144 78 5MB
English Pages 242 [237] Year 2024
Language and Dialect Death Theorising Sound Change in Obsolescent Gascon Damien Mooney
Language and Dialect Death
Damien Mooney
Language and Dialect Death Theorising Sound Change in Obsolescent Gascon
Damien Mooney School of Modern Languages University of Bristol Bristol, UK
ISBN 978-3-031-51100-4 ISBN 978-3-031-51101-1 https://doi.org/10.1007/978-3-031-51101-1
(eBook)
© The Editor(s) (if applicable) and The Author(s), under exclusive licence to Springer Nature Switzerland AG 2023 This work is subject to copyright. All rights are solely and exclusively licensed by the Publisher, whether the whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology now known or hereafter developed. The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication does not imply, even in the absence of a specific statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use. The publisher, the authors, and the editors are safe to assume that the advice and information in this book are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or the editors give a warranty, expressed or implied, with respect to the material contained herein or for any errors or omissions that may have been made. The publisher remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Cover image: Massimo Ravera/Getty Images This Palgrave Macmillan imprint is published by the registered company Springer Nature Switzerland AG The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland Paper in this product is recyclable.
James Harrington 1986–2019
Patrick J. Merrigan 1963–2019
To my best friend, James, and my loving father, Pat. Ar dheis Dé go raibh a n-anamacha dílse.
Acknowledgements
This monograph arose from a British Academy/Leverhulme Trust Small Research Grant (SG152043), Language and dialect death in Béarn, France (2016–2018), and I would like to thank the University of Bristol Research and Enterprise Division (RED), and in particular Valérie Aspin, for their support and guidance in obtaining this funding. The research presented in this book has received ethical approval from the Faculty Research Ethics Committee, Faculty of Arts, University of Bristol (OREMS ref. 2016-2791-2738). I am indebted to my contacts in Béarn, including Bernard Coustalat and Jean-Marie Puyau of the Institut Béarnais et Gascon, and Jean-Noël Commères at Ràdio País. This project provided me with an unparalleled opportunity to enter the Gascon speech community in Béarn, to learn and speak their language with them, and to share their linguistic and cultural heritage—Mercí hera a monde de m’aver arcuelhut tan calorosament. With permission, parts of Chapter 3 (Sect. 3.1) and of Chapter 4 have been reproduced in adapted form from: Mooney, D. “Occitan”; in C. Gabriel, R. Gess, & T. Meisenburg (eds.), Manual of Romance Phonetics and Phonology, Berlin, New York: De Gruyter, 2021, pp. 709–742. vii
viii
Acknowledgements
I would like to thank my mother, Jennifer Merrigan, my sister, Erica Mooney, and my friends, David Coonan, Maria Flood, James Hawkey, Jon Morris, Sophie Nolan, and Eoin Quirke. The period over which I wrote this book was marked by sadness and hopelessness and you stood with me in my grief—I will never forget that. To my little Lucy—que m’as sauvat. Finally, I dedicate this work to my best friend, James, and my father, Pat. Their battles with cancer and untimely deaths taught me much about love, about fortitude and resilience, and about searching for hope in the face of profound loss.
Contents
1
Introduction 1.1 Research Questions 1.2 Structure of the Book References
1 6 7 10
2 Theorising Language and Dialect Death 2.1 Internal Factors 2.2 External Factors 2.2.1 Language Contact 2.2.2 Dialect Contact 2.3 Extralinguistic Factors 2.4 Language Death 2.5 Dialect Death 2.6 New Speakers References
13 13 15 15 19 23 28 32 35 41
3
47 48 53
Research Context: Southern Gallo-Romance 3.1 The Dialectalisation of Southern Gallo-Romance 3.2 Language Contact with French in Béarn
ix
x
4
5
Contents
3.3 The Gascon Speech Community in Béarn 3.4 The Standardisation of Occitan 3.5 Militant Language Politics in Béarn 3.6 Occitan Revitalisation in the Educational Context References
58 60 63 65 68
Phonetic and Phonological Systems 4.1 Linguistic Background 4.1.1 Northern Occitan 4.1.2 Southern Occitan 4.1.3 Gascon 4.1.4 Occitano-Romance Supra-Dialects 4.2 Modern Phonological Inventories 4.2.1 Consonants 4.2.1.1 Syllable- and Word-Final Consonants 4.2.2 Vowels 4.2.2.1 Stressed Oral Vowels 4.2.2.2 Vowel Length 4.2.2.3 Unstressed Oral Vowels 4.2.2.4 Vowel Nasalisation 4.2.3 Glides and Diphthongs References
71 72 73 75 76 77 78 79
Methodological Considerations 5.1 Sampling Method 5.1.1 Fieldwork Sites 5.1.2 Informant Recruitment 5.2 Linguistic Variables 5.3 Corpus Construction 5.4 Data Analysis 5.4.1 Acoustic Analysis 5.4.2 Auditory Analysis 5.5 Data Processing 5.5.1 Normalisation
87 88 90 92 93 95 96 99 101 102 102 102 105 108 111 111 115 115 115
Contents
xi
5.5.2 Data Coding 5.5.3 Statistical Analysis References
118 118 121
6
Language Death: Gascon and French 6.1 The Oral Vowel System 6.1.1 Stressed Oral Vowels 6.1.2 Unstressed Oral Vowels 6.2 Transfer from French 6.2.1 Rhotics 6.2.2 Palatal Lateral 6.2.3 Voiced Plosives 6.2.4 Gender Marking 6.3 Dialect Levelling/Mixing 6.3.1 Unstressed Oral Vowels 6.3.2 Voiceless Affricate 6.3.3 Voiced Affricate 6.3.4 Voiced Apical Plosive /d/ 6.3.5 Voiced Postalveolar Fricative /ʒ/ 6.3.5.1 Voiceless Apical Fricative /-s/ 6.4 Interim Summary and Discussion References
123 124 124 129 133 134 136 138 142 143 143 144 145 147 148 149 150 153
7
Dialect Death: Gascon and Occitan 7.1 The Oral Vowel System 7.1.1 Stressed Oral Vowels 7.1.2 Unstressed Oral Vowels 7.2 Transfer from French 7.2.1 Rhotics 7.2.2 Palatal Lateral 7.2.3 Voiced Plosives 7.2.4 Gender Marking 7.3 Dialect Levelling 7.3.1 Unstressed Oral Vowels 7.3.2 Voiceless Affricate 7.3.3 Voiced Affricate 7.3.4 Voiced Apical Plosive /d/
155 156 156 158 161 161 164 165 167 167 168 168 169 170
xii
Contents
7.3.5 Voiced Postalveolar Fricative /ʒ/ 7.3.6 Voiceless Apical Fricative /-s/ 7.4 Interim Summary and Discussion References
171 171 172 177
8 Towards a New Theory of Language Death 8.1 Theorising Sound Change During Obsolescence 8.2 Principles of Linguistic Change During Language Obsolescence 8.3 Principles of Linguistic Change During Language Revitalisation 8.4 Internal, External, and Extralinguistic Factors References
179 180
9
197 202
Conclusions References
182 186 191 195
Appendix 1: Wordlist translation task
203
Appendix 2: Language Death Study—Supplementary Data
209
Index
219
List of Figures
Fig. Fig. Fig. Fig.
3.1 3.2 5.1 6.1
Fig. 6.2 Fig. 6.3 Fig. 7.1 Fig. 7.2 Fig. A2.1 Fig. A2.2 Fig. A2.3
The Gallo-Romance languages The Gallo-Romance dialect areas Sub-dialect fieldwork sites in the language death study Stressed oral vowel system for (male and female) Laruns speakers, normalised data Unstressed oral vowels for (male and female) Laruns speakers, normalised data Unstressed oral vowels for (male and female) Arzacq speakers, normalised data Stressed oral vowel system for (male and female) new speakers, normalised data Unstressed oral vowels for (male and female) new speakers, normalised data Stressed oral vowel system for (male and female) Arzacq speakers, normalised data Stressed oral vowel system for (male and female) Lembeye speakers, normalised data Stressed oral vowel system for (male and female) Nay speakers, normalised data
49 50 103 127 130 132 157 159 210 210 211
xiii
xiv
Fig. A2.4 Fig. A2.5 Fig. A2.6 Fig. A2.7
List of Figures
Stressed oral vowel system for (male and female) Ouest speakers, normalised data Unstressed oral vowels for (male and female) Lembeye speakers, normalised data Unstressed oral vowels for (male and female) Nay speakers, normalised data Unstressed oral vowels for (male and female) Ouest speakers, normalised data
211 212 212 213
List of Tables
Table 4.1 Table 4.2 Table 5.1 Table 5.2 Table 5.3 Table 5.4 Table 6.1
Table 6.2
Table 6.3
Table 6.4
The consonantal phonemes of Occitan Stressed oral vowels in Occitan Native speaker participants in the language death study New-speaker participants in the dialect death study Wordlist token counts by syllabic context Variants used in auditory analysis of categorical variables All native speakers (normalised data): regression analysis of F2 values for /a/, with ‘speaker’ and ‘word’ as random effects All native speakers (normalised data): regression analysis of F3 values for /e/, with ‘speaker’ and ‘word’ as random effects Laruns speakers (normalised data): regression analysis of F1 values for /e/ and /E/, with ‘speaker’ and ‘word’ as random effects Laruns speakers (normalised data): regression analysis of F3 values for /e/ and /E/, with ‘speaker’ and ‘word’ as random effects
79 89 104 105 110 116
125
125
127
128
xv
xvi
List of Tables
Table 6.5 Table 6.6
Table 6.7
Table 6.8 Table 6.9 Table 6.10
Table 6.11
Table 6.12 Table 6.13
Table 6.14 Table 6.15
Table 6.16 Table 6.17 Table 6.18
Table 6.19 Table 6.20
Significant formant frequency differences for front mid-vowels by place of origin Laruns speakers (normalised data): regression analysis of F1 values for /-e/ and /-O/, with ‘speaker’ and ‘word’ as random effects Laruns speakers (normalised data): regression analysis of F2 values for /-e/ and /-O/, with ‘speaker’ and ‘word’ as random effects Significant formant frequency differences for post-tonic vowels by place of origin All native speakers: frequency distribution of rhotic variable All native speakers: regression analysis of rhotic variable, with variants [apical] and [uvular]; ‘speaker’ and ‘word’ as random effects All native speakers: regression analysis of apical rhotics, with variants [ɾ] and [r]; ‘speaker’ and ‘word’ as random effects All native speakers: frequency distribution of palatal lateral variable All native speakers: regression analysis of palatal lateral variable, with variants [ʎ] and [j]; ‘speaker’ and ‘word’ as random effects All native speakers: frequency distribution of voiced bilabial plosive variable All native speakers: regression analysis of voiced bilabial plosive variable, with variants [plosive] and [lenited]; ‘speaker’ and ‘word’ as random effects All native speakers: frequency distribution of voiced apical plosive variable All native speakers: frequency distribution of voiced velar plosive variable All native speakers: regression analysis of voiced velar plosive variable, with variants [plosive] and [lenited]; ‘speaker’ and ‘word’ as random effects All native speakers: frequency distribution of gender marking variable All native speakers: frequency distribution of voiceless affricate variable
129
131
131 133 135
135
136 138
139 139
140 141 141
142 143 144
List of Tables
Table 6.21
Table 6.22 Table 6.23
Table 6.24
Table 6.25 Table 6.26 Table 7.1
Table 7.2
Table 7.3
Table 7.4
Table 7.5
Table 7.6
Table 7.7
All native speakers: regression analysis of voiceless affricate variable, with variants [plosive] and [affricate]; ‘speaker’ and ‘word’ as random effects All native speakers: frequency distribution of voiced affricate variable All native speakers: regression analysis of voiced affricate variable, with variants [palatal] and [postalveolar]; ‘speaker’ and ‘word’ as random effects All native speakers: regression analysis of voiced apical plosive variable, with variants [apical] and [palatalised]; ‘speaker’ and ‘word’ as random effects All native speakers: frequency distribution of voiced postalveolar fricative variable All native speakers: frequency distribution of voiceless apical fricative variable New speakers (normalised data): Regression analysis of F1 values for /e/ and /E/, with ‘speaker’ and ‘word’ as random effects All (native and new) speakers (normalised data): Regression analysis of F1 values for /a/, with ‘speaker’ and ‘word’ as random effects New speakers (normalised data): Regression analysis of F1 values for /-e/ and /-O/, with ‘speaker’ and ‘word’ as random effects New speakers (normalised data): Regression analysis of F2 values for /-e/ and /-O/, with ‘speaker’ and ‘word’ as random effects New speakers (normalised data): Regression analysis of F1 values for /-O/, with ‘speaker’ and ‘word’ as random effects New speakers: Regression analysis of rhotic variable, with variants [apical] and [uvular]; ‘speaker’ and ‘word’ as random effects New speakers: Regression analysis of apical rhotics, with variants [ɾ] and [r]; ‘speaker’ and ‘word’ as random effects
xvii
145 146
146
148 149 150
157
158
160
160
161
162
163
xviii
List of Tables
Table 7.8
Table A2.1
Table A2.2
Table A2.3
Table A2.4
Table A2.5
Table A2.6
Table A2.7
Table A2.8
Table A2.9
Table A2.10
Table A2.11
Table A2.12
New speakers: Regression analysis of palatal lateral variable, with variants [ʎ] and [j]; ‘speaker’ and ‘word’ as random effects Arzacq speakers (normalised data): Regression analysis of F1 values for /e/ and /E/, with ‘speaker’ and ‘word’ as random effects Arzacq speakers (normalised data): Regression analysis of F2 values for /e/ and /E/, with ‘speaker’ and ‘word’ as random effects Arzacq speakers (normalised data): Regression analysis of F3 values for /e/ and /E/, with ‘speaker’ and ‘word’ as random effects Lembeye speakers (normalised data): Regression analysis of F1 values for /e/ and /E/, with ‘speaker’ and ‘word’ as random effects Lembeye speakers (normalised data): Regression analysis of F2 values for /e/ and /E/, with ‘speaker’ and ‘word’ as random effects Lembeye speakers (normalised data): Regression analysis of F3 values for /e/ and /E/, with ‘speaker’ and ‘word’ as random effects Nay speakers (normalised data): Regression analysis of F1 values for /e/ and /E/, with ‘speaker’ and ‘word’ as random effects Ouest speakers (normalised data): Regression analysis of F1 values for /e/ and /E/, with ‘speaker’ and ‘word’ as random effects Ouest speakers (normalised data): Regression analysis of F3 values for /e/ and /E/, with ‘speaker’ and ‘word’ as random effects Lembeye speakers (normalised data): Regression analysis of F2 values for /-e/ and /-O/, with ‘speaker’ and ‘word’ as random effects Nay speakers (normalised data): Regression analysis of F1 values for /-e/ and /-O/, with ‘speaker’ and ‘word’ as random effects Nay speakers (normalised data): Regression analysis of F2 values for /-e/ and /-O/, with ‘speaker’ and ‘word’ as random effects
164
213
213
214
214
214
215
215
215
216
216
216
217
List of Tables
Table A2.13
Table A2.14
Table A2.15
Ouest speakers (normalised data): Regression analysis of F2 values for /-e/ and /-O/, with ‘speaker’ and ‘word’ as random effects All native speakers (normalised data): Regression analysis of F2 values for /-e/, with ‘speaker’ and ‘word’ as random effects All native speakers (normalised data): Regression analysis of F2 values for /-O/, with ‘speaker’ and ‘word’ as random effects
xix
217
217
218
1 Introduction
The analysis of language obsolescence and death was formalised as an independent line of enquiry within the discipline of sociolinguistics with the publication of Dorian’s (1981) Language Death. Dorian’s examination of East Sutherland Gaelic (ESG) provided the first comprehensive and extended theoretical account of language contraction and death by advancing not only a means of classifying the sociolinguistic profiles of the speakers of obsolescent languages, but by also advancing data-driven principles that provided a theoretical reference point against which to examine the nature of linguistic changes occurring in dying languages during language shift. Taking inspiration from the theoretical framework established by Dorian and the work of other scholars in the field (e.g., Schmidt 1985; Dressler 1982; Campbell and Muntzell 1989), Jones (1998) examined linguistic change in two sociolinguistically contrasting Welsh communities, allowing her not only to test and confirm many of Dorian’s predictions about that nature of linguistic change during obsolescence, but also to establish new theoretical constructs that firmly situated the study of language death within discipline of sociolinguistics. Both studies prioritised data-driven theorisation of the linguistic
© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 D. Mooney, Language and Dialect Death, https://doi.org/10.1007/978-3-031-51101-1_1
1
2
D. Mooney
and social processes of language obsolescence rather than simply documenting the language varieties in question. Taking inspiration from both of these seminal studies, this monograph seeks to advance the study of language obsolescence and death by integrating firmly into the analysis theories and methods from variationist sociolinguistics, often referred to as language variation and change (LVC) sociolinguistics. This approach aims to explore the interface between the linguistic processes and mechanisms active during language death by setting quantitative findings within the context of well-established variationist theories of language contact and language change. This study will make a formal distinction, at least initially, between the processes of linguistic change, and the mechanisms which comprise those processes that may potentially occur during, on the one hand, language death, and, on the other, dialect death (see Chapter 2). The case study will focus on Gascon, the indigenous southern Gallo-Romance variety historically spoken in southwestern France, and more specifically on the variety of Gascon spoken in the region of Béarn. Under pressure from French, Gascon has been largely ousted from its territory via the socio-political process of language shift, whereby members of a speech community cease to speak their indigenous language in favour of an incoming dominant language, a progressive process which eventually leads to language death, where the obsolescent language is eliminated completely. Gascon is at an advanced stage of this process, making its remaining speakers excellent candidates for the study of language obsolescence. During the process of language death, Gascon is undergoing two different types of linguistic change: externally motivated changes occurring in all sub-dialects as a result of contact with French and changes hypothesised to be due to dialect mixing or levelling between sub-dialects of the language (see Chapter 6). Additionally, the variety of Gascon spoken in Béarn lends itself to the study of dialect death in that it is a specific localised dialect of Gascon and, in turn, a dialect of the larger langue d’oc continuum which covers the southern third of France. Since the 1960s, Gascon has also been under threat at a subordinate level from the standardised langue d’oc variety ‘Occitan’, which has succeeded in securing a monopoly on the (limited) institutional and educational space available to regional languages in the south of France. Since the
1 Introduction
3
1980s, private Calandretas, or Occitan immersion education schools, have offered bilingual education in Occitan at primary school level. Together with the increasing availability of adult education in Occitan, the revitalisation movement has led to a perceived cleavage in Gascony between the speech of older Gascon speakers and the Occitan-educated, ‘néo-locuteurs’ (literally, ‘new speakers’). These linguistic differences are a result of the process of dialect obsolescence, whereby a specific localised variety is replaced by a non-local variety of the same language, which leads to dialect death. The presence of Occitan in the region therefore constitutes a complicating factor in this study of obsolescence, in that the dying language, Gascon, is itself subject to potential encroachment from the artificial standard. While comprehensive sociolinguistic studies of language and dialect death exist for Celtic languages in the United Kingdom (Dorian 1981; Jones 1998), quantitative research on obsolescent languages in the Francophone context is much more limited. France’s regional and minority languages (RMLs), the so-called langues de France (literally, ‘languages of France’), are relatively well documented and have been examined from a sociolinguistic perspective but the focus of these studies if often on describing the contexts in which the languages are used (e.g., Jones 1995 on Breton) or on abstract theorising about issues of authenticity, ideology, and speaker motivation (e.g., Costa and Gasquet-Cyrus 2013 on Provençal; Pivot and Bert 2017 on Francoprovençal). Relatively recent efforts on the part of predominantly English-speaking scholars have aimed to address the lack of quantitative variationist research on the langues de France. In particular, Jones has produced a wealth of quantitative literature on the varieties of Norman spoken in France and on the Channel Islands (e.g., Jones 2001, 2008, 2015), Pooley (1996) and Hornsby (2006a) have published quantitative data on Picard, and a Special Issue of the Journal of French Language Studies (Hall et al. 2019) entitled ‘Langues régionales: models and methods’ included articles on sentential negation in Gallo (Burnett 2019), interrogatives in Picard (Auger and Villeneuve 2019), morphosyntactic and morphophonological variation in Breton (Kennard 2019), metathesis of aspiration in Basque (Egurtzegi 2019), and palatal laterals in Occitan and Catalan (Mooney and Hawkey 2019). While these scholars, who have also
4
D. Mooney
published elsewhere on France’s RMLs, have gone some way towards filling a much needed gap, all of these studies suffer from at least one of two limitations: either they lack the methodological rigour, in the form of instrumental analysis and statistical modelling, that is required in modern variationist sociolinguistics, or they lack the space, depth of study, and scope to address complex theoretical issues relating to the interface between language contact and dialect contact in these dying languages. The case study of Gascon presented in this monograph will employ quantitative instrumental and statistical analysis to investigate these key theoretical issues in the study of obsolescence. France’s regional languages offer an excellent opportunity to study the theoretical processes of language and dialect death, and the relationship between them, particularly in southern regions where localised varieties are both dying out as a result of contact with French and being revitalised in the form of Occitan. In Gascony, and elsewhere in the south of France, the language varieties in contact (French, standard Occitan, local langue d’oc dialects and sub-dialects) are also more typologically similar than, for example, English and Welsh, and, since typological similarity is often considered to permit higher levels of linguistic transfer, this context will provide a fresh theoretical perspective on the mechanisms involved in linguistic change during obsolescence as a result of different types of contact. Formal analyses of the variety of Gascon spoken in Beárn are rare, the most recent of which can be found in two studies (Marchal and Moreux 1989; Kristol and Wüest 1986), which are descriptive rather than explanatory and provide no theoretical discussion of the processes of obsolescence active in the variety. There is thus a need to further document inter-dialectal and macro-level variation and change in Gascon while it still exists and to establish the factors conditioning these linguistic developments, with the aim of investigating the relationship between the processes of language and dialect death and of examining the implications of this relationship for variationist theories of language and dialect contact. Language obsolescence is often assumed to proceed along a defined structural path that is that certain linguistic levels are affected by processes of linguistic change in a defined order. The factors influencing
1 Introduction
5
these changes are often seen to be external or contact-induced but attrition at particular linguistic levels may also mirror the acquisition of those features during childhood. The hypothesis that language loss follows a defined structural path has been formalised in what is commonly referred to as the ‘structured obsolescence hypothesis’, which, in its strictest form, predicts that the lexicon is affected first, followed by phonology, syntax, and finally morphology (Dauzat 1927: 49–57). Dauzat (1927) referred to morphology as the citadelle de la langue (‘bastion of the language’), stating that this level is least affected by the processes of language obsolescence; this hypothesis has indeed been confirmed by later sociolinguistic studies such as Dorian (1978: 608) who argued that, at least with respect to some linguistic features, Gaelic in East Sutherland was dying ‘with its morphological boots on’. On the contrary, Hornsby (2006b) examined evidence for the structured obsolescence hypothesis by investigating the retention and loss of Picard linguistic transfer in the regional French of Avion, concluding that the hypothesis was a ‘myth’ because he found ‘no relationship between phonological and morphological change which would be consistent with the level-by-level obsolescence predicated by the model’ (2006b: 134). In an effort to examine in more depth the extent to which morphology is subject to contact-induced change, Jones (2018) applied Myers-Scotton’s Abstract Level and 4-M models (MyersScotton 2002; Myers-Scotton and Jake 2017) to Norman data from Jersey (Jèrriais) in order to investigate, among other things, ‘whether different morpheme types of Jèrriais are related to the production process in different ways and are, accordingly, more or less susceptible to change during the process of language obsolescence’ (2018: 399). This approach had the added advantage of setting the study of language obsolescence firmly within a theoretical framework that was developed outside of the discipline of language death, in this case code-switching in bilingual speech. Jones’ (2018) analysis of Jèrriais has shown that, even within the morphological level, contact-induced transfer from English affects different morpheme types in a relatively well-defined order, thus lending support, in even more depth, to the structured obsolescence hypothesis. This monograph will provide a detailed comparative quantitative analysis of phonetic and phonological changes in Gascon that are motivated by the processes of both language and dialect death. This approach will
6
D. Mooney
have the advantage of exploring in much more depth the effect of obsolescence at the level of phonology, while at the same time investigating the linguistic phenomenon that has since its inception been the focus of variationist sociolinguistic theory: sound change. Previous European dialectological studies have been criticised, by Labov (2007: 348), for their tendency to analyse the transfer of ‘well-known’ features from one language to another in an ‘X is replaced by Y’ fashion, rather than investigating evidence for new phonetic and phonological changes in progress. The Gascon study will, of course, examine contact-induced transfer from French but, at the same time, the phonetic and phonological analysis provided will seek to identify changes occurring independently within the language, either due to contact with other dialects of the langue d’oc or due to internally motivated sound change, integrating the study of obsolescence firmly within a variationist theoretical framework.
1.1
Research Questions
Traditional sociolinguistic studies of language obsolescence and death have not fully exploited the variationist sociolinguistic toolkit in their analysis and, at least to some extent, have focused on developing a theory of language death that is independent of well-established theories of language contact and, to a greater extent, of a wealth of variationist theorisation on the processes and mechanisms of language change more generally. Theorisation within the discipline of language death has focused, among other things, on impressionistic comparisons with ‘healthy’ languages, compression of the time scale for change, speaker profiles, and the effects of obsolescence on different linguistic levels. With the exception perhaps of Jones (1998), the interface between language death and dialect death and their relationship to wider processes of language and dialect contact and change warrants further exploration, particularly in the Francophone context. Adopting a variationist approach, and focusing on the theorisation of linguistic change processes that are due to different types of contact (as well as the interface between these processes), this monograph will seek to answer the following research questions:
1 Introduction
7
1. What phonetic and phonological changes are occurring in obsolescent Gascon as a result of contact with French? 2. To what extent is phonetic and phonological change in Gascon the result of levelling between its localised sub-dialects? 3. What effect has revitalisation in the form of Occitan had on the phonological structure and phonetic realisation of Gascon? These research questions will be addressed in two sub-studies. Firstly, the language death study (Chapter 6), which will address questions 1 and 2, will examine vocalic and consonantal variation and change in the speech of older native Gascon speakers from five fieldwork sites in the region of Béarn. Secondly, the dialect death study (Chapter 7) will examine the speech of L2 Occitan speakers, or néo-locuteurs, from Béarn, again examining research questions 1 and 2, but also providing a detailed phonological and phonetic comparison with the results of the older native speakers in the language death study, in order to respond to research question 3. Together, these two sub-studies will furnish a detailed comprehensive analysis of the processes and mechanisms of linguistic change active in Gascon as a result of contact with French, contact between sub-dialects of Gascon, and contact with standardised Occitan. The results of these studies will not only, in Chapter 8, permit an exploration of the interface between language and dialect death but will, crucially, form the basis of the first theoretical analysis of obsolescence in southern Gallo-Romance by setting these findings within the context of well-established variationist theories of language and dialect contact and change.
1.2
Structure of the Book
Through an examination of the processes and mechanisms of linguistic change active in Gascon during both language and dialect obsolescence, this book offers a unique comparison of the differential outcomes of each contact situation for phonological variation and sound change in Gascon, with two sub-studies addressing L1 Gascon and L2 Occitan speakers respectively.
8
D. Mooney
Chapter 2 provides a detailed account of the various factors (internal, external, extralinguistic) that condition language variation and change, focusing on the issue of causation. This discussion draws on wider variationist theories of language contact, dialect contact, and linguistic change before outlining the potential implications of these theories for studying the structural linguistic consequences of language and dialect death. This chapter also explores the extent to which the processes of language and dialect obsolescence (and language and dialect contact) can be considered independently in situations where a minority language is at once undergoing contraction at the hands of a dominant national language and revitalisation in the form of a standard variety, focusing on the role of new speakers. Chapter 3 comprises an external history of the dialectalisation of Gallo-Romance, focusing on Gascon and, in particular, on the Béarnais sub-dialect of Gascon, which attests the highest number of native speakers. The sociolinguistic profile of the extant Béarnais speech community is discussed in detail before turning the Occitan revitalisation movement and, in particular, the implications of this movement for the Béarnais speech community. This discussion examines the Calandreta immersion education schools and adult education initiatives that have, since the 1980s, produced new speakers of Gascon (in the form of Occitan) in Béarn. This chapter will thus provide a comprehensive overview of the internal, external, and extralinguistic factors that have influenced and may influence language change in Béarnais. Chapter 4 provides a detailed historical and synchronic account of the vocalic and consonantal inventories of standard Occitan, making detailed reference to the phonetic realisation of these phonological structures in non-standardised Gascon and providing detailed comparisons between Gascon and the other principal dialects of southern Gallo-Romance. This chapter also provides a comprehensive overview of sub-dialectal variation within the region of Gascony and, more specifically, within the region of Béarn, where high levels of sub-dialectal variation are attested. This chapter furnishes a solid comparative baseline against which to assess current trajectories of linguistic change identified in the studies of Gascon presented in the language and dialect death sub-studies.
1 Introduction
9
Chapter 5 focuses on methodology, presenting two studies which explicitly address the research questions discussed above, and outlines the methodology employed in the empirical analysis of variation and change in Gascon. The chapter outlines the phonological and phonetic variables that will constitute the focus of the analyses, as well as information pertaining to data collection (fieldwork site selection, informant sampling, corpus construction) and data analysis (acoustic phonetic techniques, auditory coding, statistical analysis). The language and dialect death studies are both sociophonetic in nature, involving primarily an acoustic phonetic analysis with rigorous adherence both to analytic best practices and to a Labovian variationist sociolinguistic research methodology. Chapter 6 presents the results of the language death study, which investigates two different types of linguistic change occurring in Gascon: those changes occurring in all sub-dialects suspected to be due to influence from French and changes hypothesised to be due to dialect mixing or levelling between sub-dialects in Béarn. Using wordlist translation data collected from 30 older native Gascon speakers, the investigation of the first type of change focuses on the points at which it diverges structurally from French; secondly, the analysis of variables whose realisation traditionally differs according to geographical location in Béarn will examine levelling and dialect mixing, as forms from the central area of the region may be adopted to replace peripheral forms. This chapter also examines the applicability of current models of language obsolescence, language contact, and language change (outlined in Chapter 2) to the linguistic transfer and dialect mixing observed. Chapter 7 examines the processes of linguistic change active in Gascon as it is subjected to varying levels of standardisation as part of the Occitan revitalisation movement, investigating, in detail, phonetic and phonological variation and change in the speech of ten ‘new speakers’ who have acquired Gascon in the Occitan-dominated educational context. The analysis begins by comparing variation and change in the speech of the new speakers with the speech of the older native speakers examined in Chapter 6, examining evidence for contact-induced transfer from French, for the retention of traditionally localised phonological features and for the acquisition of linguistic features that have survived the dialect
10
D. Mooney
levelling process. The acoustic phonetic findings are interpreted with reference both to theories of dialect contact, such as supralocalisation, and to the recent theorisation of the role of the néo-locuteur or ‘new speaker’ in minority language situations. Chapter 8 assesses the implications of the empirical findings of the language and dialect death studies for current theories of language and dialect contact and change, before outlining the basis for a new theory of language death. The theoretical discussion addresses the differential outcomes of externally motivated linguistic changes arising from a variety of contact situations to which the dying language is exposed. This chapter attempts to formalise the differences between language and dialect obsolescence, insofar as such as distinction can be made, both in terms of the linguistic mechanisms active in each context and the sociopolitical circumstances that give rise to their existence. Crucially, this chapter examines the commonalities between language and dialect obsolescence, and the interface between them, focusing on situations in which the dominant language (French) and the standard language (Occitan) may act in tandem as external motivators of linguistic change.
References Auger, Julie, and Anne-José Villeneuve. 2019. Building on an old feature in langue d’Oïl: interrogatives in Vimeu Picard. Journal of French Language Studies 29: 209–233. Burnett, Heather. 2019. Sentential negation in north-eastern Gallo-Romance dialects: Insights from the Atlas Linguistique de la France. Journal of French Language Studies 29: 189–207. Campbell, Lyle, and Martha C. Muntzell. 1989. The structural consequences of language death. In Investigating obsolescence, ed. Nancy Dorian, 186–196. Cambridge: Cambridge University Press. Costa, James, and Médéric. Gasquet-Cyrus. 2013. What is language revitalization really about? Competing language revitalization movements in Provence. In Keeping languages alive: Documentation, pedagogy and revitalisation, ed. Mari C. Jones and Sarah Ogilvie, 212–224. Cambridge: Cambridge University Press.
1 Introduction
11
Dauzat, Albert. 1927. Les patois: évolution, classification, étude. Paris: Delagrave. Dorian, Nancy C. 1978. The fate of morphological complexity in language death: Evidence from East Sutherland Gaelic. Language 54: 590–609. Dorian, Nancy C. 1981. Language death: The life cycle of a Scottish Gaelic dialect. Philadelphia: University of Pennsylvania Press. Dressler, Wolfgang. 1982. Acceleration, retardation and reversal in language decay? In Language spread , ed. Robert Cooper, 321–336. Bloomington: Indiana University Press. Egurtzegi, Ander. 2019. Metathesis of aspiration as the source of anticipatory voicelessness in Basque. Journal of French Language Studies 29: 265–279. Hall, Damien, Jonathan R. Kasstan, and David C. Hornsby. 2019. Beyond obsolescence: A twenty-first century research agenda for the langues régionales. Journal of French Language Studies 29: 155–169. Hornsby, David. 2006a. Redefining Regional French: Koinéization and dialect levelling in northern France. Oxford: Legenda. Hornsby, David. 2006b. The myth of structured obsolescence. Journal of French Language Studies 16: 125–146. Jones, Mari C. 1995. At what price language maintenance? Standardization in modern Breton. French Studies 49: 424–438. Jones, Mari C. 1998. Language obsolescence and revitalisation: Linguistic change in two sociolinguistically contrasting Welsh communities. Oxford: Clarendon Press. Jones, Mari C. 2001. Jersey Norman French: A linguistic study of an obsolescent dialect. Oxford: Blackwell. Jones, Mari C. 2008. The Guernsey Norman French translations of Thomas Martin: A linguistic study of an unpublished archive. Leuven: Peeters. Jones, Mari C. 2015. Variation and change in mainland and insular Norman: A study of superstrate influence. Leiden: Brill. Jones, Mari C. 2018. Does language loss follow a principled structural path? Evidence from Jersey Norman French. Journal of French Language Studies 28: 399–429. Kennard, Holly. 2019. Morphosyntactic and morphophonological variation in Breton: A cross-generational perspective. Journal of French Language Studies 29: 235–263. Kristol, Andres M., and Jakob Th. Wüest. 1986. Drin de tot: Travaux de sociolinguistique et de dialectologie béarnaises. Bern: Peter Lang. Labov, William. 2007. Transmission and diffusion. Language 83: 344–387.
12
D. Mooney
Marchal, Alain, and Bernard Moreux. 1989. La longueur des voyelles en béarnais et en français du Béarn. Cahiers de l’Université de Pau et des Pays de l’Adour: Langues en Béarn 15: 257–280. Mooney, Damien, and James Hawkey. 2019. The variable palatal lateral in Occitan and Catalan: Linguistic transfer or regular sound change? Journal of French Language Studies 29: 281–303. Myers-Scotton, Carol. 2002. Contact linguistics. Oxford: Oxford University Press. Myers-Scotton, Carol, and Janice L. Jake. 2017. Revisiting the 4-M model: Codeswitching and morpheme election at the abstract level. International Journal of Bilingualism 21: 340–366. Pivot, Bénédicte., and Michel Bert. 2017. Orthography creation for postvernacular languages: Case studies of Rama and Francoprovençal revitalization. In Creating orthographies for endangered languages, ed. Mari C. Jones and Damien Mooney, 276–290. Cambridge: Cambridge University Press. Pooley, Tim. 1996. Ch’timi: The urban vernaculars of northern France. Clevedon: Multilingual Matters. Schmidt, Annette. 1985. Young people’s Dyirbal . Cambridge: Cambridge University Press.
2 Theorising Language and Dialect Death
This chapter will first provide a theoretical overview of the internal, external, and extralinguistic factors that may exert a significant effect on linguistic variation and change, focusing in particular on sound change. The implications of these central objects of study in variationist sociolinguistics for the study of language obsolescence and death will subsequently be examined, exploring the distinction between language and dialect death in the context of well-established theories of language and dialect contact. The final section will investigate the relatively recent theorisation of the ‘new speaker’ in situations of language obsolescence, evaluating their role in the obsolescence (and revitalisation) process.
2.1
Internal Factors
Linguistic changes are, at the outset, most often internally motivated, that is influenced by linguistic structural factors. These changes are commonly referred to in variationist theory as changes from below or innovations: changes that occur within a language variety as a result
© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 D. Mooney, Language and Dialect Death, https://doi.org/10.1007/978-3-031-51101-1_2
13
14
D. Mooney
of language-internal factors and which, in their strictest conceptualisation, occur below the level of conscious awareness. In the context of sound change, we can make a useful distinction between phonetic and phonological change; in short, phonetic changes (referred to as syntagmatic changes) involve a change in the pronunciation of a single sound category or phoneme over time while phonological changes (referred to as paradigmatic changes) involve a change to the phonological system, usually caused by the loss of old, or creation of new, phonemes. Phonetic changes are (usually) phonetically ‘natural’, meaning that they result from repeated minor variations in pronunciation that are due to the unsynchronised timings associated with the dynamic movement of the articulators: ‘a great deal of […] change derives merely from such small adjustments in the movements of the organs of speech’ (Trask 2015: 49). Phonological change, on the other hand, can involve the neutralisation of a phonetic distinction between phonemes (phonemic merger, e.g., merger of the /e/~/E/ distinction to /e/ in wordfinal open syllables in supralocal French), or the creation of a new meaningful distinction whereby one phonemes splits into two (phonemic split, e.g., split of Old English /f/ into /f/ and /v/). Phonemic merger may, in some instances, be caused by system internal pressures related to the functional load of a phoneme; this means that one phoneme may merge with another because the number of meaningful distinctions that rely on its existence are minimal, or it has a low functional load or range of functions. Both phonetic and phonological changes can be classified as ‘unconditioned’ or ‘conditioned’: unconditioned changes apply to every instance of a given phoneme, regardless of its position in the word or of the phonetic or phonological environment (e.g., merger of /œ/ ˜ with /˜E/ in supralocal French); conditioned changes apply to certain instances of the phoneme in particular linguistic contexts (e.g., lenition of Latin /p t k/ to Spanish /b d ɡ/ in intervocalic position). Conditioned sound changes are thus subject to linguistic constraints related to the operation of language-internal structural factors on the change in question. In variationist sociolinguistics, the (initial) progression of changes from below is often conceptualised with reference to the two linguistic processes of transmission and incrementation (Labov 2007). Transmission involves the ‘unbroken sequence’ of native-language acquisition by
2 Theorising Language and Dialect Death
15
children or the faithful replication of the older generations’ language use which, Labov stresses, does not, in the case of linguistic changes in progress, need to be a complete replica. In Labov’s view, children are capable of preserving ‘linguistic descent’ even when replication is imperfect that is when language changes (Labov 2007: 346). Changes from below are thus said to be generated, over time, by the process of incrementation whereby ‘successive cohorts and generations of children advance the change beyond the level of the caretakers and roles models, and in the same direction over many generations’ (Labov 2007: 346). Incrementation therefore takes place when children associate variability in the language of previous generations with the vector of age, advancing the changes further along the same trajectory. Thus, for the changes to proceed, there must be some awareness of their existence but, in the early stages of phonetic and phonological changes from below, they are not conceived to carry overt social value, but rather to result from operation of language-internal factors.
2.2
External Factors
Externally motivated language changes are those changes that result from contact between languages or between speakers of different varieties of the same language. As such, we can make a theoretical distinction between linguistic changes that occur during language contact and dialect contact, even if this distinction is not always applicable in practice. As we will see, the processes involved in language and dialect contact may act in tandem on a linguistic variety, producing changes that cannot necessarily be attributed to a single causative external explanation.
2.2.1 Language Contact Contact between languages frequently leads to the transfer of linguistic structures between them. Weinreich notes that transfer, which he calls ‘interference’, results in ‘the rearrangement of patterns, arising from the introduction of foreign elements into the more highly
16
D. Mooney
structured domains of language, such as the bulk of the phonemic system, a large part of the morphology and syntax, and some areas of the vocabulary’ (1968: 1). The influence exerted by one language on another may vary largely depending on the linguistic level in question which is, in turn, dependent on the nature of the contact situation, particularly on whether bilingualism exists, the degree of bilingualism, and its duration (Hickey 2010: 7). Bilinguals are largely seen to be the locus of change because competence in two languages is a precondition for the adoption of material from one language into another, independently of their degree of awareness of the separateness of their linguistic (sub)systems (Hickey 2010: 9). While some traditional views see the dominant language as influencing the less dominant language, Hickey criticises this view as being overly simplistic because transfer between languages in contact is normally bilateral. The argument that transfer occurs through bilingual speech in a contact situation is not widely contested but the type and amount of interference observed can depend largely on contextual factors. Thomason (2010: 34) discusses what she terms ‘predictors’ in contact scenarios, which describe the conditions under which types of change may take place. A distinction is drawn between contact situations in which one group of speakers shifts to another language and fails to learn it fully and other situations where speakers are fully bilingual and imperfect learning plays no role. The predicted outcomes for each situation are very different with shift-induced transfer occurring primarily at the phonological and syntactic levels with little or no lexical item transfer from the shifting group’s original language into their target language. If the people involved in contact are fluent speakers of both languages, the first and predominant transferred features are lexical items. Later, after the languages have been in contact for some time, structural features may also be transferred (Thomason 2010: 36). We thus have a dichotomy between fully bilingual communities where imperfect learning plays no role, leading primarily to lexical interference while structural feature transfer lags behind, and a monolingual community shifting from one language to another, with imperfect learning through unguided second language acquisition, leading primarily to phonological and syntactic transfer, while lexical item transfer lags behind.
2 Theorising Language and Dialect Death
17
Where there are high levels of stable bilingualism, intensity of contact is typically related to the duration of contact and the level of bilingualism in the speech community (Thomason 2010: 36). These two factors are related in that greater levels of bilingualism will result from longer contact periods. In situations of shift, levels of intensity vary according to factors such as the relative sizes of the populations speaking the source and receiving languages, the degree of access to the target language by the shifting group and the length of time over which the shift occurs (Thomason 2010: 37). If the language shift occurs abruptly, a relatively large amount of phonological and syntactic transfer is likely. Where the shift occurs very slowly, little or no shift-induced transfer can be expected. However, shift-induced transfer may be found in long-term contact situations in which imperfect learning was dominant early on, followed by a high level of bilingualism over a considerable period of time and finally by the completion of the language shift to the incoming language. Having been fixed at a time when the level of bilingualism was low, the language of the shifting group may not converge towards the target language as it was spoken by the original speech community. Thomason provides the example of the French spoken in the originally Breton-speaking community of Île de Groix, where by 1970 only the oldest community members were fluent in Breton but the French spoken there was heavily influenced by Breton in both structure and vocabulary. In contact-induced change, the degree of typological difference between specific subsystems of a source language and a receiving language also helps to predict the kinds of transfer that may occur. Morphology tends to lag behind phonology and syntax in terms of those features that are transferred from one language into another, in both shift-induced transfer and transfer in which imperfect learning plays no role. In cases where languages in contact differ significantly in inflectional and other categories, typological distance is said to provide a barrier to contact-induced change, though not an absolute barrier: when the languages involved in contact are very different typologically but contact is very intense, any linguistic feature can be transferred, but transfer occurs primarily at the phonological and syntactic levels (Thomason 2010: 41). Where typological distance is small, linguistic subsystems for which contact-induced change is rare, such as inflectional
18
D. Mooney
and derivational morphology, may be affected by transfer (Thomason 2010: 40). In short, higher levels of typological similarity will allow more contact-induced transfer to occur and on more linguistic levels; greater typological distance, conversely, will reduce the amount of transfer that can occur. Thomason’s distinction between stable bilingualism and shift-induced bilingualism highlights the acknowledgement, in many language contact studies, of the importance of context for languages undergoing the obsolescence process (see Sect. 2.4) and shows the application of theoretical frameworks used in the study of language contact to the specific context of language death. Less developed, however, is the application of research on second language acquisition to the examination of transfer during long-term language contact. In order to conceptualise the abstract level at which the phonological categories of two languages are equated, Weinreich makes a distinction between the bilingual’s ‘primary’ language and their ‘secondary’ language (1968: 14). He uses these terms, rather than ‘mother tongue’ and its opposite, to highlight the fact that we cannot always assume the primary language to be the native one. For Weinreich, determining which of the two systems was learned first is irrelevant: all that matters is to identify which language is the source language and which language is the recipient language where ‘source’ and ‘recipient’ are directly equated with ‘primary’ and ‘secondary’ (1968: 74). Weinreich states that transfer will arise when a bilingual identifies a phoneme of the secondary system with one of the primary system and, in reproducing it, subjects it to the phonetic rules of the primary language (1968: 14). Thus, a contrastive analysis of the phonemes of two languages and the rules by which they are realised can yield a list of forms which we might expect to transfer in a given contact situation. Of course, both languages in contact can act as the source and recipient for different instances of transfer since there are two ways in which contact can be experienced by bilinguals: one system with reference to another, and the other way round (Weinreich 1968: 14). Flege’s research on second language acquisition using the Speech Learning Model (SLM), however, provides a framework within which to study phonetic and phonological transfer in a more systematic and theoretically informed way. The basic premise of the model is that during
2 Theorising Language and Dialect Death
19
the acquisition of a second language (L2), the process of ‘interlingual identification’ causes words learnt to be (initially) decomposed into categories based on familiar first language (L1) sounds (Flege 1995: 98). This (initial) association of L2 sounds with L1 categories is the basis for the central tenet of the SLM: the categories making up the L1 and L2 subsystems of a bilingual exist in a ‘common phonological space’ and so have the potential to mutually influence one another (Flege 2007: 366). The large majority of research undertaken using the SLM has focused predominantly on second language learners in the traditional sense, speakers who have learnt a language that is not normally spoken in their nation, and who are, in essence, speaking their L2 with a ‘foreign accent’. The central emphasis of many these studies, undertaken primarily by Flege in conjunction with a variety of other researchers, was on the effect of ‘age of learning’ on the perception and production of ‘non-nativelike’ sounds by L2 speakers of a variety of languages. This framework has, however, been fruitfully applied to the study of a situation of long-term societal and cognitive language contact, between French (as L2) and Occitan (as L1), in order to study the transfer of phonetic and phonological features from Occitan into the regional variety of French spoken in southwestern France (Mooney 2016, 2019). The application of the SLM to the study of transfer in regional French illustrates the potential benefits of applying models of language contact that are based on the analysis of short-term imperfect learning in the context of foreign language learning to longer-term contact situations. It is perhaps prudent, here, to note that the concept of ‘imperfect learning’ is at odds with recent developments in the discipline of foreign language teaching that promote a ‘positive psychology’ approach (e.g., Dewaele and Saito 2022). This approach rejects a focus on deficits in learning and problematises the monolingual ideologies that hold the imagined ‘native speaker’ variety in higher esteem.
