forms 1st Past vowel 2nd Past vowel ‘ other ’ 1st past vowel 2nd past vowel ‘ other ’ n= % n= % n= % n= % n= % n= % 2 3 100.0 0 0.0 0 0.0 - - - - - - 63 1 100.0 0 0.0 0 0.0 - - - - - - 64 9 90.0 0 0.0 1 10.0 - - - - - - 65 2 100.0 0 0.0 0 0.0 - - - - - - 118 1 100.0 0 0.0 0 0.0 - - - - - - 119 6 85.7 0 0.0 1 14.3 - - - - - - 196 Appendix I - Distribution of Grade Vowel per Text File and Verb Class <?page no="215"?> Text file Singular Forms Plural Forms 1st Past vowel 2nd Past vowel ‘ other ’ 1st past vowel 2nd past vowel ‘ other ’ n= % n= % n= % n= % n= % n= % 121 1 100.0 0 0.0 0 0.0 - - - - - - 122 - - - - - - 2 100.0 0 0.0 0 0.0 123 9 100.0 0 0.0 0 0.0 1 25.0 3 75.0 0 0.0 137 2 100.0 0 0.0 0 0.0 - - - - - - 143 0 0.0 0 0.0 1 100.0 - - - - - - 149 4 50.0 1 12.5 3 37.5 0 0.0 7 100.0 0 0.0 150 0 0.0 1 100.0 0 0.0 - - - - - - 155 8 19.0 34 81.0 0 0.0 0 0.0 4 100.0 0 0.0 158 1 100.0 0 0.0 0 0.0 0 0.0 1 100.0 0 0.0 170 0 0.0 1 100.0 0 0.0 - - - - - - 171 2 100.0 0 0.0 0 0.0 - - - - - - 172 2 100.0 0 0.0 0 0.0 0 0.0 2 100.0 0 0.0 173 1 100.0 0 0.0 0 0.0 - - - - - - 214 - - - - - - 0 0.0 0 0.0 1 100.0 231 1 100.0 0 0.0 0 0.0 - - - - - - 245 0 0.0 0 0.0 1 100.0 - - - - - - 246 1 50.0 0 0.0 1 50.0 - - - - - - 247 1 33.3 0 0.0 2 66.7 - - - - - - 260 9 100.0 0 0.0 0 0.0 0 0.0 4 100.0 0 0.0 261 9 100.0 0 0.0 0 0.0 0 0.0 3 100.0 0 0.0 262 3 100.0 0 0.0 0 0.0 1 50.0 1 50.0 0 0.0 272 1 100.0 0 0.0 0 0.0 - - - - - - 273 1 100.0 0 0.0 0 0.0 - - - - - - 276 - - - - - - 0 0.0 1 100.0 0 0.0 277 9 47.4 0 0.0 10 52.6 0 0.0 7 87.5 1 12.5 278 8 44.4 1 5.6 9 50.0 0 0.0 6 54.5 5 45.5 280 14 82.4 0 0.0 3 17.6 0 0.0 4 100.0 0 0.0 282 0 0.0 0 0.0 1 100.0 0 0.0 1 100.0 0 0.0 285 6 75.0 0 0.0 2 25.0 0 0.0 1 100.0 0 0.0 286 3 75.0 0 0.0 1 25.0 0 0.0 1 50.0 1 50.0 291 - - - - - - 0 0.0 2 100.0 0 0.0 295 3 75.0 0 0.0 1 25.0 1 100.0 0 0.0 0 0.0 296 2 100.0 0 0.0 0 0.0 - - - - - - 297 1 100.0 0 0.0 0 0.0 1 100.0 0 0.0 0 0.0 Class IIIc 197 <?page no="216"?> Text file Singular Forms Plural Forms 1st Past vowel 2nd Past vowel ‘ other ’ 1st past vowel 2nd past vowel ‘ other ’ n= % n= % n= % n= % n= % n= % 301 7 100.0 0 0.0 0 0.0 0 0.0 2 100.0 0 0.0 304 - - - - - - 0 0.0 1 100.0 0 0.0 1000 1 100.0 0 0.0 0 0.0 - - - - - - 1100 3 75.0 0 0.0 1 25.0 - - - - - - 1200 4 100.0 0 0.0 0 0.0 0 0.0 1 100.0 0 0.0 1300 11 84.6 0 0.0 2 15.4 - - - - - - 1600 3 75.0 0 0.0 1 25.0 - - - - - - 1800 0 0.0 0 0.0 1 100.0 - - - - - - 2000 5 62.5 0 0.0 3 37.5 0 0.0 0 0.0 2 100.0 2001 0 0.0 0 0.0 1 100.0 - - - - - - 2002 0 0.0 0 0.0 1 100.0 0 0.0 1 100.0 0 0.0 Total: 158 65.0 38 15.6 47 19.3 6 8.7 53 76.8 10 14.5 Class IVa Text file Singular Forms Plural Forms 1st Past vowel 2nd Past vowel ‘ other ’ 1st past vowel 2nd past vowel ‘ other ’ n= % n= % n= % n= % n= % n= % 2 - - - - - - 0 0.0 1 100.0 0 0.0 3 1 100.0 0 0.0 0 0.0 - - - - - - 4 1 100.0 0 0.0 0 0.0 2 100.0 0 0.0 0 0.0 5 0 0.0 1 100.0 0 0.0 0 0.0 2 100.0 0 0.0 6 0 0.0 1 100.0 0 0.0 0 0.0 2 100.0 0 0.0 7 1 100.0 0 0.0 0 0.0 0 0.0 2 100.0 0 0.0 8 0 0.0 1 100.0 0 0.0 0 0.0 2 100.0 0 0.0 10 1 50.0 1 50.0 0 0.0 0 0.0 1 100.0 0 0.0 64 4 80.0 1 20.0 0 0.0 1 50.0 1 50.0 0 0.0 65 2 100.0 0 0.0 0 0.0 - - - - - - 118 1 100.0 0 0.0 0 0.0 - - - - - - 119 0 0.0 1 50.0 1 50.0 0 0.0 1 100.0 0 0.0 123 0 0.0 1 100.0 0 0.0 0 0.0 1 100.0 0 0.0 136 - - - - - - 0 0.0 1 100.0 0 0.0 198 Appendix I - Distribution of Grade Vowel per Text File and Verb Class <?page no="217"?> Text file Singular Forms Plural Forms 1st Past vowel 2nd Past vowel ‘ other ’ 1st past vowel 2nd past vowel ‘ other ’ n= % n= % n= % n= % n= % n= % 137 - - - - - - 0 0.0 1 100.0 0 0.0 138 - - - - - - 1 100.0 0 0.0 0 0.0 149 3 75.0 0 0.0 1 25.0 3 75.0 0 0.0 1 0.3 150 1 100.0 0 0.0 0 0.0 1 100.0 0 0.0 0 0.0 155 28 96.6 1 3.4 0 0.0 3 75.0 1 25.0 0 0.0 158 3 75.0 1 25.0 0 0.0 - - - - - - 160 - - - - - - 0 0.0 2 100.0 0 0.0 169 1 100.0 0 0.0 0 0.0 1 100.0 0 0.0 0 0.0 175 - - - - - - 0 0.0 1 100.0 0 0.0 179 - - - - - - 0 0.0 1 100.0 0 0.0 182 2 100.0 0 0.0 0 0.0 - - - - - - 214 - - - - - - 0 0.0 1 100.0 0 0.0 229 - - - - - - 0 0.0 1 100.0 0 0.0 231 1 100.0 0 0.0 0 0.0 - - - - - - 234 2 100.0 0 0.0 0 0.0 - - - - - - 235 - - - - - - 0 0.0 1 100.0 0 0.0 241 - - - - - - 0 0.0 1 100.0 0 0.0 245 0 0.0 1 100.0 0 0.0 - - - - - - 246 1 100.0 0 0.0 0 0.0 0 0.0 2 100.0 0 0.0 247 0 0.0 1 100.0 0 0.0 0 0.0 4 100.0 0 0.0 259 0 0.0 1 100.0 0 0.0 - - - - - - 260 0 0.0 2 100.0 0 0.0 0 0.0 1 100.0 0 0.0 261 0 0.0 1 100.0 0 0.0 0 0.0 4 100.0 0 0.0 262 0 0.0 3 100.0 0 0.0 0 0.0 3 100.0 0 0.0 272 0 0.0 1 100.0 0 0.0 - - - - - - 273 0 0.0 1 100.0 0 0.0 - - - - - - 276 - - - - - - 0 0.0 1 100.0 0 0.0 277 4 44.4 2 22.2 3 33.3 0 0.0 1 100.0 0 0.0 278 2 33.3 1 16.7 3 50.0 1 16.7 5 83.3 0 0.0 280 9 90.0 1 10.0 0 0.0 0 0.0 2 100.0 0 0.0 282 3 100.0 0 0.0 0 0.0 2 50.0 2 50.0 0 0.0 285 14 100.0 0 0.0 0 0.0 5 100.0 0 0.0 0 0.0 286 18 90.0 2 10.0 0 0.0 0 0.0 16 100.0 0 0.0 291 0 0.0 1 100.0 0 0.0 0 0.0 3 100.0 0 0.0 Class IVa 199 <?page no="218"?> Text file Singular Forms Plural Forms 1st Past vowel 2nd Past vowel ‘ other ’ 1st past vowel 2nd past vowel ‘ other ’ n= % n= % n= % n= % n= % n= % 295 8 100.0 0 0.0 0 0.0 3 100.0 0 0.0 0 0.0 296 14 93.3 0 0.0 1 6.7 2 100.0 0 0.0 0 0.0 297 11 100.0 0 0.0 0 0.0 4 100.0 0 0.0 0 0.0 298 12 92.3 1 7.7 0 0.0 4 100.0 0 0.0 0 0.0 301 4 100.0 0 0.0 0 0.0 - - - - - - 1100 1 50.0 1 50.0 0 0.0 0 0.0 3 100.0 0 0.0 1200 2 100.0 0 0.0 0 0.0 - - - - - - 1300 7 100.0 0 0.0 0 0.0 0 0.0 5 100.0 0 0.0 1600 22 91.7 2 8.3 0 0.0 0 0.0 24 100.0 0 0.0 1800 - - - - - - 0 0.0 3 100.0 0 0.0 2000 0 0.0 7 100.0 0 0.0 0 0.0 1 100.0 0 0.0 2001 0 0.0 2 100.0 0 0.0 0 0.0 1 100.0 0 0.0 2002 5 62.5 3 37.5 0 0.0 0 0.0 3 100.0 0 0.0 Total: 189 78.4 43 17.8 9 3.7 33 23.2 108 76.1 1 0.0 Class V Text file Singular Forms Plural Forms 1st Past vowel 2nd Past vowel ‘ other ’ 1st past vowel 2nd past vowel ‘ other ’ n= % n= % n= % n= % n= % n= % 2 17 81.0 4 19.0 0 0.0 0 0.0 3 100.0 0 0.0 3 15 88.2 2 11.8 0 0.0 - - - - - - 4 1 50.0 1 50.0 0 0.0 0 0.0 3 100.0 0 0.0 5 0 0.0 1 100.0 0 0.0 0 0.0 2 100.0 0 0.0 6 0 0.0 1 100.0 0 0.0 0 0.0 2 100.0 0 0.0 7 0 0.0 1 100.0 0 0.0 0 0.0 2 100.0 0 0.0 8 0 0.0 2 100.0 0 0.0 0 0.0 2 100.0 0 0.0 10 - - - - - - 0 0.0 2 100.0 0 0.0 63 1 100.0 0 0.0 0 0.0 - - - - - - 64 26 92.9 2 7.1 0 0.0 6 75.0 2 25.0 0 0.0 65 4 66.7 1 16.7 1 16.7 2 50.0 2 50.0 0 0.0 118 2 9.1 18 81.8 2 9.1 0 0.0 2 100.0 0 0.0 200 Appendix I - Distribution of Grade Vowel per Text File and Verb Class <?page no="219"?> Text file Singular Forms Plural Forms 1st Past vowel 2nd Past vowel ‘ other ’ 1st past vowel 2nd past vowel ‘ other ’ n= % n= % n= % n= % n= % n= % 119 0 0.0 13 52.0 12 48.0 0 0.0 2 100.0 0 0.0 120 0 0.0 13 48.1 14 51.9 0 0.0 2 100.0 0 0.0 121 0 0.0 4 50.0 4 50.0 0 0.0 1 100.0 0 0.0 122 0 0.0 1 100.0 0 0.0 0 0.0 8 100.0 0 0.0 123 0 0.0 24 68.6 11 31.4 0 0.0 8 80.0 2 20.0 128 0 0.0 0 0.0 1 100.0 - - - - - - 137 3 50.0 3 50.0 0 0.0 1 33.3 2 66.7 0 0.0 139 6 66.7 0 0.0 3 33.3 - - - - - - 142 4 100.0 0 0.0 0 0.0 0 0.0 4 100.0 0 0.0 143 0 0.0 5 71.4 2 28.6 - - - - - - 147 0 0.0 6 85.7 1 14.3 - - - - - - 149 8 66.7 0 0.0 4 33.3 5 83.3 0 0.0 1 16.7 150 2 100.0 0 0.0 0 0.0 1 100.0 0 0.0 0 0.0 155 111 99.1 1 0.9 0 0.0 0 0.0 2 66.7 1 33.3 158 25 92.6 0 0.0 2 7.4 0 0.0 6 100.0 0 0.0 160 2 100.0 0 0.0 0 0.0 1 100.0 0 0.0 0 0.0 169 3 75.0 1 25.0 0 0.0 - - - - - - 170 1 50.0 1 50.0 0 0.0 0 0.0 1 100.0 0 0.0 171 0 0.0 0 0.0 1 100.0 - - - - - - 172 - - - - - - 0 0.0 6 100.0 0 0.0 173 0 0.0 13 86.7 2 13.3 0 0.0 10 83.3 2 16.7 174 0 0.0 3 100.0 0 0.0 - - - - - - 175 - - - - - - 0 0.0 1 100.0 0 0.0 178 - - - - - - 0 0.0 1 100.0 0 0.0 182 10 100.0 0 0.0 0 0.0 2 100.0 0 0.0 0 0.0 184 1 100.0 0 0.0 0 0.0 - - - - - - 188 4 100.0 0 0.0 0 0.0 2 100.0 0 0.0 0 0.0 214 1 100.0 0 0.0 0 0.0 0 0.0 1 100.0 0 0.0 218 45 97.8 0 0.0 1 2.2 - - - - - - 220 3 75.0 1 25.0 0 0.0 - - - - - - 222 5 83.3 1 16.7 0 0.0 - - - - - - 228 1 100.0 0 0.0 0 0.0 - - - - - - 232 0 0.0 1 100.0 0 0.0 - - - - - - 234 - - - - - - 0 0.0 1 100.0 0 0.0 Class V 201 <?page no="220"?