2.2.2 Dialect Contact When speakers of a language variety come into contact with speakers of other varieties of the same language, we often find that, over time, the differences between the varieties will be reduced. Kerswill calls this
20
D. Mooney
process ‘regional dialect levelling’ (RDL) or supralocalisation, which is claimed to be ‘leading to the loss of localised features in urban and rural varieties of English in Britain, to be replaced with features found over a wider region’ (2003: 223). The phenomenon of RDL arises from the interplay of at least two focusing mechanisms: levelling ‘proper’ and geographical diffusion. Before describing each of these processes in detail, it is first necessary to briefly define ‘speech accommodation’ which theoretically underpins both levelling ‘proper’ and geographical diffusion. Speech accommodation is a social-psychological language change mechanism whereby, in a dyadic situation, speakers converge to, or diverge from, the variety spoken by their interlocutor in order to gain social favour or distinguish themselves socially (Trudgill 1986: 2). Trudgill argues that long-term linguistic changes occurring in dialect contact situations result from multiple such short-term, face-to-face interactions between individual speakers. The linguistic forms which are accommodated to are not necessarily phonetically identical to the original form. Trudgill calls this accommodation ‘imperfection’, explaining that speakers in dyadic situations aim to reduce dissimilarity between their speech and that of their interlocutors while at the same time not imitating them slavishly (1986: 57). We may note that the occurrence of accommodation is dependent on the social network characteristics and the social mobility of the speakers involved with close-knit communities impeding change and loose social network ties favouring higher levels of interaction, more accommodation and thus, more linguistic change. Levelling ‘proper’ is defined as a process promoting the ‘reduction or attrition of marked variants’ (Trudgill 1986: 98), where marked refers to forms that are unusual or in the minority. This means that localised varieties in contact with each other become more like one another and reduce the differences between them. Levelling ‘proper’, facilitated by speech accommodation, often leads to the creation of a new variety, over a wider geographical space that is characterised by the absence of highly localised forms. At the structural level, the outcome of levelling is a ‘reduction in exponents of phonological and morphological categories’ between varieties in contact (Kerswill and Williams 2005: 1024). The linguistic outcomes of levelling ‘proper’ lead, over time, to the adoption
2 Theorising Language and Dialect Death
21
of pan-regional forms: the use of new, regionally specific but geographically widespread features that retain majority regional features at the expense of localised ones. Features that are common to a number of contiguous varieties are more likely to survive the levelling process than those variants that have a specifically localised affiliation. Geographical diffusion is defined as the process by which linguistic features spread out from a populous and economically and culturally dominant centre (Kerswill 2003: 223), which is often, but does not have to be, the capital city. At an individual level, geographical diffusion is also motivated by speech accommodation as features spread via face-to-face interactions with other speakers, from elsewhere, who have already adopted the feature in question. The spread of diffusing linguistic features is wave-like, in that they radiate outwards from a central focus reaching geographically nearby locations before those at greater distances. This wave-like diffusion is often modified by the likelihood that nearby towns and cities will adopt the diffusing features before more rural areas in between (Kerswill 2003: 223). This is termed ‘urban hierarchical diffusion’ which is explained by Trudgill’s ‘gravity model’ (1986): linguistic innovations are said to be leaping or ‘parachuting’ according to a defined hierarchical pattern, beginning in the largest urban centre and spreading to rural areas via smaller and smaller ‘satellite’ towns (Trudgill 1986: 39). Hierarchical diffusion therefore involves a demographic factor—the population size of the communities involved in the interaction—and a geographical factor— the distance between urban centres. We must make a distinction between linguistic features diffusing across geographical space and linguistic features diffusing across social groups within a defined geographical space. Both types of diffusion are underpinned by the process of speech accommodation and both are presented by Labov (2007) as secondary to the process of transmission (see Sect. 2.1), in that diffusing features are seen to replace traditional dialectal features which have been passed down from parent to child: ‘They are the result of a secondary process in which speakers of one particular dialect gain an ascendancy – commercial, political, or cultural – and the ensuing expansion of this dialect wipes out the intermediate forms of the original continuum’ (Labov 2007: 347). These linguistic changes that diffuse geographically or socially are commonly
22
D. Mooney
referred to as ‘change from above’, that is changes that are externally motivated in that they result from contact; in their strictest conception, these changes are said to proceed above the level of conscious awareness. Trudgill stresses the importance of distinguishing between two different dialect contact scenarios, both of which give rise to a reduction in the use of minority linguistic forms via levelling ‘proper’. The first scenario is equivalent to German Dialektausgleich, whereby dialect areas enlarge, and dialect diversity is lost to a certain extent; this is akin to the process of RDL discussed above. The second scenario is parallel to German Nivellierung whereby, in situations of dialect mixing, levelling is characterised by a reduction in the number of forms available in the contact mix (Trudgill 1999). The process of Nivellierung is more commonly described as ‘koinéization’: ‘the linguistic process that occurs when different dialects or closely related subsystems come into contact’ (Siegel 1993: 6). During koinéization, the focusing mechanisms of levelling ‘proper’ and simplification lead to the formation of a new dialect. Three pre-conditions must be present in order for new dialect formation of this type to occur: large-scale in-migration to a relatively compact geographical area; native speakers of different mutually intelligible dialects of the same language; face-to-face interaction between speakers of the different incoming dialects. Thus, like RDL, koinéization is also motivated by speech accommodation but dialect contact is said to have been ‘heightened’ by the presence of in-migration (Kerswill and Williams 2000: 70). It is also heavily influenced by the demography and social network characteristics of the in-migrating populations. Within the koinéization model, the outcomes of levelling ‘proper’ are, in principle, the same as those described for RDL above but the sources of variation are different: during RDL or supralocalisation, levelling ‘proper’ promotes the adoption or retention of majority regional forms at the expense of localised ones; during koinéization, on the other hand, levelling ‘proper’ promotes the retention of majority forms in the dialect mix at the expense of minority forms. Additionally, levelling ‘proper’ generally occurs within a relatively compact geographical space during koinéization and thus lacks the geographically based distribution inherent in the creation of supralocal varieties (Kerswill 2003: 236). Kerswill and Williams (2000) have shown that
2 Theorising Language and Dialect Death
23
the establishment of a new town, such as Milton Keynes, provides the perfect conditions in which to examine the outcomes of koinéization, noting that children are central to linguistic focusing during koinéization as they, over time, filter out the variation from their parents’ collective speech, with focusing occurring in either the second or third generation after initial in-migration (Kerswill and Williams 2000: 67–71). The second focusing mechanism involved in koinéization is ‘simplification’, defined by Trudgill as a process that leads to an increase in morphophonemic regularity (1986: 98). Simplification during koinéization may, in fact, be viewed as a combination of three-related processes (Trudgill 2009: 100): the regularisation of irregularities; an increase in morphological transparency; a loss of redundancy or a reduction in the repetition of information. It goes without saying that a given language variety can find itself in language contact with another typologically distinct language as well as in dialect contact with other regional or supralocal varieties of the same language; both language and dialect contact can, at the same time, cause linguistic variation and change. This interface between the processes and mechanisms involved in language and dialect will be explored in the context of obsolescence in the study of phonetic and phonological variation and change in Gascon presented in this book.
2.3
Extralinguistic Factors
Variationist sociolinguistics examines the extent to which (linguistic and extralinguistic) factors exert a variable effect on linguistic choice (in the case of variation) or on the progression of a linguistic change. These factors, also known as independent variables, are said to influence linguistic variation and change in a probabilistic manner, in that their presence or absence can increase or decrease the likelihood of using a particular linguistic variant, rather than determining categorical usage of one variant in a given context. Linguistic factors are those discussed in Sect. 2.1 (e.g., phonological environment), whereas extralinguistic factors are non-linguistic or social factors such as speaker age, sex, social class, or ethnicity, for example. The way in which extralinguistic
24
D. Mooney
factors have been operationalised in variationist studies has changed over time, as the concept of ‘social meaning’ has been repeatedly revised in variationist theory. First wave variationism was primarily concerned with determining the variable effect of macrosociological category membership on variant choice in situations of stable linguistic variation or linguistic change (from below or above). As such, social meaning was interpreted in terms of the broad correlation between ‘linguistic variables and the macrosociological categories of socioeconomic class, gender, ethnicity and age’ (Eckert 2012: 87). The first wave of variationism proposed data-driven principles of linguistic variation and change. Data collection methods that elicited spontaneous and read speech were used to ‘introduce a new quantitative empiricism into linguistics, with supportive theoretical underpinnings’ (Eckert 2012). Data from Labov’s (1966) study of the social stratification of English in New York City were used to advance testable hypotheses about the nature of language variation and change. Labov showed that the use of standard linguistic forms increased for speakers further up the socioeconomic hierarchy and that the distribution of standard and non-standard forms on the socioeconomic spectrum correlated with their distribution on the stylistic spectrum or that, when speaking more formally, informants emulated, to some extent, the patterns of variability in the social class above them. Labov also identified the lower middle class, those in the middle of the socioeconomic hierarchy, as the leaders of language change because they demonstrated the ‘crossover effect’, exhibiting hypercorrect linguistic behaviour in more formal speech styles. This meant that, when speaking formally, lower middle-class speakers not only emulated the linguistic behaviour of the social class above them, they overshot the target, thus advancing linguistic change from above. Variation on the style spectrum was primarily conceived, in the first wave, as a result of a speaker’s ‘attention paid to speech’: self-monitoring was seen to produce more standard variants in formal styles. With regard to gender, commonly conceived of simply in terms of biological sex, Labov showed that in situations of stable sociolinguistic variation and change from above, women used more standard variants than men, but in situations of change from below, women used more non-standard variants than men.
2 Theorising Language and Dialect Death
25
Labov (2001) formalised this apparent confound as the ‘gender paradox’: ‘women deviate less than men from linguistic norms when they are overtly prescribed, but more than men when the deviations are not prescribed’. In short, Labov showed that women tend to lead language change whether it proceeds from above or below, though these principles have been criticised for being overly simplistic by more recent variationist studies (see Cheshire 2002, for criticism). The use of ‘age’ as an independent variable allowed first wave studies to examine age stratification of linguistic variables, that is how the variables were used differently by different age groups, in order to identify linguistic changes ‘in apparent time’: differences in linguistic behaviour between the different age groups were interpreted, in certain contexts, as indicative of linguistic change that was taking place in real time, with younger speakers frequently being shown to lead the change. In sum, first wave studies identified young lower middle-class women as the leaders of language change and the basis of the correlation between their linguistic behaviour and the combined effect of their macrosociological category membership. The second wave of variationism shifted its focus from ‘attention paid to speech’ to the notion of ‘speaker agency’ or the ‘socioculturally mediated capacity to act’ (Ahearn 2010: 28) and examined the vernacular as an expression of local or class identity (Eckert 2012: 91). This focus on agency led researchers to examine that which speakers had relative control over, the people that they interact with on a regular basis, or their social network. The social networks approach was pioneered by and formalised in variationist sociolinguistics by Leslie Milroy: this approach still sought to establish correlations between linguistic behaviour and marcosociological categories, but shifted its focus to their local negotiation and construction. As such, it sought to look within these categories in an attempt to better account for the variation and change observed. For Milroy (1987: 178), the central object of study was the ‘preexisting social group’, rather than the individual as representative of a more abstract social category. Milroy and Milroy’s (1978) distinction between dense and loose social networks accurately explained within category variation in Belfast, where men and women from the same social class were shown to behave differently, not only on the basis of their sex; the linguistic behaviour of working class people from different
26
D. Mooney
areas of the city was shown to correlate heavily with the nature of their social networks. As such, the social networks approach provided a greater understanding of variation within traditional macrosociological categories like socioeconomic class. The second wave also heralded the integration of a more ethnographic approach to data collection in variationism which focused on the ‘community of practice’: ‘an aggregate of people who come together around mutual engagement in a common endeavour’ (Eckert and McConnell-Ginet 1992: 464). Linguistic behaviour was thus seen to emerge from engagement in this common endeavour and first wave notions such as style-shifting were no longer seen as a function of attention paid to speech, but as part and parcel of a speaker’s work to construct a meaningful social identity (Meyerhoff 2003: 401). In the communities of practice approach, linguistic behaviour was not viewed as independent of, for example, styles of dress or other social and behavioural patterns. Participant observation was used by researchers, such as Eckert (2008), to show the way in which speakers construct social meaning at a local level and, crucially, to show how this local construction of identity can draw on wider macrosociological categories which are sometimes contested and sometimes maintained by the linguistic actions of individuals. In short, the second wave did not seek to do away with the analysis of macrosociological categories, but to look within them: the goal was not ‘to dispense with global categories, but to attach them to personal and community experience in such a way that the structure of variation makes everyday sense’ (Eckert 2000: 222). The first wave of variationism considered stylistic variation as a form of intraspeaker variation (Labov 1972) along a scale of formality. Third wave variationism, on the other hand, conceptualises style in a more speaker-oriented agentive framework. A Labovian definition of style implies that a change in context can induce a style shift, but more recent conceptualisations acknowledge that a style shift can, in and of itself, change the context: ‘style-shifting is primarily a means whereby speakers alter the images of self they project for others. Sometimes these alternations are triggered by changes in the conversational context, but more often they are not; in fact, they often serve, in and of themselves, to
2 Theorising Language and Dialect Death
27
bring about contextual changes’ (Schilling-Estes 1998: 69). This reanalysis of the functions of style-shifting shifts focus from the perception of the speaker by others to the speaker’s agentive performance of self. Third wave variationism draws on the theoretical frameworks proposed by the previous waves, acknowledging correlations between linguistic variables and macrosociological category membership, social networks, etc., but shifting focus to the effect of their use in context: ‘While the individual variables available in a dialect may correlate with various aspects of social membership and practice, most of them take on interpretable social meaning only in the context of the broader linguistic styles to which they contribute’ (Eckert 2000: 213). The use of a single variant of a linguistic variable in context may thus be interpreted as a stance taken by the individual. These stances may index wider correlations between linguistic behaviour and social categorisation, but their specific meaning is only created in context. For example, Kiesling (2009) showed that use of the [-n] variant of the (-ing) variable in English was used by one participant to adopt a laid-back stance and by another to adopt an oppositional stance. This was achieved by indexing traits associated with hard-working masculinity and, more generally, with being working class. This exploitation of indexical order is central to third wave variationism: individual speakers are seen to create meaning locally by adopting a particular stance (e.g., being oppositional). This allows them to create meaning in context by indexing traits associated with, for example, being working class, without actually being a member of the macrosociological category in question. It is important to note that variationist sociolinguistics investigates the variable effect of internal, external, and extralinguistic factors on linguistic variation and change, usually by modelling quantitative data statistically to determine the factors that exert a significant influence on the variation and change observed, as well as interactions between those factors. Indeed, almost all cases of linguistic variation and change are the result of all three types of factor operating at different times or in tandem to produce the variable distribution of the variants involved.
28
2.4
D. Mooney
Language Death
Language obsolescence may be defined as a process whereby a language is ousted from its territory by another language leading to a decline in its number of speakers and ‘during which gradual reduction in use, due to domain-restriction, may result in the emergence of historically inappropriate morphological and/or phonological forms together with extensive lexical borrowing’ (Jones 1998: 5, cf. Brenzinger and Dimmendaal 1992: 3). Language death may be seen as the end point of this obsolescence process which occurs at a superordinate level between distinct language varieties: structural separation and typological dissimilarity between the dominant and obsolescent language are largely maintained. It is important to note that language obsolescence results from a combination of socio-political and linguistic changes and, indeed, the former gives rise to the latter: ‘the decline of a language does not take place in a vacuum, and the sociopolitical setting is often instrumental in precipitating language change’ (Jones and Singh 2005: 81). Sasse (1992a: 10) emphasises the necessity, in studies of language obsolescence, to distinguish three separate issues: external setting, speech behaviour, and structural consequences. It is therefore necessary to consider the socio-political factors that have given rise to the situation of obsolescence, the profile of the speech community and the language’s domains of use, and the linguistic changes that result from the specific contact scenario under investigation. The study of language death has hitherto focused on a variety of theoretical constructs relating to the socio-political, behavioural, and linguistic aspects of the obsolescence process. While language death can occur for a variety of reasons, most analyses have focused on languages undergoing ‘gradual death’ as a result of language shift ‘where a language is lost in a contact situation, with the dominant language gradually ousting the subordinate – often minority – variety in a scenario that typically involves intermediate stages of bilingualism’ (Jones and Singh 2005: 80). The types of speaker involved in the obsolescence process has also been the focus of much attention. The distinction between fluent native speakers, younger fluent speakers, rusty speakers, and semispeakers has been crucial in examining variability in speech behaviour
2 Theorising Language and Dialect Death
29
during language loss. Rusty speakers are a type of ‘imperfect’ speaker who ‘despite having a reasonable knowledge of the grammatical system of a language usually have certain gaps in this area and in their vocabulary’ (Jones and Singh 2005: 85–86; cf. Sasse 1992b: 61), while ‘semi-speakers may be distinguished from fully fluent speakers of any age by the presence of deviations […] which are explicitly labelled ‘mistakes’ by the fully fluent speakers’ (Dorian 1981: 107). The existence of different types of speakers of differing abilities led to the development of the concept of a proficiency continuum in situations of language shift where it is common to find speakers of quite different ability among the residual speaking population (Dorian 1981: 114). Even within the semi-speaker category, Dorian showed that there is a proficiency continuum ranging from competent speakers to those with limited production ability who distinguish themselves from near-passive bilinguals by their ability to manipulate words in sentences (Dorian 1981: 107). In the context of gradual death during language shift, this proficiency continuum is frequently age-graded, with younger speakers demonstrating least control over the dying language. As will be the focus of the present study, there has been much theoretical attention paid to the linguistic structural consequences of language contraction and death. While some scholars have examined the structured obsolescence hypothesis (see Chapter 1), the primary theoretical focus of studies in the discipline has been on the nature of linguistic change during obsolescence. Dying languages have been repeatedly shown to undergo the same types of linguistic change as ‘healthy’ languages. The types of changes that we observe in the structure of a dying language are not notably different from those well established in the study of language change in general, ‘but the time span for change seems compressed and the amount of change seems relatively large’ (Dorian 1981: 154). Thus, in ‘healthy’ languages, a few changes will occur almost imperceptibly over a considerable period of time, but during language obsolescence, it is possible to witness many different types of change over a matter of generations (Dorian 1981: 151). Some early studies likened the processes and mechanisms of change active during language obsolescence to pidginisation (Dressler and Wodak-Leodolter 1977: 37) or to ‘creolisation in reverse’
30
D. Mooney
(Trudgill 1976; Mühlhäusler 1980) due to a reduction in stylistic repertoire and structural complexity in dying languages, and the compressed timespan for changes to occur (Sankoff 1979), but these claims have been refuted on the basis that loss of register during obsolescence does not necessarily imply monostylism (Dorian 1994) and that ‘the simplifications occurring in the context of pidginization are far more drastic that those occurring during obsolescence’ (Jones 1998: 40). Language death theory has also focused heavily on the internal and external motivations for linguistic change, to the extent that such a distinction can be reliably established (Jones and Singh 2005: 88). Structural changes that occur in obsolescent languages may or may not be directly attributable to the influence of an encroaching language. Many changes observed in formal language structure cannot necessarily be explained by external factors (Jones 1998: 249): simplification leading to increases in regularity; the complete loss of structural elements, e.g., the loss of case distinctions in East Sutherland Gaelic (Dorian 1981: 130); increases in transparency or the use of more analytic than synthetic structures; the generalisation of unmarked categories, e.g., loss of soft mutations in Welsh. Other changes may be a result of externally motivated transfer from the dominant language (Jones 1998: 252–257): grammatical interference, e.g., syntactic calques; phonological interference; code-switching; and lexical borrowing. As such, the role of language contact in language obsolescence is not always the go-to explanation for structural change and, indeed, ‘it is also not sufficient to account either for all the general trends or for the differences in retention within given categories’ (Dorian 1981: 152–153). While we now know that ‘the types of changes in formal language structures […] are not notably different from those well established in the study of language change in general’ (Dorian 1981: 154), what are the potential implications of more recently developed theoretical constructs and methodological approaches in variationism (i.e., ‘the study of language change in general’) for our understanding of the processes and mechanisms of linguistic change during language death? A central theoretical construct in variationism that of the ‘variable rule’ has been largely neglected in studies of language obsolescence. This
2 Theorising Language and Dialect Death
31
construct has as its basis the notion of ‘orderly heterogeneity’ (Weinreich et al. 1968: 100), or the postulate that language variation and language change are constrained by a combination of (potentially interacting) internal, external, and extralinguistic factors. Variable rules are ‘abstract optional rules’ which form an integral part of a language variety’s structural description (Cedergren and Sankoff 1974: 333–334) and, from the linguist’s perspective, the variable rule can be considered as ‘the probabilistic modelling and statistical treatment of discrete choices and their conditioning’ (Sankoff 1988: 984). As such, the analysis of linguistic variation and change in dying languages must seek to determine the existence of variable rules governing the variation and change observed. Firstly, it is crucial to investigate the linguistic constraints on variation and change or, put simply, the linguistic structural environments that favour or disfavour the use of a variable linguistic feature, in order to explain the internal motivations for change, rather than adopting a more descriptive ‘X becomes Y’ approach. This avoidance of impressionistic description is also extremely important in the consideration of external motivations for change. Labov (2007) has criticised European dialectological studies for their tendency to analyse contactinduced change in a manner that investigates the transfer of well-known features from one language (or dialect) to another in a ‘X is replaced by Y’ fashion, with near complete disregard for the potential existence of a variable rule which contains information on the internal and extralinguistic constraints on such transfer. While some studies have attempted to avoid this simplistic approach (e.g., Jones 2005; ‘overt’ versus ‘covert’ transfer), there is a marked lack of studies that have examined the linguistic constraints on features transferred during language contact (see Mooney and Hawkey 2019). Equally, the examination of linguistic transfer must take account not only of transfer from the dominant language to the dying language, but also of the processes of dialect levelling and mixing (Jones 1998: 290) between the extant dialects of the threatened language (as in Jones 1998), thus incorporating relevant theories from the study of regional dialect levelling where possible. As early as the 1980s, Dorian stated, in her study of Gaelic, that ‘sociolinguistically many features of language change are not strongly
32
D. Mooney
present […] (notably marked social stratification)’ (1981: 154), indicating that the effect of extralinguistic factors on language variation and change is potentially reduced in situations of language contraction and death. It is necessary, however, to consider the potential role of independent variables such as age (see Sect. 2.6), speaker sex, and social class in conditioning language variation and change and, crucially, the potential interactions between these extralinguistic factors and language-internal factors. It is true, however, that social stratification of the type usually studied in urban variationist studies is frequently absent in the specific context of language death and that any stratification that exists is linked to an urban-rural dichotomy that, in turn, is heavily correlated with the distinction between new and native speakers (see Sect. 2.6). Finally, the study of language obsolescence would potentially benefit from a more speaker agency-oriented analytical framework, which seeks to establish correlations between language use and a speaker’s social networks or communities of practice. Equally, the extent to which speaker identity is performed, either by adopting specific stances within the dying language, or by switching between the dominant and dying language may reveal more clearly the role of the individual in language death, though this ethnographic approach is regrettably beyond the scope of the Gascon study presented in this monograph.
2.5
Dialect Death
From a typological perspective, we have seen that the level of structural similarity between languages in contact determines the nature and amount of linguistic transfer that may occur. Linked to typological similarity are the separate but related processes of language death and dialect death. Dialect death, also known as ‘dedialectalisation’, is the process by which a specific localised variety of a language becomes obsolescent and dies out. It thus occurs on a subordinate level and does not involve a ‘whole’ language. For the purpose of this discussion, I follow Dorian in defining a dialect linguistically, as a particular local variety of a larger superordinate language (1981: 8). Dialect death has been defined in a
2 Theorising Language and Dialect Death
33
variety of different ways, three of which will be discussed below, although these definitions are by no means exhaustive. Dorian described East Sutherland Gaelic as a particular local dialect of the Scottish Gaelic language that is yielding to an equally particular local dialect of the English language, East Sutherland English (1981: 8). Scottish Gaelic will not become extinct with the loss of East Sutherland Gaelic and by this definition, only a particular local variety, and not a language will have died. Dorian herself points out that this definition of dialect death is problematic: the whole Scottish Gaelic language is threatened by (Scottish) English at a superordinate level. Therefore, the competition, at a national (Scottish) level, is between two typologically distinct languages, Gaelic and English; at a local level, the question is always which language, and not which dialect, to speak. Since, in the local context, the death of East Sutherland Gaelic will mean that a language, and not a dialect, will have died, this situation may more accurately be described as one of language death. By this definition, ‘dialect death’ is just one localised step in the language shift process: the gradual elimination of all dialects of Gaelic by English. A second type of dialect death sees a particular local variety of a language being replaced by a supralocal or (standardised) national variety of the same language. In this case, dedialectalisation, at a subordinate level, leads to the replacement of local features with other non-local features of the same language. For example, Jones notes that since the introduction of Welsh-medium education in Wales, ‘there is now occurring a phenomenon whereby local Welsh is being replaced by a Welsh that includes non-local features, most of them features of Standard Welsh but some also associated with other regional dialects’ (1998: 260). In the case of Wales, Standard Welsh is the result of a standardisation process which involved dialect mixing, rather than the selection of a single dialect to form the basis of the standard. It is also worth noting that, in the case of Welsh, as with Scottish Gaelic, these local varieties are also threatened by English at a superordinate level. This type of dialect death is primarily the result of the standardisation of the dying language, which can proceed in a variety of ways, and of the implementation of this standard variety in educational contexts (see Sect. 2.6). Jones (1998: 289–290) describes this type of dialect death in terms of ‘dialect
34
D. Mooney
convergence’ or as a unidirectional process whereby the local dialects of the dying language give up their idiosyncratic features which are being replaced by those of the standard language. The process of dialect convergence is said to exhibit some interesting similarities with the process of koinéization (Jones 1998: 289) in that they both involve ‘the levelling out of minority and otherwise marked speech forms, and of simplification, which involves, crucially, a reduction in irregularities’ (Trudgill 1986: 107). We must note, however, that dialect convergence (towards standard Welsh) in Wales appears to be occurring in all local dialects and thus distinguishes itself from koinéization by its lack of restricted spatial localisation. A third definition of dialect death falls somewhere between those of Dorian and Jones. This type of dialect death, which comprises both language shift and dedialectalisation, has been discussed by Hornsby (2006) in relation to Picard and French which evolved, largely independently, from Latin and are historically bound to the regions of Ile-de-France1 and Picardy respectively. They are both classified as langues d’oïl and they exhibit many structural similarities common to varieties of northern Gallo-Romance. Following the imposition of French as the national language, language contact led to high levels of linguistic convergence towards French in Picard because of the typological similarity between the languages. This convergence has been so great that ‘it has fostered the perception that Picard is no more than a debased subvariety of French’ (Hornsby 2006: 9). On this observation, Hornsby treats Picard obsolescence as a case of dialect death rather than language death. This ‘dialect’ has subsequently been ousted from its territory by French by the process of dedialectalisation, in a process similar to that by which regional varieties of Welsh are yielding to non-local Welsh forms. This definition of dialect death is, however, dependent on delineating some form of boundary at which one language becomes another, which is notoriously difficult in the case of contact between varieties of northern Gallo-Romance.
1
The emergence of French or francien also involved high degrees contact and mixing with other northern Gallo-Romance varieties. It is thus not as straightforwardly localised as Picard (cf. Lodge 1993).
2 Theorising Language and Dialect Death
35
As with language death, we must assess the potential contribution of variationist theories of linguistic variation and change to the study of dialect death. The concept of the variable rule is again of utmost importance not only to the internal mechanisms governing linguistic changes in local dialects of dying languages but to the analysis of constraints on the transfer of linguistic material between dialects in contact. While studies of ‘healthy’ languages have tended to examine the results of regional dialect levelling over extensive areas of geographical space (e.g., Mooney 2016), and the results of koinéization in the context of new towns (Kerswill and Williams 2000), the linguistic mechanisms that underpin these theoretical processes (levelling ‘proper’; social and geographical diffusion; simplification) are undoubtedly relevant to the analysis of contact between local dialects of dying languages (e.g., dialect mixing) and contact between these local dialects, which share common structural features, and an artificially standardised ‘dialect’ (i.e., language) that is propagated, primarily, via the education system. As noted above in the discussion of language death, the role of speaker variables, or extralinguistic factors is commonplace in the study linguistic change during dialect contact (age, sex, social class, etc.) are often under-represented in studies of dialect death; investigating the effect of such factors will better inform the analysis of contact-induced change by formalising holistic variable rules that account for the nature of the transfer observed.
2.6
New Speakers
Linked to the process of dialect death is the process of standardisation, which usually involves selection of a variety to standardise, codification of the grammatical and phonological systems in dictionaries and grammars, elaboration of the lexicon for use in a variety of domains, and status planning to promote the acceptance and use of the standard language (Haugen 1966). The process of standardisation, as conceptualised by Haugen (1966), has been revised, refined, and critiqued by other scholars (e.g., Joseph et al. 2020; Ayres-Bennett and Bellamy 2021). Nonetheless, for Fishman, ‘in the modern world, standard dialects are written languages and they have definite written conventions, as far as writing
36
D. Mooney
system, orthography and grammatical structure’ (1991: 346). Dialectal variation is frequently viewed negatively by lay-speakers because it may give the impression that the speech community is fragmented both linguistically and in terms of its identity. While standardisation is often considered a solution to this problem, the selection of linguistic forms to form the basis of a standard language can emerge as a contentious and divisive issue, particularly in the context of language endangerment and revitalisation; the standard language may be seen to privilege speakers of one dialect over others (Grenoble and Whaley 2006: 102). When endangered languages exhibit high levels of diatopic variation, or when multiple dialects are seen to be in competition, several options are available during the process of linguistic standardisation: (i) the unilectal approach, or selecting one particular dialect for standardisation; (ii) the dialectal approach, or creating multiple standards; (iii) the multilectal approach, or creating a standard that contains linguistic features from a number of dialects; (iv) the common core approach, or creating a standard that emphasises linguistic features common to all dialects. Jones makes a theoretical distinction between orientation (of local dialects) towards a standard language of type (i), which she calls standardisation, and orientation towards a standard language of type (iv), which she calls ‘dialect convergence’ because ‘it differs from most documented cases of standardization in that the proliferated variety is not a regional one, elevated through prestige, but rather a non-geographically locatable amalgam in which the regional dialects have been reduced to a common core’ (1998: 290). The difference between ‘language standards’ and ‘standard language’ (Joseph 1987) is also of relevance here in that, instances of uniformity are not always related to the process of standardisation. Whatever the approach adopted, successful corpus planning may constitute a ‘powerful tool’ towards achieving the goal of reversing language shift if the standardised orthography can be read and understood by a large number of speakers (Grenoble and Whaley 2006: 130). Indeed, Pickl (2020) emphasises the need to focus on coherence between, rather than on individual linguistic features of, the dialects. In a great many situations of language death, native speakers are frequently illiterate in the dying language, usually because the received no formal instruction in this language or because the orthographic
2 Theorising Language and Dialect Death
37
and linguistic norms of the standard are not adequately understood by the speaking population. As part of language revitalisation efforts, the newly developed standard frequently functions as a tool for teaching the language and for the creation of a new speaking population, commonly referred to as ‘new speakers’, or individuals ‘with little or no home or community exposure to a minority language but who instead acquire it through immersion or bilingual education programs, revitalization projects or as adult language learners’ (O’Rourke et al. 2015: 1). These speakers constitute another theoretical reference point in the hierarchy of obsolescent language speakers discussed above. Unlike native speakers, younger fluent speakers, rusty speakers, and semi-speakers, however, recent theorisation of the new speaker category has not defined them explicitly in terms of the historical appropriateness of their language production, but instead has been more qualitative in nature, focusing on ideological authenticity, legitimacy, and power relations with other speaker types (Kasstan 2017: 2). In spite of this, frequent claims have been made, themselves perhaps ideological in nature, that the speech forms used by new speakers are ‘far removed from community norms’ (Kasstan 2017: 2). These ‘community norms’ are frequently defined, in minority language contexts, not as the historical norm, but as the ‘usage of the oldest and best speakers’ (Dorian 1981: 117). Dorian takes the view that the linguistic benchmark for the ‘norm’ in a dying language should not constitute the standard historical representations of the language but rather the internal conservative norm represented by the oldest local speakers (1981: 80) and this methodological approach is widely accepted in the field. As the language shift advances, the historical norm serves as little more than a theoretical construct representing some ‘pure’ past version of the obsolescent language. Dorian’s definition of norms is primarily based on her categorisation of fluent native speakers. At this point, it is worth noting that the concept of a ‘native’ speaker, at least with respect to the common-sense definition of the term, is frequently ill-defined and problematic (Davies 2003: 24; Dewaele and Saito 2022: 225). For our purposes, we may follow Dorian in defining ‘native’ as the oldest fluent speakers who acquired the obsolescent language as their first language (L1), as opposed to ‘new’ speakers, who are in essence, second language learners (L2), frequently
38
D. Mooney
speaking the dominant language, to which the obsolescent language is yielding, as their L1. While new speakers typically acquire the obsolescent language in a purely educational context, they can be distinguished from L2 speakers, the object of study in second language acquisition research, in socio-political terms: ‘in cases where severe endangerment is coupled with embryonic revitalisation efforts, new speakers not only represent an important proportion of the total speakers of the language, but they are also influential arbiters in emergent normative practices’ (Kasstan 2017: 3). The result of this is that new speakers frequently occupy authoritative positions in the social hierarchy of the language, alongside other fluent (frequently ‘native’ speakers), in a way which typical L2 speakers would not (Nance et al. 2016: 168). Authority is, however, frequently linked to mastery (or perceived mastery) and, much like the category of semi-speakers, current conceptualisations of new speakers view them as ranging along a proficiency continuum ‘ranging from second language learners with limited competence […] right up to expert L2 users’ (O’Rourke and Ramallo 2013: 288). The socio-demographic characteristics of new speakers are frequently markedly different from those of native speakers, at least in the Western European context: while new speakers are frequently ‘middleclass, urban-dwelling, well-educated and highly politicised’ (Jones 1998: 129), native speakers tend to be found in rural areas where they are (or were) engaged in (rural) working class occupations. As such, new speakers are frequently perceived to be spatially (and therefore ideologically) removed from the ‘speech community’ or, put simply, from native speakers and, by extension, from their linguistic norms (O’Rourke and Ramallo 2011: 139). Jones (1998: 257) has described the speech of new speakers as a potential ‘xenolect’, or as a standardised non-local variety in which dialectal variation is largely absent, noting that, in Welsh and Breton at least, the xenolect is spread through the education system. She notes, however, that, in Wales, the speech of adult L2 learners, or new speakers, retained more local or dialectal features than the speech of English mother-tongue children acquiring Welsh in an immersion education context (Jones 1998: 144); both adult and child new speakers were, however, ‘highly standardized’ relative to native-speaker dialectal norms. New speakers are likely, however,
2 Theorising Language and Dialect Death
39
to exhibit some oral dialectalisms as a result of exposure to non-local phonological and lexical features in the speech of their teacher (Jones 1998: 154), who may potentially be from another dialect area. Overall, however, it is possible, at least for highly competent new speakers, that deviance from the conservative norms of native speakers is perceived in terms of their use of standard linguistic forms (and consequent lack of dialectal forms), rather than in terms of incorrect usage, as is the case for semi-speakers. If this is the case, new speakers may be somewhat aligned with Dorian’s younger fluent speakers ‘who show departures from the most conservative norms […] while still speaking a fluent, expressive, and versatile Gaelic’ (1981: 116). In the case of Gaelic, however, Dorian states that ‘the speech community does not recognise younger fluentspeaker speech performance’ (1981: 107); the perception of deviance in the speech of new speakers therefore aligns them, despite potential accuracy in the standardised language, with younger fluent speakers, whose speech performance is not overtly recognised by native speakers. Quantitative variationist studies seeking to establish the nature of new speaker speech production, and the role of new speakers in processes of language variation and change, are thin on the ground. In Dorian’s (1981) study of Gaelic, the categories of older fluent speakers, younger fluent speakers, and semi-speakers were ordered by age and she notes that ‘changes which appear across speaker groups within the proficiency continuum are also changes in “apparent time” (Labov 1966)’ (1981: 117). While this approach attempts to replicate an important first wave variationist methodology, it is worth noting that, at least where the deviant semispeakers are concerned, these changes identified by Dorian are not canonical changes from below in that they are not always subject to gradual incrementation associated with the vector of age in the transmission process. In the absence of intergenerational transmission, comparison of native and new speaker norms could nonetheless indicate the variety of the language that will survive since, due to the fact that new speakers are almost invariably younger than native speakers, language revitalisation is frequently dependent on the survival of the xenolect. Indeed, there is evidence from third wave variationist approaches to
40
D. Mooney
suggest that new speakers may view native speaker varieties as ideologically inauthentic, orienting their ‘ideal self ’ (Nance et al. 2016: 185) towards the new speaker model. In the context of overt performance of new speaker norms, a question that remains to be answered is ‘do these forms then penetrate native-speaker networks?’ (Kasstan 2017: 8). Jones (1998: 147–148) does note, for example, that ‘historically inappropriate usage on the part of English mother-tongue pupils may be influencing the speech of their Welsh mother-tongue counterparts’, noting that, in the context of the revitalisation of French in Canada, Mougeon and Beniak have shown that ‘some linguistic aspects of restricted and semi-restricted speaker performance may “rub off ” on the speech of un-restricted speakers if the latter have become a small group relative to the former as is typical of language death and advanced language shift situations’ (1991: 14). The extent to which this is the case across situations of language obsolescence certainly warrants further instrumental and quantitative investigation. This chapter has discussed the potential impact of integrating central theoretical concepts and methodological approaches from variationist sociolinguistics into the study of language and dialect obsolescence. Dying languages have been repeatedly reported to show ‘astonishing persistence in the face of great pressure’ (Dorian 1981: 9) and are thus living languages while they still exist. It is no surprise, then, that they exhibit linguistic changes akin to those that we see in ‘healthy’ languages, even though the rate and amount of changes are frequently modified by socio-political factors. It is therefore crucial that we examine changes occurring during obsolescence with the same theoretical and methodological rigour as changes occurring in ‘healthy’ languages. Indeed, while the integration of variationist theory further into the study of language contraction and death may be fruitful, it is highly possible that this approach will, in turn, inform the central theoretical frameworks of variationism. In particular, the complex interface between language contact and dialect contact can be explored by investigating the outcomes of language and dialect death. Jones notes that ‘studying the processes of language and dialect death in this way, it may even be possible to demonstrate that the phenomena are linked, or at least, that they have some bearing on each
2 Theorising Language and Dialect Death
41
other’ (1998: 2). The existence of situations where both language and dialect death are occurring simultaneously within the same speech community provides an exciting opportunity to explore the complex mechanisms of language variation and change at an accelerated speed.
References Ahearn, Laura M. 2010. Living language: An introduction to linguistic anthropology. Oxford: Blackwell. Ayres-Bennett, W., and J. Bellamy, eds. 2021. The Cambridge handbook of language standardization. Cambridge: Cambridge University Press. Brenzinger, Matthias, and Gerrit J. Dimmendaal. 1992. Social contexts of language death. In Language death: Factual and theoretical explorations with special reference to East Africa, ed. Matthias Brenzinger, 3–5. Berlin, New York: De Gruyter. Cedergren, Henrietta J., and David Sankoff. 1974. Variable rules: Performance as a statistical reflection of competence. Language 50: 333–355. Cheshire, Jenny. 2002. Sex and gender in variationist research. In The handbook of language variation & change, ed. J. K Chambers, Peter Trudgill, and Nathalie Schilling-Estes, 423–443. Oxford: Blackwell. Davies, Alan. 2003. The Native Speaker: Myth and Reality. Clevedon: Multilingual Matters. Dewaele, J.M., and K. Saito. 2022. Positive psychology can help overcome the pernicious native speaker ideology. The European Educational Researcher 5: 225–234. Dorian, Nancy C. 1981. Language death: The life cycle of a Scottish Gaelic dialect. Philadelphia: University of Pennsylvania Press. Dorian, Nancy C. 1994. Stylistic variation in a language restricted to privatesphere use. In Small-language fates and prospects, ed. Nancy C. Dorian, 309– 328. Leiden: Brill. Dressler, Wolfgang, and Ruth Wodak-Leodolter. 1977. Language preservation and language death in Brittany. International Journal of the Sociology of Language 12: 33–44. Eckert, Penelope. 2000. Linguistic variation as social practice: The linguistic construction of identity in Belten High. Malden, MA: Blackwell.
42
D. Mooney
Eckert, Penelope. 2008. Variation and the indexical field. Journal of Sociolinguistics 12: 453–476. Eckert, Penelope. 2012. Three waves of variation study: The emergence of meaning in the study of sociolinguistic variation. Annual Review of Anthropology 44: 87–100. Eckert, Penelope, and Sally McConnell-Ginet. 1992. Think practically and look locally: Laguage and gender as community-based practice. Annual Review of Anthropology 21: 461–490. Fishman, Joshua A. 1991. Reversing language shift: Theoretical and empirical foundations of assistance to threatened languages. Clevedon: Multilingual Matters. Flege, James E. 1995. Second-language speech learning: Theory, findings, and problems. In Speech perception and linguistic experience: Issues in crosslanguage research, ed. Winifred Strange, 229–273. Timonium, MD: York Press. Flege, James E. 2007. Language contact in bilingualism: Phonetic system interactions. In Laboratory Phonology 9, ed. Jennifer Cole and José Ignacio Hualde, 353–380. Berlin: Mouton de Gruyter. Grenoble, Leonore A., and Lindsay J. Whaley. 2006. Saving languages: An introduction to language revitalization. Cambridge: Cambridge University Press. Haugen, Einar I. 1966. Language conflict and language planning: The case of modern Norwegian. Cambridge, MA: Harvard University Press. Hickey, Raymond. 2010. The handbook of language contact. Oxford: Blackwell. Hornsby, David. 2006. Redefining Regional French: Koinéization and dialect levelling in northern France. Oxford: Legenda. Jones, Mari C. 1998. Language obsolescence and revitalisation: Linguistic change in two sociolinguistically contrasting Welsh communities. Oxford: Clarendon Press. Jones, Mari C. 2005. Transfer and changing linguistic norms in Jersey Norman French. Bilingualism: Language and Cognition 8: 159–175. Jones, Mari C., and Ishtla Singh. 2005. Exploring language change. New York: Routledge. Joseph, J.E. 1987. Eloquence and power: The rise of language standards and standard languages. London: Frances Pinter. Joseph, J.E., G. Rutten, and R. Vosters. 2020. Dialect, language, nation: 50 years on. Language Policy 19: 161–182. Kasstan, Jonathan. 2017. New speakers: Challenges and opportunities for variationist sociolinguistics. Language and Linguistics Compass 11: 1–16.
2 Theorising Language and Dialect Death
43
Kerswill, Paul. 2003. Dialect levelling and geographical diffusion in British English. In Social dialectology, ed. David Britain and Jenny Cheshire, 223– 243. Amsterdam: John Benjamins. Kerswill, Paul, and Anne Williams. 2000. Creating a new town koiné: Children and language change in Milton Keynes. Language in Society 29: 69–115. Kerswill, Paul, and Anne Williams. 2005. New towns and koinéization: Linguistic and social correlates. Linguistics 43: 1023–1048. Kiesling, Scott. 2009. Style as stance: Stance as the explanation for patterns of sociolinguistic variation. In Stance: Sociolinguistic perspectives, ed. Alexandra Jaffe, 171–194. Oxford: Oxford University Press. Labov, William. 1966. The social stratification of English in New York City. Washington, DC: Center for Applied Linguistics. Labov, William. 1972. Sociolinguistic patterns. Philadelphia: University of Pennsylvania Press. Labov, William. 2001. Principles of linguistic change: Social factors. Oxford: Blackwell. Labov, William. 2007. Transmission and diffusion. Language 83: 344–387. Lodge, R. Anthony. 1993. French: From dialect to standard . New York: Routledge. Meyerhoff, Miriam. 2003. Communities of practice. In The handbook of language variation & change, ed. J.K. Chambers, Peter Trudgill, and Nathalie Schilling-Estes, 526–548. Oxford: Blackwell. Milroy, Leslie. 1987. Language and social networks. Oxford: Blackwell. Milroy, James, and Leslie Milroy. 1978. Belfast: Change and variation in an urban vernacular. In Sociolinguistic patterns in British English, ed. Peter Trudgill, 19–36. London: Arnold. Mooney, Damien. 2016. Southern regional French: A Linguistic analysis of language and dialect contact. Oxford: Legenda. Mooney, Damien. 2019. Phonetic transfer in language contact: Evidence for equivalence classification in the mid-vowels of Occitan-French bilinguals. Journal of the International Phonetic Association 49: 53–85. Mooney, Damien, and James Hawkey. 2019. The variable palatal lateral in Occitan and Catalan: Linguistic transfer or regular sound change? Journal of French Language Studies 29: 281–303. Mougeon, Raymond, and Edouard Beniak. 1991. Linguistic consequences of language contact and restriction: The case of French in Ontario, Canada. Oxford: Clarendon Press. Mühlhäusler, Peter. 1980. Structural expansion and the process of creolization. New York and London: Academic Press.