> Text file Singular Forms Plural Forms 1st Past vowel 2nd Past vowel ‘ other ’ 1st past vowel 2nd past vowel ‘ other ’ n= % n= % n= % n= % n= % n= % 244 0 0.0 1 100.0 0 0.0 - - - - - - 245 0 0.0 17 100.0 0 0.0 0 0.0 6 100.0 0 0.0 246 9 37.5 12 50.0 3 12.5 0 0.0 3 100.0 0 0.0 247 4 14.8 23 85.2 0 0.0 0 0.0 9 100.0 0 0.0 249 31 91.2 0 0.0 3 8.8 0 0.0 2 100.0 0 0.0 259 1 100.0 0 0.0 0 0.0 - - - - - - 260 5 9.8 35 68.6 11 21.6 0 0.0 11 100.0 0 0.0 261 29 56.9 19 37.3 3 5.9 0 0.0 4 100.0 0 0.0 262 8 26.7 13 43.3 9 30.0 0 0.0 2 100.0 0 0.0 270 2 100.0 0 0.0 0 0.0 - - - - - - 271 9 75.0 3 25.0 0 0.0 0 0.0 2 100.0 0 0.0 272 0 0.0 18 90.0 2 10.0 0 0.0 5 100.0 0 0.0 273 1 6.3 15 93.8 0 0.0 0 0.0 5 100.0 0 0.0 275 - - - - - - 0 0.0 1 100.0 0 0.0 276 0 0.0 14 100.0 0 0.0 0 0.0 1 100.0 0 0.0 277 15 30.0 24 48.0 11 22.0 0 0.0 12 100.0 0 0.0 278 8 18.6 17 39.5 18 41.9 1 3.1 29 90.6 2 6.3 280 35 68.6 15 29.4 1 2.0 0 0.0 16 100.0 0 0.0 282 19 86.4 3 13.6 0 0.0 2 40.0 3 60.0 0 0.0 285 126 96.2 3 2.3 2 1.5 14 51.9 12 44.4 1 3.7 286 199 75.4 64 24.2 1 0.4 3 3.8 75 96.2 0 0.0 291 2 9.1 20 90.9 0 0.0 0 0.0 5 100.0 0 0.0 295 52 92.9 3 5.4 1 1.8 4 80.0 1 20.0 0 0.0 296 43 91.5 4 8.5 0 0.0 29 96.7 1 3.3 0 0.0 297 25 80.6 6 19.4 0 0.0 16 94.1 0 0.0 1 5.9 298 87 93.5 6 6.5 0 0.0 19 82.6 4 17.4 0 0.0 301 15 100.0 0 0.0 0 0.0 1 33.3 0 0.0 2 66.7 1000 0 0.0 15 51.7 14 48.3 0 0.0 3 100.0 0 0.0 1100 27 64.3 15 35.7 0 0.0 0 0.0 5 100.0 0 0.0 1200 7 50.0 7 50.0 0 0.0 0 0.0 4 100.0 0 0.0 1300 35 50.0 31 44.3 4 5.7 0 0.0 18 100.0 0 0.0 1400 9 81.8 1 9.1 1 9.1 0 0.0 3 100.0 0 0.0 1600 166 91.2 14 7.7 2 1.1 2 3.1 60 93.8 2 3.1 1700 6 85.7 0 0.0 1 14.3 - - - - - - 202 Appendix I - Distribution of Grade Vowel per Text File and Verb Class <?page no="221"?> Text file Singular Forms Plural Forms 1st Past vowel 2nd Past vowel ‘ other ’ 1st past vowel 2nd past vowel ‘ other ’ n= % n= % n= % n= % n= % n= % 1800 0 0.0 3 75.0 1 25.0 0 0.0 2 100.0 0 0.0 2000 1 1.7 49 81.7 10 16.7 0 0.0 12 100.0 0 0.0 2001 0 0.0 8 100.0 0 0.0 0 0.0 2 100.0 0 0.0 2002 46 86.8 7 13.2 0 0.0 0 0.0 14 100.0 0 0.0 Total: 1323 63.3 609 29.1 159 7.6 111 20.7 410 76.6 14 2.6 Class V 203 <?page no="222"?> Appendix II - List of Lexels with OE a ˉ in LAEME The allocation into the three word classes is based on the respective tag in the grammel. Non-weak verbs only include infinitive ( ‘ vi ’ ) and present tense ( ‘ vps ’ ) forms. Nouns Lexel Types Tokens Texts A : D 2 2 2 A : DL 1 1 1 A : GENSLAGA {*} 1 1 1 A : NLE : PIGHOOD 1 1 1 A : NRAE : DNESS 4 4 4 A : NSAE : TA {*} 1 1 1 A : R 90 181 41 A : RFAESTNESS 1 1 1 A : RFULNESS {*} 1 1 1 A : Y EXE 1 1 1 A : Y UM 7 7 5 ABODE 5 6 4 BA : RA 1 1 1 BA : T 1 1 1 BEHA : T 3 3 2 BLA : C 3 5 1 BLA : R 1 1 1 BLOWER 1 1 1 BOAR 20 23 11 BOARSPEAR 1 1 1 BOAT 4 5 4 BONE 87 105 36 BRA : Y 1 1 1 BROADLOOKER 1 1 1 CA : F 2 2 2 CA : FARISE 1 1 1 CAPE 9 9 7 Lexel Types Tokens Texts CLA : TE 1 1 1 CLOTH 51 79 24 CLOTHES 78 105 38 CROW 7 9 4 CROWLEEK 1 1 1 DA : G 1 1 1 DOE 1 1 1 DRA : F 1 1 1 FA : CEN 1 1 1 FA : H 13 13 9 FLA : 18 22 9 FLA : N 2 3 2 FOAM 4 4 4 FOE 126 161 59 FOEMAN 12 14 8 FOERAE : DEN 1 1 1 FOESHIP 1 1 1 FORETOKEN 3 5 2 GA : LHOOD 1 1 1 GA : LNESS 24 33 11 GA : LSHIP 2 2 1 GA : R 5 6 3 GA : RA 8 9 6 GA : RABEORG 1 1 1 GA : RCLIFE 1 1 1 GEDA : L 2 2 1 GEMA : HNESS 1 1 1 <?page no="223"?> Lexel Types Tokens Texts GEMA : NA 4 4 4 GERA : D 7 14 2 GHOST 261 459 68 GNA : ST 2 2 1 GOAD 3 3 2 GOAT 11 11 8 GOATHEORD 4 4 4 GOATTICCEN 3 3 3 GO - BY - DITCH 1 1 1 GO - BY - GROUND 1 1 1 GO - MID - LAMB 1 1 1 GRA : 1 1 1 GRA : P 2 2 2 GRA : R 1 1 1 GRA : TR 2 2 1 HA : D 30 89 14 HA : LEWAE : GE 22 22 14 HA : LGA 63 88 33 HA : LWENDA 1 1 1 HA : LWENDE 1 1 1 HA : M 1 1 1 HA : MSO : CN 12 12 6 HA : T 1 1 1 HA : WERE 2 2 1 HALLOWER 1 1 1 HANACRA : W 2 2 1 HE : LAWA : G 1 1 1 HLA : W 5 5 4 HOAR 1 1 1 HOLIDOM 6 6 4 HOLIHOOD 2 2 2 HOLINESS 30 37 19 HOLYCHURCH 3 3 3 HOLYDAY 19 23 14 HOLYGHOST 173 268 47 HOLYMAN 16 43 4 HOLYNIGHT 3 3 3 Lexel Types Tokens Texts HOLYPOPE 2 2 1 HOLYROOD 3 3 3 HOLYROODDAY 3 3 3 HOLYWATER 6 6 5 HOLYWEEK 1 1 1 HOME 78 141 40 HOMECOMING 2 3 2 HOTHEARTNESS 2 2 1 INLA : TE 1 1 1 INLOAD 2 3 2 KNOWLEDGE 2 2 2 KNOWNESS 1 1 1 LA : C 57 110 24 LA : F 7 8 4 LA : M 13 21 9 LA : N 17 17 12 LA : RE : OW 4 4 2 LA : T 24 28 16 LA : TTE : OW 6 6 4 LI : FLA : D 28 68 14 LOAD 3 3 3 LOAF 16 27 8 LORD 583 1795 87 LORDDOM 3 4 2 LORDFLOCK 1 2 1 LORDHOOD 8 8 4 LORDING 16 19 12 LORDSHIP 5 6 4 LORE 118 226 61 LORESPEL 11 12 6 LORETHANE 1 1 1 LORE Y E : OW 27 32 5 LOTH 24 25 20 LOTHLESSNESS 6 7 2 LOTHNESS 1 1 1 LOWNESS 1 1 1 MA : D M 9 14 5 Nouns 205 <?page no="224"?> Lexel Types Tokens Texts MA : D MHOUSE 1 1 1 MA : GE 20 21 15 MA : GR 1 1 1 MA : L 3 3 3 MA : N 4 5 4 MA : N { O } 1 1 1 MA : NDEED 1 1 1 MA : NOATH 2 2 2 MA : NSWARA 1 1 1 MA : NSWARU 1 1 1 MISLORE 2 2 2 NA : M 1 1 1 NAFOGA : R 1 1 1 OAK 5 5 3 OAKHYRST 1 1 1 OAKSTYBB 1 1 1 OAR 3 4 1 OAT 6 7 2 OATH 49 69 24 OATHSWARU 1 1 1 ORE 2 2 2 OUTLOAD 2 3 2 OWNER 2 2 2 OWNHI : GAN 1 1 1 OWNWILL 4 4 2 POLE 1 1 1 POPE 10 10 9 RA : D 7 7 5 ROAD 2 2 2 ROADHORSE 1 1 1 ROE 6 6 5 ROEDEER 1 1 1 ROPE 23 28 14 SA : 1 1 1 SA : GOL 1 1 1 SLOE 5 5 4 SLOTH 11 13 8 Lexel Types Tokens Texts SLOWWORM 2 2 1 SNOW 19 19 13 SNOWWATER 2 2 2 SOAP 5 5 5 SOAPER 1 1 1 SORE 81 105 42 SORENESS 1 1 1 SORRINESS 23 36 13 SOUL 323 827 76 SOULHOUSE 1 1 1 SOULMURDER 1 1 1 SOWER 2 3 2 SPA : LD 1 1 1 SPA : TL 4 5 4 SPOKE 4 4 4 STONE 195 270 60 STONEBUCK 1 1 1 STONECLU : D 1 1 1 STONEKIN 1 1 1 STONEROCK 1 1 1 STONESTREET 1 1 1 THROE 9 9 7 TOAD 10 11 8 TOE 8 11 5 TOKEN 61 78 28 TOKENHOOD 1 1 1 TROTEVA : LE 1 1 1 VA : D I 5 6 3 VA : N { D } 19 20 14 VA : N { H } 11 14 9 VA : N { P } 6 12 5 VA : NDR 4 4 4 WA : G 30 34 23 WA : GRIFT 3 6 2 WA : R 1 1 1 WEAKNESS 9 9 7 WHOLESHIP 5 6 4 206 Appendix II - List of Lexels with OE a ˉ in LAEME <?page no="225"?> Lexel Types Tokens Texts WHOLESOME- NESS 1 1 1 WITHYLA : CU 1 1 1 WOAD 4 4 3 WOE 175 325 68 WOESI : Y 9 11 4 WRA : 1 1 1 Lexel Types Tokens Texts WRA : H 1 1 1 WROTH 72 113 35 WROTHERHAE : LU 11 13 8 WROTHERTIME 1 1 1 Y RA : G 21 23 15 Total 3744 6802 131 Adjectives Lexel Types Tokens Texts A : DLIG 1 1 1 A : HT 17 21 7 A : NLE : PIG 36 40 24 A : NMO : D 2 2 2 A : NRAE : D 3 3 2 A : NWILLE 7 7 6 A : RFAEST 3 3 1 A : RFAESTLI : C 1 1 1 A : RFUL 2 2 2 A : RLE : AS 12 12 10 A : RWORTH 1 1 1 ALONE 147 262 51 BLA : C 15 16 13 BLA : C { B } 1 1 1 BLA : R 5 5 5 BONELESS 3 3 3 BOTH 51 59 28 BRA : D 2 2 2 BRA : D LIGR 1 1 1 BRA : DLUNGA 1 1 1 BROAD 56 63 33 CA : F 3 4 3 FA : CENLI : C 2 2 1 FA : CNE 2 2 1 FA : D 1 1 1 FA : H 12 13 10 Lexel Types Tokens Texts FRA : - R 1 1 1 GA : L 6 8 5 GA : LLI : C 6 8 3 GEMA : H 1 1 1 GHOSTLY 98 124 28 GOLDFA : H 1 1 1 HA : D FUL 1 1 1 HA : LWENDE 16 16 8 HA : S 3 3 3 HOAR 8 12 5 HOLY 349 1120 73 HOLYMAN 1 1 1 HOLYROOD 1 1 1 HOT 50 65 35 KNOWN 1 1 1 LA : MI 2 2 2 LORDLY 1 1 1 LORELESS 1 1 1 LOTH 174 237 47 LOTHLESS 2 3 2 LOTHLY 56 63 30 LOW 59 67 32 MA : NFUL 3 3 1 MA : NKEEN 1 1 1 MISWHOLE 1 1 1 MORE 366 544 69 Adjectives 207 <?page no="226"?> Lexel Types Tokens Texts OWN 355 566 75 OWNINGLY 2 2 1 SA : KED 1 1 1 SA : RCURNE 1 1 1 SLA : POL 1 1 1 SLOTH 1 1 1 SLOW 13 13 11 SNOWBRIGHT 1 1 1 SNOWWHITE 5 5 5 SNOWY 2 2 2 SORE 74 91 46 SORRILY 1 1 1 SORRY 133 195 51 SORRYMO : D 7 7 6 SOULLESS 1 1 1 STONE 11 11 7 Lexel Types Tokens Texts STONEDEAD 2 2 2 STONY 4 4 4 THROUGHHOLY 1 1 1 UNSA : KED 1 1 1 UNWHOLE 18 20 11 VA : NLESS 2 2 2 WA : GLESS 2 2 2 WA : RIG 10 10 9 WEAK 48 59 26 WEAKLY 1 1 1 WHOLE 105 136 41 WHOLEHEADED 1 1 1 WHOLESOME 5 5 4 WOE 19 20 17 WOEFUL 6 8 4 Y RA : R 3 4 3 Total 2433 3989 120 Verbs L EXEL Types Tokens Texts A : DLIAN 1 2 1 A : GAN 234 358 68 A : GANNOT 8 8 6 A : GNIAN 7 7 5 A : NCENNED { B } 2 2 2 A : NIAN {*} 3 4 1 A : RIAN 2 2 2 A : RWEOR Y IAN 1 1 1 AETLA : Y IAN 1 1 1 ALA : Y IAN 3 3 3 AMA : NSIAN 16 16 8 AMA : NSUMIAN 1 1 1 ASA : KEN 1 1 1 ATGO 4 4 4 BEGA : LEN 1 1 1 BEGO 3 3 3 L EXEL Types Tokens Texts BEHA : TAN 62 73 34 BEKNOW 5 5 5 BELA : DIAN 2 2 1 BETOKEN 106 182 23 BLA : CIAN 2 2 1 BLA : CNEN 1 1 1 BLOW 47 48 33 BRA : DIAN 2 2 2 CLOTHE 30 32 17 CROW 2 2 2 CWA : NIAN 1 1 1 FA : GIAN 1 1 1 FORGO 24 24 20 FORHA : TAN 2 2 2 FORTHGO 1 1 1 GEA : NLAE : CAN 2 2 1 208 Appendix II - List of Lexels with OE a ˉ in LAEME <?page no="227"?