44
D. Mooney
Nance, Claire, Wilson McLeod, Bernadette O’Rourke, and Stuart Dunmore. 2016. Identity, accent aim, and motivation in second language users: New Scottish Gaelic speakers’ use of phonetic variation. Journal of Sociolinguistics 20: 164–191. O’Rourke, Bernadette, and Fernando Ramallo. 2011. The native-non-native dichotomy in minority language contexts: Comparisons between Irish and Galician. Language Problems & Language Planning 35: 139–159. O’Rourke, Bernadette, and Fernando Ramallo. 2013. Competing ideologies of linguistic authority amongst new speakers in contemporary Galicia. Language in Society 42: 287–305. O’Rourke, Bernadette, Joan Pujolar, and Fernando Ramallo. 2015. New speakers of minority languages: The challenging opportunity—Foreword. International Journal of the Sociology of Language 231: 1–20. Pickl, S. 2020. Factors of selection, standard universals, and the standardization of German revitalisers. Language Policy 19: 235–258. Sankoff, Gillian. 1979. Linguistic variation in pidgin-creole studies. Paper presented at Conference on theoretical orientations in creole studies, St. Thomas, US Virgin Islands, 28–31 March. Sankoff, David. 1988. Sociolinguistics and syntactic variation. In Linguistics: The Cambridge survey: Volume IV. Language: The socio-cultural context, ed. Frederick H. Newmeyer, 140–161. Cambridge: Cambridge University Press. Sasse, Hans-Jürgen. 1992a. Theory of language death. In Language death: Factual and theoretical explorations with special reference to East Africa, ed. Matthias Brenzinger, 7–30. Berlin, New York: De Gruyter. Sasse, Hans-Jürgen. 1992b. Language decay and contact-induced change: Similarities and differences. In Language death: Factual and theoretical explorations with special reference to East Africa, ed. Matthias Brenzinger, 59–80. Berlin, New York: De Gruyter. Schilling-Estes, Natalie. 1998. Investigating ‘self-conscious’ speech: The performance register in Ocracoke English. Language in Society 27: 53–83. Siegel, Jeff. 1993. Introduction: Controversies in the study of koinés and koinéization. International Journal of the Sociology of Language 99: 5–8. Thomason, Sarah. 2010. Contact explanations in linguistics. In The handbook of language contact, ed. Raymond Hickey, 31–47. Oxford: Blackwell. Trask, Larry. 2015. Historical linguistics. New York: Routledge. Trudgill, Peter. 1976. Creolization in reverse: Reduction and simplification in the Albanian dialects of Greece. Transactions of the Philological Society 75: 32–50. Trudgill, Peter. 1986. Dialects in contact. Oxford: Blackwell.
2 Theorising Language and Dialect Death
45
Trudgill, Peter. 1999. The dialects of England . Oxford: Blackwell. Trudgill, Peter. 2009. Sociolinguistic typology and complexification. In Language complexity as an evolving variable, ed. Geoffrey Sampson, David Gil, and Peter Trudgill, 99–109. Oxford: Oxford University Press. Weinreich, Uriel. 1968. Languages in contact: Findings and problems. The Hague: Mouton. Weinreich, Uriel, William Labov, and Marvin I. Herzog. 1968. Empirical foundations for a theory of language change. In Directions for historical linguistics: A symposium, ed. Winfred P. Lehmann and Yakov Malkiel, 95–195. Austin, TX: University of Texas Press.
3 Research Context: Southern Gallo-Romance
This chapter begins with an external overview of the history of southern Gallo-Romance and its dialects, focusing in particular on the historical development of the variety of Gascon spoken in the region of Béarn, southwestern France. Historical language contact with French, leading to the obsolescence of Gascon, is discussed as well as the modern sociolinguistic profile of the Gascon speech community in Béarn. This chapter will also outline the standardisation of southern Gallo-Romance in the form of Occitan which has, in some ways, led to ideological conflict among militants in the region of Béarn. Finally, the educational opportunities available to become a new speaker of Occitan, as part of the Occitan revitalisation movement, are critically evaluated.
© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 D. Mooney, Language and Dialect Death, https://doi.org/10.1007/978-3-031-51101-1_3
47
48
3.1
D. Mooney
The Dialectalisation of Southern Gallo-Romance
By the tenth century, the varieties descended from the Vulgar Latin spoken in Gaul had become strongly diversified along regional lines, leading to the development of three broad dialect areas for GalloRomance (see Fig. 3.1). While this regional diversity is, to a certain extent, attributable to the settlement of large numbers of non-Latin speaking migrants in Roman Gaul during the fifth and sixth centuries, Lodge notes that varying rates of romanisation in the territory and contact between Latin and various indigenous substrate languages, such as Gaulish or Aquitainian varieties, meant that even before the Germanic invasions, the Latin spoken in Gaul was not uniform (1993: 79). These external influences led to the development of three broad dialect areas and within those dialect areas, the development of sub-dialect areas (Hawkins 1993: 58). The most significant division within GalloRomance is between the langue d’oc, or ‘Occitan’, in the south, and the langue d’oïl , in the north. The third, eastern area, francoprovençal , shares features of both northern and southern dialects, as well as independent developments which distinguish it from both. Within the langue d’oc area, many dialectal and sub-dialect divisions have been proposed, as well as supra-dialectal classifications that group and re-group the langue d’oc varieties. The modern langue d’oc area is commonly divided into six main dialects (see Fig. 3.2): Gascon in the southwest; central Lengadocian; Lemosin and Auvernhat in the north of the langue d’oc region; Provençal in the southeast; Vivaro-Alpin or provençal alpin, north of the Provençal region. Additional principal dialects have been proposed, e.g., l’aquitain (Ford 1921: 1; Coustenoble 1945: 11), but this would be included within the Gascon dialect area by most other classificatory systems. The modern langue d’oc dialect area covers thirty-one French départements and extends beyond France’s national borders (Laroussi and Marcellesi 1993: 90), with langue d’oc varieties also spoken in about 16 valleys in the provinces of Cuneo and Turin in Italy (Oliviéri and Sauzet 2016: 319), in the Val d’Aran in the Generalitat de Catalunya, Spain, and in Guardia Piemontese (La Garda)
3 Research Context: Southern Gallo-Romance
49
Fig. 3.1 The Gallo-Romance languages
in Calabria, southern Italy (Bec 1963: 13; Oliviéri and Sauzet 2016: 319). The six main langue d’oc dialects have been shown, to a certain extent, to have a common structural base, but there is a vast amount of linguistic diversity attested over geographical space with varieties at opposite ends of the continuum exhibiting substantial structural differences in their spoken form. They are, for this reason, frequently grouped together on the basis of common structural developments. Bec (1963: 37), for example, proposes regrouping the langue d’oc into three supra-dialectal divisions: nord-occitan, including Lemosin, Auvernhat, and VivaroAlpin; occitan moyen, including Lengadocian and Provençal; Gascon stands alone. Oliviéri and Sauzet (2016: 319) mirror this classification, making a distinction between northern Occitan, southern Occitan, and Gascon. Sampson proposes four supra-dialectal groupings, separating Lengadocian, Provençal, and Gascon, but maintaining a nord-occitan group on the basis that Lemosin, Auvernhat, and Vivaro-Alpin ‘form something of a linguistic continuum with a good deal of overlapping of linguistic features’ (1999: 139). Wheeler (1988: 246), on the other hand, proposes a bipartite supra-dialectal division between northeastern
50
D. Mooney
Fig. 3.2 The Gallo-Romance dialect areas
Occitan, including Lemosin, Auverhnat, Vivaro-Alpin, and Provençal, and southwestern Occitan, including Lengadocian and Gascon. In Bec’s (1963) proposed supra-dialectal groupings for Occitan, nord-occitan, occitan moyen, and Gascon, he, in fact, also includes Catalan in his taxonomy. Although the classification of Occitan as Gallo-Romance and Catalan as Ibero-Romance is common, some scholars criticise the justification for this decision as superficial. Posner is not alone in maintaining that ‘the end-result looks suspiciously like areal grouping’ (1996: 24), and Judge (2007) defends the notion that Catalan and Occitan should in fact form their own separate sub-branch of Romance. Many studies, including Bec (1963) and Sumien (2006), have emphasised the common diachronic developments in, and strikingly similar synchronic structures of, Occitan and Catalan: ‘whereas a “Proto-Occitan-Catalan” is a quite plausible concept, a “Proto-Occitan-French” (excluding Catalan) certainly is not’ (Harris 1988: 16). On this basis, ‘Occitano-Romance’ has been proposed as a sub-language family in its own right, distinct
3 Research Context: Southern Gallo-Romance
51
from Gallo- and Ibero-Romance, or, in some cases, as a sub-grouping within Gallo-Romance. Bec (1963: 55–58) takes this argument further, proposing two supra-dialectal areas within Occitano-Romance, on the basis of common phonological developments in each area: alvernoméditerranéen (later arverno-méditerranéen; Bec 1973: 18–19), including nord-occitan and Provençal, and aquitano-pyrénéen, including (Pyrenean) Lengadocian, Gascon, and Catalan. This division mirrors, to some extent, the northeastern-southwestern supra-dialectal areas proposed by Wheeler (1988). Within the aquitano-pyrénéen grouping, which is said to be ‘centré autour du gascon’ (‘centred around Gascon’) (Bec 1973: 18), geographically restricted Pyrenean Lengadocian is identified as ‘le “pont” naturel entre gascon et catalan’ (‘the natural “bridge” between Gascon and Catalan’) (Bec 1973: 19), but the rest of the Lengadocian dialect is singled out as particularly conservative, emphasising the structural parallels between the Gascon dialect and Catalan. Regardless, of the dialectal supra-regional groupings, within each dialectal area, many sub-dialectal divisions and classifications have been proposed. For the three northern varieties, nord-occitan, various sub-dialects can be identified: ‘le limousin a pour sous-dialectes: le bas-limousin, le haut-limousin, le périgourdin et le marchois. L’auvergnat a pour sous-dialectes: le cantalien, le limagnien, le velaunien et le forézien. Le [Vivaro-Alpin] a pour sous-dialectes: le briançonnais, le diois, le valentinois et le vivarais’ (‘Lemosin has the following sub-dialects: le bas-limousin, le haut-limousin, le périgourdin et le marchois. Auvernhat has the following sub-dialects: le cantalien, le limagnien, le velaunien et le forézien. Vivaro-Alpin has the following sub-dialects: le briançonnais, le diois, le valentinois et le vivarais’) (Ford 1921: 1). There is substantial variation regarding these sub-dialectal divisions, both in terms of their geographical limits and, indeed, in terms of the names that the sub-dialects are given. Many further sub-divisions have also been proposed; for example, Quint (1999: 1–2), working on the a variety of Vivaro-Alpin that he calls albonnais, provides the following classification framework: Occitan (langue) > Vivaro-Alpin (dialecte) > vivarodauphinois (sous-dialecte) > vivaro-vellave (groupe de parlers) > boutiérot (sous-groupes de parlers) > albonnais (parler ). The northern limits of the nord-occitan area constitute a transition zone, often referred to as the
52
D. Mooney
croissant, between the langue d’oc and langue d’oïl , where intermediate linguistic forms are commonly found (Bec 1963: 13). Bec (1963: 45) proposes four sub-dialectal groupings within Lengadocian: languedoccien méridional (central, toulousain, fuxéen, donésanais, narbonnais); languedocien septentrional (rouergat, gévaudanais, aurrillacois), which constitutes a transition zone into the nord-occitan area; languedocien occidental (agenais, quercinois, albigeois); languedocien oriental (biterrois, montpelliérain, cévanol), which transitions into Provençal. Provençal is commonly divided into four sub-dialects: le rhodanien, spoken in the west of the province; le maritime, spoken in the central part of the province; le bas-alpin or le gavot, spoken in the lower Alps; le nissart spoken in and around Nice. Bec (1963: 47) makes a distinction between le maritime (spoken along the coast) and le central , bordered by le maritime to the south, le rhodanien to the west, and le bas-alpin to the east. Equally, the variety of Provençal traditionally spoken in Marseille is sometimes singled about as an autonomous sub-dialect (Ford 1921: 1). Le rhodanien is, of course, the basis of Frédéric Mistral’s Trésor and has acted as a reference point for many works on Occitan (e.g., Ford 1921; Coustenoble 1945). One commonality in all taxonomies is that Gascon is considered as separate from other langue d’oc dialects, primarily because it is the most divergent in terms of its phonological and morphosyntactic structure (Rohlfs 1977; Walter 1988: 153). From as early as the fourteenth century, Gascon has been referred to as ‘un lengatge estranh’, by Guilhem Molinièr in the grammatical treatise Las Leys d’Amors (1356) (Anglade 1919/1971: 164), but the explanation as to why Gascon contains such highly divergent linguistic features has been much debated. Lodge notes that numerous factors seem to be at play: the pre-Latin Aquitainian (rather than Gaulish) substratum; the post-Latin Basque superstratum resulting from the Basque invasion of the area between the Garonne and the Pyrenees from the sixth to the ninth centuries; close communication networks with Romance-speaking populations south of the Pyrenees (1993: 68). Linguistic Gascony stretches from the Pyrenees to the Atlantic Ocean and from the Garonne river to the area around Toulouse (Bec 1963: 48) with the exception of the small historically
3 Research Context: Southern Gallo-Romance
53
Basque-speaking area in the extreme southwest (see Fig. 3.2). Many subdialects of Gascon can also be identified, such as le bordelais, traditionally spoken in the Bordeaux area, lo parlar negre (‘the black language’), spoken primarily in the western part of the département des Landes, and le béarnais (see Fig. 3.2), spoken in the region of Béarn. Béarn is the historically Romance-speaking part of the modern-day Pyrénées-Atlantiques département. The number of Gascon speakers in southwestern France increases steadily from north (Bordeaux) to south (the Pyrenees) and because Béarn is the area of linguistic Gascony with the highest recorded number of Gascon speakers (Moreux 2004), Béarnais may be considered the principal surviving dialect of Gascon, and has thus become largely synonymous with it. Jean Lafitte describes the relationship between Béarnais and Gascon in the most romantic terms: ‘le gascon, une langue à part entière, le béarnais, âme du gascon’ (‘Gascon, a separate language, Béarnais, Gascon’s soul’) (1996: 1). For this reason, le béarnais, or Béarnais, will form the basis of the analyses of language a dialect death in this monograph.
3.2
Language Contact with French in Béarn
From as early as the twelfth century A.D., all official documentation in Béarn was written in the local Gascon sub-dialect, Béarnais, rather than in Latin (Keller 1985: 65). Béarnais was thus firmly established as the prestige language at this time when official documentation written elsewhere, in the geographical space we know today as France, was largely written in Latin. Despite the predominance, elsewhere, of Latin in writing over all territorial vernaculars until the sixteenth century, the presence of French in Béarn can be attested as far back as 1387, when Gaston Fébus, Count of Foix and Viscount of Béarn, wrote his Livre de Chasse. However, Fébus notes in the epilogue that French is not his mother tongue and apologises for any errors in his writing: ‘Et aussi ma langue n’est pas si bien duite de parler le franssois comme mon propre langayge’ (‘And also I am not blessed to speak French as my mother tongue’) (Tucoo-Chala 1976: 171). He wrote many other works, primarily in Béarnais, and, at this time, French was almost unheard of in
54
D. Mooney
Béarn and was certainly far from exerting any threat to, or competing in any way with, Béarnais. In 1539, Francis I signed the Ordonnances de Villers-Cotterêts into law, one provision of which was to discontinue the use of Latin in official documentation within the geographical area governed by the French crown. The result was that over the 1540–1580 period, a very specific type of bilingualism or diglossia developed in many parts of the territory which followed a simple formula: ‘le français est employé, quand on écrit, et l’idiome local, quand on parle’ (‘French is used when you write; local languages are used when you speak’) (Brun 1923: 5). All official documentation, in the territory under French rule, was translated from Latin into French during this forty-year period. The implications of this for the largely illiterate masses were minimal and the majority continued to speak their local language variety. At the time of the Ordonnances, Béarn was a politically sovereign state, forming part of the Royaume de Navarre, and thus the legal stipulations of the Ordonnances did not impose any obligation. In the sixteenth century, official documentation, in Béarn, continued to be written in Béarnais. For example, the Fors (constitutional legislature) were printed in Béarnais in 1552 (Brun 1923: 27). The queen regnant of Navarre, Jeanne d’Albret, mother of Henri IV of France, demanded that all judicial acts, patents, and official documentation be written in Béarnais (Brun 1923: 15) and, in 1568, that Catholic mass be given to the population in Béarnais (Keller 1985: 68). Despite this resistance to French at an official level in Béarn, a series of historical events took place resulting in situations favourable to its use in many domains. Jeanne d’Albret converted to Protestantism and introduced it to the region in the late sixteenth century. This introduction of the protestant faith involved high levels of in-migration to Béarn from Paris. The protestant preachers spoke French, by virtue of their regional origins and their relative education, and the nomadic nature of their work, spreading their word throughout the region, resulted in widespread contact between French and Béarnais. While it is almost certain that the nobility were using French during the protestant reform, there is evidence to suggest that the general population may have engaged in the protestant faith through Béarnais (Keller 1985: 69).
3 Research Context: Southern Gallo-Romance
55
Henri IV, Roi de Navarre, succeeded to the French throne in 1589. Following his succession, he granted continued sovereignty to Béarn and it thus remained an independent state, separate from France. Despite Béarn’s independence, the reign of Henri IV (1589–1610) was characterised by a rise in the use of French in official acts and documentation. The alternations in usage between French and Béarnais, in Béarn, were relatively infrequent with Béarnais remaining the dominant language. Béarnais remained the official language of the state and of address to the state or to those in authority and undoubtedly the language used in everyday communication by the masses. The Fors were again printed in Béarnais in 1602. By this stage, the rest of the south of France had renounced written usage of the local idiom in favour of French, a process almost certainly accelerated by the introduction of the printing press (Brun 1923: 27). The seventeenth century heralded a large number of political reforms that affected both national (Béarn) and linguistic norms in Béarn. In 1610, with the succession to the French throne of Henri IV’s son, Louis XIII, the status of Béarnais as official language of the sovereign state suffered as a result of the political unity with the French Kingdom. In 1620, Louis XIII incorporated Béarn into France by signing the Édit d’Union and ordered that all ‘procédures de notre dite cour de Parlement soient expédiés en langage françois’ (‘the processes of our Paliamentary Court will be delivered in the French language’) (Brun 1923: 32). While this legislation marks the shift from Béarnais to French at the court and in the officialdom, Brun states that it is almost certain that at this time southern populations, beyond the court, were completely ignorant of French and unable to use it (1923: 6). In addition to this, sufficient evidence exists to suggest that the nobility and the upper classes in Béarn were using French, though their mother tongue was Béarnais (Brun 1923: 19; Keller 1985: 70). As a result of this, the people in Béarn who were speaking French were still highly dependent on Béarnais and spoke a French that was highly influenced by their mother tongue (Keller 1985: 90). From 1620 onwards, French was the official language of Béarn but there is evidence to suggest that even in the officialdom the language shift was far from complete for a substantial time period following the
56
D. Mooney
incorporation of Béarn into France. The Fors were re-printed in Béarnais throughout the seventeenth and eighteenth centuries (1625, 1673, 1715, 1723, 1781) (Keller 1985: 70). Written documentation in the rest of Gascony, Provence, and Languedoc had been in French since 1540. While writing in French gradually began to overtake Béarnais during the eighteenth century, the amount of documentation written in Béarnais was substantial, particularly in more rural areas such as the Pyrenean valleys (Brun 1923: 52) and Béarnais remained the language of oral communication and some documentation until at least the eve of the Revolution. Brun states, in discussing the shift from Béarnais to French, ‘et ce dénouement a été préparé et consommé non par l’intrusion de la monarchie française en Béarn, mais par celle des Béarnais dans les affaires de France’ (‘and this end point was reached not because of a forced intrusion in Béarn on the part of the French monarchy, but by the increasing integration of the Béarnais people in France’s business’) (1923: 17). The incorporation of Béarn into France did not, in and of itself, encourage the Béarnais people to speak French; in fact, it had little effect on the language of everyday communication. However, the increasing contact with Paris indirectly encouraged those who looked outside of the region to use French at the expense of Béarnais. Upon the publication of the Convocation des États Généraux in 1789, many Béarnais nobles and bourgeois expressed nationalistic (Béarn) sentiments, stating that the Édit d’Union (1620) was illegal as it had never been ratified by Béarn and had been imposed by force (TucooChala 2009: 115). In fact, some notaries continued to write in Béarnais until 1815. However, from 1815 onwards, ‘les Béarnais se contentèrent de suivre, toujours avec retard et avec modération, les implusions politiques venues de Paris’ (‘the Béarnais people slowly followed the political influence of the Parisians’) (Tucoo-Chala 2009:15). Post-Revolution, there is very little information available on the subsequent stages of language shift in Béarn. Keller discusses the increase of literacy in French as an indicator of the rate of decline of Béarnais over the course of the nineteenth century (1985: 71). Baron Dupin’s (1826) La carte de la France scolaire, indicates that, in Pyrénées-Atlantiques, only one in fifteen boys of school-going age had received any formal education suggesting that at most 6% of the youngest generation may have spoken French.
3 Research Context: Southern Gallo-Romance
57
Victor Drury’s (1864) similar survey found that between 70% and 80% of children attending school had some knowledge of French in Keller (1985: 73). This does not, however, account for the vast majority of children who were not attending school. The eventual shift from Béarnais to French was supported by the Lois Jules Ferry (1881) and the introduction of national free and compulsory education in 1881–1886. The period immediately following the Lois Jules Ferry was not necessarily characterised by large levels of fluency and literacy in French among the general population. Weber takes the end of the First World War as the watershed in the fortunes of the traditional Romance dialects and regional languages of France (1979: 79), the shared experience of conscription having raised national consciousness and underlined the value of French as a lingua franca (Hornsby 2006: 125). Moreux and Moreux (1989) note, however, that though the first signs of disaffection with Béarnais, in rural areas, appeared in the generation born during the First World War, the predominant daily use of French in Béarn began only with the generation born during the Second World War. The general shift from a rural to an urban-focused society in the twentieth century also supported the use of French in everyday communication. There are indications that the vitality of Béarnais continued for some time, particularly in rural areas. For example, Pottier’s (1968) Carte du bilinguisme en milieu rural shows the majority of the rural areas of the Midi exhibiting a bilinguisme usuel (‘normal bilingualism’) while only the Pyrenean borderlands and contiguous areas are credited with the label bilinguisme intense (‘intense bilingualism’) (Kristol and Wüest 1985: 2). The exact intensity of bilingualism in Béarn is open to interpretation. We can glean from this, however, that language contact in Béarn continued up until the mid-twentieth century, at least in rural areas. Indeed, in the mid-1980s, Kristol and Wüest, proclaimed that there were several rural sites in Béarn where intense bilingualism was still found and where we can study the language of the last generation of native speakers who arrived at school aged five, unable to speak French (1985: 4). The contact period between Béarnais and French seems to date from somewhere between 1826 and 1864 until sometime between the two World Wars, a period of approximately eighty to one hundred years.
58
D. Mooney
Many of the last generation of Béarnais-French bilinguals are thus still alive today. The cessation of intergenerational transmission that occurred during the twentieth century, and concomitant completion of language shift to French, can thus be observed in apparent time, particularly in the urban and semi-urban areas of Béarn.
3.3
The Gascon Speech Community in Béarn
From the late nineteenth century onwards, and indeed over the course of the twentieth century, the Béarnais sub-dialect of Gascon found itself in an increasing state of language obsolescence (as did all varieties of the langue d’oc ). While there is some evidence to suggest that the last generation of native speakers were born in rural areas up until the eve of the Second World War (Moreux 2004: 25), the state of the language in the latter half of the twentieth century displays the characteristics described by language obsolescence models: the language not being taught to children in the home; the number of speakers declining very rapidly, the entire population being bilingual, with French preferred in almost all situations; little or no literacy in the language (Bauman 1980). In Dorian’s terms, Béarnais is and has been for some time, in a situation of ‘gradual death’ (1981: 107), but this definition specifies, however, an age-governed proficiency continuum in the obsolescent variety: young speakers tend to be least proficient and older speakers most proficient with at least one generation of semi-speakers. Béarnais, however, has no marked age-governed proficiency continuum: native speakers are relatively easily locatable among Béarn-born inhabitants over the age of sixty-five in rural areas, but below this age, they are much more difficult to find and very few speakers are to be found under the age of forty (see Jones [2001: 3] for a parallel situation in Jersey Norman French). In the entire historically Gascon speaking region, the number of speakers, at all levels of proficiency, varies from (at most) 3% of the population in Bordeaux to 30% in Béarn, approximately 500,000 speakers
3 Research Context: Southern Gallo-Romance
59
in total1 (Moreux 2004: 25). These numbers include fluent native speakers and semi-speakers. In contrast to Moreux (2004), however, Martel (2007) estimated that there were around 500,000 langue d’oc or Occitan speakers across the south of France in total. Nonetheless, in some areas of Béarn there are said to be many native speakers, an ‘enormous pool of people with passive language competence’ (Moreux 2004: 29) and some parents who still transmit the language to their children; this was, at least, the situation at the turn of the twenty-first century. For the Béarn region, Moreux cites 16% of people aged over fourteen as saying that they spoke Béarnais well (fluent speakers) and 14% as saying that they spoke a little, giving a total of 30% for speakers of all levels of proficiency, or about 75,000 speakers. If we cautiously examine only the fluent speakers (approximately 40,000 speakers), we must note that over 50% of these speakers are over the age of sixty and rural-dwellers and only 3.5% are between the ages of fourteen and twenty-four. All of the Béarnais speakers noted by Moreux were born in Béarn or in linguistic Gascony with INSEE (1994) showing that, in the late twentieth century, only a little over half of the people living in Béarn over the age of fourteen were actually born there. Finally, we may note that in Pau and its surrounding areas, where almost half of the population of Béarn now lives, Béarnais is almost totally absent (Moreux 2004: 31). Any Béarnais speakers present in Pau are all over the age of fifty and a minority even in this age group. Béarnais is spoken only in closed tight-knit networks, usually made up of older people, who are very emotionally attached to their language, but few of whom make it an ideological issue or are preoccupied by its future (Moreux 2004: 25). Béarnais is now seen as a community-based and not a family-based language: ‘la famille n’est pas le lieu priviligié où on le parle le plus souvent, […] le béarnais est la langue de la communauté restreinte’ (‘the family is not the privileged domain where it is most frequently spoken, […] Béarnais is the language of a restricted community of speakers’) (Müller et al. 1985: 103). Kristol and Wüest (1985), 1
All speaker numbers are taken from Moreux (2004) who uses three reports entitled Pratique(s), (présence) et représentations de l’occitan (ou langue occitane) published between 1994 and 1997 after surveys were undertaken for the Conseils Généraux of the Pyrénées-Atlantiques département and for the Aquitaine Conseil Régional .
60
D. Mooney
referring to the domains of use of Béarnais, summarise the situation as follows: there are a limited number of domains where the language can be spoken, such as the market, for example, to the extent that using Béarnais in an unfamiliar context, such as an interview, can violate the speakers sense of appropriate setting; Béarnais is thus highly restricted in its domains of use. Nowadays, grandparents, especially grandfathers, may teach their grandchildren playful uses of Béarnais. However, these grandparents, who now seek to transmit some Béarnais, were the same parents who were careful not to speak any Béarnais to their children to avoid the shame of them starting school ‘sans comprendre un mot de français’ (‘not understanding a word of French’) (Moreux 2004: 37). Müller et al. (1985) observed that linguistic exogamy has been a major factor in driving language shift, in both rural and urban areas. In 1983, they observed that in Osse-en-Aspe, a small rural village in the Pyrenean Vallée d’Aspe, almost all of the non-Béarnais speakers had had at least one parent of non-Béarnais origin. By contrast, in families with both parents of Béarnais origin, intergenerational transmission was total for all informants surveyed. In the urban setting, notably in Oloron-Sainte-Marie, the largest town in western Béarn, and in Bédous, a large town nearby, non-transmission in families having two Béarnais-speaking parents was the cause of half the cases of monolingual French speakers, the other half being caused by linguistic exogamy (Müller et al. 1985). We must exercise caution in interpreting these results due to limited sample sizes and sampling techniques that targeted first language speakers, but they indicate that there was a clear rural–urban divide in relation to intergenerational transmission as recently as the 1980s. In many of the rare cases where intergenerational transmission occurred at home, it was often imperfect and a highly marginal endeavour.
3.4
The Standardisation of Occitan
The langue d’oc, including Gascon and its sub-dialects, has been standardised in the form of Occitan, though the term ‘Provençal’ has, in the past, been used to refer collectively to all dialects. Today, two main grammatical and/or orthographical standards exist for writing the language,
3 Research Context: Southern Gallo-Romance
61
the Mistralian norm and the Classical norm, the latter being the primary system used in the limited educational, media, and public spaces afforded to these obsolescent language varieties. In addition to these orthographical systems, the Febusian norm, discussed in Sect. 3.5 below, is widely used by certain militant associations in Gascony and Béarn. The Mistralian norm was invented by Frédéric Mistral (1830–1914) and was specifically created to write in the Provençal dialect; this system is phonetically transparent from a French perspective, in that it replicates many sound-spelling correspondences present in standard French orthography such as the use of ou for the phoneme /u/. This has a ‘transfer of literacy’ advantage in that many Provençal or Occitan speakers are (or were) literate in French from the end of the nineteenth century onwards. The Mistralian norm is also considered to be prestigious due to its association with the large body of literary works published in Provençal by Mistral himself, who won the Nobel Prize for Literature in 1904. Its main disadvantage, however, is that it is best suited to writing in the Provençal dialect and was therefore perceived to be maladapted to writing in the other dialects of the langue d’oc; indeed, when it is still used today, it is almost primarily for writing Provençal and Nissart, a sub-dialect of Provençal. As a partial reaction to the predominance of Mistralian norm, the Sociéte d’Études Occitanes was founded in 1935 by Joseph Anglade and Valère Bernard, with Louis Alibert as its secretary. This association was replaced, in 1945, by the Institut d’Estudis Occitans, with both associations establishing the term ‘Occitan’ to refer collectively to all langue d’oc varieties, a function previously fulfilled by the term ‘Provençal’. The standardisation process for Occitan began in 1935 with the publication of Louis Alibert’s (1935) Gramatica occitana segon los parlars lengadocians (‘Occitan grammar of Lengadocian dialects’), supplemented by further corpus planning efforts in the form of Alibèrt’s (1965) Dictionnaire occitan-français selon les parlers languedocians (‘Occitan-French dictionary for Lengadocian dialects’). The orthographical standard proposed by these works was based on high medieval (1100–1350) representations of central langue d’oc varieties together with some modern Lengadocian. Lengadocian is often considered to be the most conservative Occitan
62
D. Mooney
dialect (Wheeler 1988: 246) as it is the centre of the langue d’oc area— a lowest common denominator of sorts. It is for this reason that it was selected as the basis for the standard: ‘le languedocien a vraiment, dans l’ensemble occitanophone, […] vocation de parler directeur et référentiel’ (‘Lengadocian is, in the entire Occitan-speaking area, the most authoritative reference point’) (Bec 1973: 20). During the 1970s, standardisation efforts were extended by adapting Alibert’s standard orthographical system to Gascon (by Pierre Bec) and Provençal (by Robert Lafont). The standard Occitan orthographical system is based on the principle of a phonological diasystem, an abstract, standardised phonological reference point against which we can compare variation observed in the dialects. The aim of the diasystem, in short, is to at once represent all of the dialects and none of the dialects in particular (Bec 1973: 24). The concept of the diasystemic standard implies, to a certain extent, that the standard Occitan language is polynomic or pluricentric (Sumien 2006), that is that, while there is an orthographical standard, morphosytactic and phonological variation are recognised, accommodated, and valued in spoken varieties by the language planning initiatives. The reality, however, is that the Occitan standard finds itself in (increasing) ideological conflict with a host of more localised language planning initiatives which use orthographical systems, including the Mistralian norm, that are thought, by their proponents, to more accurately and faithfully reflect the linguistic structures of their respective dialects. For example, the Escolo dóu Po norm in the Italian Occitan valleys, the Bonnaudian norm of the Auvernhat dialect, and the Febusian norm, mentioned above, for Gascon, are all used extensively alongside the Classical norm. Akin to the Mistralian norm, these more localised orthographical systems tend to promote literacy by facilitating the transfer of literacy from French; this often leads to the misconception that these systems incorrectly apply French orthography to the langue d’oc, which is an oversimplification of these corpus planning endeavours. While the reality of the Occitanist agenda, at least regarding orthography, is that it aims to support and accurately represent all varieties of Occitan, this
3 Research Context: Southern Gallo-Romance
63
has not always been clearly conveyed to lay-speakers and other more grassroots language planners, leading to the frequent misconception the standard Occitan is an artificial language that all southern regions are expected to adopt.
3.5
Militant Language Politics in Béarn
The process of language shift often leads to the emergence of political campaigns or pressure groups, such as Ofis ar Brezhoneg in Brittany or Conradh na Gaeilge in Ireland, which aim to call attention to the plight of their language and to demand revitalisation efforts. This section examines the militant discourses that, on one side, promote Gascon and Béarnais as languages in their own right and, on the other, refute their autonomy and view Gascon and Béarnais as localised dialects of the Occitan language. In 1896, the Escole Gastoû Fébus was created and spurred a movement for the survival and development of the Gascon and Béarnais language and culture. The aim of this movement was to defend the specificity of Gascon against the predominant ‘Provençal’ glottonym of the time. The Febusian movement was ‘very solidly Catholic, rural, conservative and nationalist [French], making only moderate regional demands involving no militant actions’ (Moreux 2004: 39). It gave rise to literary, journalistic, and theatrical works and enabled these works to reach a larger public. The movement used the Béarnais spoken around Pau as a standard, created orthographical rules, and developed the tools necessary for linguistic codification, namely a dictionary and a grammar (Moreux 2004: 39). The Febusian movement continued to promote Gascon and Béarnais up until the end of the 1980s, but after 1960, it had to compete with the relatively new Occitan movement when the Institut d’Estudis Occitans (IEO) arrived in Béarn. The IEO, which had existed elsewhere in the south of France since the eve of the Second World War, differed from the Félibrige movement in Provence in its more radical regionalist demands and leftist tendencies, often going as far as claiming independence and becoming extremely left wing after 1968 (Moreux 2004: 40).
64
D. Mooney
By the late 1980s and early 1990s, the presence of the Occitan standard was clearly visible in the public sphere and the Occitan orthography was generalised to all domains in publishing and any activity receiving public subsidy such as educational textbooks, newspapers, magazines, and bilingual road signs. Since the early 1990s, two reactions to the Occitan paradigm have emerged: ‘the first scholarly, and based on linguistic arguments, concerns all of Gascony (Gasconism). The second, more “grass-roots”, originated and spread mostly in Béarn (Béarnism)’ (Moreux 2004: 45). In 1993, Jean Lafitte, a member of the IEO, along with several others, called Gascon ‘une langue à part entière’ ‘a separate language’, not simply a ‘dialecte de l’occitan’ ‘a dialect of Occitan’, and began writing the Dictionnaire du gascon modern, which was largely based on the Dictionnaire du béarnais et du gascon modernes (Palay 1980). This modern Gasconism followed on from the ideology of the Escole Gastoû Fébus and, on the whole, refused any political involvement in that it was not concerned with ‘taking on’ French centralisation. It did, however, campaign actively for the inclusion of Gascon in the European Charter for Regional or Minority Languages and indeed for the very ratification of the charter by the French state. Lafitte also founded an academic journal, LigamDiGam, which was written exclusively in Febusian orthography. In fact, the largest publishing company in Gascony, Éditions Pyrémonde, began to publish regularly in Febusian orthography, thus ending the quasimonopoly of the Occitan Classical orthographic norm in publishing (through the rival publishing house Per Noste). In 2002, the Gasconist movement united with a Béarnist organisation, Pays de Béarn et Gascogne (PBG), founded in 1995 by Jean-Marie Puyau, to form the Institut Béarnais et Gascon (IBG) or Enstitut Biarnés e Gascoû. The members of the IBG originally demanded the right to name their own language, to preserve it as it was and to affirm their identity, not only symbolically, in orthography and bilingual signage, but also in economic terms. It was an openly anti-Occitanist organisation, campaigning for the use of the term ‘Béarnais’ in the public sphere and for the inclusion of both Béarnais and Gascon in the European Charter. While a wide variety of glottonyms are used to refer to southern Gallo-Romance, in Béarn and in general, it is worth noting that native speakers often attest simply
3 Research Context: Southern Gallo-Romance
65
that they speak le patois or, sometimes, they will refer to the name of the local variety, e.g., le béarnais or le montpelliérain. The IBG continues to use adapted versions of the Fébusian orthography, laid down by the Escole Gastoû Fébus, in press and publishing. Since the Gasconist and Béarnist movements united against their common adversary, Occitanism, they have gained a certain public visibility though they have yet penetrated Occitan’s educational and institutional monopoly (Moreux 2004: 5). The Béarnist movement, however, is scarcely concerned with issues of revitalisation and is generally interested in the documentation and celebration of the speech of older native speakers: there is neither active recruitment of younger members nor any concrete encouragement of intergenerational transmission (see Sect. 3.6), with an ‘almost fatalistic regard for the future of the language’ (Moreux 2004: 54).
3.6
Occitan Revitalisation in the Educational Context
The Occitan movement has, to some extent, spearheaded the revitalisation movement for Gascon in the region of Béarn (and throughout Gascony). Many children now have the opportunity to learn Gascon in immersion education or bilingual education at primary school level and there is also some limited provision at secondary level, though the language learned is almost invariably referred to as Occitan. There are also a wide variety of adult education provisions which target learners of all ages, but in particular young adults. As such, the Occitan movement can be seen to place the creation of néo-locuteurs, or new speakers, at the centre of its revitalisation efforts. Active competence in Gascon among younger generations is rare, but since the 1980s, immersion education schools, called Calandretas have been offering education through Occitan at primary school level. Today, these schools provide non-profit, secular, and free education throughout the south of France. The Pyrénées-Atlantiques département has been particularly progressive in terms of promoting primary competence among the young (Moreux and Puyau 2002: 12). Of eleven
66
D. Mooney
Calandretas in linguistic Gascony, nine are now located in PyrénéesAtlantiques (https://calandretaaq.wordpress.com/a-propos/). The Calandretas provide children, in theory, with the opportunity to strengthen ties with older generations in the region even if the language that they are taught at school is perceived to be different from the one they may potentially hear at home. The effect of the nine Calandretas on the linguistic competence of the younger generations in the Béarn region is almost negligible, however, with only 393 children in total attending these schools in 2017 (www.capoc.fr). Outside of the Calandreta system, some primary school students traditionally had access to classes in cultural awareness and basic Gascon which were delivered by itinerant teachers called caminaires, who would be responsible for a certain number of schools within a defined geographical area. The Pyrénées-Atlantiques département was also at the forefront of this initiative, with classes reaching around 4000 children in 1999–2000 (Moreux 2004: 33), though these classes have largely ceased and the focus was never on attaining active linguistic competence among those children who speak no Gascon, but on valorising the lost language of children who come from traditionally Gascon speaking families. In 2017, 813 students attended bilingual schools, écoles bilingues, where instruction is provided equally in French and Occitan, and a further 801 students had access to an initiative called enseignement renforcé (‘reinforced tuition’) in (French) public schools, which provides three consecutive hours of instruction in Occitan per week. In 2019, the Pyrénées-Atlantiques département also started providing various forms of Occitan-French bilingual education in public schools, offered by the Centre d’Animacion Pedagogica en Occitan (CAP’OC), which follow the curricula laid down by the Éducation nationale. There is one second-level Collègi Calandreta in Pau, which provides immersion education up to GCSE-equivalent level, making Pau one of the perceived hubs of the Occitan revitalisation movement, alongside Toulouse and Montpellier, where second-level immersion education is also offered. In 2014, there were also three publicly funded FrenchOccitan bilingual education collèges (in Morlass, Lasseube, and Laruns) and 12 collèges that offered ‘optional’ instruction in Occitan (all provided by CAP’OC). There are also five lycées offering (some) instruction in
3 Research Context: Southern Gallo-Romance
67
Occitan up to baccalauréat (A-Level equivalent). Beyond the primary school level, the focus of secondary or third-level education is rarely to increase the number of Gascon speakers but more specifically to train Occitan teachers (Moreux 2004: 34). Blanchet (2004: 155) notes that there is a tendency on the part of the Éducation nationale system to emphasise the notion of langues d’oc in the plural, so that, when secondlevel students take the Occitan examination for the baccalauréat, they must indicate the variety in which they wish to be examined. Third-level candidates for the CAPES d’occitan-langue d’oc teacher-training qualification, however, can be examined on texts and documents from any one of the langues d’oc dialects. While it is still possible to study Occitan at degree level in many southern universities (such as in Bordeaux, Toulouse, and Montpellier), the Université de Pau et des Pays de l’Adour no longer offers primary degrees in Occitan or Occitan studies. Alongside the Calandreta-educated new speakers, the adult education initiatives of the Centre de formacion professionala Occitan (cfpÒC) in Nouvelle-Aquitaine have contributed substantially to the revitalisation movement and the creation of new speakers. The cfpÒC was established in Béarn in 1999 and proposes to offer education in Occitan ‘toujours dans sa variante locale’ (‘always in the local dialect’) (www.cfpoc.com). The aim of the instruction of new speakers is therefore to teach phonological and grammatical structures that are associated with the variety of the langue d’oc spoken locally: ‘C’est donc tout naturellement dans sa variante béarnaise qu’il sera enseigné en Béarn’ (‘It goes without saying that the Béarnais variety will be taught in Béarn’) (www.cfpoc.com). The cfpÒc offers a wide variety of services, including teacher training for both Calandretas and public bilingual education, 25 weekly evening classes for adults called Setmanèrs throughout the Béarn region, week-long and daylong Parlar immersion schools, individual tuition, and distance learning. The cfpÒC also provide European CEFRL certification from A1 to C1 in Occitan. The Gascon variety of the langue d’oc emerged from the dialectalisation of the Latin spoken in Gaul and comprises an array of sub-dialects which share common structural characteristics and exhibit localised geographical variation, of which Béarnais, spoken in the region of Béarn, is the principal surviving sub-dialect, both in terms of speaker numbers
68
D. Mooney
and notoriety. Long-term language contact with French, intensifying after the introduction of free and compulsory education in the late nineteenth century, has led to the obsolescence of Gascon and, indeed, of all of France’s regional languages. In rural areas, the process of language shift to French accelerated in the interwar period, leaving behind a fragmented Gascon speech community in Béarn (and elsewhere), where the language was (and now is) only spoken in very limited domains of use. The standardisation of the langue d’oc in the form of Occitan began in 1945 and the Occitanist movement that sprung from the standardisation process arrived in Béarn in the 1960s where it was met with resistance from preexisting grassroots movements. This ideological conflict, centred around naming the language and the use of different orthographical systems, continues to this day. While local militant groups, such as the IBG, focus on the preservation of existing Gascon speakers and their cultural practice, the Occitan movement has been somewhat more progressive in pushing the revitalisation of Gascon forward by providing ample opportunities within the region for the creation of new speakers.
References Alibèrt, Louis. 1935. Gramatica occitana segón los parlars lengadocians. Toulouse: Societat d’Estudis Occitans. Alibèrt, Louis. 1965. Dictionnaire occitan-français, d’après les parlers languedociens. Toulouse: Societat d’Estudis Occitans. ´ Anglade, Joseph (1919/1971), Edition de “Las Leys d’Amors” , vol. 3, Toulouse, Privat, 1919. Reprint New York/London, Johnson Reprint Corporation, 1971. Bauman, James J. 1980. A guide to issues in Indian language retention. Washington, DC: Center for Applied Linguistics. Bec, Pierre. 1963. La langue occitane. Paris: Presses Universitaires de France. Bec, Pierre. 1973. Manuel pratique d’occitan moderne. Paris: Picard. Blanchet, Philippe. 2004. Review article: Uses and images of ‘Occitan’: An occitanist view of the world. International Journal of the Sociology of Language 169: 151–159.
3 Research Context: Southern Gallo-Romance
69
Brun, Auguste. 1923. L’Introduction de la langue française en Béarn et en Roussillon. Paris: Champion. Coustenoble, Hélène N. 1945. La phonétique du provençal moderne en terre d’Arles. Hertford: Austin. Dorian, Nancy C. 1981. Language death: The life cycle of a Scottish Gaelic dialect. Philadelphia: University of Pennsylvania Press. Ford, Harry E. 1921. Modern Provençal phonology and morphology: Studied in the language of Frédéric Mistral . New York: Columbia University Press. Hawkins, Roger 1993. Regional variation in France. In French today: Language in its social context, ed. Carol Sandars, 55–84. Cambridge: Cambridge University Press. Harris, Martin. 1988. The Romance languages. In The Romance languages, ed. Martin Harris and Nigel Vincent, 246–278. London and Sydney: Croom Helm. Hornsby, David. 2006. Redefining Regional French: Koin´eization and Dialect Levelling in Northern France. Oxford: Legenda. Jones, Mari C. 2001. Jersey Norman French: A linguistic study of an obsolescent dialect. Oxford: Blackwell. Judge, Anne. 2007. Linguistic policies and the survival of regional languages in France and Britain. Basingstoke: Palgrave Macmillan. Keller, Judith. 1985. Histoire de la langue béarnaise en Béarn. In Drin de tot: Travaux de sociolinguistique et de dialectologie béarnaises, ed. André M. Kristol and Jakob Th. Wüest, 63–74. Bern: Peter Lang. Kristol, Andres M., and Jakob Th. Wüest. 1985. Drin de tot: Travaux de sociolinguistique et de dialectologie béarnaises. Bern: Peter Lang. Lafitte, Jean. 1996. Le gascon, langue à part entière, et le béarnais, âme du gascon. LiGam DiGam: Cadèrn de lingüistica e de lexicografia gasconas, 4. Laroussi, Foued, and Jean-Baptiste Marcellesi. 1993. The other languages of France. In French today: Language in its social context, ed. Carol Sandars, 85–104. Cambridge: Cambridge University Press. Lodge, R. Anthony. 1993. French: From dialect to standard . New York: Routledge. Martel, Philippe. 2007. Qui parle occitan? A propos d’une enquête. Langues et cité 10: 3. Moreux, Bernard. 2004. Béarnais and Gascon today: Language behaviour and perception. International Journal of the Sociology of Language 169: 25–62. Moreux, Bernard, and Colette Moreux. 1989. La transmission du béarnais en milieu rural aujourd’hui. In Langues en Béarn, ed. Bernard Moreux, 235– 256. Toulouse: Presses Universitaires du Mirail.