> L EXEL Types Tokens Texts GO 271 551 72 GRA : PIAN 9 9 9 GRA : TA 3 3 2 GROAN 12 12 9 HA : DIAN 21 31 10 HA : LGIAN 37 40 25 HA : LIAN 1 1 1 HA : LSIAN 13 16 12 HA : RIAN 4 4 2 HA : SIAN 1 1 1 HA : TAN { C } 50 84 32 HA : TAN { N } 52 79 32 HA : WIAN 1 1 1 ILA : TET 3 3 3 KNOW 157 195 53 LA : KEN 7 27 1 LA : Y IAN 8 8 7 MA : NSWEAR 1 1 1 MISGO 1 1 1 NA : KENEN 1 1 1 OFGO 6 6 3 ONKNOW 1 1 1 OUTGO 2 2 2 OVERGO 32 34 20 L EXEL Types Tokens Texts RA : D A 2 2 1 ROAR 8 8 6 SLAY 38 56 30 SLOW 3 3 3 SMA : KEN 2 4 1 SOW 32 44 16 SPA : TLIAN 2 2 2 STONE 3 3 2 STONECAST 2 3 1 THROW 7 10 7 THROW { TW } 1 1 1 TOKEN 109 162 27 TOKNOW 2 2 1 TOTHROW 1 1 1 TOWA : WAN 1 1 1 UNA : NIAN 2 2 1 UNDERGO 3 3 2 UNGO 1 1 1 UNHA : DIAN 2 2 2 UPGO 1 1 1 WA : FIAN 1 1 1 WA : NIAN 42 43 25 WEAKEN 1 1 1 YMBEGO 1 1 1 Total 1539 2262 110 Verbs 209 <?page no="228"?> Appendix III - Distribution of <a> and <o> per Text File Text file Mapping <a> <o> other Total n= % n= % n= % n= 2 onlyo 3 3.4 86 96.6 0 0.0 89 3 moreo 21 26.6 55 69.6 3 3.8 79 4 moreo 4 5.3 72 94.7 0 0.0 76 5 onlya 50 96.2 2 3.8 0 0.0 52 6 morea 63 94.0 4 6.0 0 0.0 67 7 moreo 19 24.4 59 75.6 0 0.0 78 8 moreo 7 9.2 68 89.5 1 1.3 76 10 moreo 5 7.8 59 92.2 0 0.0 64 11 other 0 0.0 2 25.0 6 75.0 8 12 other 0 0.0 1 12.5 7 87.5 8 13 other 0 0.0 0 0.0 1 100.0 1 14 onlyo 0 0.0 1 100.0 0 0.0 1 15 onlyo 0 0.0 4 100.0 0 0.0 4 16 onlya 1 100.0 0 0.0 0 0.0 1 17 onlyo 0 0.0 3 100.0 0 0.0 3 18 onlyo 0 0.0 1 100.0 0 0.0 1 63 onlyo 0 0.0 8 100.0 0 0.0 8 64 morea 470 82.5 97 17.0 3 0.5 570 65 morea 105 91.3 9 7.8 1 0.9 115 66 moreo 4 15.4 22 84.6 0 0.0 26 67 onlyo 0 0.0 6 100.0 0 0.0 6 118 onlya 284 99.6 0 0.0 1 0.4 285 119 onlya 160 98.8 2 1.2 0 0.0 162 120 onlya 89 97.8 0 0.0 2 2.2 91 121 onlya 157 100.0 0 0.0 0 0.0 157 122 onlya 61 98.4 0 0.0 1 1.6 62 123 onlya 157 99.4 0 0.0 1 0.6 158 124 mixed 4 40.0 2 20.0 4 40.0 10 125 onlyo 0 0.0 1 100.0 0 0.0 1 129 onlya 2 100.0 0 0.0 0 0.0 2 131 onlya 5 100.0 0 0.0 0 0.0 5 133 onlya 34 100.0 0 0.0 0 0.0 34 <?page no="229"?> Text file Mapping <a> <o> other Total n= % n= % n= % n= 135 onlya 17 100.0 0 0.0 0 0.0 17 136 moreo 4 26.7 11 73.3 0 0.0 15 137 moreo 3 12.0 22 88.0 0 0.0 25 138 onlyo 0 0.0 4 100.0 0 0.0 4 139 other 0 0.0 3 42.9 4 57.1 7 140 moreo 4 14.8 23 85.2 0 0.0 27 141 onlyo 0 0.0 6 100.0 0 0.0 6 142 onlyo 4 4.2 92 95.8 0 0.0 96 143 onlya 13 100.0 0 0.0 0 0.0 13 146 moreo 1 12.5 7 87.5 0 0.0 8 147 moreo 2 18.2 9 81.8 0 0.0 11 148 mixed 1 50.0 1 50.0 0 0.0 2 149 onlya 18 100.0 0 0.0 0 0.0 18 150 moreo 11 12.9 73 85.9 1 1.2 85 151 onlyo 0 0.0 9 100.0 0 0.0 9 155 moreo 13 8.5 130 85.0 10 6.5 153 157 morea 4 66.7 0 0.0 2 33.3 6 158 onlyo 1 1.6 62 98.4 0 0.0 63 159 mixed 3 60.0 2 40.0 0 0.0 5 160 moreo 5 16.7 25 83.3 0 0.0 30 161 moreo 5 11.6 38 88.4 0 0.0 43 162 moreo 2 28.6 5 71.4 0 0.0 7 163 onlyo 0 0.0 3 100.0 0 0.0 3 169 mixed 21 50.0 20 47.6 1 2.4 42 170 morea 12 92.3 1 7.7 0 0.0 13 171 moreo 2 33.3 4 66.7 0 0.0 6 172 moreo 4 5.1 69 88.5 5 6.4 78 173 moreo 62 15.5 334 83.5 4 1.0 400 174 onlyo 0 0.0 3 100.0 0 0.0 3 175 other 0 0.0 0 0.0 2 100.0 2 176 onlyo 0 0.0 8 100.0 0 0.0 8 177 mixed 1 50.0 1 50.0 0 0.0 2 178 moreo 0 0.0 4 66.7 2 33.3 6 179 morea 2 66.7 1 33.3 0 0.0 3 180 mixed 1 50.0 1 50.0 0 0.0 2 Appendix III - Distribution of <a> and <o> per Text File 211 <?page no="230"?> Text file Mapping <a> <o> other Total n= % n= % n= % n= 181 onlyo 0 0.0 1 100.0 0 0.0 1 182 moreo 5 9.4 48 90.6 0 0.0 53 183 onlya 7 100.0 0 0.0 0 0.0 7 184 morea 8 80.0 2 20.0 0 0.0 10 185 morea 3 75.0 1 25.0 0 0.0 4 188 onlya 36 97.3 0 0.0 1 2.7 37 189 mixed 15 51.7 13 44.8 1 3.4 29 214 moreo 4 6.9 54 93.1 0 0.0 58 218 onlyo 0 0.0 51 100.0 0 0.0 51 220 moreo 3 5.8 48 92.3 1 1.9 52 222 onlyo 0 0.0 26 100.0 0 0.0 26 227 moreo 1 33.3 2 66.7 0 0.0 3 228 onlyo 0 0.0 8 100.0 0 0.0 8 229 onlyo 0 0.0 20 100.0 0 0.0 20 230 onlya 5 100.0 0 0.0 0 0.0 5 231 morea 12 85.7 0 0.0 2 14.3 14 232 moreo 1 25.0 3 75.0 0 0.0 4 233 onlyo 0 0.0 1 100.0 0 0.0 1 234 moreo 0 0.0 5 83.3 1 16.7 6 235 onlyo 0 0.0 1 100.0 0 0.0 1 236 onlyo 0 0.0 3 100.0 0 0.0 3 238 onlyo 0 0.0 6 100.0 0 0.0 6 239 onlyo 0 0.0 9 100.0 0 0.0 9 241 mixed 4 50.0 3 37.5 1 12.5 8 242 mixed 11 35.5 20 64.5 0 0.0 31 243 onlyo 0 0.0 2 100.0 0 0.0 2 244 moreo 1 33.3 2 66.7 0 0.0 3 245 moreo 21 7.3 261 91.3 4 1.4 286 246 moreo 7 4.9 132 92.3 4 2.8 143 247 onlyo 0 0.0 109 99.1 1 0.9 110 248 moreo 9 29.0 22 71.0 0 0.0 31 249 moreo 7 11.9 51 86.4 1 1.7 59 258 onlyo 0 0.0 2 100.0 0 0.0 2 259 onlyo 0 0.0 2 100.0 0 0.0 2 260 onlya 226 99.6 1 0.4 0 0.0 227 212 Appendix III - Distribution of <a> and <o> per Text File <?page no="231"?> Text file Mapping <a> <o> other Total n= % n= % n= % n= 261 onlya 153 100.0 0 0.0 0 0.0 153 262 onlya 121 100.0 0 0.0 0 0.0 121 263 onlyo 0 0.0 11 100.0 0 0.0 11 264 onlyo 0 0.0 7 100.0 0 0.0 7 265 moreo 2 20.0 8 80.0 0 0.0 10 266 onlyo 0 0.0 3 100.0 0 0.0 3 267 onlyo 0 0.0 1 100.0 0 0.0 1 268 onlya 1 100.0 0 0.0 0 0.0 1 269 onlyo 0 0.0 6 100.0 0 0.0 6 270 onlyo 0 0.0 1 100.0 0 0.0 1 271 onlyo 0 0.0 23 95.8 1 4.2 24 272 onlya 299 98.4 0 0.0 5 1.6 304 273 onlya 266 98.9 2 0.7 1 0.4 269 274 onlyo 0 0.0 1 100.0 0 0.0 1 275 morea 16 94.1 1 5.9 0 0.0 17 276 morea 171 94.5 7 3.9 3 1.7 181 277 morea 120 75.5 20 12.6 19 11.9 159 278 morea 123 72.8 15 8.9 31 18.3 169 279 moreo 2 12.5 14 87.5 0 0.0 16 280 onlyo 7 3.9 174 96.1 0 0.0 181 282 moreo 4 5.9 62 91.2 2 2.9 68 285 moreo 41 13.4 263 85.9 2 0.7 306 286 onlyo 13 2.2 580 97.8 0 0.0 593 291 moreo 99 21.9 327 72.2 27 6.0 453 292 onlyo 0 0.0 1 100.0 0 0.0 1 294 onlyo 0 0.0 2 100.0 0 0.0 2 295 morea 172 94.0 9 4.9 2 1.1 183 296 onlya 238 99.2 1 0.4 1 0.4 240 297 onlya 248 98.4 3 1.2 1 0.4 252 298 onlya 378 99.5 0 0.0 2 0.5 380 300 moreo 3 27.3 8 72.7 0 0.0 11 301 onlya 360 100.0 0 0.0 0 0.0 360 302 morea 5 71.4 2 28.6 0 0.0 7 304 morea 26 83.9 3 9.7 2 6.5 31 1000 onlya 236 99.2 1 0.4 1 0.4 238 Appendix III - Distribution of <a> and <o> per Text File 213 <?page no="232"?> Text file Mapping <a> <o> other Total n= % n= % n= % n= 1100 moreo 32 10.3 278 89.4 1 0.3 311 1200 moreo 16 6.0 251 94.0 0 0.0 267 1300 moreo 98 12.7 671 87.0 2 0.3 771 1400 moreo 48 28.4 116 68.6 5 3.0 169 1600 onlyo 8 1.1 747 98.9 0 0.0 755 1700 moreo 7 16.3 36 83.7 0 0.0 43 1800 moreo 9 6.5 128 92.1 2 1.4 139 2000 onlya 482 98.2 6 1.2 3 0.6 491 2001 onlya 195 98.0 4 2.0 0 0.0 199 2002 onlyo 5 1.6 312 98.1 1 0.3 318 Total 6311 6549 193 13 053 214 Appendix III - Distribution of <a> and <o> per Text File <?page no="233"?> Appendix IV - Linguistic Maps Map 3.1: Basic map of all LAEME locations as used in this study. <?page no="234"?> Map 4.1: Past singular forms with the 2nd past grade vowel in verbs of Class V. 216 Appendix IV - Linguistic Maps <?page no="235"?> Map 4.4: Past singular forms with the 2nd past grade vowel in verbs of Class IVa. Appendix IV - Linguistic Maps 217 <?page no="236"?> Map 4.5: Past singular forms with the 2nd past grade vowel in verbs of Class II. 218 Appendix IV - Linguistic Maps <?page no="237"?> Map 4.6: Past plural forms with the 1st past grade vowel in verbs of Class V. Appendix IV - Linguistic Maps 219 <?page no="238"?> Map 4.7: Past plural forms with the 1st past grade vowel in verbs of class IVa. 220 Appendix IV - Linguistic Maps <?page no="239"?> Map 4.8: Past plural forms with the 1st past grade vowel in verbs of class II. Appendix IV - Linguistic Maps 221 <?page no="240"?> Map 5.1: Distribution of <a> and <o> in the LAEME CTT. 222 Appendix IV - Linguistic Maps <?page no="241"?> Map 5.2: Distribution of <a> and <o> in the West Midlands per period (a) - (f). Appendix IV - Linguistic Maps 223 <?page no="242"?> Map 5.3: Distribution of <a> and <o> in OE a ˉ + w (a) and OE a ˉ + ɣ (b). Map 5.4: Distribution of <a> and <o> in GHOST . 224 Appendix IV - Linguistic Maps <?page no="243"?> Map 6.1: Spatial distribution of 1st person dual forms. Map 6.2: Spatial distribution of 2nd person dual. Appendix IV - Linguistic Maps 225 <?page no="244"?