70
D. Mooney
Moreux, Bernard, and Jean-Marie Puyau. 2002. Dictionnaire français-béarnais. Monein: Éditions Pyrémonde. Müller, Franziska, Renata Huber, Béatrice Nützi, and Daniel Nussbaum. 1985. Enquête sociolinguistique dans trois communes du Béarn. In Drin de tot: Travaux de sociolinguistique et de dialectologie béarnaises, ed. André M. Kristol and Jakob Th. Wüest, 75–115. Bern: Peter Lang. Oliviéri, Michèle, and Patrick Sauzet. 2016. Southern Gallo-Romance (Occitan). In The Oxford guide to the Romance languages, ed. Adam Ledgeway and Martin Maiden, 319–349. Oxford: Oxford University Press. Palay, Simin. 1980. Dictionnaire du béarnais et du gascon modernes. Pau: Marrimpouey Jeune. Posner, Rebecca. 1996. The Romance languages. Cambridge: Cambridge University Press. Pottier, Bernard. 1968. La situation linguistique en France. In Le langage, ed. André Martinet, 1144–1161. Paris: Encyclopédie de la Pléiade. Quint, Nicolas. 1999. Le parler occitan ardéchois d’Albon: Canton de SaintPierreville, Ardèche: Description d’un parler alpin vivaro-vellave du boutiérot moyen. Paris: L’Harmattan. Rohlfs, Gerhard. 1977. Le gascon. Études de philologie pyrénéenne, 3rd ed. Tübingen: Niemeyer (1st ed. 1935). Sampson, Rodney. 1999. Nasal vowel evolution in Romance. Oxford: Oxford University Press. Sumien, Domergue. 2006. La standardisation plucentrique de l’occitan: novel enjeu sociolinguistique, développement du lexique et de la morphologie. Turnhout: Brepols. Tucoo-Chala, Pierre. 1976. Gaston Fébus: Un grand prince d’Occident au XVIe siècle. Pau: Marrimpouey Jeune. Tucoo-Chala, Pierre. 2009. Petite histoire du Béarn du Moyen Âge au XX e siècle. Monein: Éditions Pyrémonde. Walter, Henriette. 1988. Le francais dans tous les sens. Paris: Robert Laffont. Weber, Eugene. 1979. Peasants into Frenchmen: The modernization of rural France 1870–1914. London: Chatto and Windus. Wheeler, Max W. 1988. Occitan. In The Romance languages, ed. Martin Harris and Nigel Vincent, 246–278. London and Sydney: Croom Helm.
4 Phonetic and Phonological Systems
This chapter details the phonetics and phonology of Occitan and its varieties, focusing both on the commonalities between phonological systems over geographical space and the high levels of variation observable. It begins with a historical phonological overview of the development of the primary dialects, before discussing the consonantal and vocalic inventory, with cross-dialectal comparisons included in each section, and focusing, in particular, on phonetic and phonological variation in the region of Béarn. As we have seen, the term ‘Occitan’ and the orthographical standard are frequently a source of ideological conflict between Occitanist organisations and grassroots movements in the regions that focus on individual varieties, often considering them to be languages in their own right, e.g., L’Institut Béarnais et Gascon in Béarn. Nonetheless, I will use the term ‘Occitan’ and the standard ‘classical’ orthographical system here, without their associated affiliations, as they are the most commonly employed in the academic sphere.
© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 D. Mooney, Language and Dialect Death, https://doi.org/10.1007/978-3-031-51101-1_4
71
72
4.1
D. Mooney
Linguistic Background
Occitan displays many historical developments in its linguistic evolution that distinguish it in phonological terms from the other Romance languages and, in particular, from Catalan and French. This section will serve to summarise these developments and, in 4.1.1–4.1.4, to outline the major phonological developments, from a historical perspective, within the main supra-dialectal areas. Occitan exhibits the following characteristics in its phonological structure when compared with the other Romance languages (adapted from Bec 1963: 24; cf. Ronjat 1930–1937): • The rounded mid-vowels /ø/, /œ/, and /o/, and the low vowel, /ɑ/, are largely absent from the phonological inventory; • Palatalisation of stressed long Latin u¯ has resulted in a close front rounded vowel phoneme /y/, common to other varieties of GalloRomance, e.g., duru(m) > dur /dyr/ ‘hard’; • Latin vowel + nasal sequences are largely maintained as a partially nasalised oral vowel with a following consonantal segment, e.g., tempus > temps /tens/ ‘time’; • Diphthongisation, during the Middle Ages, of stressed Latin e and o in specific environments (primarily before palatal consonants and less commonly before velar consonants ([w] or [k]), e.g., vetulu(m) > vielh /bjeʎ/ ‘oldm.sg ’; lectu(m) > lieit /ʎejt/ ‘bed’; • No diphthongisation of Vulgar Latin /e/ and /o/, e.g., debere > déver /ˈdebe/ ‘have toinf ’; flore > flor /flu/ ‘flower’; • Vulgar Latin /o/ closes to [u], e.g., dolore(m) > dolor /duˈlu/ ‘pain’; • Latin stressed a is maintained, e.g., pratu(m) > prat /pɾat/ ‘meadow’; capra(m) > cabra /ˈkabɾO/ ‘goat’; • Retention of Latin final unstressed -a (lost in standard French), most frequently as labialised /-O/, e.g., porta(m) > pòrta /ˈpOɾtO/ ‘door’; • No syncope in phrase-internal or phrase-final syllables (cf. French), e.g., Oc. una petita femna sus la finèstra / y.nO.peˈti.tOˈfEn.nO.sys.la.fiˈnEs.tɾO/ ‘a little woman in the window’ (12 syllables); Fr. une petite femme sur la fenêtre /yn.ptit.fam.syʁ.la.fnEtʁ/ (6 syllables);
4 Phonetic and Phonological Systems
73
• No proparoxytons (words stressed on the antipenultimate syllable); stress usually on the penultimate syllable. In this aspect Occitan is essentially opposed to Spanish and Italian (Bec 1963: 28), e.g., Oc. pagina /paˈd͡ʒinO/, It. pagina /ˈpad͡ʒina/ ‘page’; • /O/ closes to [u] in pre-tonic position, e.g., pòrc /ˈpOɾk/ ‘pig’ vs. Gasc. porqueria /puɾkeˈɾijO/ ‘mess’ or ‘pig shed’. In addition to these eleven phonetic/phonological traits that characterise Occitan, Bec (1963: 24–30) also details five morphological, one syntactic, and two lexical features (as described by Ronjat 1930–1937). While Occitan is commonly grouped with French as a Gallo-Romance variety, it is clear that, in some ways, it shares more linguistic affinity with the other Romance languages, such as Catalan (see Sect. 3.1).
4.1.1 Northern Occitan This section describes the key historical phonological developments in the northern Occitan supra-dialectal grouping, or what Bec (1963) has called nord-occitan, including Lemosin, Auvernhat, and VivaroAlpin. These three dialects exhibit common features that justify their supra-dialectal classification (adapted from Bec 1963: 37): • Palatalisation of Latin ca and ga sequences to /t͡ʃa/ and /d͡ʒa/ respectively, e.g., cantat > canta [ˈt͡ʃantO] ‘sing3sg ’; gallina(m) > galina [d͡ʒaˈlinO] ‘chicken’; • Deletion of Latin intervocalic -d- (via [z]) in Vivaro-Alpin (cf. Francoprovençal), e.g., credemus > credem [kɾeˈjEm] ‘believe1pl ’; videmus > vedem [veˈjEm] ‘see1pl ’, with subsequent epenthesis of [j] to avoid hiatus. This development also affects past participles with intervocalic [d] < -t-, e.g., cantada [t͡ʃanˈtajO] ‘sing pst.ptcp. f.sg ’; • Retention of /v/ as a contrastive phoneme, distinct from /b/, except in southern Auvergne, e.g., lavare > lavar [laˈva] ‘washinf ’, not *[laˈba]; • Noun pluralisation not marked or marked by vowel length and/or quality rather than by /-s/ morpheme, expect in Vivaro-Alpin, e.g., Lim. vacha [ˈvat͡ʃO] ‘cowsg ’; vachas [ˈvat͡ʃɑː] ‘cowpl ’, not *[ˈvat͡ʃOs].
74
D. Mooney
• /l/ vocalisation in final position, which also occurs in Provençal and Gascon, e.g., sale(m) > sal [saw] ‘salt’; intervocalic /l/ may be realised as [ɾ], [G], or [w], e.g., pala(m) > pala [ˈpaɾO], [ˈpaGO], or [ˈpawO] ‘blade’; • /s/ is deleted in pre-tonic syllables when followed by the voiceless plosives /p t k/, e.g., castellu(m) > chastèl [t͡ʃaˈtEw] ‘castle’; spiare > espiar [eˈpja] ‘lookinf ’; schola(m) > escòla [eˈkOlO] ‘school’; • First person singular inflectional morpheme /-i/ is realised as [-e] in Lemosin and Auverhnat, e.g., chante [t͡ʃante] ‘sing1sg ’, and as [-u] in Vivaro-Alpin, e.g., chanto [t͡ʃantu]. In bas-limousin, the Latin ct > /t͡ʃ/ change, e.g., factu(m) > /fat͡ʃ/ ‘dopst.ptcp.m ’, has largely taken place, with /t͡ʃ/ commonly realised as [t͡s], e.g., fach [fat͡s]. In marchois, however, in the extreme north of the Lemosin area, /t͡ʃ/ is realised as [jt] (Bec 1963: 41; see Sect. 4.1.4). Lemosin also maintains plural marking, especially for feminine nouns, with /-Os/ realised as [-ɑː], involving a change in vowel quality and compensatory lengthening due to the loss of final /-s/, e.g., la jalina [d͡ʒaˈlinO] ‘hensg ’; las jalinas [d͡ʒaˈlinɑː] ‘henpl ’. In some varieties, plural marking involves a change in the stressed syllable, e.g., jalinas [d͡ʒaliˈnɑː] ‘chickenpl ’. Auvernhat is said to be less conservative than Lemosin (Bec 1963: 42), primarily due to palatalisation of /s z t d l n k ɡ/ before /i/ and /y/ and the palatalisation of /p b f v/ before /i/, e.g., libre [ljibɾe] ‘book’; dire [djiɾe] ‘sayinf ’. In bas-auvergnat, diphthongs are reduced, e.g., paire [ˈpeɾe] ‘father’; aiga [ˈiɡO] ‘water’, and fully nasalised vowels from Latin vowel + nasal sequences are common (see Sect. 4.2.2.4). In haut-auvergnat varieties, /s/ is maintained in medial pre-tonic codas before the voiceless plosives, e.g., castellu(m) > chastèl [t͡saˈstE(r)] ‘castle’. Finally, Vivaro-Alpin varieties distinguish themselves from other nordoccitan dialects by exhibiting some Francoprovençal features (e.g., deletion of intervocalic -t-) and the realisation of /l/ as [r] or [ɾ] before labial consonants and intervocalically (Bec 1963: 43), e.g., balma [ˈbaɾmO] ‘cave’. Final consonants are frequently maintained and /s/ is variable in pre-tonic syllable codas before /p t k/. Extremely rare in the Occitan
4 Phonetic and Phonological Systems
75
domain is the retention of final [-r] in verbal infinitives, e.g., cantar [t͡ʃantar] ‘singinf ’ making Vivaro-Alpin, in this respect, one of the most conservative Occitan dialects vis-à-vis Latin.
4.1.2 Southern Occitan Southern Occitan, or occitan moyen, includes the Lengadocian and Provençal dialects. This supra-dialectal area is noted to include the most conservative varieties (Bec 1963: 44) and is characterised by the following phonological developments (adapted from Bec 1963: 44): • Latin ca and ga sequences are not palatalised, e.g., capra(m) > cabra [ˈkabɾO] ‘goat’; gallu(m) > gal [ɡal] ‘rooster’; • /s/ is maintained in clusters with /p t k/, with some realisations as [h] in the north of the Lengadocian domain; • Pre-tonic a maintains its quality as [a], with some velarisation to [O] in the north of the domain, e.g., castèl [kasˈtEl] ‘castle’, but [kOsˈtEl] in Rouergue; • No consonantal palatalisation, including /s/, except in the Périgord; • Diphthongs and triphthongs are often maintained, e.g., paire [pajɾe] ‘father’; puei [pɥej] ‘then’ (cf. puei [pEj] in Toulouse); • Intervocalic /l/ is maintained, e.g., bèla [ˈbElO] ‘beautifulf.sg ’. These examples demonstrate that the features that characterise southern Occitan involve primarily the maintenance of traditional phonological forms, making this supra-dialectal area one of the most conservative from a structural perspective. Bec (1963: 44) notes that the Lengadocian dialect is most conservative and labels this dialect Occitan ‘par excellence’. The following examples illustrate the primary differences between Lengadocian and Provençal from a historical phonological perspective (adapted from Bec 1963: 45–47): • Retention of final /-s/ as the noun pluralisation morpheme in Lengadocian, e.g., pòrtas [ˈpOɾtOs] ‘doorpl ’, but loss in Provençal, e.g., [ˈpOɾtO];
76
D. Mooney
• Final /-n/ < Latin intervocalic -n- is frequently deleted in Lengadocian, e.g., pan [pa] ‘bread’, but is maintained in Provençal, e.g., [pãn]; • Final /-l/ is not vocalised, e.g., ostal [uˈstal] ‘house’, which distinguishes Lengadocian from Gascon and Provençal, e.g., [uˈstaw]; • Labio-dental /v/ has merged with /b/ in Lengadocian, e.g., lavar [laˈβa] ‘washinf ’, but is maintained in Provençal, e.g., [laˈva]; • Final consonants are largely realised in Lengadocian, e.g., vengut [beŋˈɡyt] ‘come pst.ptcp.m.sg ’, but dropped in Provençal, e.g., [veŋˈɡy]. • Definite articles are los [lus] ‘them.pl ’ and las [las] ‘thef.pl ’ in Lengadocian, but polyvalent lei [li] or [lej] in Provençal.
4.1.3 Gascon Gascon is the most divergent of the Occitan dialects; it exhibits a wide range of phonological and morphosyntactic developments that distinguish it from the rest of southern Gallo-Romance. The key phonological developments (adapted from Bec 1963: 48) will be dealt with in this section, with more detailed information on sub-dialectal variation in Gascon reserved for Sect. 4.2: • Latin f became a fully aspirated glottal fricative, [h], in Gascon, word-initially before a vowel and in intervocalic position, where it remained [f ] in other dialects, e.g., farina(m) > haria [haˈɾijO] ‘flour’; calefacere > cauhar [kawˈha] ‘heatinf ’; • Intervocalic -n- in (primarily) late-closed syllables is lost in Gascon, e.g., fenestra(m) > hièstra [ˈhjEstɾO] ‘window’; • -ll- becomes an apical tap or trill [ɾ r] in Gascon feminine nouns where it becomes [l] in other dialects (Molyneux 2002: 26), e.g., pulla(m) > pora [puˈɾO] ‘chicken’; in masculine nouns, -ll- becomes a palatalised affricate [t͡ç] when it occurs in coda position as a result of apocope (Grosclaude 1986: 9), e.g., castellu(m) > castèth [kasˈtet͡ç] ‘castle’; in most modern varieties of Gascon, this has further simplified to [t], e.g., [kasˈtet], with more conservative varieties using [t͡ʃ], e.g.,
4 Phonetic and Phonological Systems
•
•
• •
77
[kasˈtet͡ʃ] in Couserans and Commenges, or [tj ], e.g., [kasˈtetj ] in parts of Béarn. -mb- and -nd- sequences became [m] and [n] respectively in Gascon while they remained [mb] and [nd] in the rest of the Occitan dialects, e.g., cumba(m) > coma [ˈkumO] ‘valley’; retunda(m) > arredona [arːeˈdunO] ‘round’; For r- in word-initial position, the syllable onset is strengthened in Gascon to a trill prefixed by [a-], which does not occur in Occitan, e.g., rege(m) > arrei [aˈrej] ‘king’. This is suspected to be due to contact with Basque, e.g., errege ‘king’, and is also comparable to Spanish, where initial r- is always realised as a trill [r], not a tap [ɾ]. Final /-l/ is vocalised, e.g., ostal [usˈtaw] ‘house’; Latin qu- sequences are maintained as /kw/ and do not evolve to simplex /k/, e.g., quattuor > quatre [ˈkwate] ‘four’; note also that /r/ is frequently elided in final post-tonic clusters when it precedes the unstressed vowel /e/, e.g., alteru(m) > autre [ˈawte] ‘othersg ’.
4.1.4 Occitano-Romance Supra-Dialects The two major supra-dialectal groupings within Occitano-Romance, proposed by Bec (1963, 1973), alverno-méditerranéen, including northern Occitan, Provençal, and parts of Lengadocian, and aquitanopyrénéen, including Pyrenean Lengadocian, Gascon, and Catalan are characterised primarily in terms of their differing phonological developments (adapted from Bec 1973: 18–19): • Latin -ct- evolved to [jt] in aquitano-pyrénéen, e.g., factu > fait [fajt] ‘dopst.ptcp.m.sg ’ or [hEjt], but palatalised to [t͡ʃ] in alvernoméditerranéen, e.g., fach [fat͡ʃ]; • The Occitan diphthong /aj/ is realised as [Ej] in the majority of aquitano-pyrénéen dialects, e.g., qu’ei [kEj] ‘have1sg ’ in Gascon; • The alverno-méditerranéen (and standard Occitan) phoneme /d͡ʒ/ is realised variably as [ʒ] or [j] in aquitano-pyrénéen, e.g., getar /d͡ʒeˈta/ ‘throwinf ’ is pronounced [ʒeˈta] or [jeˈta];
78
D. Mooney
• The phonemes /b/ and /v/ have merged (to /b/) in aquitano-pyrénéen, e.g., vaca [ˈbakO] ‘cow’, but remain distinct in much of alvernoméditerranéen, e.g., [ˈvakO]; • Stable word-final and intervocalic consonants in aquitano-pyrénéen, e.g., cantat [kanˈtat] ‘singpst.pctp.m.sg ’ in Gascon, but [kanˈta] in Provençal; cantada [kanˈtaðO] ‘singpst.pctp.f.sg ’ in Gascon, but [t͡ʃanˈtajO] in Vivaro-Alpin; • Palatalisation intervocalically of /s/ to [ʃ] in words such as paréisser /paˈɾejse/ ‘appearinf ’, e.g., paréisher [paˈɾeʃe] in Gascon; parèixer [paˈɾEʃe] in Catalan; some varieties of Lengadocian palatalise the fricative but maintain the preceding glide which triggered the palatalisation, e.g., [paˈɾejʃe]. • First person singular inflection morpheme is /-i/ in aquitano-pyrénéen, e.g., canti [ˈkanti] ‘sing1sg ’, but /-e/ is more common in alvernoméditerranéen, e.g., [ˈkante], with the exception of provençal maritime, e.g., [ˈkanti], and Vivaro-Alpin, e.g., [ˈt͡ʃantu]. The Lengadocian dialect has the unique quality of containing a mix of alverno-méditerranéen features (e.g., /t͡ʃ/ and /d͡ʒ/, /-e/1sg in the northeastern area) and aquitano-pyrénéen forms (e.g., loss of /v/, stable final consonants, /-i/1sg in southwestern area). In fact, the only feature that it does not share with another dialect area is the final /-l/ which is not vocalised to [w] in Lengadocian, e.g., sal [sal] ‘salt’, not *[saw].
4.2
Modern Phonological Inventories
Lengadocian is often considered to be the most conservative dialect (Wheeler 1988: 246), the centre of the Occitan-Romance area—a lowest common denominator of sorts. It was regarded the most authoritative reference variety (Bec 1973: 20) and it is for this reason that it was selected as the basis for the standard. The Occitan orthographic system, known commonly as the grafia classica (‘classical orthography’), is based on the principle of a phonological diasystem. A diasystem is essentially an abstract, standardised phonological reference point against which we can
79
4 Phonetic and Phonological Systems
compare variation observed in the dialects. The aim of the diasystem— and the resultant orthography—is to at once represent all of the dialects and none of the dialects in particular (Bec 1973: 24). As we have seen, there are problematic ideological constructs associated with this diasystemic approach (see Sect. 3.4) but, nonetheless, the phonological analysis presented in this overview will use the diasystem and its standard orthography throughout as a means of facilitating detailed comparison between and across dialectal areas.
4.2.1 Consonants The Occitan diasystem contains 20 distinct consonantal phonemes (see Table 4.1), including /v/, which is only found in some varieties of alverno-méditerranéen, and excluding the glide phonemes, /j/, /w/, and /ɥ/, which are discussed below (see Sect. 4.2.3). There is some debate over the phonemic status of /t͡s/, /ʃ/, /ŋ/, and /ʁ/ (Wheeler 1988: 248), which will be dealt with in the discussion below. Bec (1973: 51) considers the inventory in Table 4.1 to represent the maximum consonantal diasystem, while also proposing a minimum diasystem of 15 phonemes that results from the following common phonological mergers in some major dialect varieties: /ʃ/ ~ /t͡ʃ/ > /ʃ/ or / tʃ/; /t͡ʃ/ ~ /d͡ʒ/ > /t͡s/; /ʎ/ ~ /j/ > /j/; /r/ ~ /ɾ/ > /ʁ/; /v/ ~ /b/ > /b/. Bec (1973: 51) also makes reference to the voiceless alveolar affricate /t͡s/, which merges with /s/ in dialects with a reduced phonological inventory. Table 4.1 The consonantal phonemes of Occitan Bilabial Plosive Affricate Nasal Trill Tap Fricative Lateral
p
Labiodental
b
Alveolar t
m
f
(v)
s l
d n r ɾ z
Postalveolar t͡ʃ
d͡ʒ
Palatal
Velar k
ñ
(ʃ) ʎ
ɡ
80
D. Mooney
The Occitan voiceless plosives /p t k/ derive from Latin voiceless plosives in word-initial position and in syllable-initial position word-medially when following a consonant, e.g., campu(m) > camp ‘field’. They also derive from geminate voiceless plosives in intervocalic position, e.g., cuppa(m) > copa /ˈkupO/ ‘champagne glass’. The voiced plosives /b d ɡ/ have evolved from intervocalic Latin voiceless (or from voiced geminate) plosives, e.g., lupa(m) > loba /ˈlubO/ ‘wolff.sg ’ (Oliviéri and Sauzet 2016: 325). Voiced plosives originally occurred word-finally, e.g., caput > cab /kab/ ‘head’, but final devoicing became established from the twelfth century onwards, e.g., cap [kap] (Oliviéri and Sauzet 2016: 325). Latin ca and ga sequences gave rise to /k/ and /ɡ/ respectively, but these phonemes vary in their phonetic realisation over geographical space. In northern Occitan dialects, /k/ is palatalised to [t͡ʃ], e.g., chabra [t͡ʃabɾO] ‘goat’; the affricate /d͡ʒ/ occurs initially, e.g., jalina [d͡ʒaˈlinO] ‘hen’, and after a consonant, e.g., longa [ˈlund͡ʒO] ‘longf.sg ’, but is realised as [j] after a vowel, e.g., paiar [paˈja] ‘payinf ’ (Wheeler 1988: 250–251). In Vivaro-Alpin, and more specifically in the albonnais sub-dialect, these initial palatalised forms are alveolar, rather than postalveolar, e.g., cantare > chantar [t͡sOnˈtE] ‘singinf ’; gallu(m) > gal [d͡zar] ‘rooster’ (Quint 1999: 14). In fact, Vivaro-Alpin displays a system of palatalisation that results in the neutralisation of phonemic contrast between some plosive pairs. The /t/ ~ /k/ contrast is neutralised to [c] before /i/ and /y/, e.g., qui [ci] ‘who’; tu [cy] ‘you2sg ’; the /d/ ~ /ɡ/ contrast is neutralised to [ɟ] in the same phonological context, e.g., dina [ˈɟinO] ‘dine3sg ’; guida [ˈɟidO] ‘guide’ (Quint 1999: 8). Coustenoble (1945: 76) notes that [ɟ] also occurs as an allophone of /ɡ/ before [i e je jE] in rhodanien Provençal, e.g., seguida [seˈɟidO] ‘followpst.ptcp.f.sg ’. In Gascon, /b d ɡ/ have contextually conditioned approximant allophones [β̞ ð̞ G̞] when they occur intervocalically, alone or in a cluster with [ɾ r l z], e.g., saba [ˈsaβO] ‘know3sg ’, marga [ˈmaɾGO] ‘sleeve’ (Oliviéri and Sauzet 2016: 325). This also occurs across word boundaries, e.g., lo buòu [luˈβwOw] ‘the ox’. In western Gascon, the / b/ phoneme, realised as [b], derives from b and v and intervocalic p. In Eastern and Central Gascon, initial b and v also converge to [b] (alternating allophonically with [β] after a vowel) but intervocalic -p-
4 Phonetic and Phonological Systems
81
yields [β], e.g., sapere > saber [saˈβe] (‘knowinf ’). Intervocalic /b/ that derives from b or v, however, yields [w], e.g., lavabat > lavava [laˈwawO] (‘wash3sg.impf ’). In some Pyrenean valley varieties, these contextually conditioned allophonic variants of intervocalic /b d ɡ/ are replaced by their corresponding voiceless plosives [p t k] when they occur before [a O y u] (Moreux and Puyau 2002: 25), e.g., cabra [ˈkɾaβO] ‘goat’ is pronounced [ˈkɾapO]. There is also considerable variation in southwestern varieties between voiced and voiceless plosive pairs, [p b], [t d], [k ɡ], when following nasal consonants [m] and [n], e.g., cramba [ˈkɾampO] or [ˈkɾambO] ‘bedroom’. The Occitan diasystem has three nasal consonant phonemes: /m/, / n/, and /ñ/, e.g., lama [ˈlamO] ‘blade’; lana [ˈlanO] ‘wool’; lanha [ˈlañO] ‘grief ’ (Oliviéri and Sauzet 2016: 326). The velar [ŋ] and labiodental [ɱ] nasals exist as allophones of the other nasals when followed by a velar or labiodental consonant respectively (Coustenoble 1945: 82), e.g., longa [ˈluŋɡO] ‘longf.sg ’; comfortable [kuɱfuɾˈtable] ‘comfortable’. In wordfinal position, the dialects show considerable variability (Wheeler 1988: 249; Oliviéri and Sauzet 2016: 326): word-final /m n ñ/ are realised as [n] in Lengadocian (Salvat 1973: 7; Sampson 1999: 149), e.g., fum [fyn] ‘smoke’, but to [ŋ] in Provençal (Ford 1921: 31; Coustenoble 1945: 1), e.g., banh [baŋ] ‘bath’. Provençal [ŋ] is long in stressed syllables when it is preceded by [e E ø], e.g., novembre [nuˈveŋːbɾe] ‘November’, but short elsewhere, e.g., camin [kaˈmiŋ] ‘pathway’ (Coustenoble 1945: 67). In Lengadocian, nord-occitan, and some varieties of Gascon (e.g., Béarnais and Bigordan), word-final /n/ is frequently deleted, e.g., vin [bi] ‘wine’ and, in nord-occitan, this often leads to the development of nasal vowels (see Sect. 4.2.2.4). Some, but not all, Vivaro-Alpin dialects, e.g., albonnais (Quint 1999: 18), also delete /n/ word-finally, e.g., camin [t͡sOˈmi] ‘pathway’. In Gascon, /m/, /n/, and /ñ/ can all occur word-finally, e.g., fum [hym] ‘smoke’, but the reflex of Latin intervocalic -n- is lost, e.g., luna > luna [l˜yO] ‘moon’; this deletion also frequently involves the development of nasal vowels (see Sect. 4.2.2.4) and has also occurred in Galician, Alpine dialects of south-eastern France and north-western Italy, dialects of Sardinian and Corsican, and early Romanian (Sampson 1999: 145). Sampson (1999: 153) notes that the Gascon varieties spoken in the Landes department, referred to as lo parlar negre, have the maximum
82
D. Mooney
number of nasal consonants that may appear in word-final position: [m -n -ñ -ŋ]. In these dialects, intervocalic n yields [ŋ] (after apocope) whereas final [n] derives from -nd and -nt clusters which have become final, e.g., venit > ven [ˈbeŋ] ‘come3sg ’ vs. ventum > vent [ˈben] ‘wind’. The trill /r/ and the tap /ɾ/ are in contrastive distribution in intervocalic positions, e.g., poret /puˈɾet/ ‘chicken’ ~ porret /puˈret/ ‘leek’, but not contrastive in other contexts such that ‘an archiphoneme could be set up for all other positions’ (Cardaillac Kelly 1973: 32). Some analyses choose to treat this contrast as a single coronal rhotic phoneme with a long-short length contrast (Oliviéri and Sauzet 2016: 326). The apical rhotics are not, however, in strictly complementary distribution in non-intervocalic contexts: the distribution of [r] and [ɾ] is somewhat constrained by their position within the syllable and with respect to word boundaries with a tendency for [r] to occur word-initially and as an onset after [n], and [ɾ] to occur in onset clusters and in the syllable coda, but this distribution is by no means categorical (Cardaillac Kelly 1973: 32; Mooney 2014: 345). It is most prevalent in ‘the conservative west’ (Wheeler 1988, 250), or in aquitano-pyrénéen Occitan; in other varieties, such as Provençal and its sub-dialects, both /r/ and /ɾ/ are traditionally realised as [ʁ] or [ʀ], neutralising contrasts in intervocalic positions (Bec 1973: 48). In some sub-dialects, however, such as rhodanien, only the trill is replaced by a uvular consonant and the intervocalic contrast is maintained (Coustenoble 1945: 93), e.g., pòre [ˈpOɾe] ‘pore’ ~ pòrre [ˈpOʁe] ‘leek’. The uvular realisation of the rhotic consonant has been attested in Provençal varieties since at least the eighteenth century (Stéfanini 1969: 167, cited in Oliviéri and Sauzet 2016: 326) and as such it is unlikely that this feature is the result of transfer from French since at this time the apical rhotic most likely constituted the French norm. In other varieties, such as Gascon, however, there is evidence to suggest that apical rhotics are being replaced by uvular rhotics as a result of contact with French (Mooney 2018), with various constraints on transfer such as phonological environment, position in the syllable, and the speakers’ place of origin within Gascony. Finally, Proto-Occitan word-final [r] is systematically deleted in infinitives (and some suffixes such as -ièr < -arius), e.g., cantar /kanˈta/ ‘singinf ’, though this is often retained in Vivaro-Alpin, e.g., [t͡ʃanˈtar].
4 Phonetic and Phonological Systems
83
Latin f remained /f/, realised as a labiodental fricative [f ] throughout the Occitan area, e.g., ferrum > fèr(re) [ˈfEr(e)] ‘iron’, with the exception of Gascony, where it is realised as a fully aspirated voiceless glottal fricative [h], e.g., hèr [hEr]. When [f ] occurs in Gascon, it is due to borrowing from French, e.g., fresa [ˈfɾEzO] < Fr. fraise ‘strawberry’. When the f > [h] change occurs in an onset cluster with a rhotic consonant, the [h] is deleted, giving the sequence f > [h] > 0 (Wheeler 1988: 250), e.g., formaticu(m) > Oc. formatge [forˈmad͡ʒe] ‘cheese’, Gasc. hromatge [ruˈmad͡ʝe]; in fact, /h/ in initial position is also subject to (variable) deletion (Field 1978: 83). In some Occitan varieties, the Latin b ~ v contrast is maintained, e.g., vin bon /vin bun/ ‘wine goodm.sg ’, but in many others, these phonemes have merged to /b/, a process called ‘betacism’: ‘today, the betacizing area covers all of Gascony, almost all of Languedoc, and a large portion of Auvergne’ (Oliviéri and Sauzet 2016: 325). The /v/ phoneme is more common in varieties of alvernoméditerranéen (Bec 1973: 37) and is attested in rhodanien Provençal (Ford 1921: 55–56), albonnais (Quint 1999: 13), Chiomonte (Sibille 2012: 2235) Vivaro-Alpin, and in Lemosin (Oliviéri and Sauzet 2016: 325). Ford (1921: 79) also notes that in rhodanien, words beginning with [u] or [ɥe] have a tendency to develop an initial [v], e.g., onze / ˈund͡ze/ [ˈvuŋd͡ʒe] ‘eleven’. In Gascon generally, /b/ is the modern reflex of Latin initial b- and v-, and intervocalic -p-, -b-, and -v- (Oliviéri and Sauzet 2016: 325), e.g., que trobava lo vin bon [ketɾuˈβaβOluˈβ˜ıˈβu˜ ] ‘he found the wine good’. In some varieties of Gascon, intervocalic -pevolved to /b/ and intervocalic -b- and -v- evolved to /w/, leading to an opposition between /b/ and /w/ between vowels, with the latter corresponding to alverno-méditerranéen /v/ (Oliviéri and Sauzet 2016: 325), e.g., [ketɾuˈβawOluˈβ˜ıˈβu˜ ]. The spread of betacism from Gascon means that Lengadocian is now predominately betacising as well. The fricatives /s/ and /z/ are usually apico-alveolar in Occitan; the contrast is neutralised in preconsontal position, with [s] before voiceless consonants and pauses and [z] or [ʒ] before voiced consonants (Bec 1973: 42; Field 1978: 82), e.g., mesclar [mesˈkla] ‘mixinf ’; esdracar [ezdraˈka] ‘wring outinf ’. /s/ and /z/ may be postalveolar in nord-occitan varieties (Wheeler 1988: 248), e.g., saison /seˈzun/ [ʃeˈʒu˜ ] ‘season’, where, additionally, pre-consonantal /s/ is subject to weakening and is frequently
84
D. Mooney
realised as [x], [h], [j], vowel lengthening, or 0, with [j] as the majority variant (Wheeler 1988: 251), e.g., escòla [ehˈkOlO] ‘school’, [ejˈkOlO], [eˈkOlO] etc. In Vivaro-Alpin, and more specifically in the sub-dialect of albonnais, /s/ is realised as [ʃ] in word- and syllable-initial position, as well as intervocalically, e.g., sopa [ˈʃupO] ‘soup’; perseguièr [perˈʃje] ‘peach tree’; caçaire [t͡soˈʃajre] ‘hunter’, and as [h] pre-consonantally and pre-pausally, e.g., chasque [ˈt͡sahke] ‘every’; cantas [ˈt͡sOntEh] ‘sing2sg ’. In word-final position, /s/ is realised as [ʒ] before a vowel, e.g., los arbres [luˈʒarbɾEh] ‘the treepl ’, as [h] before /p t ts k/, e.g., los chals [luhˈt͡sOwh] ‘the cabbagepl ’, and as [j] before other consonants, e.g., los libres [lujˈlibɾEh] ‘the bookpl ’. In Lengadocian, word-final /s/ is realised as [j] before consonants other than voiceless plosives (Field 1978: 101; Wheeler 1988: 252), e.g., las claus [lasˈklaws] ‘the keypl ’; las femnas [lajˈfennOs] ‘the womanpl ’; before voiceless plosives, word-final /s/ may be realised as [h], e.g., [lahˈklaws]. The phonemic status of /ʃ/ is not established for the majority of Occitan dialects; as we have seen, [ʃ] primarily occurs as a geographical and/or allophonic variant of /s/. Equally, [ʒ] is attested as a variant of /z/ but the sound is not contrastive. [ʒ] in Gascon can correspond to an affricate, e.g., Oc. jamès /d͡ʒaˈmEs/ ‘never’ and Gasc. [ʒaˈmEs]. In southwestern Lengadocian (i.e., aquitanopyrénéen), [ʃ] appears as a contextually conditioned allophone of /s/ following /j/, e.g., peis /pejs/ [pejʃ] ‘fish’ (Oliviéri and Sauzet 2016: 325). In Gascon, however, /ʃ/ is a phoneme; peis is realised as [peʃ], without contextual conditioning, and minimal pairs can be found to establish the /ʃ/ ~ /s/ contrast, e.g., peis /peʃ/ ‘fish’ ~ pes /pes/ ‘weight’. [ʃ] also occurs as an allophone of /s/ in Gascon, following palatal consonants (Field 1978: 90), e.g., uelhs /weʎs/ [weʎʃ] ‘eyepl ’. The voiced fricative /z/ only occurs in intervocalic position in standard Occitan and derives from intervocalic -d-, -c-, and -s- (Wheeler 1988: 251), e.g., audire > ausir /awˈzi/ ‘hearinf ’; racemu(m) > rasim /raˈzim/ ‘grape’; causa(m) > causa /ˈkawzO/ ‘thing’. Some dialects, such as Gascon, maintain [d] for Occitan /z/ < -d-, e.g., ausir [awˈdi], and sometimes for /z/ < -c-, rasim [araˈdim]. Intervocalic /z/ has largely been lost in Provençal, camisia(m) > camisa /kaˈmizO/ [kaˈmjO] ‘shirt’, with glide formation of the stressed vowel and stress shift to the final vowel.
4 Phonetic and Phonological Systems
85
The Occitan diasystem has two postalveolar affricates, /t͡ʃ/ and /d͡ʒ/, e.g., chicar /t͡ʃiˈka/ ‘chew tobaccoinf ’; jamai /d͡ʒaˈmaj/ ‘never’, which vary widely across dialects both in terms of their phonetic realisation and their phonemic status. In northern varieties, especially Lemosin, Auvernhat, and northern Lengadocian, /t͡ʃ/ and /d͡ʒ/ are depalatalised to [t͡s] and [d͡z]; in the northernmost part of the Occitan area, the affricates are realised as [s] or [θ] and [z] or [ð], respectively (Wheeler 1988: 248; Oliviéri and Sauzet 2016: 325). In areas where /t͡ʃ/ is realised as [s] and /d͡ʒ/ as [z], /s/ and /z/ are palatalised to [ʃ] and [ʒ] (Wheeler 1988: 248), e.g., Lim. cerchar /serˈt͡ʃa/ [ʃerˈsa] ‘look forinf ’. In varieties that preserve both phonemes, such as Provençal, /t͡ʃ/ may be realised at [t͡ʃ] or [t͡s] and /d͡ʒ/ as [d͡ʒ] or [d͡z] (Coustenoble 1945: 86; Bec 1973: 44); in rhodanien, Coustenoble (1945: 86) notes that the affricates are depalatalised to [t͡s] and [d͡z]. In much of the Lengadocian dialect, the affricates are merged to [t͡s] (Bec 1973: 44), e.g., chicar [t͡siˈka] ‘chew tobaccoinf ’, dotze [ˈdut͡se] ‘twelve’. Oliviéri and Sauzet (2016: 326) note considerable variation within Lengadocian: in Béziers, for example, / t͡ʃ/ and /d͡ʒ/ merge in terms of voicing, but do not depalatalise. The phonemic status of /t͡s/, as distinct from /t/ + /s/ or from /t͡ʃ/, is marginal (Wheeler 1988: 248); [t͡s] is primarily involved in a morphophonological alternation with [z], e.g., crotz [kɾut͡s] ‘cross’ ~ crosar [kɾuˈza] ‘crossinf ’. In western Lengadocian varieties, /d͡ʒ/ may be realised as [ʒ] (Field 1978: 80; Wheeler 1988: 248); this is due to the fact that Gascon realises /d͡ʒ/ as [ʒ] or [j], e.g., jamai [ʒaˈmEj], [jaˈmEj] ‘never’. Latin -ct- became /t͡ʃ/ in most of the Occitan area, but is realised as /jt/ in Gascon, western Lengadocian, Auvernhat, and Vivaro-Alpin (Wheeler 1988: 251; Quint 1999: 17) (see Sect. 4.1.4). Some varieties of Gascon, such as Béarnais, have two palatalised affricate phonemes: voiceless /t͡ç/ and voiced /d͡ʝ/. In masculine nouns, root-final -ll- became voiceless / t͡ç/ when it occurred in coda position as a result of apocope (Grosclaude 1986: 9), e.g., castellu(m) > castèl [kasˈtEt͡ç] ‘castle’; within Gascony, [t] and [t͡ʃ] are common as variants of /t͡ç/ (Mooney 2014: 348). Castet (1895: 16) notes that palatalised [t͡ç] may occur as a variant of /t/ in past participles in the mountain dialect of Couserans, e.g., cantat [kanˈtat͡ç]
86
D. Mooney
‘singpst.ptcp.m.sg ’, perhaps by analogy. -tc- sequences, on the other hand, which occurred as result of elision, developed into /d͡ʝ/ in Gascon, e.g., villaticum > vilatge [biˈlad͡ʝe] ‘village’. Finally, in loanwords from French, /ʃ/ is often rendered as an affricate in Occitan, e.g., cheval > chabal [Ùaˈβal] ‘horse’ in Lengadocian; chivau [Ùiˈvaw] in Provençal, but as /ʃ/ in Gascon (Oliviéri and Sauzet 2016: 325), e.g. chivau [ʃiˈβaw], perhaps due to the phonemic status of /ʃ/ in Gascon. Occitan /l/ is a dental lateral approximant, which may be velarised in coda position (Bec 1973: 46), where it commonly vocalises to [w] in dialects other than conservative Lengadocian, e.g., calet > Leng. cal [kal] ‘be necessary3sg ’; Gasc. que cau [keˈkaw]. Within Lengadocian, however, some sub-dialects show the /l/ > [w] change in codas before dentals, while others exhibit the change before consonants other than dentals (Wheeler 1988: 249). There is also evidence for contextual conditioning in other dialects. For example, in the rhodanien dialect of Provençal, vocalisation occurs in all contexts except after /u/ and /y/, e.g., lo sol [luˈsul] ‘the sun’ (with the suffixed form solèu also occurring); note, however, that lo só [luˈsu] occurs in other dialects, including Gascon. In the albonnais sub-dialect of VivaroAlpin, /l/ is preserved in singular nouns, but vocalised in plurals, e.g., castèl [tsOhˈtEr] ‘castlesg ’; castèls [tsOhˈtEw] ‘castlepl ’; these examples also demonstrate the common realisation of /l/ as [r] or [ɾ] in Vivaro-Alpin (Wheeler 1988: 253). In Auvernhat, intervocalic, syllable-initial /l/ may vocalise, e.g., pala [ˈpawO] ‘shovel’, or be realised as a velar [ɡ] or labiodental [v], e.g., [ˈpaɡO], [ˈpavO], and in Gascon final /l/ < -ll- is realised as [t͡ç], [t͡ʃ], or [t], e.g., castèl [kasˈtEt͡ç] ‘castle’. Gascon /t͡ç/ in masculine nouns alternates morphophonologically with [ɾ] in feminine nouns, e.g., bèth [bEt͡ç] ‘beautifulm.sg ’; bèra [bEɾO] ‘beautifulf.sg ’ (Field 1978: 79). Finally, the palatal lateral may occur word-initially, intervocalically, and word-finally in the Occitan diasystem, but not medially in coda position and it is subject to a variety of constraints in the dialects themselves. Gascon traditionally preserves /ʎ/ in all positions, e.g., lhevar [ʎeˈβa] ‘raiseinf ’; tribalhar [tɾiβaˈʎa] ‘workinf ’; uelh [weʎ] ‘eye’. There are relatively high rates of retention in intervocalic position across dialects (with
4 Phonetic and Phonological Systems
87
the exception of Provençal), but the palatal approximant [j] frequently emerges as an allophone of /ʎ/ in intervocalic position, e.g., tribalhar [tɾibaˈja]. In final position, /ʎ/ is realised as [l] in Lengadocian (Field 1978: 102) and as [j] or [w] in Provençal, e.g., genolh > Leng. [dʒeˈnul] ‘knee’, Prov. [dʒeˈnuj]; filh > Leng. [fil] ‘son’, Prov. [fjew] (Wheeler 1988: 249). The [w] variant in Provençal is the result of first depalatalisation to [l] and then vocalisation to [w] (Ford 1921: 20). In Vivaro-Alpin, the [ʎ] variant is preserved with high rates intervocalically, and palatal laterals may arise from the palatalisation of /l/ in initial /ɡl/ and /kl/ clusters, e.g., la glèisa [laˈɡʎEjzO] ‘church’; la clau [laˈkʎaw] ‘key’.