> Map 6.3: Forms of the 2nd pl. pers. pron. with g-, ȝ and ᵹ -. 226 Appendix IV - Linguistic Maps <?page no="245"?> Map 6.4: Location of dual forms in LAEME. Map 6.5: LAEME text files with reinforcement of the dual through quantifiers. Appendix IV - Linguistic Maps 227 <?page no="246"?> Map 7.1: Unetymological past grade vowels in class V in the (South) East. Map 7.2: OE a ˉ > ME ǭ in the South-West Midlands. 228 Appendix IV - Linguistic Maps <?page no="247"?> Index AB language 117, 173 ablaut 24, 26 - 33 Ælfric ’ s Grammar and Glossary 140, 152, 154 analogical extension 34 - 35, 171 analogy 25, 28, 31, 34 - 36, 51, 53 - 54, 57, 97 - 98, 129, 148, 161, 174 anchor texts 9, 11, 62 Ancrene Riwle 89, 98, 118 Ancrene Wisse 116 - 117 Anglian 33, 41 - 43, 53, 55 Anglo-Frisian brightening see first fronting Anglo-Norman 31 - 32 Ayenbite of Inwyt 10, 64, 123 breaking 42 - 43 Candet Nudatum Pectus 119 class mixing see hybridization compound 7, 39, 42, 45, 71 - 72, 124, 127 - 129 Computer Developed Linguistic Atlas of English (CLAE) 84 consonant cluster 27, 32, 42, 44 - 45, 76, 96, 116, 124 - 130, 134, 172, 176 contraction, contracted forms 28, 98, 125, 127, 130, 134, 173 contracted verbs 33, 74 Corpus of Middle English Prose and Verse 144, 169 Cursor Mundi 107 - 108 diachronic variation 7, 41, 126 dialect boundary 6, 9, 82 dialect continuum 2, 20, 66, 163 dialectology see Middle English Dialectology diatopic (spatial) variation 7, 21, 32, 41, 46, 100, 126, 156, 161, 169 diffusion 11 - 20, 25, 36, 48, 68 - 69, 83, 87, 106, 110 - 111, 116 - 119, 123, 134, 156 - 176 - contra-hierarchical model 13, 164 - cultural hearth model 13, 164 - gravity (urban hierarchical) model 13, 18 - 19, 164 - lexical 37 - wave (contagion) model 12 - 13, 15, 163 - 164, 174 dual 49 - 59, 80, 136 - 156, 174 eLALME 10, 48, 115 first fronting 39, 41 - 42 fit-technique 9 - 10, 20, 68, 159, 163 Foot-Final Lengthening 44 formulaic expressions 140, 152, 154 French 3, 32, 78 Genesis and Exodus 57, 66, 95, 108, 109, 115, 139, 141, 146, 148, 151, 162 Gothic 29, 49, 51 - 53 gradation 28 - 33 grammel 70 - 73, 87 Great Vowel Shift 46 <h>-insertion 142 Hali Meiðhad 137 Havelok 57, 100, 105, 115, 139, 143, 151, Heliand 52 Helsinki Corpus 60 homorganic cluster lengthening (HOL) 42 - 43, 45, 124 hybridization (class mixing) 25, 33 - 35, 88, 98 Icelandic 51 Icelandic First Grammarian 133 isogloss 48, 64, 82 Katherine Group 116 - 117, 121, 137, 139 Kentish 10, 41, 53, 61, 123, 125, 135, 149, 157 - 158, 174 Kentish Sermons 50, 123 LAEME see Linguistic Atlas of Early Middle English <?page no="248"?> LALME see Linguistic Atlas of Late Mediaeval English Lambeth Homilies 116, 139, 145, 152 Latin 3, 4, 11, 40, 49, 52, 61 - 62, 64, 67, 78, 106, 113 - 114, 151 Lay Subsidy Rolls 7, 69 La ȝ amon ’ s Brut 55, 57, 67, 103, 118, 137, 139 - 140" 146, 150, 156, 174 language change see linguistic change lengthening 30, 38, 41 - 45, 95 leveling 25, 34 - 36, 88, 157 - 158, 163, 170 - 172, 176 lexical diffusion see diffusion, lexical lexical quantifiers 57 - 58, 154 - 155, 168 lexel 70 - 71, 87 Linguistic Atlas of Early Middle English (LAEME) 10 - 11, 15, 20 - 21, 60 - 80, 83 - 84, 163 - 164, 176 Linguistic Atlas of Late Mediaeval English (LALME) 8 - 11, 15, 20, 22, 48, 60, 62 - 63, 83 - 84, 86, 118, 173 Linguistic Atlas of Older Scots (LAOS) 10 Linguistic Atlas of the Gulf States (LAGS) 82 linguistic change 2, 4, 12, 14 - 15, 61, 82, 156, 175 - 176 Linguistic Survey of Scotland 9 London 15, 17 - 19, 68, 82, 103, 159, 162 Lord ’ s Prayer 125 map - choroplethic map 83 - descriptive map 81 - 84 - dialect map 8, 86 - display map 81 - dot map 8, 11, 84 - feature map 8, 64, 69 - interpretive map 81 - isogloss map 7, 63 - 64, 82 - item map 8, 15 - linguistic map 18, 69, 81 - 86, 174, 176, 215 - 229 - qualitative map 83 - quantitative map 83 Mercian 41, 47, 53, 157 Middle English dialectology 5 - 11, 21, 60, 65 Middle English Dictionary 6 Middle English Open Syllable Lengthening (MEOSL) 44, 46 Middle High German 52 Norman French 3 North Frisian 50, 52 northern leveling 25, 157 Northumbrian 42, 113 OE ā > ME ǭ 37 - 49, 77 - 80, 112 - 135, 159 - 160, 172 - 173 Old French 4, 62 Old Frisian 38, 52 Old High German 29, 52 Old Norse 7, 29, 31 - 32, 51, 53 - 54, 70, 107 - 108 onomastic evidence 6 - 7, 40, 47 - 49, 114, 117, 134 Ormulum 55, 57, 60, 64, 100 - 101, 104, 114, 139 - 140, 142, 144, 147, 153, 156, 160 - 161, 169, 174 Paston Letters 23, 95 personal pronoun - dual forms see dual - 2nd person plural forms 138, 146, 148 - 3rd person plural forms 15, 162 Peterborough Chronicle 3, 10, 29, 100 - 101, 103 - 104, 114 place-names see onomastic evidence Poema Morale 72, 102 - 104, 114, 116 - 117, 123 pre-Conquest document 115 - 116, 134, 137, 150 - 154, 160, 168 Proclamation of Henry III 11 Proverbs of Alfred 62 - 63, 137, 140, 156 Proverbs of Hending 132 reduplication 24, 27 - 29 retraction 37, 39 - 40, 43 Rushworth Gospels 53 Sawles Warde 55, 137 - 138 scribal error 1, 47, 88, 102, 147, 174 scribal practice 4, 74 second fronting 41 shortening 38, 43 - 45, 78, 122, 124 - 130, 134, 172 - 173, 176 230 Index <?page no="249"?> Shortening before Consonant Clusters (SHOCC) 44, 124 - 128, 176 Soul ’ s Address to the Body 139, 152 South English Legendary 61, 65 spatial diffusion see diffusion spatial variation see diatopic variation St Katherine 97, 103, 137 St Juliana 137 St Margaret 137 strong verbs 24 - 36, 73 - 77, 88 - 111, 157 - 158, 170 - 172 - class I 26 - 27, 30, 33, 88, 98 - 99, 106 - 109, 111, 192 - 193 - class II 24, 26 - 27, 30, 33 - 34, 77, 89, 97 - 98, 105 - 106, 108, 171 - 172, 193 - 195 - class III 26 - 27, 28, 30, 33, 76, 95 - 96, 98 - 99, 103 - 107, 109, 195 - 198 - class IV 26 - 28, 33, 94 - 95, 99, 102 - 103, 158, 171 - 172, 198 - 200 - class V 26 - 28, 33, 88 - 95, 97, 99 - 103, 108 - 111, 164, 167, 171, 200 - 203 - class VI 26 - 29, 33 - 34, 98, 107 - class VII 26 - 27, 3040, 109 Survey of English Dialects 7 Survey of Middle English Dialects 7 - 8 Swedish 52 temporal variation see diachronic variation text language 64 - 69, 114, 116, 118, 121, 137, 158, 160, 164, 173 The Creed 114 The Owl and the Nightingale 50, 55, 66 - 67, 118, 121, 125, 135, 137, 139, 142 - 143, 149, 156, 173 - 174 transition zone 82 - 83, 116, 159, 169 transmission gap 3 - 5 Tremulous Worcester Hand 116, 137, 152, 160 Trinity Homilies 67, 114 Trisyllabic Shortening (TRISH) 44 - 45, 127 - 129 uniformitarian principle 12 Vices and Virtues 37, 57, 86, 100 - 101, 103, 114, 127, 139 - 140, 147, 159 - 160, 174 visual economy 84 - 86 weak verbs 24 - 25, 30 - 32, 42, 88, 98, 107 - 111, 167, 172 West Saxon 3 - 4, 33, 40 - 41, 43, 46 - 47, 53 - 55, 58, 70, 76, 101 West Saxon Gospels 53, 149, 174 Will of Mantat 140, 146, 151, 160 Yiddish 52 Index 231 <?page no="250"?> The present study examines the diffusion of three linguistic variables in Early Middle English with special focus on the East versus West Midlands divide, namely the reduction from four to three stems in the gradation of strong verbs, variation between Middle English <a> and <o> and the decline of the dual forms of the personal pronoun. The data is retrieved from version 2.1 of the corpus of tagged texts of the Linguistic Atlas of Early Middle English (LAEME) (Laing and Lass 2008-), in which two thirds of the 167 semi-diplomatically transcribed corpus files are localized, permitting innovative approaches to Early Middle English dialectology, such as investigations into spatial diffusion phenomena. The present study offers suggestions as to how modern diffusion models can be adjusted and applied to historical data. It also discusses the usability of LAEME for this specific purpose and develops a set of plausible hypotheses on spatial diffusion patterns in Early Middle English. At the same time, the study addresses the main issues of studying medieval manuscripts and of working with historical corpora, and it illustrates how maps prove to be a useful tool in the visual representation of linguistic change across time and space.