4.2.1.1 Syllable- and Word-Final Consonants In standard Occitan, based on Lengadocian, only sonorants and /s/ are permitted in non-final syllable codas (Oliviéri and Sauzet 2016, 326), e.g., factor [fa(t)ˈtu] < Fr. facteur /fakˈtœʁ/ ‘postman’, where /k/ assimilates in place or is omitted. In word-final position, /j/, /w/, /r/, /N/, /s/ and, sometimes, /f/ are permitted in codas; there is a strong tendency in nord-occitan and Provençal to delete word-final obstruents (Wheeler 1988: 252–253). In Lemosin, /s/ in coda position is deleted with compensatory lengthening or it debuccalises to [h], /l/ vocalises to [w], and nasal consonants are deleted, with only [j] and [r] permitted in coda position (Oliviéri and Sauzet 2016: 326); similarly, Vivaro-Alpin only allows five consonants, [h ʁ n j w] (Quint 1999: 8). In wordfinal codas, Bec’s (1963, 1973) distinction between alverno-méditerranéen and aquitano-pyrénéen Occitan best accounts for the distribution of final consonants (see Sects. 3.1 and 4.1.4). In aquitano-pyrénéen dialects, words can end in obstruents and in clusters, e.g., lo còp [luˈkOp] ‘the time’, la sèrp [laˈsEɾp] ‘the snake’; in alverno-méditerranéen dialects, final obstruents are deleted and deleted from clusters, e.g., lo còp [luˈkO], la sèrp [laˈsEr]. Eastern Lengadocian allows final obstruents, e.g., [luˈkOp], but not final clusters, e.g., [laˈsEr], except in nominal plurals with the morpheme /-s/, e.g., las sèrps [lasˈsErs] ‘the snakepl ’ (Oliviéri and Sauzet 2016: 327); some Gascon varieties allow triple consonant clusters with plural /-s/, e.g., [las sEɾps]. Some Provençal dialects, such as nissart,
88
D. Mooney
permit diasystemic consonant clusters by moving the obstruent into the onset of a syllable whose nucleus is a paragogical vowel (Oliviéri and Sauzet 2016: 327), e.g., la sèrp [laˈseɾpe]. Where they are retained, final voiced plosives /b d ɡ/ are devoiced to [p t k] (Field 1978: 80), e.g., saber /saˈbe/ ‘knowinf ’, sap / sap/ ‘know3sg ’; amic /aˈmik/ ‘friendm ’, amiga /aˈmiɡO/ ‘friendf ’. Before vowels, final fricatives are voiced, e.g., los arbres [luˈzaɾbɾes] ‘the treepl ’; /t͡ʃ/ can be voiced or voiceless, while /v/ is commonly vocalised in final position, e.g., nòu [nOw] ‘newm ’, nòva [ˈnOβO] ‘newf ’ (Bec 1973: 36, 40; Field 1978: 115; Wheeler 1988: 252). Only Gascon maintains full contrast between the laterals /l/ and / ʎ/ and the nasals /m n ñ/ in final position, with Lengadocian and Provençal neutralising these contrasts to [l] and [n], and [w] and [ŋ], respectively (Coustenoble 1945: 1; Bec 1973: 45; Wheeler 1988: 252). Quint (1999: 8) notes that final /m/ does occur in verbal paradigms in Vivaro-Alpin, e.g., avèm [OˈvEm] ‘have1pl ’, but that it is increasingly replaced by [n], e.g., aurem [OwˈʁEn] ‘have1pl.fut ’, and that there is therefore no /m/ ~ /n/ contrast in final position. Place contrasts for coda obstruents /p t t͡ʃ k/ are maintained, especially in Lengadocian and Vivaro-Alpin (Wheeler 1988: 252), before vowels, but before consonants these contrasts are generally neutralised, e.g., còps [kOts] ‘timepl ’, occitan [utsiˈta] ‘Occitan’, except for the contrast between /s/ and /f/ (Wheeler 1988: 252). In dialects with /t͡s/, it is common for [t͡s] to simplify to [s]; since [t͡s] can arise in coda position from /t/ + /s/, /p/ + /s/, and /k/ + /s/, this results in an abundance of forms with syllablefinal [s]. Final plosives also assimilate in manner to following consonants, e.g., ròc mòl [rOmˈmOl] ‘soft rock’, triggering a geminate, and consonants between consonants are frequently deleted (Wheeler 1988: 252), e.g., bèls miralhs [ˈbElmiˈrals] ‘beautifulm.pl mirrorpl ’.
4.2.2 Vowels The full Occitan vowel system is presented in Table 4.2 and contains seven phonemic oral vowels in stressed syllables: /i y e E a O u/, preserving the common Western-Romance vowel system (Sampson 1999: 140).
4 Phonetic and Phonological Systems
89
This vowel system has arisen from two unconditioned changes, /u/ > /y/ and /o/ > /u/, which have occurred in all dialects (Wheeler 1988: 247), e.g., comun /kuˈmyn/ < comm une(m). In pre-tonic syllables, this ¯ system is reduced to /i y e a u/ and, in post-tonic syllables, to /i e O u/. Some analyses consider stressed /O/, e.g., sòl /sOl/ ‘ground’, as distinct from post-tonic word-final /-O/, e.g., sòla /ˈsOlO/ ‘sole’, preferring to use the symbol /o/ (Coustenoble 1945: 4; Wheeler 1988: 247; Quint 1999: 5) or /e/ (Bec 1973: 35) for the latter; the analysis presented here will not consider them to represent independent phonemes but the variability in each context is treated in detail in 4.2.2.1 and 4.2.2.3 respectively. For rhodanien, Coustenoble (1945: 4) replaces /y/ with /ø/ (see Sect. 4.2.2.1) and adds /œ/ to the inventory; for Vivaro-Alpin, Quint (1999) also includes /œ/, both giving an eight-term vocalic inventory. In most Occitan varieties, the tonic syllable of lexical words may be final or penultimate, though proparoxytons are permitted in nissart Provençal, e.g., dimenge /diˈmend͡ʒe/ ‘Sunday’ vs. Nissart [diˈmeneɡe]. The position of tonic stress on the final or penultimate syllable is subject to inter-dialectal variation, though the general pattern can be described as follows (Mooney 2014: 347–348): (i) stress is lexically defined for any word ending in a vowel other than /a/; (ii) words ending in /a/ are always oxytonous, e.g., cantar /kanˈta/ ‘singinf ’; (iii) words ending in post-tonic /-O/ are paroxytonous, e.g., cambra /ˈkambɾO/ ‘bedroom’. Words ending in a consonant are oxytonous, with the following exceptions: (a) verb conjugations ending in 3rd person plural /-n/ are paroxytonous, e.g., cantan /ˈkantOn/ ‘sing3pl ’; (b) verb conjugations ending in 2nd person singular /-s/ are paroxytonous in the majority of tenses, e.g., venes / ˈbenes/ ‘come2sg ’; (c) for nouns and adjectives, the plural morpheme / Table 4.2 Stressed oral vowels in Occitan Oral close close-mid open-mid open
Front
Back
Unrounded
Rounded
Rounded
i e E a
y
u O
90
D. Mooney
-s/ does not modify the stress pattern found in the singular form, e.g., Gasc. la gojata /laɡuˈjatO/ ‘the young girlsg ’; las gojatas /lazɡuˈjatOs/ ‘the young girlpl ’. In short, if a lexical word ends in a heavy syllable (i.e., one with a diphthong or a final consonant), stress is final; otherwise, stress is penultimate. Therefore, closed syllables (i.e., words ending in consonants) attract primary stress, but many dialects of Occitan have a strong tendency to drop word-final consonants (see Sect. 4.2.1.1). This has led to a number of words ending in stressed monophthongal vowels (Wheeler 1988: 251; Oliviéri and Sauzet 2016: 238), e.g., donar /duˈna/ ‘giveinf ’, Prov. qualitat [kaliˈta] ‘quality’. Synchronically, therefore, the general rule is that words ending in an underlying consonant are stressed on the final syllable (cf. Bec 1973, 56). Lexical exceptions to the general stress pattern include words like aquò /aˈkO/ ‘that’ and aicì /ajˈsi/ ‘here’ (Oliviéri and Sauzet 2016: 238), loanwords such as velò [veˈlO] ‘bicycle’, and some words ending in -ol /-ul/ (Salvat 1973: 16), e.g., apòstol / aˈpOstul/ ‘apostle’. Additionally, inflection morphemes do not alter the stress pattern of words to which they are affixed, e.g., canta /ˈkantO/ ‘sing3sg ’, cantas /ˈkantOs/ ‘sing2sg ’. In central Lemosin, however, stress is systematically re-assigned to the final long or closed syllable, even for inflectional morphemes (Oliviéri and Sauzet 2016: 238), e.g., cantas [t͡ʃanˈtaː] ‘sing2sg ’ or [sanˈtaː]. While nissart Provençal allows antipenultimate stress in general, e.g., pèrsegue [ˈpEseɡe] ‘peach’, Aranese Gascon only allows it in learned words (Oliviéri and Sauzet 2016: 238), e.g., musica [ˈmyzikO] ‘music’. In some dialects, such as albonnais VivaroAlpin, stress alone distinguishes grammatical person (Quint 1999: 5), e.g., fenisse [feˈniʃe] ‘finish1sg ’ ~ fenissètz [feniˈʃe] ‘finish2pl ’.
4.2.2.1 Stressed Oral Vowels The /i/ phoneme remains relatively stable across dialects; /y/, however, demonstrates some variability. The /y/ phoneme evolved from Latin tonic long u, ¯ but the date of change from /u/ to /y/ is not known (Coustenoble 1945: 14). This development occurred throughout Gallo-Romance, including French, but did not take place in Catalan. The /y/ vowel is known to be involved in free variation with /i/
4 Phonetic and Phonological Systems
91
and /e/, and to replace /ø/ in phonologically integrated loanwords from French (Müller and Martin 2012: 158). The acoustic quality of /y/ is said to be highly variable, showing some lowering and centralisation (Müller and Martin 2012: 158). Indeed, the most frequent variant of this phoneme is [ø], e.g., muscle /myskle/ [ˈmøskle] ‘mussel’: ‘This system is found from Arles to an area on the edge of Catalan-speaking domains which has neither [y] nor [ø]’ (Oliviéri and Sauzet 2016: 322). Coustenoble (1945), working on the rhodanien Provençal dialect of Arles, chooses to use the symbol /ø/ for the phoneme, noting that /y/ can be realised as [e] before a nasal consonant, e.g., un /yn/ [eŋ] ‘one’, demonstrating both opening (to [ø]) and unrounding (to [e]) in this context. As mentioned above, Coustenoble (1945) also cites /œ/ as a phoneme of rhodanien; this corresponds to /ɥE/ in the diasystem, e.g., vuèja /vɥEd͡ʒO/ [vœːd͡zO] ‘emptyf.sg ’. When /y/ finds itself in a stressed syllable closed by /j/ or /ɾ/, it is realised as [œ] in Auvernhat (Field 1978: 119), e.g., durmir [dyɾˈmi] ‘sleepinf ’, duerm [dœɾ] ‘sleep3sg ’. The third high vowel /u/ is relatively stable in stressed position. There are some Occitan varieties, however, where all close vowels diphthongise to falling diphthongs such as in Saint-Agnès, near the Italian border (Oliviéri and Sauzet 2016: 323), e.g., fil [føj] ‘thread’. The /e/ ~ /E/ contrast is largely maintained in final closed syllables, e.g., set /set/ ‘thirst’ ~ sèt /sEt/ ‘seven’, and in non-final syllables, e.g., crema /kɾemO/ ‘burn3sg ’ ~ crèma /kɾEmO/ ‘cream’ (Oliviéri and Sauzet 2016: 322). Bec (1973: 57) notes that maintenance of this contrast is more robust in aquitano-pyrénéen varieties, weak in alvernoméditérranéen, and near absent in Provençal. In nord-occitan, the /e/ ~ /E/ contrast is largely dephonologised, with [e] and [E] functioning as contextually conditioned allophones of the same phoneme (Bec 1973: 57; Wheeler 1988: 247; see Sect. 4.2.2.3), with [e] in open syllables and [E] in closed syllables. In rhodanien, stressed /e/ may be rounded to [ø] before nasal consonants, e.g., femna [føːmO] ‘woman’ (Coustenoble 1945: 22); this may involve neutralisation of the diasystem’s /e/ ~ /y/ contrast, e.g., fen /fen/ [føŋ] ‘hay’ ~ fum /fym/ [føŋ] ‘smoke’. In the socalled parlar negre of Gascon, in the Landes department, stressed /e/ is realised as [œ], such that the /e/ ~ /E/ contrast is maintained by rounding rather than by vowel aperture, e.g., set [sœt] ~ sèt [sEt] (Oliviéri and
92
D. Mooney
Sauzet 2016: 322). In Auvernhat, /e/ may be closed to [i], e.g., cercle [ˈsiɾkle] ‘circle’. Stressed /a/ is frequently labialised to [O] before nasals in nord-occitan and northern Lengadocian dialects (Field 1978: 114; Quint 1999: 9; Oliviéri and Sauzet 2016: 322), e.g., chanta [ˈt͡ʃOntO] ‘sing3sg ’. There are some lexicalised exceptions to this in Vivaro-Alpin (Quint 1999: 9), in particular the words for multiples of ten in the numeral system, e.g., cinquanta [ʃiŋˈkantO] ‘fifty’. Word-final stressed /a/ is also routinely realised as [E] in the albonnais sub-dialect of Vivaro-Alpin, e.g., donar [duˈnE] ‘giveinf ’; this feature is systematic in albonnais but also frequently found in the variety of Lemosin spoken around the town of Nontron and in briançonnais, but only in verbal infinitives (Quint 1999: 10). In rhodanien Provençal, /a/ is realised as [æ] in final open syllables, e.g., cofar [kwiˈfæ] ‘style hairinf ’; in non-final stressed syllables, the realisation is frequently [ɑː] (Coustenoble 1945: 27–30), cofada [kwiˈfɑːdO] ‘style hairpst.ptcp.f.sg ’. The /O/ shows little variability in stressed syllables, though it is usually longer in monosyllabic words (Müller and Martin 2012: 157), e.g., còp [kOːp] ‘time’, and, in Provençal and Rouergat, is diphthongised in closed syllables, e.g., pòrc [pwOɾk] ‘pig’, [pwEɾk], [pwaɾk] (Oliviéri and Sauzet 2016: 322); in Vivaro-Alpin, /O/ is also frequently diphthongised to [wO] (Quint 1999: 10; Müller and Martin 2012: 157).
4.2.2.2 Vowel Length In Lemosin and Auvernhat, phonemic vowel length has developed as a result of the loss of the plural /-s/ morpheme (Oliviéri and Sauzet 2016: 323); the deletion of /-s/ for plural nouns has resulted in compensatory vowel lengthening, such that singular and plural nouns are distinguished by vowel length, e.g., pè [pe] ‘footsg ’ ~ pès [peː] ‘footpl ’. Vowel length also serves to distinguish second and third person singular in verbal paradigms, e.g., cantas [ˈsãtOː] ‘sing2sg ’ ~ canta [ˈsãtO] ‘sing3sg ’; this may be accompanied by stress reassignment, e.g., cantas [sãˈtaː] ~ canta [ˈsãtO] (Oliviéri and Sauzet 2016: 323; cf. also Meinschaefer 2020). In rhodanien Provençal, vowel length is phonetically conditioned,
4 Phonetic and Phonological Systems
93
not contrastive (Coustenoble 1945: 98–100): in final open syllables, stressed and unstressed vowels are short, e.g., dormir [ˈdurmi] ‘sleepinf ’; in medial syllables (open and closed), stressed vowels are long and unstressed vowels are short, e.g., divina [diˈviːnO] ‘divinef.sg ’; in final closed syllables, vowels are short before [-ŋ] and long before /-s/ and / -r/, e.g., partirés [paʁtiˈʁeːs] ‘leave2sg.cond ’, mar [maːʁ] ‘sea’, particularly in verbal inflection morphemes for person and in oral narratives. In the Vivaro-Alpin dialect of Chiomonte, Italy, when the stressed vowel in the sequence ˈVCV is long, the following consonant is singleton; when the stressed vowel is short, the consonant is geminate (Sibille 2012: 2234–2235), e.g., pasta [ˈpaːtO] ‘pastry’, pata [ˈpatːO] ‘paw’.
4.2.2.3 Unstressed Oral Vowels Post-tonic /i/ occurs primarily in 1st person singular verbal morphology (Wheeler 1988: 247), e.g., canti /ˈkanti/ ‘sing1sg ; it is frequently realised as [-e] in alverno-méditerranéen dialects and as [-u] in Vivaro-Alpin. Posttonic /i/ also occurs in a large number of learned words, borrowed from Latin (Ford 1921: 66), e.g., memòri /meˈmOɾi/ ‘collective memory’. In unstressed syllables, both pre- and post-tonic, the phonemic contrast between /e/ and /E/ is neutralised to /e/ (Bec 1973: 31), e.g., pèl /pEl/ ‘skin’, pelar /peˈla/ ‘skininf ’. In Vivaro-Alpin, pre-tonic /e/ is realised as [e] (Sibille 2012: 2235) or [œ] (Field 1978: 126), frequently leading to syncope in the dialect of Chiomonte, Italy, e.g., semana [seˈmanO] ‘week’ > [ˈsmanO]. In albonnais Vivaro-Alpin, [E] occurs in post-tonic syllables as an allophone of /e/ before [h] < /s/ (Quint 1999: 6), e.g., ères [EɾEh] ‘be2sg.impf . In rhodanien Provençal, post-tonic /e/ is often realised as [e], particularly in the Vaucluse sub-dialect (Coustenoble 1945: 24); in the same dialect, final /e/ is deleted when followed by a vowel (Coustenoble 1945: 122), e.g., un brave ome [øŋbʁaːˈvOːme] ‘a courageousm.sg man’. In the parlar negre of Gascon, post-tonic /e/ is realised [œ] (Mooney 2014: 348), e.g., càder [ˈkaðœ] ‘fallinf ’. Post-tonic /u/ has a relatively restricted distribution, occurring primarily in third person plural verb forms (Wheeler 1988: 247), e.g., lièjon /ˈljEd͡ʒun/ ‘read3pl ’.
94
D. Mooney
Pre-tonic /a/ has the majority variant [a], but in dialects including Lemosin, northern Lengadocian and Vivaro-Alpin, this is frequently realised as [O] (Quint 1999: 9; Oliviéri and Sauzet 2016: 324) (see Sect. 4.1.2), e.g., parlam [pOrˈlan] ‘speak1pl ’. The post-tonic /a/ phoneme of Old Occitan has, in the modern language, evolved to /-O/; note that some analyses choose to treat final post-tonic /-O/ as distinct from stressed /O/, preferring the notation /a/ (Oliviéri and Sauzet 2016: 324), /o/ (Coustenoble 1945: 33), or /e/ (Bec 1973: 35), because of the high level of variability in the phonetic realisation of the post-tonic vowel, though there is no phonological basis (e.g., the existence of minimal pairs) to do so (Coustenoble 1945: 33). /O/ cannot occur in pre-tonic syllables, where it alternates with [u], e.g., pòrc [pOɾk] ‘pig’, Gasc. porqueria [puɾkeˈɾijO] ‘mess’ or ‘pig shed’, though it does occur in some borrowings from French (Bec 1973: 33), e.g., auto /OˈtO/ ‘car’. The realisation of post-tonic /O/ varies geographically, with the following variants: [O] is found in the majority of Occitan dialects, e.g., pòrta [ˈpOɾtO] ‘door’; [a] or [ɑ] is found in lodévois and montpélliérain Lengadocian (Bec 1973: 34), nissart Provençal (Oliviéri and Sauzet 2016: 324), in the Gascon of Pontacq, Asson (Bouzet 1928: 9), and the Biros valley (Castet 1895: 15), e.g., [ˈpOɾta]; [u] is found in Couserans Gascon (Castet 1895: 15) and in some sub-dialects of the Médoc and the Ariège, as well as in vaudois and the variety spoken in the southern Alpilles (Bec 1973: 34), e.g., [ˈpOɾtu]; [e] is found in rhodanien Provençal (Ford 1921: 24; Coustenoble 1945: 34) e.g., [ˈpOɾte]; [œ] is found in parlar negre Gascon (Mooney 2014: 348), e.g., [ˈpOɾtœ]. In the parlar negre, therefore, the contrast between word-final post-tonic /-e/ and /O/ is neutralised as both are realised as [-œ], e.g., càder [ˈkaðœ] ‘fallinf ’ ~ cada [ˈkaðœ] ‘every’. Other dialects present a more complex array of variants in final position. In Lemosin, [O] occurs in singular feminine nouns and 3rd person singular verbal morphology, while [aː] occurs in plurals and 2nd person singular verbal endings (Bec 1973: 35), e.g. filha [ˈfijO] ‘daughtersg ’, filhas [ˈfijaː] ‘daughterpl ’; canta [ˈt͡ʃantO] ‘sing3sg ’, cantas [ˈt͡ʃantaː] ‘sing2sg ’. The same pattern can be observed in Vivaro-Alpin but, additionally, stressed /a/ can also be realised as [O] in singular nouns when it finds itself in an open syllable as a result of word-final consonant
4 Phonetic and Phonological Systems
95
deletion (Field 1978: 123), e.g., prat [pɾO] ‘meadowsg ’; this [O] alternates with [aː], which marks feminine plurality, as in Lemosin, e.g., prats [pɾaː] ‘meadowpl ’. In Lunel Lengadocian, post-tonic /O/ is realised as [O] after phrasal stress and otherwise as [a] (Oliviéri and Sauzet 2016: 324), e.g., la taula lònga [laˌtawlaˈlOŋɡO] ‘the longf.sg table’, but la lònga taula [laˌlOŋɡaˈtawlO]. In Bessan Lengdocian, post-tonic /O/ is realised as [O] in closed syllables and as [a] in open syllables (Roque-Ferrier 1878, cited in Oliviéri and Sauzet 2016: 324), e.g., las taulas lòngas [lasˌtawlOsˈlOŋɡOs] ‘the longf.pl tablepl ’, but la taula lònga [laˌtawlaˈlOŋɡa] ‘the longf.sg tablesg ’.
4.2.2.4 Vowel Nasalisation Standard Occitan does not have any phonemic nasal vowels; Latin vowel + nasal sequences are generally preserved as a phonemic oral vowel followed by a nasal consonant, e.g., *femu(m) > fum /fym/ ‘smoke’. In southern Occitan, there are no phonological nasal vowels (Oliviéri and Sauzet 2016: 326), with Lengadocian being particularly conservative in this respect (Sampson 1999: 147). Phonemic oral vowels do, however, have nasalised allophonic variants when they appear before nasal consonants (Sampson 1999: 139). The degree of nasalisation of vowels in this context is attested to be at most partial and ‘only really perceptible in the latter part of the vowel’ (Sampson 1999: 147; Oliviéri and Sauzet 2016: 326). In Provençal, both Coustenoble (1945: 66) and Ford (1921: 22) note that nasalisation is only allophonic, but there is more evidence for the weakening of nasal consonant codas than in Lengadocian, particularly following /a/, leaving behind an unconditioned nasal vowel, [ã] (Sampson 1999: 150), e.g., grand /ɡɾan/ [ɡɾã] ‘bigm.sg ’. In fact, in southern Occitan, final /-n/ frequently drops, but typically leaves an oral vowel in an open syllable (Field 1978: 101; Sampson 1999: 151), e.g., panem > /pan/, Leng. [pa] ‘bread’. In some dialects of nord-occitan, such as Lemosin, phonemically nasal vowels have developed in open syllables as a result of the weakening and loss of nasal consonant codas (Sampson 1999: 147; Oliviéri and Sauzet 2016: 151); this is frequently accompanied by vowel opening
96
D. Mooney
such that /iN yN uN/ are realised as [˜e œ ˜ õ] respectively (Sampson 1999: 151). In Lemosin, nasal vowels are ‘systematically long’ and the absence or presence of nasality is involved in morphophonological alternations, e.g., chantan [ˈsãːtãː] ‘sing3pl ’, chanta [ˈsãːtO] ‘sing3sg ’ (Oliviéri and Sauzet 2016: 323). In Auvernhat, Sampson (1999: 152) notes that non-open vowels are more resistant to nasalisation than open vowels. In Vivaro-Alpin, retention of nasal consonant codas is the norm (Sampson 1999: 150; Quint 1999: 6) and as such there are no phonemically nasal vowels in this variety. In the Vivaro-Alpin sub-dialect of Chiomonte, Italy, nasality is preserved in post-tonic position in words with a Latin etymon ending in -ine(m), -ene(m), -inu(m) (Sibille 2012: 2235), e.g., homine(m) > ome [ˈOmã] ‘man’; iuvene(m) > jove [ˈʒuvã] ‘youngm.sg ’; frassinu(m) > fraisse [ˈfrEjsã] ‘ash tree’. In Gascon, the dropping of Latin intervocalic -n- has led to the existence of a phonemic nasal vowel system, /˜ı y˜ u˜ e˜ ã/, in some sub-dialects, such as Béarnais (Mooney 2014: 346), e.g., vinum > vin /b˜ı/ ‘wine’; cata-unum > cadun /kaˈd˜y/ ‘everyonem.sg ’; masione(m) > maison /ˈmajz˜u/ ‘house’, plenu(m) > plen /pl˜e/ ‘fullm.sg ’; pane(m) > pan /pã/ ‘bread’ (Sampson 1999: 154). In other varieties of Gascon, however, these vowels have denasalised, e.g., vin [bi], while in still others, there has been restructuring as a nasal vowel + [-ŋ] (Sampson 1999: 154), e.g., vin [biŋ].
4.2.3 Glides and Diphthongs Standard Occitan has three glide phonemes, /j w ɥ/. Minimal pairs that show /j/ to be in contrast with /i/ are readily available (Wheeler 1988: 250), e.g., trairà /trajˈra/ ‘pull out3sg.fut ’ vs. traïra /traiˈra/ ‘betray3sg.fut ’. In the Gascon of west-central Bearn and the Pyrenean valleys and, to a lesser extent, the central Pau region, /j/ is replaced by [ʒ] in all positions (wordinitially, intervocalically and postconsonantally), e.g., jo [ʒu] ‘me’. The phonemic status of /w/ is not as clear (Wheeler 1988: 250); in the Occitan diasystem, it occurs rarely in syllable onsets, and when it appears in codas, it is usually the result of vocalisation of /l/, /ʎ/, or
4 Phonetic and Phonological Systems
97
/v/. In Gascon, initial /kw/ and /ɡw/ are retained, whereas these are simplified to /k/ and /ɡ/ in other dialects, e.g., quatre /katɾe/ ‘four’ > Gasc. [ˈkwate]. The /ɥ/ phoneme is absent in Gascon; in other dialects, it never occurs in final position (Bec 1973: 49). Wheeler (1988: 250) notes that the phonemic status of /ɥ/ is ‘very dubious’ and the sequences such as /yj/ and /ɥi/ are not consistently distinguished. In some varieties, such as Rouergat Lengadocian and Vellave Auvernhat (Field 1978: 108; 116–117), glides in stressed syllables alternate with corresponding vowels when stress is shifted in inflected forms, e.g., dona [ˈdwOnO] ‘give3sg ’, donar [duˈna] ‘giveinf ’; dobrir [duɾˈbi] ‘openinf ’; duèrb [dɥEɾ] ‘open3sg ’. In Vivaro-Alpin, final glides in singular nouns, e.g., pomièr [pumEj] ‘apple treesg ’, often alternate morphologically with rhotics to mark pluralisation, e.g., pomièrs [pumEr] ‘apple treepl ’ (Field 1978: 123). In the diasystem, rising and falling diphthongs are formed with a vowel preceded by /j/, /w/, or /ɥ/, or with a vowel followed by /j/ or /w/. The following rising diphthongs are possible: /je/, e.g., sobrietat /subɾjeˈtat/ ‘sobriety’; /jE/, e.g., ièr /jE/ ‘yesterday’; /ja/, e.g., embestiar /embesˈtja/ ‘annoyinf ’; /ju/, e.g., violon /vjuˈlun/ ‘violin’; / jO/, e.g., iòga /jOˈɡa/ ‘yoga’; /ji/, e.g., saunegi /sawˈneji/ ‘dream1sg ’; /jy/, e.g., jumpar /jymˈpa/ ‘rock, swinginf ’; /ɥe/, e.g., cuer /kɥer/ ‘leather’; /ɥi/, e.g., aduire /aˈdɥiɾe/ ‘bringinf ’; /ɥO/, e.g., fuòc /fɥOk/ ‘fire’; /ɥu/, e.g., cuol /kɥul/ ‘arse’. The diphthong /jE/ is used in the standard Occitan suffixes -ièr /-ièra [-jE]/[-jEɾO] which may be realised as [-E]/[-EɾO] in Gascon and Auvernhat or as [-Ej]/[-EjɾO] in northern sub-dialects of Gascon (Oliviéri and Sauzet 2016: 323); in Auvergne, this /jE/ > [E] realisation is also attested in stems, e.g., mièlhs [mEj] ‘better’. /jy/ is realised as [jø] in rhodanien Provençal (Coustenoble 1945: 58). Even in dialects that retain /ɥ/, Bec (1973: 54–55) notes that the majority of realisations of /ɥO/ and /ɥu/ are as [jO] and [ju] respectively, across dialects, e.g., fuòc [fjOk] ‘fire’; cuol [kjul] ‘arse’, but Quint (1999: 10) documents fùoc [fɥo] ‘fire’ in VivaroAlpin. In dialects such as Gascon, where /ɥ/ is absent, the following rising diphthongs are attested: /we/, e.g., uelh [weʎ] ‘eye’; /wE/, e.g., cuèlher [ˈkwEʎE] ‘fetchinf ’; /wa/, e.g., quatre [ˈkwate] ‘four’, /wi/, e.g., guidar [ɡwiˈða] ‘guideinf ’; /wO/, e.g., coïcar [kwiˈka] ‘whineinf ’, since
98
D. Mooney
/ɥ/ is lost. The diphthongs [wO], [wa], and [wE] can also arise from the diphthongisation of /O/, e.g., pòrta [ˈpwOɾtO] ‘door’. Possible falling diphthongs are (Bec 1973: 54–55): /ej/, e.g., veire /ˈvejɾe/ ‘seeinf ’; /Ej/, e.g., pèira /ˈpEjɾO/ ‘stone’; /aj/, e.g., maire / ˈmajɾe/ ‘mother’; /uj/, e.g., coire /ˈkujɾe/ ‘copper’; /Oj/, e.g., còire /ˈkOjɾe/ ‘cookinf ’; /ew/, e.g., beu /bew/ ‘drink3sg ’; /Ew/, e.g. nèu /nEw/ ‘snow’; /aw/, e.g., Gasc. nau /naw/ ‘nine’; /iw/, e.g., viu /biw/ ‘livelym.sg ’; /Ow/, nòu /nOw/ ‘newm.sg ’. In rhodanien Provençal, /Ej/ is variably realised as [i] (Coustenoble 1945: 44), e.g., èime [ˈiːme] ‘spirit’, /Ew/ is variably realised as [ew] (Coustenoble 1945: 37), e.g., solèlh [sulew] ‘sun’, /aw/ as [ɑw] or [Ow] in stressed syllables and as [ow] or [u] in unstressed syllables (Coustenoble 1945: 50), e.g., faudau [fuˈdɑw] ‘apron’, and / Ow/ is variably realised as [Ow], [ow], or [u] (Coustenoble 1945: 43; 50). However, Coustenoble also posits phonemic /ow/ for rhodanien, which contrasts with /Ow/ and /u/, e.g., pòutra [ˈpOwtɾO] ‘young female donkey’ ~ pautra [ˈpowtɾO] ‘burgundy’; sosleu [souˈlEw] ‘raising’ ~ solèu [suˈlEw] ‘sun’. The /aj/ diphthong is frequently realised as [Ej] in aquitanopyrénéen dialects; [ej] is attested in Vivaro-Alpin (Quint 1999: 12). /aw/ preserves the same falling Latin dipthong (Wheeler 1988: 247), but can also arise from vocalisation, e.g., sal [saw] ‘salt’; /aw/ is variably realised as [ow] or [Ow] in unstressed position in Vivaro-Alpin (Quint 1999: 12), e.g., se pausar [ʃepowˈʒe] ‘stop oneselfinf ’. /Ow/ is realised variably as [Ew] in Auvernhat (Field 1978: 118), e.g., nòu [nEw], and as [aw] in Gascon, e.g., [naw]. To the series of falling diphthongs with /w/, Bec (1973: 54) adds /yw/ for Gascon, e.g., cuol [kyw] ‘arse’. Triphthongs are formed with both a pre- and post-vocalic glide. The following combinations are possible (Bec 1973: 54–55): /ɥOw/, e.g., uòu /ɥOw/ ‘egg’; /ɥej/, e.g., nueit /nɥejt/ ‘night’; /ɥOj/, e.g., cuòissa /ˈkɥOjsO/ ‘thigh’; /jej/, e.g., fieira /fjejɾO/ ‘fairf.sg ’; /jaw/, e.g., mial [mjaw] ‘honey’ in nord-occitan. Bec (1973: 55) also notes that [jOw] occurs as a variant of /ɥOw/. For dialects such as Gascon, without /ɥ/, /ɥ/ is substituted by [w] or sometimes [j]. Additionally, for rhodanien Provençal, Coustenoble (1945: 62) posits /jew/ and /jEw/ as additional phonemes, e.g., lo sieu /luˈsjew/ ‘yoursm.sg ’, lo fièu /luˈfjEw/ ‘the fief ’, [jOw] as a variant of /ɥOw/, and [jEj] or [jøj] as variants of /ɥej/.
4 Phonetic and Phonological Systems
99
In this chapter, detailed phonetic and phonological characteristics of Occitan and its varieties have been presented, with focus placed both on the similarities between phonological systems diatopically and on the wealth of observable variation in this language. Gascon has been shown to stand out from other dialects as highly divergent in comparison to other Occitan dialects and within sub-dialects of Gascon, relatively high levels of geographical variation can be observed. The remainder of this book will focus on the Gascon spoken in Béarn, or the principal surviving dialect of Gascon, Béarnais and, in particular, on its sound system, with the aim of challenging or confirming traditional accounts of its phonetic and phonological structure as outlined here.
References Bec, Pierre. 1963. La langue occitane. Paris: Presses Universitaires de France. Bec, Pierre. 1973. Manuel pratique d’occitan moderne. Paris: Picard. Bouzet, Jean. 1928. Manuel de grammaire béarnaise. Pau: Marrimpouey Jeune. Cardaillac Kelly, Reine. 1973. A descriptive analysis of Gascon. The Hague and Paris: Mouton. Castet, Abbé Guillaume. 1895. Études grammaticales sur le dialecte gascon du Couserans. Foix: Gadrat aîné. Coustenoble, Hélène N. 1945. La phonétique du provençal moderne en terre d’Arles. Hertford: Austin. Field, Thomas T. 1978. Modern Occitan phonology and dialectology. PhD dissertation, Cornell University. Ford, Harry E. 1921. Modern Provençal phonology and morphology: Studied in the language of Frédéric Mistral . New York: Columbia University Press. Grosclaude, Michel. 1986. La langue béarnaise et son histoire: étude sur l’évolution de l’occitan du Béarn. Orthez: Per Noste. Molyneux, Roger-Gordon. 2002. Grammar and vocabulary of the language of Béarn: for beginners (Abridged and translated from the works of Vastin Lespy). Monein: Éditions Pyrémonde. Mooney, Damien. 2014. Illustrations of the IPA: Béarnais (Gascon). Journal of the International Phonetic Association 44: 343–350. Mooney, Damien. 2018. Quantitative approaches for modelling variation and change: A case study of sociophonetic data from Occitan. In Manual of
100
D. Mooney
Romance Sociolinguistics, ed. Wendy Bennett and Janice Carruthers, 59–87. Berlin and New York: De Gruyter. Moreux, Bernard, and Jean-Marie Puyau. 2002. Dictionnaire français-béarnais. Monein: Éditions Pyrémonde. Müller, Daniella, and Sydney Martin. 2012. A preliminary acoustic study of the Occitan vowel system. In Études de linguistique gallo-romane, ed. Mario Barra-Jover, Guylaine Brun-Trigaud, Jean-Philippe. Dalbera, Patrick Sauzet, and Tobias Scheer, 149–159. Vincennes: Presses Universitaires de Vincennes. Oliviéri, Michèle, and Patrick Sauzet. 2016. Southern Gallo-Romance (Occitan). In The Oxford guide to the Romance languages, ed. Adam Ledgeway and Martin Maiden, 319–349. Oxford: Oxford University Press. Quint, Nicolas. 1999. Le parler occitan ardéchois d’Albon: Canton de SaintPierreville, Ardèche: Description d’un parler alpin vivaro-vellave du boutiérot moyen. Paris: L’Harmattan. Roque-Ferrier, Alphonse. 1878. Un fragment de poëme en langage de Bessan (Hérault). RLaR 14: 24–31. Ronjat, Jules. 1937. Grammaire historique des parlers provençaux modernes III . Montpellier: Société des Langues Romanes. Salvat, Joseph. 1973. Grammaire occitane des parlers languedociens. Toulouse: Privat. Sampson, Rodney. 1999. Nasal vowel evolution in Romance. Oxford: Oxford University Press. Sibille, Jean. 2012. Le parler occitan de Chiomonte (Italie): situation linguistique et sociolinguistique. In 3e congrès mondial de linguistique française, ed. Franck Neveu, Valelia Muni Toke, Peter Blumethal, Thomas Klinger, Pierluigi Ligas, Sophie Prévost, and Sandra Teston-Bonnard, 2231–2248, EDP Sciences, Institut de linguistique française. Stéfanini, Jean. 1969. Un provençaliste marseillais: l’abbé Féraud (1725–1807). Aix-en-Provence: Ophrys. Wheeler, Max W. 1988. Occitan. In The Romance languages, ed. Martin Harris and Nigel Vincent, 246–278. London and Sydney: Croom Helm.
5 Methodological Considerations
This chapter outlines the data collection, analysis, and processing undertaken for the language death and dialect death studies. The language death study examines the speech of thirty native Gascon speakers in Béarn, analysing evidence for L2-to-L1 phonetic and phonological transfer from French, as well as evidence for dialect mixing between subvarieties of Gascon within the Béarn region. The dialect death study examines the speech of ten new Gascon speakers, who have acquired their language in the educational context. The dialect death study provides a comparison of the speech of these new speakers with that of the native speakers in the language death study, while also investigating the retention of localised phonetic forms and, crucially, evidence for innovative linguistic change from below. The chapter will begin by outlining informant sampling for both studies, before discussing the linguistic variables analysed and detailing the method of data elicitation. The acoustic and auditory analyses performed on the data, as well as the statistical techniques employed, will then be presented.
© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 D. Mooney, Language and Dialect Death, https://doi.org/10.1007/978-3-031-51101-1_5
101
102
5.1
D. Mooney
Sampling Method
5.1.1 Fieldwork Sites The variety of Gascon spoken in Béarn (Béarnais) can, linguistically speaking, be divided into five sub-dialectal areas, with geographically based phonetic and phonological variation attested between them. These areas are: (i) the Pyrenean valleys; (ii) the north; (iii) Vic-Bilh in the north east; (iv) the central plain; (v) the west, bordering the Basque country. The fieldwork sites selected for the language death study therefore aimed to recruit informants from the following towns, which fall within the sub-dialectal areas outlined above (see Fig. 5.1): (i) Laruns in the Vallée d’Ossau; (ii) Arzacq-Arraziquet in the north; (iii) Lembeye in the north east; (iv) Nay in the central plain; (v) Oloron, Navarrenx, and Orthez in the west. For the dialect death study, new speakers were recruited from a variety of areas within the region, with the majority (9/ 10) residing in Pau, the region’s capital city. The commonly observed phenomenon, whereby new speakers form part of the ‘urban intelligensia’, was also borne out in this study, with the majority of new-speaker participants working in the Occitan educational or media spheres.
5.1.2 Informant Recruitment The sample structure for the language death study, stratified by biological sex and sub-dialectal area, with three speakers per cell, is outlined in Table 5.1. For the dialect contact study, ten younger new speakers were recruited, again stratified by biological sex (see Table 5.2). Each of the thirty speakers from the language death study was also included in the sample for the dialect death study, in order to explore the effect of speaker type (native or new) on language use and, given that new speakers are categorically younger than the native speakers, as a potential exploration of change in apparent time. Milroy and Gordon note that in minority language situations, ‘random selection from a sample frame such as an electoral register will
5 Methodological Considerations
103
Fig. 5.1 Sub-dialect fieldwork sites in the language death study
be both inadequate and inefficient’ (2003: 27) because random sampling methods will often only locate monolingual speakers of the dominant language. It is thus necessary in such situations to explore other sampling methods. Informants for the language death study were sourced using a ‘quota sampling’ technique. Quota sampling, or ‘judgment sampling’ involves ‘a process of selection based on familiarity with the community in question, targeting sectors best representing the overall community’ (Taylor 1996: 29) whereby ‘the researcher identifies in advance the types of speakers to be studied and then seeks out a quota of speakers who fit the specified categories’ (Milroy and Gordon 2003: 30). In order to fill the cells for the quota/judgement sample, I adopted a
104
D. Mooney
Table 5.1 Native speaker participants in the language death study Speaker I.D
Sex
Birth year
Place
B C D E F G H I L M N O Q R S T U V W X Y AA AB AC AD AG AH AE AF AI
M M M F F F M M M F F F M M M F F F M M M F F F M M M F F F
1949 1964 1964 1932 1935 1926 1947 1941 1933 1939 1945 1938 1950 1948 1935 1937 1935 1946 1936 1953 1924 1934 1939 1943 1946 1946 1935 1939 1951 1948
Laruns Laruns Laruns Laruns Laruns Laruns Arzacq Arzacq Arzacq Arzacq Arzacq Arzacq Lembeye Lembeye Lembeye Lembeye Lembeye Lembeye Nay Nay Nay Nay Nay Nay West West West West West West
‘friend-of-a-friend’ technique, also known as a ‘snowball technique’ or ‘network sampling’. The technique involves entering the community and contacting potential informants via an intermediary and/or utilises the social networks of participants in the study to recruit potential new participants (Milroy and Gordon 2003: 32). Potential informants for the language death study were initially proposed by contacts at the L’Institut Béarnais et Gascon (IBG), Pau, and these informants were then, in turn, asked to recruit friends and family members who met the specified inclusion criteria. The new speakers in the dialect death study were
5 Methodological Considerations
105
Table 5.2 New-speaker participants in the dialect death study Speaker I.D
Sex
Birth year
Place
AJ AN AP AS AU AK AL AM AQ AT
M M M M M F F F F F
1979 1990 1972 1983 1979 1978 1996 1995 1989 1986
Pau Pau Pau West Pau Pau Pau Pau Pau Pau
recruited via Occitan Facebook groups online and via two interviews that the researcher participated in on the Occitan-language radio station Ràdio País, based in Pau. The dialect death study participants were either Calandreta-educated, worked in the area of Occitan education, or produced written or audiovisual media in Occitan; unlike the participants in the language death study, the dialect death study participants were all literate in Occitan, using the classical orthography.
5.2
Linguistic Variables
The linguistic variables examined in the language and dialect death studies included both continuous and categorical variables. The continuous variables analysed were the seven stressed oral vowels, /i/, /y/, /e/, /E/, /a/, /u/, /O/, and two unstressed post-tonic vowels /-e/ and /-O/. All stressed oral vowels were examined in interconsonantal (C_C) position; post-tonic vowels were examined in word-final (C_#) position. The former are assumed, in the literature, to be stable across varieties of Béarnais, Gascon, and Occitan more generally though, as we have seen, the quality of the /-e/ and /-O/ has been shown to vary widely over geographical space, even within the relatively compact region of Béarn (see Sect. 4.2.2.3). These continuous variables were measured instrumentally (see Sect. 5.4.1) and as such, have no discrete identifiable variants but, in order to provide a meaningful comparison with the
106
D. Mooney
existing literature on quality of the post-tonic vowels, the interpretation of the results will acknowledge their traditional treatment as categorical variables, with /-e/ having variants [-e] and [-œ], and /-O/ having variants [-O] and [-œ]. The Gascon spoken in Béarn, Béarnais, can be seen as a coherent variety with sub-dialects exhibiting common structural features and geographically localised dialect-specific features. In addition to the nine continuous variables discussed above, the language and dialect studies considered a number of categorical variables that were expected to exhibit contact-induced transfer from French, sub-dialectal variation within Béarn, or both. The examination of the (r) rhotic variable considered the distribution of the traditional historically appropriate variants [r] and [ɾ] in a range of syllable types (V_V; #_V; #C_V; V_#; V_C#), as well as investigating the presence of contact-induced uvular variants that were transferred from French: [ʁ], [ʁ̞], or [ʀ]. While Gascon traditionally realises /ʎ/ as [ʎ] in all positions, previous studies (e.g., Mooney and Hawkey 2019) have shown that a contact-induced variant [j] is also present in the /ʎ/ variable’s distribution. It appears, broadly speaking, that the [ʎ] variant is favoured where the phonemic cognate in French is /l/, a lateral, and that the [j] variant is favoured when the phonemic cognate in French is /j/. This means that, in Gascon, [ʎ] has been shown to be statistically more likely to occur in word-initial (#_V) position and in final clusters with /-s/ (V_[s]#), and that [j] is more likely to occur in intervocalic (V_V) and word-final (V_#) position. In the language and dialect death studies, the /ʎ/ variable was therefore examined in these syllabic contexts. Occitan word-final /l/ is commonly replaced by /t͡ç/ in Gascon, though there is inter-dialectal variation within Béarn in terms of the phonetic realisation of this phoneme (see Sect. 4.2.1). As such, the /t͡ç/ variable may, in the context of Béarn be realised as [t͡ç], [t͡ʃ] or [t], or additionally as the standard form [l]. Intervocalic Occitan /t͡ʃ/ < Latin –tc- is commonly replaced in Gascon by /d͡ʝ/, with variants [d͡ʝ] and [t͡ʃ]. These two localised affricate variables were also included in both the language and dialect death studies. In intervocalic position and in clusters with [r] and [l], the voiced plosives /b/, /d/, and /ɡ/
5 Methodological Considerations
107
are lenited in Béarn to fricatives or approximants at the same place of articulation: [β], [ð], and [G]. These allophonic variants are not common to all varieties of Occitan and, indeed, do not occur in French; evidence for their absence may therefore indicated contact-induced change. For /b/ and /ɡ/, the variants [b] or [β] and [ɡ] or [G] are respectively identifiable. /d/ may be realised as [d], [ð] or [z] intervocalically, with the latter constituting a geographically localised variant in the Vic-Bilh area of Béarn (speakers from Lembeye). The /ʒ/ variable may be realised in the Gascon of Béarn as [j] or [ʒ] over geographical space (see Sect. 4.2.1), with [ʒ] common in west-central Béarn and the Pyrenean valleys, while word-final /-s/ may be realised as [-s] or [-j] in the words james, mes, and qu’es (see Sect. 4.2.3), and so both variables were included in the analysis in order to investigate dialect mixing and levelling. Finally the language and dialect death studies considered the grammatical (gender) variable, focusing on sixteen lexical items where the grammatical gender (masculine or feminine) differed between French and Gascon. In sum, the language and dialect death studies considered nine continuous variables, /i/, /y/, /e/, /E/, /a/, /u/, /O/, /-e/, and /-O/, and ten categorical variables, (r), /ʎ/, /t͡ç/, /d͡ʝ/, /b/, /d/, /ɡ/, /ʒ/, /-s/, and (gender). In the language death study, the investigation of contactinduced change will focus on the points at Gascon diverges structurally from French, examining acoustically the vocalic inventory in its entirety /i y e E a O u/, focusing particularly on the maintenance of mid-vowel contrasts, as well as examining categorical consonantal variables: loss of apical rhotics [r, ɾ] in favour of French [ʁ], loss of palatal lateral [ʎ] in favour of French [j], and despirantisation of voiced plosive /b d ɡ/ allophones [β ð G]. Secondly, the language death study will consider variables whose realisation traditionally differs according to geographical location in Béarn (e.g., /-e/, /-O/, /t͡ç/, /d͡ʝ/, /d/, /ʒ/, /-s/) in order to identify evidence for levelling and dialect mixing, as forms from the central area of the region may be adopted to replace peripheral forms. The dialect death study will consider the same variables as the language death study, comparing the speech of new speakers to older native speakers, focusing on the vocalic inventory, rhotics, the palatal lateral, and the
108
D. Mooney
voiced plosives. The dialect death study will also examine the loss or retention of traditionally localised Béarnais phonological features in the speech of the older Gascon speakers and younger Occitan néo-locuteurs and consider differences between speaker types in the maintenance of historically appropriate gender marking on nouns.
5.3
Corpus Construction
Participants in both studies were recorded individually using a Marantz PMD661MKII Solid State Sound Recorder. The recordings were initially stored on a Secure Digital (SD) 4 GB Memory Card before being transferred to a Macbook portable computer and backed-up to a password protected Dropbox folder. All sound files were recorded and stored in wav format on two channels (stereo) with one channel for the interviewer and the other for the participant. The sampling rate used for the recordings of this corpus was 44.1 kHz (uncompressed linear PCM recording), using a 16-bit PCM sample size. Recording levels were routinely checked throughout the recordings to ensure that no high (or ‘over’) amplitudes in the speech signal were clipped. Two battery-powered AT803 condenser microphones, with a frequency response of 30–20,000 Hz, were connected by 3-pin XLR input cables to the Marantz Recorder. Condenser microphones offer greater response and sensitivity and a louder output (Cieri 2011: 28). The AT803 microphone is designed to be worn as a lavalier which provides good quality recordings through a combination of near and unobtrusive placement. The microphone was positioned approximately six inches below the speaker’s chin using a clothing clip. Care was taken to anticipate movements that may cause the microphone to rub against or be covered by clothing and to not place the microphone in the shadow of the speaker’s chin. Every effort was made to choose a recording setting which avoided the possibility of recording echo and/or extraneous noise such as, for example, holding interviews in smaller rooms with soft furnishings, in the participants’ homes. In both studies, participants performed a translation task where they were asked, in Gascon, to translate a list of 155 French words, presented
5 Methodological Considerations
109
on an iPad, into Gascon. These words contained instances of the phonological variables outlined in Sect. 4.2 and the systematic nature of the study allowed the same variables (and lexical items) be examined in the speech of all informants. The use of the wordlist translation task to elicit data for Gascon phonological variables is open to criticism: informants are more likely to engage in transfer when translating from one language to another than when producing casual speech. Since all informants were translating, however, they were confronted with the same potential stimulus for transfer. The words were presented in a pseudo-randomised order which was the same for all speakers. The entire wordlist was entered into a spreadsheet in Microsoft Excel and was randomised using the “=RAND()” function which aligned the wordlist to a column of random numbers and then sorted it into numerical order, thus randomising the order of the stimuli. Care was taken to ensure that no two words containing instances of the same variable occurred next to one another. Table 5.3 outlines the variables by phonological (syllabic) context contained in the wordlist; the entire wordlist can be consulted in Appendix 1. In acoustic experiments carried out in laboratory conditions, it is common practice to have informants read a wordlist three times, in three random orders ‘to counteract presentation order effects such as list intonation which affects word length and may influence individual segments’ (Di Paolo and Yaeger-Dror 2011: 15). This was not possible in the translation task as the informants were not reading a wordlist, but providing translations, and to ask them to repeat the words would have been unnatural and counter-productive, forcing them pay more attention to their articulations, thus eliciting less vernacular-like tokens. Between the translations of each word, there was often meta-commentary on the lexical item produced and there was always a short delay as the next slide on the iPad was shown to them by the interviewer. Because of the use of the iPad, the informants were not aware of the length of the list, thus minimising potential listing effects, nor were they reading aloud. No fillers were inserted into the randomised list as it contained nineteen different target variables which functioned as distractors for each other: ‘the tokens of one variable may serve as fillers for the others, in the sense that they distract the participant from focusing on any
110
D. Mooney
Table 5.3 Wordlist token counts by syllabic context Variable
Context
Tokens
(i), (y), (e), (E), (a), (u), (O) (-e), (-O) (r)
C_C C_# V_V #_V C_V V_# _C# [n]_ #_V V_V V_# V_[s]# V_# V_V [R]_V [z]_V V_[a] V_[y] V_[u] V_[O] V_V [R]_V V_[a] V_[y] V_[u] V_[O] V_V [R]_V [l]_V V_[a] V_[u] V_[y] V_[O] V_V #_V V_V V_ # article (m.) article (f.)
5 5 5 5 5 5 5 1 4 5 5 5 5 5 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 1 5 5 3 8 8
(ʎ)
(t͡ç) (d͡ʝ) (b)
(d)
(ɡ)
(ʒ) (s) (gender)
5 Methodological Considerations
111
one variable of interest’ (Di Paolo and Yaeger-Dror 2011: 14–15). The words chosen for the wordlist translation task were French words with an expected target translation in Gascon. For example, if an informant was presented with the French word ‘chambre’ /ʃɑ˜ bʁ/, it was expected that they would provide the translation crampa /ˈkrampO/. The target words were drawn largely from the Atlas Linguistique et Ethnographique de la Gascogne (ALEG) (Séguy and Allières 1954–1973) and included words for which the geographical sub-dialectal variation, outlined in Sect. 4.2, was attested (see Appendix 1).
5.4
Data Analysis
The wordlist for each of the forty speakers was segmented in Praat version 5.2.21 (Boersma 2001; Boersma and Weenink 2012) using accompanying TextGrid files. When the translated Béarnais word provided approximated the ‘target’ translation expected for a given French stimulus (see Appendix 1), the target variable embedded in that word was included in the analysis. Lexical items that did not conform to the target word and thus did not contain the target variable were excluded from the analysis. For example, for the stimulus ‘le liseron’, the Gascon target word expected was la lheta, containing the /ʎ/ and (gender) variables. If a speaker, however, provided the Gascon word lo liseron, also a French borrowing, this lexical item was excluded from the analysis.
5.4.1 Acoustic Analysis All continuous variables were analysed acoustically in Praat: the seven stressed oral vowels, /i/, /y/, /e/, /E/, /a/, /u/, /O/, and the two unstressed post-tonic vowels /-e/ and /-O/. Foulkes et al. note that an auditory analysis of vowels is relatively course-grained, imposing a pre-defined set of broad phonetic categories on the data which ‘may miss finegrained detail that is not easily perceivable, but which may nevertheless be identifiable instrumentally’ (2011: 60). It is for this reason that the present study applies primarily instrumental techniques for
112
D. Mooney
the analysis of vocalic variables. The acoustic analysis employed here involves the continuous measurement of characteristic vocalic features (formant frequencies). A crucial aspect of acoustic analysis is that systematic procedures for instrumental measurement be established and applied consistently across the data set to produce meaningful results. What follows is a detailed description of these procedures and their application for the coding of all vocalic dependent variables in the language death and dialect death studies. Segment duration was measured instrumentally for all vocalic variables with the primary aim of identifying the midpoint of each vowel. Establishing the onset and offset of a segment has received much attention in the literature. Di Paolo et al. note that duration is usually measured from vowel onset to vowel offset, with the transition being considered part of the vowel, rather than the consonant, while some measurement routines incorporate glides and liquid durations into the vowel (2011: 98). Both inspection of the spectrogram and the waveform can help establish the existence of a segment boundary, but, where possible, measurements should be defined with reference to the waveform due to finer-grained detail in the visual depiction of the sound (Foulkes et al. 2011: 63). In any case, the nature of the preceding and following segments can complicate the measurement procedures involved in determining segment boundaries. Allen notes that there is a general agreement that boundaries between vowels and voiceless consonants are easier to locate than those between vowels and voiced consonants, nasals, liquids, and glides (1978: 1117). For vowels preceded by voiced and voiceless plosives, the onset measurement may be taken at the release burst of the plosive, at the first evidence of non-fricated vocal cord vibration following the release or at the beginning of the first glottal cycle in which high-frequency energy could be observed. In relation to vowels following voiceless aspirate plosives, the end of a positive VOT value is said to be the onset of striations in the second formant of the following vowel (Klatt 1975), while Cho and Ladefoged cite the beginning of the ‘first full vibration of the vocal folds’ (1999: 215). Di Paolo et al. define vowel onset in these instances as ‘the first cycle of a regular repeating pattern in the sine wave function following the release of the preceding consonant’ (2011: 91) which
5 Methodological Considerations
113
corresponds to the beginning of dark bands on the spectrogram representing at least the first three formants. Offset is similarly judged as the end of the last complete sinusoidal pattern. For vowels followed by plosives, measurement can be taken at the last glottal vibration before some particular event, such as closure, associated with the following consonant (Allen 1978: 1179) or at the release of the following plosive, thus including the closure phase within the vowel duration (see Llamas et al. 2009: 392). Other preceding and following segments with sharply defined edges, such as taps and trills, constitute comparable environments where the measurement criteria outlined above can be employed. In the context of preceding and following plosives, taps or trills, vowel duration was measured at the onset and offset of regular repeating glottal vibration; aspiration following plosive release in the preceding environment and consonantal closure in the following environment were systematically excluded from vowel duration measurements. For vowels preceded or followed by fricatives, fricative noise was never included in the vowel duration. Vowel onset and offset were determined at the points where aperiodic turbulence ended or began (or changed dramatically in intensity) and where higher formant structure became visible or invisible in the spectrogram window. For vowels followed by a nasal consonant, the boundaries between the vowels and following nasal consonant were determined at the point where sharp spectral changes occurred at the beginning of the period of oral closure. In this case, it was most useful to appeal to evidence in the spectrogram, notably a sudden drop in amplitude above F0, in order to establish the measurement point for the vowel offset. Similarly, the boundaries between vowels and liquid consonants may be identified where the amplitude signal drops radically due to the added approximant articulation (Lawson et al. 2011: 79). Measuring the onset or offset of vowels that are preceded or followed by the glides [j] and [w] is particularly difficult as there is no abrupt boundary between these segments and the vowel and these segments are characterised by having little or no steady-state portion. In this case, Allen recommends that the boundary be drawn consistently at the point of inflection for the transition between glides and vowels (1978: 1179). The acoustic analysis of the stressed and unstressed vowel variables investigated vowel quality on the basis of F1, F2, and F3. The procedure
114
D. Mooney
for measuring F1, F2, and F3 for a given vowel involves two important decisions: the location in the vowel spectrogram to measure the formants; the method used to extract the formant values. The measurement of the formants can be taken at a single point in the vowel or at multiple points, to reflect the formant’s dynamic quality. Di Paolo et al. note that two basic approaches to single-point measurement have predominated in sociolinguistics: the ‘point of inflection’ method and the ‘midpoint’ method whereby both consider one point to represent the central tendency of the vowel (2011: 90). The ‘point of inflection’ method measures the first point in the vowel that is uninfluenced by the preceding or following consonant/segment, also known as the steadystate. Where formant values are not simultaneously steady, measurement is taken at the peak amplitude and the cursor is positioned at the highest point of energy within the visible dark band (Di Paolo et al. 2011: 91). For the ‘midpoint’ method, the absolute centre of the vowel, halfway between the vowel onset and vowel offset, is chosen. This method assumes consonantal effects to be minimal at the midpoint. Multiplemeasurement techniques may also be used which involve measuring formant frequencies at the vowel onset and offset as well as at several points within the moving or steady-state trajectory (Di Paolo et al. 2011: 91). The first, second, and third formants were measured for each vowel token included in the analysis. These formants were estimated in Praat using the LPC (Linear Predictive Coding) algorithm, with a maximum of 4,000 Hz for male speakers and 4,500 Hz for female speakers. This instrumental adjustment based on biological sex was included as formant trackers may accurately track three formants below 4,500 Hz for female speakers, but may be less accurate for male speakers who might have four formants in the 4,500 Hz range (Clopper 2011: 195; Llamas et al. 2009: 392). The vowel onset and offset were first labelled in the Praat TextGrid and the value of F1, F2, and F3 was extracted manually at the vowel midpoint. The traditional articulatory correlates of the formant frequencies are: F1 and tongue height; F2 and tongue frontness/backness; F2 and F3 and lip-rounding. Various studies have, however, shown these values to be only approximate measures of articulatory configurations because
5 Methodological Considerations
115
other factors, such as the distance of the point of maximum constriction from the glottis, can also have an effect on all formant frequencies (e.g., Fant 1960): these factors are not fully accounted for in the correlations outlined above. For example, a rise in the value of F2, taken to be indicative of fronter tongue position will also entail acoustic lowering of F1 that cannot be taken as indicative of a more close lingual configuration (Rosner and Pickering 1994: 25) or a pharyngeal constriction will result in a high value for F1 that is not related to lingual configuration in the oral cavity (Rosner and Pickering 1994: 29). In sum, modelling the acoustic vowel space by plotting F1 against F2 may not be a wholly accurate indication of articulatory configuration during vowel production. Taking account of these complicating factors and in line with the widespread interpretation of these correlations in sociophonetic studies of language variation and change, the language and dialect death studies nonetheless use the first three formant frequencies to model vowel quality.
5.4.2 Auditory Analysis All ten categorical variables were categorised auditorily using the coding schema outlined in Table 5.4. Where necessary, this impressionistic analysis was supplemented by visual inspection of the acoustic spectrogram in Praat.
5.5
Data Processing
5.5.1 Normalisation The F1, F2, and F3 values for all continuous vocalic variables were subject to a normalisation procedure in order to ensure the reliability of the results. Different speakers exhibit variation in the formant values they produce for a given phonological vowel because of physiological differences in their vocal tracts. These differences are most notable when comparing speakers of different biological sexes because
116
D. Mooney
Table 5.4 Variants used in auditory analysis of categorical variables Variable
Variants
Variant description
(r)
[r] [ɾ] uvular
(ʎ)
[ʎ] [j] [t͡ç] [t͡ʃ] [t] [l] [d͡ʝ] [d͡ʒ] [t͡ʃ] [b] [β] [d] [ð] [z] [ɡ] [G] [j] [ʒ] [s] [j] correct incorrect
Voiced alveolar trill Voiced alveolar tap [ʁ] voiced uvular fricative [ʁ̞] voiced uvular approximant [ʀ] voiced uvular trill Voiced palatal lateral approximant Voiced palatal approximant Voiceless palatal affricate Voiceless postalveolar affricate Voiceless dental plosive Voiced alveolar lateral approximant Voiced palatal affricate Voiced postalveolar affricate Voiceless postalveolar affricate Voiced bilabial plosive Voiced bilabial fricative/approximant Voiced dental plosive Voiced dental fricative Voiced alveolar fricative Voiced velar plosive Voiced velar fricative Voiced palatal approximant Voiced postalveolar fricative Voiceless alveolar fricative Voiced palatal approximant Historically appropriate gender Historically inappropriate gender
(t͡ç)
/d͡ʝ/
(b) (d)
(ɡ) (ʒ) (-s) (gender)
of the greater size of the male vocal tract and the lower fundamental frequency of the source sound. There exists, however, no constant difference between the varying dimensions of the male and female vocal tracts. In order to account for this sexual dimorphism, for the fact that no two speakers’ vocal tracts share the same dimensions, and to reliably compare vowel tokens across speakers and sexes, all data were normalised. Normalisation aims to eliminate variation which is caused by anatomical differences while preserving variation that is sociolinguistically significant. Normalisation procedures are categorised as follows (following Flynn 2011: 3): vowel-intrinsic or extrinsic; formant-intrinsic or extrinsic; speaker-intrinsic or extrinsic. Vowel-intrinsic procedures use
5 Methodological Considerations
117
information from a single vowel category to normalise a given token while vowel-extrinsic procedures use information from multiple vowel categories to normalise each token. Similarly, formant-intrinsic methods use information from one formant to normalise a given token. For example, all F1 values are taken into account when normalising a given F1 token. On the other hand, formant-extrinsic procedures use multiple formants (F1, F2, F3, etc.) to normalise a single F1 value. Finally, speaker-intrinsic techniques use information from a single speaker to normalise their formant values, while speaker-extrinsic methods use information from all speakers in the sample population to normalise an individual’s vowels. Comparative studies on the reliability of various normalisation techniques have shown that the most reliable procedures are classified as vowel-extrinsic, formant-intrinsic, and speaker-intrinsic (Adank et al. 2004; Flynn 2011). Thus, successful normalisation procedures use multiple vowel values within one formant for an individual speaker to normalise their data set. The earliest example of a vowel-extrinsic, formant-intrinsic, and speaker-intrinsic normalisation procedure was proposed by Lobanov (1971), which expresses values relative to the hypothetical centre of a speaker’s vowel space. While many newer normalisation procedures exist, such as the Watt and Fabricius (2002) method and the Nearey (1977) method, I have chosen to use the Lobanov method for the following reasons: Fabricius et al. (2009) showed that Lobanov’s method outperformed both the Watt and Fabricius and the Nearey methods to a statistically significant extent; Adank et al. (2004) determined the Lobanov method to be the best for reducing the effect of anatomical variation while maintaining phonological and sociolinguistic variation; Flynn confirmed that ‘its existing use in the sociolinguistic world is […] warranted’ (2011: 22). The Lobanov method involves calculating z-scores for individual data points which has the effect of transforming the original distribution to one in which the mean becomes zero and the standard deviation becomes 1. A z-score quantifies the original value relative to the number of standard deviations that the score is from the mean of the distribution. Z-scores (z ) were obtained by first calculating the mean (μ) of F1, F2, and F3 values separately (formant-intrinsic) across vowel types (vowel-extrinsic) for all of an
118
D. Mooney
individual speaker’s data (speaker-intrinsic). The mean value of the relevant formant was then subtracted from each individual token (x ) for F1, F2, and F3. The result was then divided by the speaker’s standard deviation (σ ) of the mean frequency for the relevant formant (procedure described in Chen 2008: 635). The formula for z-score calculation can be summarised as: z = (x − μ)/σ; this was implemented using the (=STANDARDIZE) formula in Microsoft Excel. The resultant z-scores are all centred around a mean of (0, 0) for each individual speaker, enabling comparison across speakers as the mean of all of their normalised distributions is the same.
5.5.2 Data Coding Normalised F1, F2, and F3 data for the nine vocalic variables and impressionistic variants for the ten categorical variables were recorded in a series of variable-specific Excel spreadsheets for all speakers, in both the language and dialect death studies. All tokens for all nineteen variables were additionally coded for a series of independent factors, linguistic, and extralinguistic, that are hypothesised to constrain the variation under investigation: • Linguistic independent variables (lexical item, syllable type, preceding phonological environment, following phonological environment) • Extralinguistic independent variables (speaker, year of birth, speaker type, biological sex, place of origin)
5.5.3 Statistical Analysis For all linguistic variables, the normalised formant frequency data (F1, F2, F3) and impressionistically categorised data were modelled statistically; the statistical modelling technique used, in both instances,
5 Methodological Considerations
119
was linear regression, for continuous and categorical variables, respectively. The regression analysis was carried out in the R environment (version 2.15.3) using the Rbrul (version 2.06) text-based interface (Johnson 2008) which makes use of existing functions in the R environment, particularly the model-fitting functions glm and glmer (Bates and Sarkar 2008). Rbrul is a variable rule programme which evaluates the effects of multiple language-internal and/or social factors on either a binary linguistic ‘choice’ or on a continuous dependent variable. One of the assumptions underlying traditional regression analysis is that the individual observations in the data set are independent of each other (Johnson 2009: 363), or that each individual token is an independent observation. In linguistic data sets, however, tokens are grouped according to the speakers who have produced them: as soon as an individual speaker contributes more than one observation to the data set, they become a source of variation that must be accounted for in the statistical model (Tagliamonte and Baayen 2012: 7). Failure to account for the variation introduced into the data set by individual speakers can lead variable rule programmes to grossly overestimate the significance of extralinguistic effects, outputting statistically significant results that are likely to be due to individual variation combining with chance (Johnson 2009: 363). Additionally, individual lexical items can potentially favour or disfavour a particular linguistic outcome. It is also necessary to control for this possibility by taking account of variation introduced to the data set by including tokens that have been segmented from a variety of different words. Taking account of variation introduced by differing lexical items leads to more accurate conclusions about internal effects such as phonetic/phonological context (Johnson 2009: 378). The specific technique employed for all statistical models in the language and dialect death studies is a mixed-effects regression analysis. Mixed-effects models make a distinction between two types of factor that can affect a response (dependent) variable. Firstly, fixed effects are factors that are replicable in another study, for example, speaker sex (male/ female), stress (tonic/atonic), etc. Random effects, on the other hand, are factors drawn from a larger population which are not completely replicable (Johnson 2009: 365), such as individual speakers and different lexical items. Including a speaker random effect in the regression analysis
120
D. Mooney
takes into account that some individuals may favour a given linguistic outcome while others may disfavour it (Johnson 2009: 365). The mixed model will only return a significant result for a given factor, such as speaker sex, if the effect is strong enough to rise above the inter-speaker variation in the model. Likewise, including lexical item as a random effect takes into account the variation introduced into the model by individual words and only returns a significant result for internal or linguistic independent variables when their effect is large enough to outweigh inter-lexical-item variation. If individual speakers and individual words are not included in the regression model as random effects, the results of the analysis will only be relevant for the individuals and words sampled and p-values may be too small and misleading to generalise to the larger population (Tagliamonte and Baayen 2012: 7). In all regression models presented in the language and dialect death studies, ‘speaker’ and ‘word’ are included as random effects. We may note, at this point, that including preceding and following phonological environment as independent variables in the same regression model as syllable type are potentially multicollinear, where multicollinearity refers to cases where two or more independent variables are correlated with each other. There is also a certain amount of multicollinearity introduced into the model by the nature of the wordlist data. Since all speakers are producing the same, or similar, words, the preceding and following phonemes are largely correlated. Interdependencies such as those between independent predictors should never be considered together (Tagliamonte and Baayen 2012: 24–25) because unsolvable computational problems often arise resulting in various kinds of error messages from the statistical modelling programme. If the variable rule programme is successful in returning an output for a model containing multicollinear predictors, the coefficient estimates for the linear regression will not be precise because of difficulty experienced in parcelling out the effect of the multicollinear predictors on the response variable. For this reason, the regression analyses presented in Chapters 6 and 7 include either ‘syllable type’ or preceding and following phonological environment as independent variables in the analysis; the model with the lowest deviance and which therefore accounts best for the variation observed in the data is reported in each case.
5 Methodological Considerations
121
References Adank, Patti M., Roel Smits, and Roeland van Hout. 2004. A comparison of vowel normalisation procedures for language variation research. Journals of the Acoustical Society of America 116: 3099–3107. Allen, George D. 1978. Vowel duration measurement: A reliability study. Journal of the Acoustical Society of America 63: 1176–1185. Bates, Douglas, and Deepayan Sarkar. 2008. lme4: linear mixed-effects models using S4 classes. http://cran.r-project.org/. Boersma, Paul. 2001. Praat: A system for doing phonetics by computer. Glot International 5: 341–345. Boersma, Paul, and David Weenink. 2012. Praat: Doing phonetics by computer. http://www.praat.org/. Chen, Yiya. 2008. The acoustic realisation of vowels of Shanghai Chinese. Journal of Phonetics 36: 629–648. Cho, Taehong, and Peter Ladefoged. 1999. Variations and universals in VOT: Evidence from 18 languages. Jphon 27: 207–229. Cieri, Christopher. 2011. Making a field recording. In Sociophonetics: A student’s guide, ed. Marianna Di Paolo and Malcah Yaeger-Dror, 24–35. Abingdon: Routledge. Clopper, Cynthia G. 2011. Checking for reliability. In Sociophonetics: A student’s guide, ed. Marianna Di Paolo and Malcah Yaeger-Dror, 188–197. Abingdon: Routledge. Di Paolo, Marianna, and Malcah Yaeger-Dror. 2011. Field methods: Gathering data, creating a corpus, and reporting your work. In Sociophonetics: A student’s guide, ed. Marianna Di Paolo and Malcah Yaeger-Dror, 7–23. Abingdon: Routledge. Di Paolo, Marianna, Malcah Yaeger-Dror, and Alicia Beckford Wassink. 2011. Analyzing vowels. In Sociophonetics: A student’s guide, ed. Marianna Di Paolo and Malcah Yaeger-Dror, 87–106. Abingdon: Routledge. Fabricius, Anne H., Dominic Watt, and Daniel E. Johnson. 2009. A comparison of three speaker-intrinsic vowel formant frequency normalisation algorithms for sociophonetics. Language Variation and Change 21: 413–435. Fant, Gunnar. 1960. Acoustic theory of speech production. The Hague: Mouton. Flynn, Nicholas. 2011. Comparing vowel formant normalisation procedures. York Working Papers in Linguistics 11: 1–28.
122
D. Mooney
Foulkes, Paul, Gerard Docherty, and Mark J. Jones. 2011. Analyzing stops. In Sociophonetics: A student’s guide, ed. Marianna Di Paolo and Malcah YaegerDror, 58–71. Abingdon: Routledge. Johnson, Daniel E. 2008. Rbrul. http://www.ling.upenn.edu/~johnson4/ Rbrul.R. Johnson, Daniel E. 2009. Getting off the Goldvarb standard: Introducing Rbrul for mixed-effects variable rule analysis. Language and Linguistics Compass 3: 359–383. Klatt, Dennis H. 1975. Voice onset time, frication and aspiration in wordinitial consonant clusters. Journal of Speech and Hearing Research 18: 686– 706. Lawson, Elenor, Jane Stuart-Smith, James M. Scobbie, Malcah Yaegar-Dror, and Margaret Maclagan. 2011. Liquids. In Sociophonetics: A student’s guide, ed. Marianna Di Paolo and Malcah Yaeger-Dror, 72–86. Abingdon: Routledge. Llamas, Carmen, Dominic Watt, and Daniel E. Johnson. 2009. Linguistic accommodation and salience of national identity markers in a border town. Journal of Language and Social Psychology 28: 381–407. Lobanov, Boris M. 1971. Classification of Russian vowels spoken by different speakers. Journal of the Acoustical Society of America 49: 606–608. Milroy, Lesley, and Matthew Gordon. 2003. Sociolinguistics: Method and interpretation. Oxford: Blackwell. Mooney, Damien, and James Hawkey. 2019. The variable palatal lateral in Occitan and Catalan: Linguistic transfer or regular sound change? Journal of French Language Studies 29: 281–303. Nearey, Terrance M. 1977. Phonetic feature systems for vowels. Unpublished PhD thesis, University of Alberta. Rosner, Brian S., and Brian Pickering. 1994. Vowel perception and production. Oxford: Oxford University Press. Séguy, Jean, and Jacques Allières 1954–1973. Atlas linguistique et ethnographique de la Gascogne. Paris: CNRS. Tagliamonte, Sally, and R. Harold Baayen. 2012. Models, forests and trees of York English: Was/were variation as a case study for statistical practice. Language Variation and Change 24: 135–178. Taylor, Jill. 1996. Sound evidence: Speech communities and social accents in Aixen-Provence. Bern: Peter Lang. Watt, Dominic, and Anne Fabricius. 2002. Evaluation of a technique for improving the mapping of multiple speakers’ vowel spaces in the F1~F2 plane. Leeds Working Papers in Linguistics and Phonetics 9: 159–173.
6 Language Death: Gascon and French
This chapter presents the results of the language death study, analysing the acoustic and auditory data for thirty native Gascon speakers across five fieldwork sites in Béarn. The analysis begins with the oral vowel system, including both stressed and unstressed vowels, focusing in particular on the maintenance of the front mid-vowel contrast over geographical space. L2-to-L1 transfer, from French to Gascon, is then examined, with the explicit aim of identifying the systemic constraints on transfer from the dominant language to the obsolescent language, rather than simply identifying individual features of the dominant language that have been adopted in the dying language. Finally, this chapter investigates evidence for levelling or dialect mixing between the subdialects of Gascon spoken in Béarn, analysing the loss of highly localised phonetic and phonological variants as the local language, and its speech community, contracts.
© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 D. Mooney, Language and Dialect Death, https://doi.org/10.1007/978-3-031-51101-1_6
123
124
6.1
D. Mooney
The Oral Vowel System
The analysis begins with an examination of the Gascon stressed oral vowel system over geographical space, including an investigation of the maintenance of the front mid-vowel contrast. The examination of the unstressed vowels focuses primarily on the realisation of and phonetic distinction between the post-tonic vowels /-e/ and /-O/ in the sub-dialects of Gascon spoken in Béarn.
6.1.1 Stressed Oral Vowels Gascon, and all other varieties of Occitan, has the following historically appropriate stressed oral vowels: /i y u e E O a/. Differences in the phonetic realisation of these vowels across sub-dialects of Gascon in the region of Béarn are not well documented. To examine the extent to which the sub-dialects spoken in Laruns, Arzacq-Arraziguet, Lembeye, Nay, and the ‘Ouest’ have common phonetic realisations for these vowels, statistical analyses of the normalised F1, F2, and F3 data for all speakers were undertaken for each vowel. These analyses included the following independent variables: preceding phoneme; following phoneme; place of origin; speaker sex. No statistically significant differences were established between the sub-dialects for /i/, /y/, /u/, /E/, or /O/. ‘Place of origin’ was, however, shown to be a significant predictor of F2 for /a/ and of F3 for /e/. The results of the F2 model for /a/ is presented in Table 6.1. The regression model in Table 6.1 shows that speakers in Laruns have /a/ vowels that are the most anterior, or more front, in the vowel space, with speakers of the ‘Ouest’ sub-dialect having the most retracted /a/ vowels. Speakers from Arzacq-Arraziguet, Nay, and Lembeye fall somewhere in-between, with the magnitude of the difference between them being small. The sub-dialects demonstrated no significant F1 or F3 differences for /a/. Table 6.2 presents the regression analysis for normalised F3 values for /e/: speakers from Laruns and Arzacq-Arraziguet have the highest F3 values, while the sub-dialects of the ‘Ouest’, Lembeye, and Nay
125
6 Language Death: Gascon and French
Table 6.1 All native speakers (normalised data): regression analysis of F2 values for /a/, with ‘speaker’ and ‘word’ as random effects Dependent variable = F2 of /a/ N = 144 Grand mean = –0.551
R2 : 0.352 Degrees of freedom: 10
Factor group
Factor
Coefficient
N
p-value
Preceding phoneme
/t/ /p/ Laruns Arzacq Nay Lembeye Ouest Female Male
+0.101 –0.101 +0.132 +0.018 +0.018 –0.063 –0.106 +0.056 –0.056
88 56 28 30 28 29 29 70 74
0.00
Place of origin
Speaker sex
0.01
0.01
favour lower F3 values. ‘Place of origin’ was not returned as a significant predictor of F1 or F2 for /e/. Overall, the sub-dialects of Gascon in Béarn can be seen as relatively coherent in terms of their phonetic realisation of the seven-term stressed oral vowel system. The most common source of variability in the phonemic inventory of Gascon (and Occitan generally) is the loss of the phonemic contrast between /e/ and /E/, which is said to be ‘largely maintained’ in final closed and non-final syllables (Oliviéri and Sauzet 2016: 322), with the contrast being more robust in aquitano-pyrénéen Occitan varieties, Table 6.2 All native speakers (normalised data): regression analysis of F3 values for /e/, with ‘speaker’ and ‘word’ as random effects Dependent variable = F3 of /e/ N = 144 Grand mean = 0.196
R2 : 0.199 Degrees of freedom: 13
Factor group
Factor
Coefficient
N
p-value
Place of origin
Laruns Arzacq Ouest Lembeye Nay
+0.444 +0.215 –0.199 –0.212 –0.247
29 29 29 29 28
0.00
Non-significant factor groups were: preceding phoneme (p = 0.229), speaker sex (p = 0.316), and following phoneme (p = 1)
126
D. Mooney
including Gascon (Bec 1973: 57). As such, the analysis of the stressed oral vowels includes a statistical analysis of the maintenance of the /e/ ~/E/ contrast, to confirm the traditional accounts of Gascon presented in the literature. Indeed, it is worth noting that the contact language, French, also exhibits instability in the maintenance of contrast between /e/ and /E/, though Mooney (2019: 71) has shown that this contrast is largely maintained on the F1 dimension (vowel height) in the regional French of bilingual Occitan-French speakers in Béarn, based on an independent data set to the one considered here. Additionally, the /e/~/E/ contrast is said to be maintained by rounding in lo parlar negre of the Landes, which borders Arzacq-Arraziguet, whose local variety of Gascon shares many features with lo parlar negre. The realisation of /e/ as [œ] in lo parlar negre, essentially produces the mid-vowel contrast /œ/~/E/ (Oliviéri and Sauzet 2016: 322). The normalised F1 and F2 data for all stressed oral vowels for male and female speakers from Laruns, in the Vallée d’Ossau, are presented in Fig. 6.1. The general distribution of the vowels in the F1-F2 plane is as expected, noting some overlap in the distributions of /i/ and /y/ and in the distributions of /e/ and /E/. To establish the extent to which the /e/~/E/ phonemic contrast is maintained phonetically in Laruns, the normalised data for F1, F2, and F3 were submitted to statistical analysis, including the following independent variables: historically appropriate phoneme; syllable type; speaker sex. The aim of these analyses is to establish the extent to which formant values can be predicted by the historically appropriate phoneme, /e/ or /E/, when phonological environment has been taken into account. The analysis of the F2 data showed that the historically appropriate phoneme did not significantly predict F2 values in the data set (p = 0.32), showing that the Laruns speakers make no distinction between /e/ and /E/ based on vowel frontness/backness. The Laruns speakers were shown, however, to make a significant phonetic distinction between /e/ and /E/ on the basis of F1 and F3; the statistical models of F1 and F3 are presented in Tables 6.3 and 6.4, respectively. The F1 distinction between /e/ and /E/ is highly significant (p = 0.00) and of a relatively high magnitude (±0.485), indicating that the traditional height distinction between the front mid-vowels is
127
6 Language Death: Gascon and French
Fig. 6.1 Stressed oral vowel system for (male and female) Laruns speakers, normalised data
robust. This height distinction is supported by an equally significant F3 difference between the vowels (p = 0.00). The normalised data plots for the other four fieldwork sites can be consulted in Appendix 2: Figure A2.1 for Arzacq-Arraziguet in the north of Béarn; Figure A2.2 for Lembeye in the Vic Bilh, eastern Béarn; Figure Table 6.3 Laruns speakers (normalised data): regression analysis of F1 values for /e/ and /E/, with ‘speaker’ and ‘word’ as random effects Dependent variable = F1 N = 53 Grand mean = –0.13
R2 : 0.554 Degrees of freedom: 8
Factor group
Factor
Coefficient
N
p-value
Phoneme
/E/ /e/ C_C C_#
+0.485 –0.485 +0.447 –0.447
24 29 41 12
0.00
Syllable type
Non-significant factor groups were: speaker sex (p = 0.126)
0.03
128
D. Mooney
Table 6.4 Laruns speakers (normalised data): regression analysis of F3 values for /e/ and /E/, with ‘speaker’ and ‘word’ as random effects Dependent variable = F3 N = 53 Grand mean = 0.37
R2 : 0.379 Degrees of freedom: 8
Factor group
Factor
Coefficient
N
p-value
Phoneme
/e/ /E/
+ 0.32 –0.32
23 30
0.00
Non-significant factor groups were: syllable type (p = 0.357) and speaker sex (p = 0.487)
A2.3 for Nay, in the central plain; Figure A2.4 for the ‘Ouest’ speakers, in western Béarn. For each of these fieldwork sites, the normalised F1, F2, and F3 data for /e/ and /E/ were analysed statistically to examine the maintenance of the /e/~/E/. The normalised data plots for the other four fieldwork sites can be consulted in Appendix 2: Figure A2.1 for Arzacq-Arraziguet in the north of Béarn; Figure A2.2 for Lembeye in the Vic Bilh, eastern Béarn; Figure A2.3 for Nay, in the central plain; Figure A2.4 for the ‘Ouest’ speakers, in western Béarn. For each of these fieldwork sites, the normalised F1, F2, and F3 data for /e/ and /E/ were analysed statistically to examine the maintenance of the /e/~/E/ contrast. The results of these analyses are summarised in Table 6.5 and statistical models in which the historically appropriate phoneme (/e/ and /E/) was returned as a significant predictor can be consulted in Appendix 2. The speakers in Arzacq-Arraziguet and in Lembeye make a significant phonetic distinction between /e/ and /E/, with ‘phoneme’ returned as a statistically significant predictor of F1, F2, and F3 at both fieldwork sites (see Appendix 2, Tables A2.1–A2.6). For Arzacq-Arraziguet, the significant result for F1 shows that these speakers do not replicate the system of the adjacent parlar negre, where the height distinction between /e/ and / E/ is lost due to /e/ being realised as [œ]. In Nay, speakers make a significant distinction between /e/ and /E/ on the F1 dimension (see Appendix 2, Table A2.7), but ‘phoneme’ was not shown to be a significant predictor of F2 and F3. Finally, speakers from the ‘Ouest’ distinguished between /e/ and /E/ on the basis of F1 and F2 (see Appendix 2, Tables A2.8 and A2.9), but not F3. Overall, the traditional vowel height distinction between /e/ and /E/ is maintained in all five sub-dialects, confirming the
6 Language Death: Gascon and French
129
Table 6.5 Significant formant frequency differences for front mid-vowels by place of origin
p
p
p
accounts of Oliviéri and Sauzet (2016: 322) and Bec (1973: 57); there appears to be no sound change underway which involves a merger of the front mid-vowels. Considering all of the findings presented here, the subdialects of Gascon spoken in Béarn can be viewed, in terms of the stressed oral vowel system, as part of a phonemically coherent variety exhibiting common structural features, with some low-level phonetic variation over geographical space.
6.1.2 Unstressed Oral Vowels The seven-term stressed oral vowel system is reduced to five vowels in pre-tonic syllables, /i y e a u/, and to four vowels in post-tonic syllables, /i e O u/. Previous analyses of the unstressed vowels have shown word-final post-tonic /-e/ and /-O/ to exhibit the highest levels of variability over geographical space and, for this reason, they will form the basis of the analysis presented here. In some varieties of Gascon, including lo parlar negre and adjacent varieties in the north of Béarn, both /-e/ and /-O/ are realised as [œ] (Mooney 2014: 348), resulting in a neutralisation of the
130
D. Mooney
contrast between the vowels in post-tonic position, e.g., càder [ˈkaðœ] ‘fallinf ’~cada [ˈkaðœ] ‘every’. Figure 6.2 presents the /-e/ and /-O/ data for native speakers in Laruns: the overall pattern indicates phonetic separation between the vowels and, indeed, the regression analyses of F1 and F2 reveal the historically appropriate phonemes to be significant predictors of both vowel height and vowel frontness/backness (see Tables 6.6 and 6.7). There was no significant difference between the F3 values for /-e/ and /-O/. By contrast, the /-e/ and /-O/ data for the Arzacq-Arraziguet speakers, in northern Béarn, exhibit a strikingly overlapping distribution in Fig. 6.3. Regression analyses of F1, F2, and F3 determined that the historically appropriate phoneme, /-e/ or /-O/, was never a significant predictor of any formant frequency considered. This shows that, for the Arzacq speakers, /-e/ and /-O/ have merged in post-tonic position, exhibiting the well-established phonemic merger of the adjacent Gascon
Fig. 6.2 Unstressed oral vowels for (male and female) Laruns speakers, normalised data
131
6 Language Death: Gascon and French
Table 6.6 Laruns speakers (normalised data): regression analysis of F1 values for /-e/ and /-O/, with ‘speaker’ and ‘word’ as random effects Dependent variable = F1 N = 52 Grand mean = 0.011
R2 : 0.175 Degrees of freedom: 9
Factor group
Factor
Coefficient
N
p-value
Phoneme
/-O/ /-e/
+0.23 –0.23
27 25
0.03
Non-significant factor groups were: preceding phoneme (p = 0.254) and speaker sex (p = 0.266)
Table 6.7 Laruns speakers (normalised data): regression analysis of F2 values for /-e/ and /-O/, with ‘speaker’ and ‘word’ as random effects Dependent variable = F2 N = 52 Grand mean = –0.091
R2 : 0.835 Degrees of freedom: 9
Factor group
Factor
Coefficient
N
p-value
Phoneme
/-e/ /-O/ /z/ /b/ /d/ /t/
+0.633 –0.633 +0.214 +0.049 +0.042 –0.304
25 27 5 9 9 29
0.00
Preceding phoneme
0.03
Non-significant factor groups were: speaker sex (p = 0.608)
variety, lo parlar negre. The normalised data plots for the other three fieldwork sites can be consulted in Appendix 2: Figure A2.5 for Lembeye; Figure A2.6 for Nay; Figure A2.7 for the ‘Ouest’ speakers. The results of the F1, F2, and F3 regression analyses, including ‘phoneme’ as an independent variable, are summarised in Table 6.8 for speakers at all fieldwork sites. In Lembeye, the F2 values of /-e/ and /-O/ were significantly different (see Appendix 2; Table A2.10), indicating that the expected front-back distinction between the vowels is maintained. This F2 distinction between the vowels is also significant for Nay (see Appendix 2; Table A2.12) and the ‘Ouest’ (see Appendix 2; Table A2.13), with Nay speakers also realising a significant F1 difference between /-e/ and /-O/ (see Appendix 2; Table A2.11). Overall, the phonemic contrast between /-e/ and /-O/ is maintained in Béarnais, with the exception of
132
D. Mooney
Fig. 6.3 Unstressed oral vowels for (male and female) Arzacq speakers, normalised data
the sub-dialect spoken in Arzacq-Arraziguet, which exhibits the phonological merger of /-e/ and /-O/. This merger results from the conditioned phonetic changes by which both /-e/ and /-O/ become [œ]. Two final regression analyses considered /-e/ and /-O/ separately, but for speakers from all five fieldwork sites, with the aim of establishing any significant phonetic differences in the realisation of the vowels over geographical space in Béarn. For /-e/, ‘place of origin’ was shown to be a significant predictor of only F2 (see Appendix 2; Table A2.14). Laruns, Nay, and Lembeye were shown by the regression model to favour higher F2 values, or more front vowels, while the ‘Ouest’ speakers, and in particular those from Arzacq, had more centralised /-e/ vowels. For Arzacq, this centralisation of /-e/ is consistent with the vowel overlap between /-e/ and /-O/ in Fig. 6.3. For /-O/, only F2 was significantly influenced by the ‘place of origin’ of the speaker (see Appendix 2; Table A2.15). Laruns, Nay, and Lembeye were shown to have the most posterior, or back, vowels, while the /-O/ vowels for Arzacq-Arraziguet were significantly
6 Language Death: Gascon and French
133
Table 6.8 Significant formant frequency differences for post-tonic vowels by place of origin
p
p
p
centralised. Considering both analyses together, the vowel separation between /-e/ and /-O/ is maximal for Laruns, Nay, and Lembeye, and minimal for Arzacq, as expected.
6.2
Transfer from French
As outlined in Chapter 2, we can usually distinguish between language change phenomena in dying languages that can be attributed to contact with the dominant language, and those which cannot. In this section, evidence for the transfer of linguistic features and rules from French in the speech of the Gascon native speakers will be examined, thus exploring linguistic change that is attributable to contact with the dominant language.
134
D. Mooney
6.2.1 Rhotics The historically appropriate rhotic consonants of Gascon are apical, notably [r] and [ɾ]. These sounds are traditionally contrastive in intervocalic contexts but, in other contexts, some analyses consider them to be variants of an archiphoneme, often denoted /r/. These variants are not, however, allophones in the strictest sense, in that they do not occur traditionally in strict complementary distribution outside of intervocalic contexts. They do, however, exhibit a probabilistic tendency to occur in certain contexts: [r] word-initially and as an onset after [n]; [ɾ] in onset clusters and in coda position (Mooney 2014: 345). The standard French rhotic consonant is realised as a voiced uvular fricative [ʁ], though uvular approximants and trills are also attested. As such, the primary distinction between the contact languages, Gascon and French, is between apical and uvular rhotics, respectively. There is evidence to suggest that, due to prolonged contact with French, Gascon apical rhotics are variably replaced by uvular rhotics in all positions, with various constraints on transfer, including phonological environment, position in the syllable, and the speaker’s place of origin within Gascony (Mooney 2018). Table 6.9 provides the frequency distribution for all Gascon rhotic data considered in the language death study, divided by sub-dialectal area within Béarn. The percentage realisations of [ɾ], [r], [ʁ], and the null phoneme are broadly comparable across sub-dialects, but there is clear evidence for the presence of uvular rhotics in Gascon. To examine the variable distribution of these uvular rhotics, transferred from French, the apical/uvular distinction was analysed statistically (see Table 6.10), excluding the zero variants; syllable type was shown to be a highly significant predictor of the adoption of uvular rhotics from French to Gascon. Uvular [ʁ] is favoured in medial and final consonant clusters, and in simple codas; more traditional apical realisations ([r] and [ɾ] combined) are favoured in onset clusters and intervocalically. Word-initial positions had a relatively neutral effect on the choice between uvular and apical rhotics. This pattern largely replicates the general pattern outlined by Mooney (2018), considering the Gascon of other speakers from the region: ‘for syllable type, dorsal realisations are shown to be favoured
135
6 Language Death: Gascon and French
in simple syllable codas […] and disfavoured in complex onset clusters’ (Mooney 2018: 73). Excluding the uvular data for all speakers and considering only the data for [r] and [ɾ], Table 6.11 presents the results of the regression analysis of the variable distribution of the trilled and tapped apical rhotics in Gascon. Beginning with syllable type, the trilled variant [r] is significantly favoured in medial and final consonant clusters, and word-initially; the tapped variant [ɾ] is favoured in complex onsets and intervocalically. Simple codas have a neutral effect on the distribution; this is perhaps not surprising since simple codas (V_#) favour uvular variants (see Table 6.10). Overall, speakers from Arzacq, the ‘Ouest’, Lembeye, and Laruns favour trilled over tapped variants, and speakers from Nay favour tapped variants in all positions. Table 6.9 All native speakers: frequency distribution of rhotic variable Place/Variant
[ɾ]
[r]
[ʁ]
zero
N
Laruns Arzacq Lembeye Nay Ouest
34% 34% 39% 43% 36%
14% 21% 16% 4% 19%
48% 39% 43% 49% 40%
4% 6% 2% 4% 5%
135 126 134 134 135
Table 6.10 All native speakers: regression analysis of rhotic variable, with variants [apical] and [uvular]; ‘speaker’ and ‘word’ as random effects Dependent variable = uvular/apical rhotic Response variant = uvular N = 635
R2 : 0.616 Degrees of freedom: 13
Factor group
Factor
Coefficient
N
p-value
Syllable type
V_C# V_# V_C #_V V_V C_V
+1.492 +0.609 +0.468 –0.042 –0.996 –1.531
57 110 87 118 119 128
0.00
Non-significant factor groups were: speaker sex (p = 0.722) and place of origin (p = 0.967)
136
D. Mooney
Table 6.11 All native speakers: regression analysis of apical rhotics, with variants [ɾ] and [r]; ‘speaker’ and ‘word’ as random effects Dependent variable = [ɾ]/[r] rhotic Response variant = [r] N = 344
R2 : 0.516 Degrees of freedom: 13
Factor group
Factor
Coefficient
N
p-value
Place of origin
Arzacq Ouest Lembeye Laruns Nay V_C# #_V V_C V_# V_V C_V
+0.818 +0.498 +0.340 +0.285 –1.941 +1.143 +0.939 +0.617 –0.088 –0.784 –1.826
70 74 73 65 62 17 61 38 47 69 112
0.00
Syllable type
0.04
Non-significant factor groups were: speaker sex (p = 0.275)
The Gascon speakers in the language death study variably realise the rhotic variable as uvular, with the following constraints on transfer from French: transfer is more probable in medial and word-final consonant clusters (V_C and V_C#), and word-finally in simple codas (V_ #), confirming the results of previous analyses with a different sample (Mooney 2018: 73). In the environments where apical variants are favoured, [r] is favoured word-initially (#_V), while [ɾ] is favoured intervocalically (V_V) and in complex onsets (C_V). This confirms traditional accounts of the distribution of the apical rhotics, which cite [r] word-initially and [ɾ] in onset clusters (Mooney 2014: 345). The variable rule for transfer from French to Gascon is /r/ > [ʁ]/V_C#; V_#; V_ C, and it is worth noting that, in the case of the Gascon rhotics, transfer from French complexifies, rather than simplifying, the system.
6.2.2 Palatal Lateral Gascon preserves the palatal lateral /ʎ/ in all positions: word-initially, e.g., lhevar [ʎeˈβa] ‘raiseinf ’, intervocalically, e.g., tribalhar [tɾiβaˈʎa] ‘workinf ’, word-finally, e.g., uelh [weʎ] ‘eye’. This violates the general
6 Language Death: Gascon and French
137
Occitan phonotactic constraint of /ʎ/ being impermissible in coda position; this is characteristic of the conservative Occitano-Romance supra-dialectal grouping of aquitano-pyrénéen (Oliviéri and Sauzet 2016: 327); /ʎ/ may also occur in a tautosyllabic consonant cluster with / -s/ in coda position, where /-s/ is the inflectional morpheme for the pluralisation of common nouns, e.g., uelhs [weʎʃ] ‘eye’, where /-s/ is also palatalised. In other Occitan varieties, the palatal approximant [j] frequently emerges as an allophone of /ʎ/ in intervocalic position, e.g., tribalhar [tɾibaˈja]. In Cardaillac Kelly’s (1973) study of the Gascon variety spoken in Donzac (Tarn-et-Garonne), however, she notes that /ʎ/ has ‘no variants and no distributional limitations’ (1973: 32), even noting that [ʎ] transfers variably into the regional variety of French spoken by her bilingual informants, where it replaces French /j/ in intervocalic position (1973: 69), e.g., veille /vej/ [vEʎe]. This suggests that [j] is not used as a variant of /ʎ/ in Gascon. In the case of Gascon-French bilinguals, word-initial /ʎ/ in Gascon is phonemically cognate with /l/ in French, e.g., lever / leˈve/ ‘raiseINF ’, while intervocalic and word-final /ʎ/ is phonemically cognate with /j/ in French, e.g., travailler /tʁavaˈje/ ‘workinf ’ and œil / œj/ ‘eye’. The phonemic cognate of Gascon /ʎ/ in a final cluster with ͡ /-s/ is not [j] in French, but zero e.g., yeux /jø/ ‘eyes’. In Old French, /-lts ͡ ͡ / is attested in this position, realised as [lG ts ], e.g., ialz [ialG ts ] ‘eyes’ (see Mooney and Hawkey, 2019: 298, for discussion). In general, however, these phonemic correspondences between the languages in contact indicate potential points of transfer from French to Gascon. Table 6.12 presents the frequency distribution for variants of /ʎ/ by sub-dialect. In all sub-dialects, [j] is the majority variant, challenging Cardaillac Kelly’s (1973) assessment of potential transfer from French for Gascon /ʎ/. Some 50 years later, the transferred variant from French is dominant across all sub-dialects of Gascon in Béarn. This compressed time scale for linguistic change and the amount of change observed is characteristic of language death. Laruns shows the highest rate of transfer (81%), while the other sub-dialects are relatively comparable in terms of the amount of transfer observed (between 55% for Arzacq and 63% for Ouest speakers). In a small number of cases, /ʎ/ was depalatalised to [l] in initial position.
138
Table 6.12
D. Mooney
All native speakers: frequency distribution of palatal lateral variable
Place/Variant
[ʎ]
[j]
[l]
N
Laruns Arzacq Lembeye Nay Ouest
18% 44% 38% 42% 37%
81% 55% 62% 58% 63%
1% 1% 0% 0% 0%
102 99 105 103 101
To investigate the existence of constraints on the transfer of [j] from French, the [ʎ] and [j] data for all sub-dialects were submitted to regression analysis, the results of which are presented in Table 6.13. Syllable type was returned as a significant predictor of /ʎ/ being realised as [j]: the palatal approximant is favoured in intervocalic and word-final position, with the palatal lateral favoured in initial position and word-final clusters with /-s/. The ordering of the regression coefficients indicates that the distribution of the variants is indeed influenced by contact with French, in that the probability of /ʎ/ being realised as [j] is increases when the phonemic cognate of /ʎ/ in Gascon is /j/ in French. This is somewhat complicated in the V_C# context, where the phonemic cognate of /ʎ/ in French is zero. These results confirm previous analyses of the palatal lateral in contact with French in Gascon and Catalan, ‘with the French distributional constraints for the palatal glide being largely reproduced by older native speakers in both languages’ (Mooney and Hawkey, 2019: 298). The variable rule for transfer from French can be formalised as /ʎ/ > [j]/V_V; V_#; while transfer from French can occur in all positions, there are clear structural constraints involved, rather than a simplistic ‘replace X by Y’ mechanism of contact-induced, or externally motivated, change.
6.2.3 Voiced Plosives Gascon /b d ɡ/ are traditionally lenited in intervocalic position, either alone or in a cluster with [ɾ r l z] (Oliviéri and Sauzet 2016: 325) and this lenition process also applies at word boundaries (see Sect. 4.2.1). These allophonic variants do not occur regularly in all varieties of Occitan and they do not occur in French for intervocalic /b d ɡ/. The historically
139
6 Language Death: Gascon and French
Table 6.13 All native speakers: regression analysis of palatal lateral variable, with variants [ʎ] and [j]; ‘speaker’ and ‘word’ as random effects Dependent variable = palatal lateral Response variant = [j] N = 508
R2 : 0.594 Degrees of freedom: 11
Factor group
Factor
Coefficient
N
p-value
Syllable type
V_V V_# V_C# #_V
+1.503 +0.228 –0.540 –1.192
156 142 115 95
0.00
Non-significant factor groups were: speaker sex (p = 0.567) and place of origin (p = 0.614)
appropriate realisations of /b d ɡ/ in intervocalic position are [β ð G] and, as such, plosive realisations [b d ɡ] may indicate transfer from French. Table 6.14 presents the frequency distribution for variants of /b/ in intervocalic position, with historically appropriate [β] as the majority variant in Laruns (56%), Lembeye (55%), and the ‘Ouest’ (52%), though the percentage realisations of [b] and [β] are roughly 50/50 in each sub-dialect. Indeed, the regression model in Table 6.15 indicates that ‘place of origin’ has no significant effect on the variation observed (p = 0.87). There is limited evidence for fortition to [p] in Arzacq and Nay, which is well-attested (Moreux and Puyau 2002: 25), and of lenited variants with a change in place of articulation ([ð], [G], and [v]) in Laruns and Arzacq. /b/ is further lenited to [w] in a small number of cases in all sub-dialects but Lembeye, and to [j] in Arzacq. In order to explore the factors that favour the potential transfer for [b] from French, a regression analysis was undertaken to explore the Table 6.14 variable
All native speakers: frequency distribution of voiced bilabial plosive
Place/Variant
[b]
[β]
[ð]
[G]
[p]
[v]
[w]
[j]
N
Laruns Arzacq Lembeye Nay Ouest
38% 45% 45% 53% 45%
56% 45% 55% 41% 52%
1% 1% 0% 0% 0%
3% 0% 0% 0% 0%
0% 3% 0% 3% 0%
0% 2% 0% 0% 0%
2% 2% 0% 3% 3%
0% 2% 0% 0% 0%
64 66 75 72 69
140
D. Mooney
Table 6.15 All native speakers: regression analysis of voiced bilabial plosive variable, with variants [plosive] and [lenited]; ‘speaker’ and ‘word’ as random effects Dependent variable = /b/ plosive/lenited Response variant = lenited N = 339
R2 : 0.273 Degrees of freedom: 17
Factor group
Factor
Coefficient
N
p-value
Following phoneme
/O/ /a/ /u/ /y/ /e/
+1.516 +1.490 –0.657 –0.731 –1.619
72 90 50 39 88
0.00
Non-significant factor groups were: preceding phoneme (p = 0.06), place of origin (p = 0.871), and speaker sex (p = 0.874)
variation between the plosive [b] and lenited variants. For this analysis, fortified variants and those with a different place of articulation were excluded and [β w j] were recoded together as ‘lenited’. Following phoneme was the only independent variable returned by the analysis as significant: [b] is favoured before /u/, /y/, and /e/, while lenited variants are favoured before /O/ and /a/. This variable rule to describe transfer from French can be formalised as /b/ > [b]/V_/e/; V_/y/; V_/u/. Table 6.16 presents the frequency distribution for intervocalic /d/ in Gascon. The key distinction in this section is between full plosive realisations and lenited variants. As such, [ɡ] and [G] variants have been excluded from the analysis as they are not lenited variants of /d/. The single token of /t/ was also excluded as it involves fortition, not lenition. The [ð], [z], [ʒ], and [j] variants were recoded as ‘lenited’, with the following lenition cline: /d/>[ð]>[z]>[ʒ]>[j]. The variation over geographical space is discussed fully in the context of dialect mixing in Sect. 6.3.4. The regression analysis for [d] and ‘lenited’ variants returned no significant predictors: following phoneme (p = 0.19); speaker sex (p = 0.24); preceding phoneme (p = 0.45); place of origin (p = 0.60). While the plosive variant occurs between 58% and 43% of the time across dialects, there does not seem to be any significant constraints on transfer of the plosive variant from French.
141
6 Language Death: Gascon and French
Table 6.16 variable
All native speakers: frequency distribution of voiced apical plosive
Place/Variant
[d]
[ð]
[ɡ]
[G]
[t]
[z]
[ʒ]
[j]
N
Laruns Arzacq Lembeye Nay Ouest
54% 58% 44% 51% 43%
28% 42% 33% 32% 51%
3% 0 4% 1% 1%
13% 0 1% 15% 5%
0 0 1% 0 0
0 0 8% 0 0
2% 0 0 0 0
0 0 9% 1% 0
68 59 64 57 61
The frequency distribution for /ɡ/ is outlined in Table 6.17, where the majority variant is lenited [G] across all of the sub-dialects, followed by [ɡ]. Fortified [k] is also attested in all sub-dialects, with the exception of Laruns. The voiced postalveolar affricate [d͡ʒ] and fricative [ʒ] also occur, exhibiting variation in both place and manner of articulation. The variants [sk], [ʃ], and [ʃk] occurred primarily in the target word paregut ‘appearpst.pctp.m.sg ’, which has the lexical variant pareissut. /ɡ/ was realised as [ɾ] in a small number of cases in Nay. The data for /ɡ/ are modelled statistically in Table 6.18, considering only the distinction between [ɡ] and [G]. Preceding phoneme is revealed to significantly influence the variation in the data, with only one environment favouring the plosive: preceding /l/, though we must interpret this result with caution as the token counts are low. Overall, there is no consistent pattern of transfer across the analyses of the voiced plosives in intervocalic position or in intervocalic clusters. Fully realised plosives are present for all variables and this may be due to contact with French. For /b/, the historically inappropriate plosive is favoured by following /e/, /y/, and /u/. For /d/, there are no constraints Table 6.17 variable
All native speakers: frequency distribution of voiced velar plosive
Place/ Variant
[ɡ]
[G]
[k]
[d͡ʒ]
[ʒ]
[sk]
[ʃ]
[ʃk]
[ɾ]
zero
N
Laruns Arzacq Lembeye Nay Ouest
25% 15% 16% 18% 19%
67% 55% 58% 57% 54%
0% 8% 9% 2% 5%
3% 5% 0% 0% 2%
0% 5% 6% 6% 6%
5% 1% 0% 5% 0%
0% 9% 2% 2% 8%
0% 2% 9% 5% 6%
0% 0% 0% 3% 0%
0% 0% 0% 2% 0%
55 66 67 61 63
142
D. Mooney
Table 6.18 All native speakers: regression analysis of voiced velar plosive variable, with variants [plosive] and [lenited]; ‘speaker’ and ‘word’ as random effects Dependent variable = /ɡ/ plosive/lenited Response variant = lenited N = 239
R2 : 0.478 Degrees of freedom: 16
Factor group
Factor
Coefficient
N
p-value
Preceding phoneme
/i/ /a/ /ɾ/ /e/ /l/
+1.043 +0.478 +0.451 +0.052 –2.025
80 53 21 76 9
0.04
Non-significant factor groups were: speaker sex (p = 0.104), following phoneme (p = 0.23), and place of origin (p = 0.921)
on the potential transfer of [d]. For /ɡ/, the plosive variant occurs less often, in clusters where the phoneme is preceded by /l/. The amount of variability present in the data for each of these phonemes inevitably complicates the analysis of transfer but provides an excellent opportunity to observe the wealth of variation in these unstandardised varieties.
6.2.4 Gender Marking Speakers of all sub-dialects were asked to produce Gascon lexical items that differ in grammatical gender from their French equivalents (see Appendix 1). The frequency distribution for correctness or historical appropriateness is outlined in Table 6.19. Producing the correct gender for the Gascon noun ranges from 83% in Nay to 69% in Lembeye. A regression analysis of ‘correctness’ found that there was no significant difference between the sub-dialect correctness scores (p = 0.17), controlling for inter-speaker variation. Across sub-dialects, gender marking is historically appropriate 75% of the time. Of the 16 words included in the analysis, only four were produced with the correct gender 100% of the time: un ahar ‘matter’ for French une affaire; la lèbe ‘hare’ for French le lièvre; la léit ‘milk’ for French le lait; un arrelòtge ‘clock’ for French une horloge. Five further words had high correctness scores, ranging from 80 to 100%: lo hum ‘smoke’
6 Language Death: Gascon and French
Table 6.19 variable
143
All native speakers: frequency distribution of gender marking
Place/Variant
Correct
Incorrect
N
Laruns Arzacq Lembeye Nay Ouest
77% 74% 69% 83% 71%
23% 26% 31% 17% 29%
82 90 88 90 80
(97%); la sau ‘salt’ (96%); lo par ‘pair’ (96%); lo limac ‘slug’ (90%); la sèrp ‘snake’ (83%). Five words had correctness scores in the 40–80% range: lo cremalh ‘house-warming’ (72%); lo deute ‘debt’ (70%); ua ungla ‘nail’ (64%); la mensonja ‘lie’ (56%); lo grèish ‘fat’ (43%). The two lowest correctness scores were la sang ‘blood’ (44%) and la bilheta ‘ticket’ (21%). Transfer of grammatical gender from French is evident for nearly all of the words analysed, though the effect is in the minority across subdialects and it does not seem that particular lexical domains favour or disfavour transfer.
6.3
Dialect Levelling/Mixing
This section examines the retention of geographically localised phonetic and phonological variation across the sub-dialects of Gascon spoken in Béarn, with the aim of exploring the processes of dialect contact and mixing between the sub-dialects. Previous analyses of dying language, such as Jones (1998), have indicated that highly localised or peripheral forms may be lost in favour of more widespread variants. This outcome is characteristic of levelling during dialect contact and would not be directly attributable to contact with French.
6.3.1 Unstressed Oral Vowels The merger of the post-tonic vowels /-e/ and /-O/ to [œ] is characteristic of Arzacq-Arraziguet and of its neighbouring sub-dialect, lo parlar negre, spoken traditionally in the Landes. The data presented for Arzacq
144
D. Mooney
in Sect. 6.1.2 show clearly that this merger has been maintained overtime and is not aligned with all other sub-dialects considered, which make a significant phonetic (and phonological) distinction between /-e/ and /-O/. Thus, for the post-tonic vowels, this minority phonological variation has not been subject to levelling due to contact with other subdialects. This is perhaps because the Arzacq speakers are also in contact with the adjacent parlar negre varieties, which also have the merger. It is also worth noting that the reversal of phonological mergers is relatively uncommon, if not impossible (Garde 1961; Labov 1994).
6.3.2 Voiceless Affricate The /t͡ç/ phoneme of the Gascon spoken in Béarn, with geographical variants [t] and [t͡ʃ], correspond to final /l/ in standard Occitan. In the Pyrenean sub-dialect of Laruns, the majority variant is the postalveolar affricate [t͡ʃ], followed by [t͡ç]; a small minority of lexical items were also realised with singleton [t] or with voiced [d͡ʒ] (see Table 6.20). In Arzacq, and to a lesser extent in the ‘Ouest’, the majority variant is palatal [t͡ç], followed by singleton [t]. Both Nay, in the central plain, and Lembeye, in the east of Béarn, have singleton [t] for the vast majority of variants of /t͡ç/, though affricated [t͡s] is also present in Lembeye, and [t͡ʃ] is attested in Nay. Nay and Lembeye are the ‘central’ dialects, where we would expect to see simplification of /t͡ç/ to [t], though we can see that this simplified form is indeed used variably in the more geographically peripheral dialects, indicating potential dialect mixing between sub-dialects. Table 6.20 variable
All native speakers: frequency distribution of voiceless affricate
Place/Variant
[t͡ç]
[t]
[t͡s]
[t͡ʃ]
[d͡ʒ]
N
Laruns Arzacq Lembeye Nay Ouest
27% 75% 0% 0% 50%
4% 25% 88% 92% 32%
0% 0% 12% 0% 0%
65% 0% 0% 8% 18%
4% 0% 0% 0% 0%
23 24 24 24 22
145
6 Language Death: Gascon and French
Table 6.21 All native speakers: regression analysis of voiceless affricate variable, with variants [plosive] and [affricate]; ‘speaker’ and ‘word’ as random effects Dependent variable = plosive/affricate Response variant = plosive ([t]) N = 116
R2 : 0.928 Degrees of freedom: 9
Factor group
Factor
Coefficient
N
p-value
Place of origin
Nay Lembeye Ouest Arzacq Laruns Male Female
+6.359 +5.527 –2.455 –3.402 –6.029 + 2.282 –2.282
24 24 22 24 22 58 58
0.00
Speaker sex
0.00
To examine the distribution of affricated variants and singleton [t], the data were submitted to statistical analysis (see Table 6.21); place of origin was returned as a significant predictor of the occurrence of [t], with the plosive realisation favoured in the sub-dialects of Nay and Lembeye, confirming the observations above. When [t] occurs in the other sub-dialects, it is strongly favoured by male speakers. If the [t] variant is adopted from the central dialects via a wave-like diffusion model, effectively as a change from above, then this result counters traditional conceptions of the role of women in language change. It is worth noting that, across sub-dialects, [t] is the majority variant in Béarn (49%) and therefore the adoption of this central variant into peripheral dialects may be the result of levelling ‘proper’.
6.3.3 Voiced Affricate For the voiced palatal affricate /d͡ʝ/, the palatal [d͡ʝ] variant is dominant in Lembeye (100%) and Arzacq (97%), with 3% of occurrences simplified to singleton [d] in Arzacq (see Table 6.22). In central Nay, and in the ‘Ouest’ sub-dialect, both palatal [d͡ʝ] and postalveolar [d͡ʒ] occur readily, with a preference for the palatal variant in Nay. In the Pyrenean subdialect of Laruns, palatal [d͡ʝ] is not attested; [ʒ] is the majority variant (73%), itself a simplified form of the postalveolar affricate [d͡ʒ] (27%).
146
D. Mooney
The lenited [ʒ] variant is also attested in a small number of cases in the ‘Ouest’, and /d͡ʝ/ is further lenited to small number of occurrences of [j] in Nay. It is striking that geographical variation for the voiced affricate / d͡ʝ/ is much more localised than for /t͡ç/, where a higher level of dialect mixing was evidence between varieties. The equivalent of /d͡ʝ/ in Occitan is /t͡ʃ/, realised as [t͡ʃ], and it is worth noting that this variant is not attested in the Gascon of native speakers in Béarn. The regression model in Table 6.23 makes a distinction between the traditional palatal variant [d͡ʝ] and postalveolar realisations, recoding [d͡ʒ] and [ʒ] together as ‘postalveolar’; [d] and [j] were excluded from the analysis. The voiced palatal affricate [d͡ʝ] is unsurprisingly shown by the model to be highly favoured in the sub-dialects of Arzacq and Lembeye. All other sub-dialects show a preference for postalveolar variants; indeed, for Laruns, they are categorical. For Nay and the ‘Ouest’ Table 6.22 able
All native speakers: frequency distribution of voiced affricate vari-
Place/Variant
[d͡ʝ]
[d͡ʒ]
[ʒ]
[d]
[j]
N
Laruns Arzacq Lembeye Nay Ouest
0% 97% 100% 61% 47%
27% 0% 0% 36% 47%
73% 0% 0% 0% 6%
0% 3% 0% 0% 0%
0% 0% 0% 3% 0%
26 30 29 33 30
Table 6.23 All native speakers: regression analysis of voiced affricate variable, with variants [palatal] and [postalveolar]; ‘speaker’ and ‘word’ as random effects Dependent variable = palatal/postalveolar Response variant = postalveolar N = 146
R2 : 0.999 Degrees of freedom: 9
Factor group
Factor
Coefficient
N
p-value
Place of origin
Laruns Ouest Nay Arzacq Lembeye
+76.261 +41.637 +37.770 –70.688 –84.979
26 30 32 29 29
0.00
Non-significant factor groups were: speaker sex (p = 0.147) and preceding phoneme (p = 0.974)
6 Language Death: Gascon and French
147
sub-dialects, we see a much more mixed picture, though both of these dialects fall geographically between the southern Laruns, on one hand, and northern Arzacq/north-eastern Lembeye, on the other. This indicates that the forms of the peripheral north and south sub-dialects are increasingly mixed in the centre of the region. This is in contrast to the potential diffusion outwards from the centre of the singleton [t] variant of the /t͡ç/ phoneme.
6.3.4 Voiced Apical Plosive /d/ Table 6.16 outlines the frequency distribution for intervocalic /d/ (see Sect. 6.2.3). Standard Occitan has /z/ in this position, from Latin -d-, -c-, and -s- (Wheeler, 1988: 251), but the majority of Gascon dialects have /d/. In Béarn, Lembeye is oft-cited as having [z] for Gascon /d/. In Arzacq, only [d] or the expected lenited variant [ð] are used; indeed, these are the majority realisations in all sub-dialects, including Lembeye. There is some lexico-phonological variation where /d/ is realised as [ɡ] or lenited [G], though these variants occur predominantly in the words credut ‘believe pst.pctp.m.sg ’ and sedut ‘sit pst.pctp.m.sg ’, which have lexical variants cregut and segut, respectively. As such, these are not actually variants of /d/. In Lembeye, it is striking that [z] occurs only 8% of the time, with [j] (9%) and [t] (1%) also occurring, the former indicating further lenition and the latter fortition. Overall, the highest level of variability can be seen in Lembeye, but it is clear from this distribution that levelling ‘proper’ between the sub-dialects of Gascon in Béarn has favoured the loss of highly localised [z] for /d/ in Lembeye. Excluding [ɡ], [G], and the isolated [t] variant, Table 6.24 presents a regression model for the distribution of dental ([d] and [ð]) and palatalised ([z], [ʒ], [j]) variants. As expected, the Lembeye sub-dialect significantly favours the use of palatalised variants, with Arzacq and the Ouest not using palatalised variants at all. In the sub-dialects in which they occur, palatalised variants are favoured by female speakers; for Lembeye, this would mean that the female speakers are more conservative, vis-à-vis the localised norm, than the male speakers. This observation counters the assumption that, in dialectology, non-mobile
148
D. Mooney
Table 6.24 All native speakers: regression analysis of voiced apical plosive variable, with variants [apical] and [palatalised]; ‘speaker’ and ‘word’ as random effects Dependent variable = dental/palatalised Response variant = palatalised N = 281
R2 : 0.98 Degrees of freedom: 9
Factor group
Factor
Coefficient
N
p-value
Place of origin
Lembeye Nay Laruns Arzacq Ouest Female Male
+10.076 +6.900 +6.842 –11.797 –12.021 +0.928 –0.928
60 48 57 59 57 136 145
0.00
Speaker sex
0.01
Non-significant factor groups were: syllable type (p = 0.06)
older rural males speakers (NORMs) exhibit the most historically appropriate or conservative linguistic forms. It is also possible, however, since the use of the dying language is heavily restricted to male speakers, that female speakers are not using the language on a daily basis and are therefore not exposed to the levelling pressures that may be active between sub-dialects.
6.3.5 Voiced Postalveolar Fricative /ʒ/ The frequency distribution of /ʒ/ confirms (see Sect. 4.2.1) that [ʒ] is the most common variant in the Pyrenean valleys (Laruns), with high levels of occurrence in west-central Béarn (Nay and the ‘Ouest’). The palatal approximant [j] is the majority variant in Lembeye, Arzacq, and to a lesser extent, in Nay and the ‘Ouest’. The alverno-méditerranéen (and standard Occitan) equivalent of Gascon /ʒ/ is the phoneme /d͡ʒ/; the [d͡ʒ] variant is attested 21% of the time in Laruns, though this likely a fortified or older form of [ʒ], as opposed to being the result of diffusion of the Occitan norm from elsewhere. The affricate variant is also realised as palatalised [d͡ʝ] in all but the Laruns sub-dialect. This palatalisation of Occitan /d͡ʒ/ may be analogical in that it reproduces the phonetic realisation of Gascon /d͡ʝ/; indeed, the [d͡ʝ] variant of /ʒ/ is
6 Language Death: Gascon and French
149
Table 6.25 All native speakers: frequency distribution of voiced postalveolar fricative variable ͡ ͡ Place/variant [ʒ] [j] [dʝ] [d͡ʒ] [tʃ] N Laruns Arzacq Lembeye Nay Ouest
75% 4% 2% 19% 31%
4% 75% 83% 63% 59%
0% 21% 13% 12% 2%
21% 0% 2% 6% 6%
0% 0% 0% 0% 2%
52 56 60 57 58
most common in Arzacq, where /d͡ʝ/ is realised as [d͡ʝ] categorically (see Table 6.22). There was also one instance of fortified [t͡ʃ] in the ‘Ouest’ sub-dialect. For /ʒ/, in Gascon, [ʒ] and [j] has long been cited as the primary variants in Béarn, but it is clear from this analysis that affricated variants account for a substantial proportion of the observations across sub-dialects (Table 6.25).
6.3.5.1 Voiceless Apical Fricative /-s/ The final /-s/ variant only occurs in a restricted set of lexical items, including james ‘never’ and mes ‘but’, which are considered here (see Sect. 4.2.3). Final /-s/ is realised predominantly as [-s] in Pyrenean Laruns, central Nay, and north-eastern Lembeye, though [-j] is also attested. The distribution of [-s] and [-j] are relatively equally distributed in Arzacq and the ‘Ouest’ sub-dialect. As such, no sub-dialect exclusively uses one variant and it appears that there is dialect mixing, with a preference for [-s] in certain areas. There is minor evidence for palatalisation of /-s/ to [-ʃ] in Lembeye and affrication to [-t͡s] in Laruns, but token counts are low overall. A statistical analysis of this frequency distribution revealed that the variation observed could not reliably be predicted by sub-dialect (p = 0.12), indicating that place of origin of the speaker does not influence the choice of variant (Table 6.26).
150
D. Mooney
Table 6.26 All native speakers: frequency distribution of voiceless apical fricative variable ͡ Place/Variant [s] [j] [ʃ] [ts] N Laruns Arzacq Lembeye Nay Ouest
6.4
73% 50% 58% 73% 45%
18% 50% 33% 27% 55%
0% 0% 9% 0% 0%
9% 0% 0% 0% 0%
11 12 12 11 11
Interim Summary and Discussion
The seven-term stressed oral vowel system was shown to be remarkably stable in the speech of the Gascon speakers from Béarn. In particular, there was no evidence for a merger of /e/ and /E/, though the actual phonetic realisation of the distinction between /e/ and /E/ was conditioned extralinguistically by the geographical origin of the speaker within the region of Béarn. As such, there was no evidence for externally motivated contact phenomena in the Gascon stressed oral vowel system such as, for example, a merger of the front mid-vowels due to contact with (supralocal) French, which neutralises this contrast variably. The pronunciations of /a/ and /e/ were also shown to vary over geographical space on the basis of F2 and F3, respectively. The phonetic distinction between the unstressed post-tonic vowels /-e/ and /-O/ was also maintained across the sub-dialects in Béarn, with the exception of Arzacq-Arraziguet, where the merger of these phonemes to approximately [œ] was confirmed by the analysis. This merger is a classic change from below in Gascon, though it is not a recent development as the Gascon spoken in Arzacq shares this well-established feature with the adjacent parlar negre variety. There is also robust evidence for externally motivated linguistic transfer from French and, crucially, for language-internal constraints on this transfer. In the case of /r/, transfer from French results in systemic complexification in Gascon, with variable rules governing the distribution of the historically appropriate rhotics, [r] and [ɾ], as well as historically inappropriate [ʁ]. The analysis of the Gascon palatal lateral /ʎ/ also established an increased probability of replacement of [ʎ] by [j] in syllabic environments where Gascon /ʎ/ was phonemically cognate
6 Language Death: Gascon and French
151
with French /j/. As such, the realisation of /ʎ/ as [j] is also governed by a variable rule which introduces more complexity into the system. Transfer from French in Gascon, at least for these variables, is not characterised by systemic simplification, by the wholesale loss of structural elements, or by an increase in regularity, as we might expect in situations of language death. Instead, the outcome of transfer from French in Gascon is systemic complexification due to the introduction of variable rules, an increase in variability, not regularity, and the introduction of new, historically inappropriate, structural elements, causing historically appropriate linguistic forms to become variable. There is also evidence for transfer from French for the /b/, /d/, /ɡ/, and gender marking variables, in that the presence of historically inappropriate forms was attested. This transfer also introduced variability into the phonological (and morphological) system, though there was no evidence for the wholesale or categorical replacement of a historically appropriate Gascon form with an external (French) variant. For /b d ɡ/, transfer of plosives to replace lenited intervocalic variants was not consistent across phonemes, both in terms of the frequency of inappropriate forms and in terms of constraints on transfer. Transfer rates were highest for /d/ but there were no constraints identified; transfer rates were lower for /b/ and /ɡ/ and historically inappropriate plosive variants were favoured by internal factors (following and preceding phonological environment, respectively). Again, the existence of linguistic material from French in Gascon complexifies the system by introducing variability but, for these variables, the constraints on transfer were less clear. It could be argued, however, that the variable transfer for plosive realisations from French to replace lenited variants of /b d ɡ/ constitutes a partial generalisation of unmarked categories, as has often been attested in situations of language death. As Gascon contracts under pressure from French, there is also evidence for language change mechanisms associated with the processes of regional dialect levelling and koinéization, namely levelling ‘proper’, diffusion, and simplification. The resultant forms are not due to transfer from the dominant language, French, but due to contact and mixing between dialects of Gascon and, potentially, due to contact with other Occitan varieties. As noted above, the localised phonological merger of post-tonic
152
D. Mooney
vowels persists in Arzacq and there is no evidence of levelling favouring the loss of this minority form, though this may also be linked to the irreversibility of phonological mergers (Garde 1961). For the consonantal variables, however, contact between the sub-dialects of Gascon in Béarn has not led to the loss of highly localised forms, but has results instead to an increase in variability over geographical space. Highly localised variants, specific to individual sub-dialects, are not completely lost as a result of dialect contact; these variants are often now present in other subdialects which they are not traditionally associated, even if they occur with low frequency. For example, the [t] variant of /t͡ç/ has been adopted from central sub-dialects, such as Nay, into more peripheral sub-dialects, but this does not result in the categorical adoption of [t] in these subdialects; instead it results in an increase in optionality or variability for /t͡ç/, where [t] is simply added to the pool of optional variants alongside more traditional variants for that sub-dialect. For /d͡ʝ/ on the other hand, northern (Arzacq) and southern (Laruns) sub-dialects show the clearest separation in terms of their phonetic realisation, with interim, more central dialects, exhibiting mixed systems, with more variability. There is some evidence that levelling ‘proper’ has favoured a reduction in the use of [z] for /d/ in Lembeye, though there is not a complete loss of this highly localised variant and it is not used in other sub-dialects. The retention with low frequency of [z] in Lembeye was also shown to be favoured by female speakers; this means that female speakers are more conservative, vis-à-vis the localised norm, than male speakers, countering the traditional assumption that NORMs are representative of the most historically appropriate norm in situations of language death. For /ʒ/, traditional [ʒ] and [j] sub-dialects were still evident, but there was no categorical usage, indicating (limited) dialect mixing. There was also some evidence for possible diffusion of Occitan [d͡ʒ], though this may simple be a fortified or vestigial version of [ʒ]; [d͡ʝ] was also attested as a variant of /ʒ/, everywhere but Laruns, perhaps by analogy with /d͡ʝ/ via [d͡ʒ]. The overall result of mixing between [ʒ] and [j] sub-dialects, and of the introduction of [d͡ʒ] and [d͡ʝ], is an increase in variability over geographical space. Results for the /-s/ variable also indicate noncategorical usage of [-s] and [-j], with mixing between sub-dialects. The
6 Language Death: Gascon and French
153
overall outcome of dialect contact for the variables in Sect. 6.3 is not the loss of minority forms or systemic simplification, but rather the nonlocalisation of minority or localised forms, resulting in more optionality and variability over geographical space and in all sub-dialects.
References Bec, Pierre. 1973. Manuel pratique d’occitan moderne. Paris: Picard. Cardaillac Kelly, Reine. 1973. A descriptive analysis of Gascon. The Hague/Paris: Mouton. Garde, Paul. 1961. Réflexions sur les différences phonétiques entre les langues slaves. Word 17: 34–62. Jones, Mari C. 1998. Language obsolescence and revitalisation: Linguistic change in two sociolinguistically contrasting Welsh communities. Oxford: Clarendon Press. Labov, William. 1994. Principles of Linguistic Change, Vol. 1: Internal Factors. Oxford: Blackwell. Mooney, Damien. 2014. Illustrations of the IPA: Béarnais (Gascon). Journal of the International Phonetic Association 44: 343–350. Mooney, Damien. 2018. Quantitative approaches for modelling variation and change: A case study of sociophonetic data from Occitan. In Manual of Romance Sociolinguistics, ed. Wendy Bennett and Janice Carruthers, 59–87. Berlin/New York: De Gruyter. Mooney, Damien. 2019. Phonetic transfer in language contact: Evidence for equivalence classification in the mid-vowels of Occitan-French bilinguals. Journal of the International Phonetic Association 49: 53–85. Mooney, Damien, and James Hawkey. 2019. The variable palatal lateral in Occitan and Catalan: Linguistic transfer or regular sound change? Journal of French Language Studies 29: 281–303. Moreux, Bernard, and Jean-Marie Puyau. 2002. Dictionnaire français-béarnais. Monein: Éditions Pyrémonde. Oliviéri, Michèle, and Patrick Sauzet. 2016. Southern Gallo-Romance (Occitan). In The Oxford guide to the Romance languages, ed. Adam Ledgeway and Martin Maiden, 319–349. Oxford: Oxford University Press. Wheeler, Max W. 1988. Occitan. In The Romance languages, ed. Martin Harris and Nigel Vincent, 246–278. London/Sydney: Croom Helm.
7 Dialect Death: Gascon and Occitan
This chapter presents the results of the dialect death study, analysing the acoustic and auditory data for ten new Gascon speakers, or néolocuteurs, and comparing the new-speaker data to that of the thirty native speakers examined in Chapter 6. The analysis addresses in turn the oral vowel system, L1-to-L2 transfer from French, and the loss or retention of localised Gascon phonological features. The focus of this chapter will be identifying language change taking place as part of revitalisation efforts for Gascon, and more widely for Occitan; the identification of constraints on variation and change will remain central to the analysis. There is a long-standing assumption that there is a linguistic cleavage between the speech of native and new speakers of Gascon and the analyses presented here will assess the validity of this assumption empirically.
© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 D. Mooney, Language and Dialect Death, https://doi.org/10.1007/978-3-031-51101-1_7
155
156
7.1
D. Mooney
The Oral Vowel System
This section presents data for the Gascon stressed oral vowel system /i y u e E O a/ and for the phonetic realisation of the post-tonic vowels /-e/ and /-O/, comparing the results to native-speaker data presented in Sect. 6.1.
7.1.1 Stressed Oral Vowels Figure 7.1 presents the normalised F1-F2 data for the Gascon stressed oral vowel system /i y u e E O a/ for the ten new speakers, both male and female. The general distribution of the vowels in the F1-F2 plane is as expected, with some overlap between /i/, /y/, /e/, and /E/ and some notably fronted tokens of /u/. Despite apparent overlapping distributions for /e/ and /E/, the new speakers do make a significant phonetic distinction between these vowels on the F1, or vowel height, dimension, but not on the basis of F2 or F3. Table 7.1 presents the regression model of F1 values for the front mid-vowels and the historically appropriate phoneme is returned as a significant predictor of F1, indicating phonetic separation between /e/ and /E/ in terms of vowel height. Older native Gascon speakers from all five sub-dialect areas made this same distinction on the basis of F1 (see Table 6.5), with one area making an additional F2 distinction, one area making an additional F3 distinction, and two areas making significant F1, F2, and F3 distinctions for /e/ and /E/. As such, the primary distinction between the front mid-vowels of the older native speakers is replicated by the new speakers; it seems that the secondary (F2) and tertiary (F3) phonetic distinctions over geographical space are not present in the system of the new speakers. It is worth noting that the new speakers are predominantly from the regional capital, Pau, which lies in the central plain, near Nay; Nay speakers only distinguished between /e/ and /E/ on the basis of F1 or vowel height (see Fig. 6.5). Indeed, the vowel height distinction is the standard or prescribed phonetic difference between the two vowels.
157
7 Dialect Death: Gascon and Occitan
Fig. 7.1 Stressed oral vowel system for (male and female) new speakers, normalised data
Table 7.1 New speakers (normalised data): Regression analysis of F1 values for /e/ and /E/, with ‘speaker’ and ‘word’ as random effects Dependent variable = F1 N = 91 Grand mean = –0.159
R2 : 0.343 Degrees of freedom: 8
Factor group
Factor
Coefficient
N
p-value
Phoneme
/E/ /e/
+0.297 –0.297
44 44
0.01
Non-significant factor groups were: speaker sex (p = 0.119) and syllable type (p = 0.872)
To explore differences between native and new speakers in the actual pronunciation of the stressed oral vowels, regression analyses of the normalised F1, F2, and F3 data for both speaker types were undertaken for each of the seven vowels. These analyses included the following independent variables: preceding phoneme; following phoneme; speaker type; speaker sex. No statistically significant differences were established
158
D. Mooney
Table 7.2 All (native and new) speakers (normalised data): Regression analysis of F1 values for /a/, with ‘speaker’ and ‘word’ as random effects Dependent variable = F1 for /a/ N = 192 Grand mean = 1.81
R2 : 0.196 Degrees of freedom: 7
Factor group
Factor
Coefficient
N
p-value
Speaker type
Native New
+0.078 –0.078
144 48
0.05
Non-significant factor groups were: preceding phoneme (p = 0.428) and speaker sex (p = 0.812)
between the sub-dialects for /i/, /y/, /u/, /e/, /E/ or /O/. ‘Speaker type’ was, however, shown to be a significant predictor of F1 for /a/, with the results of the regression analysis presented in Table 7.2. The magnitude of the regression coefficients are small, but the significant effect returned is that /a/ has higher F1 values, or is a slightly lower vowel, for native speakers. Alternatively, in an apparent-time model, this result may indicate slight /a/-raising in the younger generation, though the effect is relatively weak. Overall, there is a remarkable stability in the stressed oral vowel system of the new speakers. Their phonetic realisations of the phonemic categories for /i/, /y/, /u/, /e/, /E/, and /O/ are statistically the same as the native speakers, with very minor differences for /a/. The new speakers also maintain phonetic separation between /e/ and /E/, even though there is a widespread tendency in supralocal varieties of their L1 to neutralise this contrast in certain syllabic environments. Even in the absence of unbroken intergenerational transmission, the new speakers of Gascon demonstrate an ability to learn the language of the older generation and to faithfully replicate the system of the native speakers (see Labov 2007: 346; Mooney 2016: 328). In Labov’s terms, the new speakers are capable of preserving ‘linguistic descent’ (2007: 346), even in the absence of parent-to-child transmission, at least for the stressed oral vowels.
7.1.2 Unstressed Oral Vowels Figure 7.2 presents the normalised data for /-e/ and /-O/; while there is some evident overlap in the distribution of the vowels, there is visual
7 Dialect Death: Gascon and Occitan
159
Fig. 7.2 Unstressed oral vowels for (male and female) new speakers, normalised data
separation between them in the expected pattern, with /-e/ being more front than /-O/. New speakers make a significant phonetic distinction between /-e/ and /-O/ on the basis of both F1 and F2, but not F3. Table 7.3 presents the results of the regression analysis of F1, with the historically appropriate phoneme as a significant predictor of the variability observed. In this model, /-O/ is shown to be significantly lower in the F1-F2 plane than /-e/, which aligns with their canonical labels as open-mid and close-mid, respectively. The results of the F2 models are presented in Table 7.4, where /-e/ is shown to be significant more front in the F1-F2 plane than /-O/, as would be expected. There is also an effect of preceding phoneme for F2, with voiced consonants favouring higher F2 values. By making a significant phonetic distinction between /-e/ and /-O/ on the basis of F1 and F2, new speakers replicate the system of the native speakers from Nay and Laruns (see Table 6.8); speakers from Lembeye and the ‘Ouest’ distinguished between the vowels on the F2 dimension
160
D. Mooney
Table 7.3 New speakers (normalised data): Regression analysis of F1 values for /-e/ and /-O/, with ‘speaker’ and ‘word’ as random effects Dependent variable = F1 N = 70 Grand mean = 0.29
R2 : 0.287 Degrees of freedom: 8
Factor group
Factor
Coefficient
N
p-value
Phoneme
/-O/ /-e/
+0.205 –0.205
43 27
0.02
Non-significant factor groups were: speaker sex (p = 0.129) and preceding phoneme (p = 0.281)
Table 7.4 New speakers (normalised data): Regression analysis of F2 values for /-e/ and /-O/, with ‘speaker’ and ‘word’ as random effects Dependent variable = F2 N = 70 Grand mean = –0.117
R2 : 0.462 Degrees of freedom: 8
Factor group
Factor
Coefficient
N
p-value
Phoneme
/-e/ /-O/ /z/ /d/ /t/
+0.141 0.141 +0.299 +0.074 –0.373
27 43 9 16 45
0.03
Preceding phoneme
0.01
Non-significant factor groups were: speaker sex (p = 0.339)
alone, and speakers from Arzacq-Arraziguet make no phonetic distinction between /-e/ and /-O/. As such, new speakers avoid the minority system used in Arzacq, characteristic of lo parlar negre, and once again replicate the system of the central plain, represented by the speakers from Nay. While the Nay system appears to be replicated faithfully, the actual pronunciation of the individual vowels was submitted to regression analysis to establish the existence of any significant differences in pronunciation between new and native speakers. For /-e/, there were no significant F1, F2, or F3 differences in pronunciation between new and native speakers, as was the case for stressed /e/ (see Sect. 7.1.1). For /-O/, speaker type was a significant predictor of F1, but not of F2 or F3. Table 7.5 presents the regression model of F1 for /-O/, showing that new speakers have significantly higher F1 values, of lower vowels,
161
7 Dialect Death: Gascon and Occitan
Table 7.5 New speakers (normalised data): Regression analysis of F1 values for /-O/, with ‘speaker’ and ‘word’ as random effects Dependent variable = F1 for /-O/ N = 168 Grand mean = 0.277
R2 : 0.314 Degrees of freedom: 7
Factor group
Factor
Coefficient
N
p-value
Speaker type
New Native
+0.133 –0.133
43 125
0.04
Non-significant factor groups were: preceding phoneme (p = 0.211) and speaker sex (p = 0.974)
than native speakers. This result may indicate increased phonetic separation between /-e/ and /-O/ in the speech of the new speakers, because of gradual /-O/-lowering. Overall, the primary front-back (F2) distinction between /-e/ and /-O/ is replicated by the new speakers, with an additional height distinction present in the speech of native speakers from both Nay and Laruns. Again, it appears that new speakers are able to accurately produce the system of older generations even in the absence of an unbroken sequence of parent-to-child transmission.
7.2
Transfer from French
Section 6.2 investigated L2-to-L1 transfer from French in speech of native Gascon speakers. The mother tongue of the new speakers is, however, French, and so an investigation of the same contact phenomena here constitutes an analysis of L1-to-L2 transfer.
7.2.1 Rhotics For the Gascon /r/ variable, new speakers produced the variants with the following frequencies (n = 215): [ɾ] (58%); [r] (9%); [ʁ] (27%); zero (6%). The majority variant is tapped [ɾ] and there is also evidence for historically inappropriate uvular [ʁ], though the latter occurs with a lower frequency than for native speakers (44% average; see Table 6.9), perhaps in an effort to avoid using variants perceived to be French.
162
D. Mooney
To examine the variable distribution of these uvular rhotics, the apical/ uvular distinction was analysed statistically (see Table 7.6), excluding the zero variants, and recoding [ɾ] and [r] as ‘apical’. Syllable type is returned as a significant predictor of the adoption of uvular rhotics from French to Gascon, which can be formalised in the following variable rule: /r/ > [ʁ]/V_C#; V_#. Uvular variants are favoured, for the new speakers, in word-final consonant clusters and simple codas. For older native speakers (see Table 6.10), uvular variants were favoured in final consonant clusters and simple codas, but also in medial clusters (V_C). For new speakers, the medial cluster syllable type significant disfavours the adoption of uvular rhotics from French; this environment has the lowest probability of observing uvular rhotics in the speech of new speakers. As such, the new speakers’ variable rule governing transfer from French to Gascon is only a partial replication of the native-speaker variable rule: /r/ > [ʁ]/V_C#; V_#; V_C. This loss of structural or systemic constraints is characteristic of imperfect learning during the process of diffusion: ‘a transportation of the general phonetic basis […], but not a faithful copy’ (Labov 2007: 360). In short, the new speakers demonstrate an ability to reproduce parts of the native-speaker system, but the lack of intergenerational transmission can lead to a situation whereby linguistic features of the system are adopted ‘off the shelf ’ (Milroy 2007), without faithful replication of associated structural and social constraints on variability (Mooney 2016: 330). Table 7.6 New speakers: Regression analysis of rhotic variable, with variants [apical] and [uvular]; ‘speaker’ and ‘word’ as random effects Dependent variable = uvular/apical rhotic Response variant = uvular N = 203
R2 : 0.553 Degrees of freedom: 10
Factor group
Factor
Coefficient
N
p-value
Syllable type
V_C# V_# C_V #_V V_V V_C
+1.349 +0.902 –0.116 –0.626 –0.733 –0.776
20 33 57 38 32 23
0.04
Non-significant factor groups were: speaker sex (p = 0.452)
163
7 Dialect Death: Gascon and Occitan
Excluding the uvular data and considering only the data for [r] and [ɾ], Table 7.7 presents the results of the regression analysis of the variable distribution of the trilled and tapped apical rhotics for the new speakers. Again, syllable type is returned as a significant predictor of the variant distribution, with trilled [r] favoured in word-initial and intervocalic position, noting that the frequency of trilled [r] is very low overall for new speakers. For native speakers, [r] is favoured in medial and final consonant clusters (V_C and V_C#) and word-initially (#_V). Both native and new speakers share the tendency for [r] to be realised word-initially, but the variable systems otherwise differ. For new speakers, tapped [ɾ] appears to be replacing [r] in its traditional (variable) environments. Overall, the systemic constraints on the transfer of [ʁ] from French are similar for new and native speakers, though the new-speaker system lacks some of the detail of the native speakers. In the environments where historically appropriate apical variants are favoured, [r] is favoured initially and intervocalically, and [ɾ] is favoured in non-final consonant clusters. The constraints governing the distribution of the apical rhotics is not a faithful replication of the traditional system of the native speakers (see Sect. 6.2.1). Table 7.7 New speakers: Regression analysis of apical rhotics, with variants [ɾ] and [r]; ‘speaker’ and ‘word’ as random effects Dependent variable = [ɾ]/[r] rhotic Response variant = [r] N = 145
R2 : 0.468 Degrees of freedom: 9
Factor group
Factor
Coefficient
N
p-value
Syllable type
#_V V_V V_# V_C# V_C C_V
+1.787 +0.408 –0.010 –0.308 –0.801 –1.075
30 26 19 10 18 42
0.04
Non-significant factor groups were: speaker sex (p = 0.425)
164
D. Mooney
7.2.2 Palatal Lateral For the /ʎ/ variable, new speakers produced the variants with the following frequencies (n = 165): [ʎ] (25%); [j] (72%); [l] (3%). The rate of transfer of [j] from French is higher than the average of 64% for native speakers (see Table 6.12). A small number of tokens (n = 6) were depalatalised to [l] and these have been excluded from the subsequent analysis. To investigate the existence of constraints on the transfer of [j] from French, the [ʎ] and [j] data were submitted to regression analysis, the results of which are presented in Table 7.8. The significant effect of syllable type shows that [j] is favoured intervocalically, in final clusters with /-s/ and in word-final position, whereas [ʎ] is favoured word-initially. In Sect. 6.2.2, it was established that word-initial /ʎ/ in Gascon is phonemically cognate with /l/ in French, while intervocalic and wordfinal /ʎ/ are phonemically cognate with /j/ in French. For native speakers, intervocalic and final positions were shown to favour the transfer of [j] from French (see Table 6.13) and the word-initial position favoured retention of [ʎ]. One anomaly in the analysis of the native-speaker data was that the V_C# environment, when /ʎ/ was in a final cluster with /-s/ was also shown to favour retention of [ʎ], even though the phonemic cognate of Gascon /ʎ/ in a final cluster with /-s/ is not [j] in French, but zero e.g., yeux /jø/ ‘eyes’. This anomaly appears to be resolved in the speech of the new speakers, where V_C# also favours [j], increasing Table 7.8 New speakers: Regression analysis of palatal lateral variable, with variants [ʎ] and [j]; ‘speaker’ and ‘word’ as random effects Dependent variable = palatal lateral Response variant = [j] N = 159
R2 : 0.759 Degrees of freedom: 7
Factor group
Factor
Coefficient
N
p-value
Syllable type
V_V V_C# V_# #_V
+2.461 +1.426 +1.363 –5.251
56 43 44 16
0.01
Non-significant factor groups were: speaker sex (p = 0.138)
7 Dialect Death: Gascon and Occitan
165
the regularity of the system and generalising the high-probability use of [j] to this environment. As for the rhotic variable, the new speakers reproduce broadly the system of the native speakers but, in this case, the system is characterised by a simplified variable rule: /ʎ/ > [ʎ]/#_V, with [j] elsewhere.
7.2.3 Voiced Plosives Gascon /b d ɡ/ are traditionally lenited in intervocalic position, either alone or in a cluster with [ɾ r l z] (Oliviéri and Sauzet 2016: 325). The existence of plosives, as opposed to lenited variants, in this position has been argued to constitute transfer from French (see Sect. 6.2.3). For the /b/ variable, new speakers produced the variants with the following frequencies (n = 115): [b] (77%); [β] (21%); [p] (1%); [w] (1%). The majority variant for these speakers is historically inappropriate [b], which occurs as a much higher rate than for native speakers (average 45%; see Table 6.14). As such, the raw frequency data for /b/ indicate that there is more transfer from French in speech of the new speakers. While there are some minority variants, namely fortified [p] and vocalised [w], there are less variants overall when compared with the frequency distribution for native speakers (see Table 6.14). A regression analysis of plosive [b] versus lenited [β] or [w], and excluding [p], revealed no effect of syllable type (p = 0.15), preceding phoneme (p = 0.30), following phoneme (p = 0.78), or speaker sex (p = 0.88). Therefore, while the frequency of transfer appears to be higher for /b/ for new speakers, there were no clear constraints on this transfer. By contrast, for native speakers, transfer was favoured by certain following phonological environments such that /b/ > [b]/V_/e/; V_/y/; V_/u/ (see Table 6.15). For the /d/ variable, new speakers produced the variants with the following frequencies (n = 103): [d] (76%); [ð] (21%); [ɡ] (1%); [G] (1%); [z] (1%). Again, the frequency of the historically inappropriate [d] variant is much higher than for the native speakers (average 50%; see Table 6.16). [ɡ] and [G] variants have again been excluded from this analysis as they are not lenited variants of /d/, but occur in lexical variants
166
D. Mooney
of credut ‘believe pst.pctp.m.sg ’ and sedut ‘sit pst.pctp.m.sg ’ (see Sect. 6.3.4). As such, for new speakers, the primary distinction is between transferred [d] and lenited ([β] or [z]). Overall, new speakers demonstrate much less variability than native speakers, for whom [t], [ʒ], and [j] were also variants (see Table 6.16). The regression analysis for [d] versus lenited ([β] and [z]) returned no significant predictors of the variability observed, as was the case for the native speakers, indicating that there are no structural constraints on transfer from French. Finally, for the /ɡ/ variable, new speakers produced the variants with the following frequencies (n = 113): [ɡ] (55%); [G] (24%); [ʃ] (13%); [d͡ʒ] (3%); [ʒ] (3%); [d͡ʝ](1%); [k] (1%). As for /b/ and /d/, there is an increase in the frequency of the historically inappropriate plosive [ɡ] (55%), compared with an average of 19% for native speakers (see Table 6.17). The [ʃ] variant occurred primarily in the target word paregut ‘appearpst.pctp.m.sg ’, which has the lexical variant pareissut, and so has been excluded from this analysis. Various palatalised forms were also noted, as well as fortified [k]; all of these forms were attested in the speech of the native speakers, with the exception of [d͡ʝ] (n = 1) which was realised by Speaker AS. Overall, there are less possible variants for the new speakers, when compared with the native speakers, but not to the same extent as for /b/ and /d/. The regression analysis of [ɡ] versus lenited [G] revealed no significant constraints on the distribution of the variants, and therefore no constraints on the transfer for [ɡ] from French. In sum, the speech of the new speakers is characterised by a marked increase in transfer from French, with [b d ɡ] replacing [β ð G] in intervocalic position, though not categorically. What distinguishes new and native speakers is the frequency of occurrence of the plosive variants and, in the case of /b/ at least, the loss of structural constraints on transfer in the speech of the new speakers. For /b/ and /d/ in particular, the new speakers demonstrate markedly reduced variability or optionality for the variables considered; focusing, paradigm simplification, and the reduction in exponents of phonological categories are hallmarks of levelling ‘proper’. In apparent-time, new speakers appear to be leading an externally motivated change that involves the replacement of intervocalic lenited forms with plosives at the same place of articulation.
7 Dialect Death: Gascon and Occitan
167
7.2.4 Gender Marking For the gender marking variable, new speakers produced correct gender 64% of the time and the incorrect gender 36% of the time (n = 147). The correctness score for new speakers is lower than the average score of 75% for native speakers (see Table 6.19); this difference is statistically significantly different (p = 0.03), with speaker and word as random effects; native speakers are more likely to know the correct nominal gender than new speakers. Of the 16 words included in the analysis, only three were produced with the correct gender 100% of the time: un ahar ‘matter’; la lèbe ‘hare’; la sau ‘salt’. For the native speakers, un ahar and la lèbe were also 100% correct and la sau was realised correctly as feminine 97% of the time. Four further words had high correctness scores, ranging from 80 to 100%: la sang ‘blood’ (90%); lo limac ‘slug’ (88%); la léit ‘milk’ (80%); la sèrp ‘snake’ (80%). The scores for these words are comparable to those of the native speakers, with the exception of la sang, which native speakers realised correctly as feminine only 44% of the time. Six words had correctness scores in the 40–80% range: un arrelòtge ‘clock’ (78%); lo hum ‘smoke’ (70%); lo grèish ‘fat’ (56%); lo deute ‘debt’ (50%); lo par ‘pair’ (44%); ua ungla ‘nail’ (44%). The scores for these words were markedly lower than the equivalent scores for native speakers (see Sect. 6.2.4). The three lowest correctness scores were for lo cremalh ‘house-warming’ (17%), la bilheta ‘ticket’ (10%), and la mensonja ‘lie’ (10%). The influence of French is much clearer for the new speakers, with higher rates of transfer of grammatical gender from French and, therefore, a higher frequency overall, as seen for the voiced plosive variables in Sect. 7.2.3.
7.3
Dialect Levelling
This final section of the dialect death study examines the use, loss, and retention, in the speech of the new speakers, of geographically localised phonetic and phonological variation. The analysis examines the use of general Gascon features, standardised Occitan features, and the variable presence of highly localised forms traditionally limited to certain
168
D. Mooney
sub-dialects of Gascon in Béarn. The speech of new speakers is often characterised, at least ideologically, by the use of forms that are perceived to be standardised (see Sect. 2.6) but Jones (1998: 154) does note that, for new speakers with near native fluency, as is the case here, their speech may contain phonological dialectalisms, often dependent on the variety spoken by their teacher or by the specific educational context in which they acquired the language. It is also possible, of course, that contact between new speakers, and indeed contact between new and native speakers, results in dialect mixing of the type identified in Sect. 6.3.
7.3.1 Unstressed Oral Vowels The contrast between the unstressed vowels /-e/ and /-O/ is stable in the speech of the new speakers and there is no evidence for the adoption of highly localised patterns, such as the merger to [œ] in Arzacq. New speakers distinguish between these vowels in terms of both vowel height (F1) and vowel frontness/backness (F2); this phonetic distinction aligns with the native speakers from Nay, in the central plain, and Laruns, in the Pyrenean Vallée d’Ossau. The primary F2 distinction between the unstressed vowels is the majority form among the sub-dialects of the native speakers and, because of this, new speakers can be seen to replicate the statistical norm in Béarn, avoiding highly localised systems.
7.3.2 Voiceless Affricate For /t͡ç/, new speakers produced the variants with the following frequencies (n = 38): [t͡ʃ] (44%); [t] (34%); [t͡ç] (16%); [t͡s] (3%); [E͡w] (3%). The majority variant for /t͡ç/ is the voiceless postalveolar affricate [t͡ʃ]. This form was never used by speakers from Arzacq and Lembeye and was used only 18% of the time by speakers from the ‘Ouest’ and 8% by speakers from Nay. [t͡ʃ] was, of course, the majority form in Laruns (65%). The singleton [t] variant also occurs in the speech of the new speakers with high frequency; this variant was the dominant variant in Lembeye (88%) and Nay (92%) and was attested
7 Dialect Death: Gascon and Occitan
169
variably in all sub-dialects; indeed, in terms of frequency, singleton [t] is the majority variant in Béarn as a whole. The traditional local form, or the voiced palatal affricate [t͡ç], is used with low frequency by the new speakers; this was never used by speakers from Lembeye or Nay and was used as the dominant form in Arzacq (75%). Standard Occitan would have /-l/ or vocalised [-w] in this position, e.g., Gasc. martèth ‘hammer’, Oc. martèl [maɾtEl]/[maɾtE͡w]. There is limited evidence for this and only one new speaker, from Pau, produced one token of [E͡w]. Overall, new speakers are using localised Gascon forms [t͡ʃ], [t], and [t͡ç], but forms are drawn from all sub-dialects. This is more characteristic of dialect mixing, rather than levelling, as levelling would promote the majority form, [t]. It is also possible that the [t͡ʃ] variant is an attempt to realise [t͡ç], but that transfer from French causes /t͡ç/ to be produced as an instance of French /t͡ʃ/, which itself is borrowed from English. It is additionally possible that the [t͡ʃ] variant constitutes transfer from Occitan: Occitan has a phoneme /t͡ʃ/, which corresponds to Gascon /d͡ʝ/ (see Sect. 7.3.3), but not to Gascon /t͡ç/. As such, it is possible that the prevalence of [t͡ʃ] for /t͡ç/ involves a set of phonetic realisation rules that cause /t͡ç/ to be produced (phonetically) as an instance of the nearest L1 (French) category or as an instance of the phonetically nearest standard Occitan category.
7.3.3 Voiced Affricate For /d͡ʝ/, new speakers produced the variants with the following frequencies (n = 53): [d͡ʝ] (49%); [d͡ʒ] (31%); [t͡ʃ] (9%); [t͡ç] (9%); [ʒ] (2%). There is limited evidence for diffusion of standard Occitan [t͡ʃ] (see Sect. 7.3.2) and there are some [t͡ç] variants attested. The [t͡ç] variant is either a voiceless variant of [d͡ʝ] or a hypercorrect palatalised variant of [t͡ʃ]; Sect. 7.3.2 established the possibility that standard Occitan /t͡ʃ/ and French /t͡ʃ/ may be equated with Gascon /t͡ç/, causing them to be realised incorrectly, or historically inappropriately, as instances of each other even though they are not phonemic cognates: During the acquisition of an L2, in this case Gascon, the
170
D. Mooney
process of ‘interlingual identification’ (Flege 1995: 98) causes words learnt in the second language to be (initially) decomposed into categories based on familiar L1, or French, sounds (Mooney 2019: 54). In any case, the majority variants for the new speakers are palatal [d͡ʝ] and, to a lesser extent, postalveolar [d͡ʒ]. For the native speakers, [d͡ʝ] was also the majority variant, used exclusively in Lembeye and almost always in Arzacq (97%); [d͡ʝ] was also used with high frequency in Nay (61%) and the ‘Ouest’ (47%), but never in Laruns, where the majority variant was [ʒ] (73%). A small number of new-speaker tokens (n = 3) were realised as [ʒ], though this is potentially due to transfer from French because French /ʒ/ corresponds to Gascon /d͡ʝ/, e.g., Fr. marriage /maʁjaʒ/ ‘marriage’, Gasc. maridatge /maɾiˈdad͡ʝe/. Overall, the newspeaker realisations of /d͡ʝ/ are predominantly localised Gascon forms, with a preference for the form most common across sub-dialects of Gascon in Béarn, [d͡ʝ]. As such, while there is evidence for dialect mixing, as was the case for /t͡ç/, the data for /d͡ʝ/ are more consistent with levelling ‘proper’ in that it favours the retention of majority forms.
7.3.4 Voiced Apical Plosive /d/ For the /d/ variable, new speakers produced the variants with the following frequencies (n = 103): [d] (76%); [ð] (21%); [ɡ] (1%); [G] (1%); [z] (1%). As discussed in Sect. 7.2.3, the intervocalic allophone of /d/, namely [ð], is variable replaced by plosive [d] due to transfer with French; the frequency of this transfer is markedly higher than for native speakers. Additionally, the [ɡ] and [G] variants are lexically defined and not variants of /d/. There is only one token of [z], produced by Speaker AU in the word pedolh [pezuʎ] ‘louse’; this variant is, as previously noted, characteristic of the Vic-Bilh area of Béarn and specifically of the dialect of Lembeye. Overall, for the /d/ variable, the speech of native speakers is characterised by high levels of transfer from French and by a decrease in variability overall, the first being an outcome of language contact and the second of potential levelling or focusing over time.
7 Dialect Death: Gascon and Occitan
171
7.3.5 Voiced Postalveolar Fricative /ʒ/ For the /ʒ/ variable, new speakers produced the variants with the following frequencies (n = 92): [j] (58%); [ʒ] (26%); [d͡ʒ] (13%); [d͡ʝ] (3%). [j] was also shown to be the majority variant for all sub-dialects, with the exception of Laruns, for which [ʒ] was the majority variant (see Table 6.25); the [ʒ] variant was also, of course, present variably in all other dialects. The [d͡ʒ] and [d͡ʝ] variants were also attested as minority variants in the speech of the older native speakers, with [d͡ʒ] primarily in Laruns and [d͡ʝ] primarily in Arzacq, Lembeye, and Nay. The new speakers of Gascon appear to reproduce the system of the native speakers, with comparable variants used and a relatively comparable frequency distribution. It is worth noting, however, that Gascon /ʒ/ corresponds in the phonemic diasystem to standard Occitan /d͡ʒ/. In Sect. 6.3.5, it seemed prudent to suggest that the presence of [d͡ʒ] in the speech of the Laruns speakers was more likely to be an instance of fortition of [ʒ] than diffusion of the Occitan norm. For these new speakers, however, we must acknowledge that the use of [d͡ʒ] may be indicative of standardisation and, perhaps, that [d͡ʝ] is a hypercorrect form of [d͡ʒ], due to analogy with Gascon /d͡ʝ/.
7.3.6 Voiceless Apical Fricative /-s/ Finally, for /-s/, new speakers produced the variants with the following frequencies (n = 20): [-s] (50%); [-j] (50%). The minority variants, [ʃ] and [t͡s], reported for the native speakers are not present in the speech of the new speakers (see Table 6.26). The [-s] variant was in the majority when all sub-dialects were considered together and the place of origin of the native speakers did not significantly predict the use of [-s] and [-j]. Since the use of [-s] and [-j] are not highly localised in Béarn, and therefore not associated with specific sub-dialects, the new speakers appear to use a system where the [-s] and [-j] variants are in free variation. When we look more closely at the lexical items, however, [-s] is used 100% of the time in the word mes ‘month’ and [-j] is used 100% of the time in
172
D. Mooney
the word james ‘never’. This was not the case for the native speakers, for whom both variants occurred in both words, but [-j] was the majority form for james (63%) and [-s] was the majority form for mes (85%). This system of variability in the speech of the native speakers is subject to simplification in the speech of the new speakers, where the variable elements [-s] and [-j] have become lexicalised in the words mes and james, respectively.
7.4
Interim Summary and Discussion
Gascon is classified as stage 8b ‘Nearly Extinct’ on the Expanded Graded Intergenerational Disruption Scale (EGIDS), originally developed by Fishman (1991): ‘The only remaining speakers of the language are members of the grandparent generation or older who have little opportunity to use the language’ (Lewis and Simons 2010: 110). The existence of new speakers of Gascon, in the form of Occitan, challenges this classification in that there are now younger fluent speakers of the language who have acquired it in educational contexts. The EGIDS, however, has as its central focus the maintenance or loss of intergenerational transmission, which is now completely absent. Labov (2007) defines ‘transmission’ in terms of a family tree model, stating that the continuity of languages is the result of children’s ability to learn the language of older generations, including structural and social constraints, and to reproduce faithfully this language (Labov 2007: 346). It is this unbroken sequence of nativelanguage acquisition by children that Labov terms transmission and, in the strictest application of the term, this ‘unbroken sequence’ is absent in the Gascon speech community. The analysis of the stressed oral vowels system showed, however, that new speakers have faithfully replicated the phonological system of the native speakers and that, with the possible exception of slightly lowering /a/, the new speakers’ actual pronunciations of all vowels in the system are not significantly different from those of the native speakers. Additionally, the new speakers’ phonetic separation of the front mid-vowels seems to replicate the system of speakers from Nay, in the central plain, adjacent to Pau, where most new speakers live. Despite the absence of intergenerational transmission and
7 Dialect Death: Gascon and Occitan
173
of native-language acquisition by children, the new speakers’ oral vowel system provides evidence for the ability of new speakers to reproduce the native-speaker patterns with a high degree of structural accuracy. For the unstressed vowels, new speakers again largely reproduce the majority system of the native speakers and the phonetic distinction between the vowels is characteristic of native-speaker varieties in the central plain. Where Labov (2007) defined the transmission process in terms of a family tree model, he says that diffusion involves the transfer of linguistic features across branches of the family tree (Labov 2007: 347). Labov views the process of diffusion as secondary to transmission in that diffusing features are seen to replace traditional dialectal features which have been passed down from parent to child (2007: 347). For Labov, the process of diffusion leads to a loss of structural and social constraints on variability, because of the imperfect language-learning abilities of adults, making a distinction between ‘the nearly error-free transmission from parent to child, and the less accurate diffusion across the population’ (Labov 2014: 1). The analysis of the Gascon /r/ variable in the speech of the new speakers showed that, while new speakers replicate the general system of the native speakers, some structural constraints on transfer of uvular [ʁ] from French are lost. As noted in Sect. 6.2.1, this loss of structural or systemic constraints is characteristic of imperfect learning during the process of diffusion. Additionally, the systemic constraints on the distribution of the historically appropriate apical rhotics were not a faithful replication of the native-speaker system. What new and native speakers did have in common, however, is that the transfer of uvular [ʁ] from French increases variability in the system due to the introduction of a variable rule for transfer. The existence of phonemic cognates in the languages of a bilingual has been evoked at points in this chapter as providing a basis for transfer of linguistic material from one language to another. The basis of this argument is that transfer will normally arise when a bilingual identifies a phoneme of the secondary system with one of the primary system and, in reproducing it, subjects it to the phonetic rules of the primary language (Weinreich 1968: 14). Flege’s (1988, 1990, 1991) Speech Learning Model (SLM) combines, in a single explanatory framework, the linguistic mechanisms by which sounds or units of sound may transfer
174
D. Mooney
from one language to another during bilingualism. During the acquisition of an L2, the process of ‘interlingual identification’ causes words learnt in the second language to be (initially) decomposed into categories based on familiar L1 sounds (Flege 1995: 98). Phonemes in the L1 and L2 that correspond structurally are termed ‘cognate phonemes’ and are said to be linked by ‘equivalence classification’, which may lead to ‘phonetic category assimilation’. When a new category is not formed for an L2 sound because it is ‘too similar’ to an L1 counterpart, the L1 and L2 sounds will assimilate, leading to a ‘merged’ L1-L2 category (Flege 2005). Within the merged category, the L2 sound will continue to resemble the L1 sound, and the L1 sound will begin to resemble the L2 sound (Flege 2007: 368). In Sect. 6.2.2, transfer from French was shown, for native speakers, to be governed by a system of constraints that favoured [j] over [ʎ] in syllabic positions where Gascon /ʎ/ was phonemically cognate with French /j/ in the bilingual grammar; where Gascon /ʎ/ was cognate with French /l/, the [ʎ] variant occurred with a higher probability. This argument was somewhat complicated by the fact that, when Gascon /ʎ/ was cognate with nothing in French, in final clusters with /-s/, [ʎ] was also favoured. The general pattern of constraints on the transfer of [j] from French was also found in the speech of the new speakers, but they were also shown to favour [j] in the anomalous environment, corresponding to French zero. As such, the new speakers faithfully replicate the basic system of the native speakers, but the new speakers’ system resolves the anomaly in final consonant clusters and is governed by a simple variable rule that favours retention of [ʎ] in initial position and transfer of French [j] in all other positions. As for the rhotic variable, transfer from French also increases variability in the system but, for /ʎ/, the exceptionless variable rule simplifies the original constraints on transfer from French; this is evidence of both structural simplification and of an increase in regularity. It may, however, also be considered imperfect learning of the structural constraints, characteristic of the process of diffusion. The analysis of the voiced plosive variables for the new speakers revealed, for /b/, /d/, and /ɡ/, a general increase in historically inappropriate plosive forms, transferred from French, in intervocalic position. For /b/ and /d/ in particular, there was a marked decrease in variability
7 Dialect Death: Gascon and Occitan
175
when compared with the native speakers’ frequency distributions. This reduction in variability is consistent with the linguistic focusing mechanism of levelling ‘proper’, where new speakers appear to have reduced the number of possible variants of /b/ and /d/ in this position. There was also some evidence for levelling of variants of the /ɡ/ variable, but this was less extensive. Additionally, for /b/, new speakers did not replicate the systemic constraints on transfer from French that were present in the speech of the native speakers (see Table 6.15). For the gender marking variable, there was also an increase in transfer from French, aligning with /b d ɡ/ in this respect; new speakers were shown to be less likely to know the historically appropriate grammatical gender of the noun than native speakers and this difference was statistically significant. The Gascon new speakers were shown, in Sect. 7.3, to use localised historically appropriate Gascon forms for the majority of the variables considered; they had not completely abandoned Gascon forms in favour of ‘standard’ Occitan, or Lengadocian, forms, as is often suggested anecdotally. For the /t͡ç/ variable, for example, Occitan [-l] was never used and vocalised [E͡w] was used in one token. The majority of the variants used were well-attested Gascon forms, notably [t͡ç], [t], and [t͡ʃ], drawn from across the sub-dialects in Béarn. There was no consistent evidence for levelling, which would have promoted [t] as the majority form in the speech of the native speakers. Instead, the frequency distribution of /t͡ç/ for the new speakers is more consistent with dialect mixing and therefore with the maintenance, not the reduction, of variability. Unlike most native speakers, the new speakers showed a preference for the [t͡ʃ] variant; this variant does occur with high frequency in the native-speaker sub-dialect of Laruns (see Table 6.20), though its high frequency in the new-speaker variety is unlikely to be a replication of this system, given their location in Pau. It is possible that /t͡ç/ is realised as an instance of a phonetically similar category in their L1, French /t͡ʃ/ or, alternatively, that /t͡ç/ is realised as an instance of a phonetically similar standard Occitan category, /t͡ʃ/, though Occitan /t͡ʃ/ is phonemically cognate with Gascon /d͡ʝ/ (see Sect. 7.3.3). It is equally possible that the existence of [t͡ʃ] as an authentic sub-dialectal variant, together with transfer from French and from standard Occitan, has resulted in this preference for [t͡ʃ].
176
D. Mooney
Again, for the /d͡ʝ/ variable, there is limited evidence of new speakers abandoning local Gascon forms in favour of standard Occitan forms, such as [-t͡ʃ]; when they do occur, however, they are sometimes realised as [t͡ç], which may indicate hypercorrect behaviour, or the realisation of standard Occitan /-t͡ʃ/ as an instance of Gascon /t͡ç/. In any case, the majority forms are Gascon in origin, notably [d͡ʝ] and [d͡ʒ]. [d͡ʝ] was employed most often by new speakers, as was the case for native speakers on average. By comparison with dialect mixing identified for /t͡ç/, the new-speaker system for /d͡ʝ/ is more consistent with the process of levelling; there is an overall reduction in variability and retention of the majority form in the speech of the native speakers. This linguistic focusing was also evidence for the /d/ variable; there was only one token of sub-dialectal [z] and therefore a general increase in variants of this variable in general. The transfer of [d] from French, to replace lenited [ð] in intervocalic position, was pervasive and the frequency distribution of /d/ in this position showed evidence for both levelling ‘proper’, resulting in a reduction in variability overall, and for high levels of L1-to-L2 transfer, when compared with the native speakers. For the /ʒ/ variable, new speakers used the same variants as native speakers and with a comparable frequency. For native speakers, the presence of the [d͡ʒ] variant was noted to be either a fortified or vestigial variant of [ʒ], or evidence for diffusion of the Occitan /d͡ʒ/, which is phonemically cognate with Gascon /ʒ/. For the new speakers, the presence of the [d͡ʒ] variant in their speech is most likely due to their exposure to standard Occitan in educational settings. It is also possible that this diffusing Occitan form can be realised in a hypercorrect manner, as an instance of Gascon /d͡ʝ/, or as [d͡ʝ]. It is possible that this hypercorrection is an attempt to localise the borrowed form or it may simply be an analogical levelling process. Finally, for the /-s/ variable, the majority variants of the native speakers are also used by the new speakers and minority variants appear to have been levelled out of the system. For new speakers, the primary variants, [-s] and [-j] are not linked to specific sub-dialects, and for new speakers, these variants are lexicalised. Overall, new-speaker varieties of Gascon are characterised by an increase in transfer from French, by the retention of non-standard
7 Dialect Death: Gascon and Occitan
177
Gascon-specific forms, and by the loss of highly localised forms. There is some evidence for diffusion of the Occitan standard and for phonetic category assimilation of L1, local L2 and standard L2 forms. Additionally, there is evidence for the simplification of variable rules, possibly due to a loss of structural detail during the process of diffusion, and for a reduction in variability overall, as a result of linguistic focusing and levelling ‘proper’.
References Fishman, Joshua A. 1991. Reversing language shift. Clevedon: Multilingual Matters. Flege, James E. 1988. The production and perception of speech sounds in a foreign language. In Human communication and its disorders: A review 1988, ed. Harris Winitz, 224–401. Norwood, NJ: Ablex. Flege, James E. 1990. The intelligibility of English vowels spoken by British and Dutch talkers. In Intelligibility in speech disorders: Theory, measurement, and management, ed. Raymond D. Kent, 157–232. Amsterdam: John Benjamins. Flege, James E. 1991. Age learning effects the authenticity of voice-onset time (VOT) in stop consonants produced in a second language. The Journal of the Acoustical Society of America 85: 395–411. Flege, James E. 1995. The phonetic study of bilingualism. In European studies in phonetics and speech communication, ed. Gerrit Bloothooft, Valerie Hazan, Dieter Huber, and Joaquim Llisterri, 98–103. Utrecht: OTS Publications. Flege, James E. 2005. The origins and development of the speech learning model. Presented at the First Acoustical Society of America Workshop on L2 Speech Learning, Simon Fraser University. Flege, James E. 2007. Language contact in bilingualism: Phonetic system interactions. In Laboratory Phonology 9, ed. Jennifer Cole and Jose I. Hualde, 353–380. Berlin: Mouton de Gruyter. Jones, Mari C. 1998. Language obsolescence and revitalisation: Linguistic change in two sociolinguistically contrasting Welsh communities. Oxford: Clarendon Press. Labov, William. 2007. Transmission and diffusion. Language 83: 344–387.
178
D. Mooney
Labov, William. 2014. The role of African Americans in Philadelphia sound change. Language Variation and Change 26: 1–19. Lewis, M. Paul., and Gary F. Simons. 2010. Assessing endangerment: Expanding Fishman’s GIDS. Revue De Linguistique Romane 55: 103–120. Milroy, Lesley. 2007. Off the shelf or under the counter? On the social dynamics of sound changes. In Managing chaos: Strategies for identifying change in English, ed. Christopher. M. Cain and Geoffrey Russom, 149–172. Berlin: Mouton de Gruyter. Mooney, Damien. 2016. Transmission and diffusion: Linguistic change in the regional French of Béarn. Journal of French Language Studies 26: 327–352. Mooney, Damien. 2019. Phonetic transfer in language contact: Evidence for equivalence classification in the mid-vowels of Occitan-French bilinguals. Journal of the International Phonetic Association 49: 53–85. Oliviéri, Michèle, and Patrick Sauzet. 2016. Southern Gallo-Romance (Occitan). In The Oxford guide to the Romance languages, ed. Adam Ledgeway and Martin Maiden, 319–349. Oxford: Oxford University Press. Weinreich, Uriel. 1968. Languages in contact: Findings and problems. The Hague: Mouton.
8 Towards a New Theory of Language Death
This chapter examines the theoretical implications of the findings in the language and dialect death studies, with the aim of advancing datadriven principles that account for language variation and change during the obsolescence process. It begins by defining the scope of a revised or expanded theory of language death in this context and moves on to formalise empirically motivated theoretical principles that account for the outcomes of language and dialect contact, and of language revitalisation, for obsolescent Gascon. The final section of this chapter examines holistically the role of language-internal, external, and extralinguistic factors on phonetic and phonological variation and change during the obsolescence process, with the aim of more clearly articulating the nature of the interface between language contact and dialect contact during language death.
© The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 D. Mooney, Language and Dialect Death, https://doi.org/10.1007/978-3-031-51101-1_8
179
180
8.1
D. Mooney
Theorising Sound Change During Obsolescence
In this chapter, we saw that previous theoretical frameworks for language death have focused on the external setting, or the specific socio-political context that gives rise to language death, on speech behaviour and the language’s domains of use, and on the structural consequences, or linguistics changes taking place in the dying language as it undergoes the obsolescence process. The external setting that has given rise to language death for Gascon and an overview of the remaining speech community are detailed in Chapter 3. As such, the theoretical proposals presented in this chapter will address only phonetic and phonological structural consequences, as these were the focus of the language death (Chapter 6) and dialect death (Chapter 7) studies. This chapter will advance a revised theorisation of language change during obsolescence, taking as a starting point Weinreich et al.’s ‘modest’ version of a theory of language change: ‘a theory of language change would merely assert that every language constantly undergoes alteration, and it would formulate constraints on the transition from one state of a language to an immediately succeeding state’ (1968: 99–100). As such, the theoretical principles advanced in Sects. 8.2 and 8.3 are advanced to account for linguistic change as the language dies out, or during its ‘transition’ from one state to another. For Weinreich et al., a ‘stronger’ version of this theory would predict ‘from a description of a language state at some moment in time, the course of development which that language would undergo within a specified interval’ (1968: 99). They note that few historians of language would claim that such a theory is possible but it is hoped that the empirical foundations established for language change in Gascon will constitute a set of baseline hypotheses against which to examine the structural consequences of language death in other situations of gradual death and language revitalisation. Weinreich et al. note that these empirical foundations make it possible to compare whether other linguistic changes observed elsewhere ‘produce[…] a language state that violates or, more significantly, conforms with the statistical norms’ (1968: 127).
8 Towards a New Theory of Language Death
181
In Weinreich et al.’s (1968: 99) ‘modest’ theory of language change, they note that such a theory would need to establish constraints on language change and to accommodate variable usage. Labov (2007: 348) has long criticised discussions of language change in European dialectology for examining linguistic variation and change in relatively simplified terms, focusing on isolated individual dependent variables in an ‘X replaces Y’ fashion, without a full analysis of the structural and sociolinguistic constraints which govern the change under investigation. As such, the theorisation of language change during obsolescence presented in this chapter emphasises the importance of framing the outcomes of language and dialect contact as variable rules which incorporate information on the constraints that govern the changes observed. This approach constitutes a ‘refinement in the theory of language structure’ (Weinreich et al. 1968: 126), in that it proposes ‘fresh’ constraints on change and permits a potential ‘reclassification’ of observed changes according to new principles. Weinreich et al. note that this ‘hypothecation of constraints on change’ (1968: 126) is of utmost importance in the refinement of sociolinguistic theory. This hypothecation acknowledges and privileges the central tenet of sociolinguistic theory, that linguistic change is characterised by orderly differentiation, or structural heterogeneity (Weinreich et al. 1968: 187). The theoretical principles presented in this chapter aim to operationalise this central tenet of variationist theory in situations of both language and dialect contact during language death. The ‘theory’ of language death presented in this chapter will necessarily be incomplete, as the empirical motivations which underpin the theoretical principles proposed are based on a specific type of linguistic data and specific contact scenarios. Sasse, quite rightly notes that ‘as yet there is no theory of language death’ (1992: 7); the theorisation presented here, however, does aim to more fully integrate the well-established and ever-evolving theory of language change into the study of language decline and death. In the sections that follow, I summarise the testable, data-driven principles which I hope will act as a starting point for future investigations of linguistic change during obsolescence and which constitute the empirical foundation for a theoretical reframing of language change during obsolescence.
182
8.2
D. Mooney
Principles of Linguistic Change During Language Obsolescence
The results of the language death study (Chapter 6) are formalised in the following data-driven principles, which account for externally motivated linguistic changes in Gascon, taking place as a result of language contact with French (Principle 1), dialect contact between sub-dialects of Gascon (Principle 2), and of both processes acting together (Principle 3): • Principle 1: The transfer of linguistic features from the dominant language to the obsolescent language is variable, not categorical, and is governed by language-internal constraints. o Principle 1.1: The language-internal constraints on transfer are at least partially determined by equivalence classification of the phonological systems of the languages in contact. o Principle 1.2: The variable transfer of linguistic features from the dominant language results in systemic complexification, in the form of new variable rules, in the obsolescent language. • Principle 2: Contact between sub-dialects of the dying language leads to dialect mixing or to the appearance of localised variants in subdialects with which they are not traditionally associated. o Principle 2.1: The outcome of dialect mixing is localised variants becoming non-local and therefore an increase in variability over geographical space or an increase in the number of variant choices for a given linguistic variable. • Principle 3: The combined effect of transfer from the dominant language and of dialect mixing between sub-dialects of the obsolescent language is an increase in variability. This section circumscribes, explains, and justifies each of these empirically motivated theoretical principles in turn, with the aim of providing a holistic overview of the outcomes of language and dialect contact, as well as the combined outcome of both processes acting simultaneously on the dying language.
8 Towards a New Theory of Language Death
183
Principle 1: The transfer of linguistic features from the dominant language to the obsolescent language is variable, not categorical, and is governed by language-internal constraints. The results of the language death study clearly established the existence of variable rules which govern the transfer of phonetic and phonological features from French to the speech of the Gascon native speakers. The variable rules formalise the probabilistic influence of primarily languageinternal constraints on this transfer and thus avoid presenting an overly simplistic account of transfer from the dominant to the dying language, such as ‘X from French replaces Y in Gascon’. It is not simply enough to identify the existence of variability; we must formalise the variability in the form of a variable rule ‘with enough precision to allow us to incorporate them into our analyses of linguistic structure’ (Weinreich et al. 1968: 169). While the motivations for this transfer are clearly external in nature, the actual process is constrained variably by internal factors in the dying language. The variable rules take the form A > g[B] / _ X; g[B] = f (C, D, E) (Weinreich et al. 1968: 170), where A is a feature of the dying language, g[B] is the variably transferred feature, such that g[B] = f (C, D, E), where C, D, and E are linguistic or extralinguistic factors that condition or constrain the variability. For example, the variable rule for transfer of French [ʁ] into Gascon is: /r/ > g[ʁ] / V_C#; V_#; V_C; g[ʁ] = f (syllable type). For the transfer of French [j] into Gascon, the rule would be: /ʎ/ > g[j] / V_V; V_#; g[j] = f (syllable type). For the /b/ variable, the rule is /b/ > g[b] /V_/e/; V_/y/; V_/u/; g[b] = (following phoneme). As the language dies out, there is an increase in heterogeneity as a result of language contact, but this heterogeneity is clearly structured by language-internal constraints (i.e., it is orderly heterogeneity). Principle 1.1: The language-internal constraints on transfer are at least partially determined by equivalence classification of the phonological systems of the languages in contact. The language-internal constraints on transfer from the dominant language identified in Principle 1 are to some extent dependent on the structural relationship between the phonologies of the languages in contact in the bilingual phonological space. This relationship is largely determined by the process of equivalence classification, whereby the phonemes of a bilingual’s languages are equated (or not equated) as
184
D. Mooney
being structurally equivalent (see Sect. 7.4). For example, for the /ʎ/ variable, the phonemic correspondences between French /j/ and Gascon /ʎ/ were cited as potential points for the transfer of French [j] into Gascon. The resultant analysis (see Sect. 6.2.2) showed that the probability of Gascon /ʎ/ being realised as an instance of French [j] increases when the phonemic cognate of /ʎ/ is /j/ in French. When Gascon /ʎ/ corresponds to French /l/ or to zero, the probability of transfer is significantly reduced. Principle 1.2: The variable transfer of linguistic features from the dominant language results in systemic complexification, in the form of new variable rules, in the obsolescent language. The variable transfer of features from French introduces new rules into the dying language and these variable rules introduce systemic complexification into Gascon. Transfer from French to Gascon is not, at least for these variables, characterised by systemic simplification, by the wholesale loss of structural elements, or by an increase in regularity, as we might expect in situations of language death. Instead, the outcome of transfer from French in Gascon is systemic complexification due to the introduction of variable rules, and increase in variability, not regularity, and the introduction of new, historically inappropriate, structural elements, causing historically appropriate linguistic forms to become variable. For all of the variables considered in Sect. 6.2, transfer from French was identified, but there was no evidence for the wholesale or categorical replacement of a historically appropriate Gascon form with an external (French) feature; the system of obsolescent Gascon accommodates both forms and, in most cases, has rules that constrain the variability introduced by transfer. Principle 2: Contact between sub-dialects of the dying language leads to dialect mixing or to the appearance of localised variants in sub-dialects with which they are not traditionally associated. The most widely recognised outcome of contact between dialects of the same language is a reduction in the differences between the dialects in contact; this reduction in differences is seen to be the result of levelling, diffusion, and simplification. The data presented in Sect. 6.3 demonstrated, however, that contact between the sub-dialects of Gascon in Béarn has not led to a loss of highly localised forms, which would of
8 Towards a New Theory of Language Death
185
course decrease the level of difference between the sub-dialects. Instead, the outcome of dialect contact was characterised by an increase in variability over geographical space. Highly localised phonetic and phonological forms, traditionally associated with specific sub-dialects, were shown to be present in other sub-dialects with which they are not traditionally associated, even if they occur with low frequency. For the /s/ variable, for example, where the variants [-s] and [-j] are traditionally associated with certain sub-dialects, it was established that the variation observed could not be reliably predicted by a speaker’s place of origin. Overall, dialect contact between the sub-dialects of Gascon is better described as dialect mixing, rather than dialect levelling. Principle 2.1: The outcome of dialect mixing is localised variants becoming non-local and therefore an increase in variability over geographical space or an increase in the number of variant choices for a given linguistic variable. The process of dialect mixing identified in Principle 2 increases variability and optionality over geographical space, as new variants appear in sub-dialects for which the feature was not previously variable. This finding counters the outcomes of dialect contact described by Jones (1998) for Welsh: ‘the levelling out of minority and otherwise marked speech forms, and of simplification, which involves, crucially, a reduction in irregularities’ (Trudgill 1986: 107). There is some potential evidence for this in that the [t] variant of /t͡ç/ has spread beyond central subdialects into geographically peripheral sub-dialects; this variant is the majority variant in Béarn and therefore the adoption of this central variant into peripheral dialects may be the result of levelling ‘proper’. The outcomes of dialect mixing identified in the language death study are, however, more in line with the ‘multilayer formulation’ described by Weinreich et al.: ‘[variants] are jointly available to all (adult) members of the speech community (1968: 159)’ but ‘it is only when a pair of dialects are jointly available to a group that switches back and forth between them […] that the multilayer formulation is relevant to an understanding of language change’ (1968: 163). As such, contact between the subdialects of the dying language appears to facilitate access to new variants for speakers of specific sub-dialects.
186
D. Mooney
Principle 3: The combined effect of transfer from the dominant language and of dialect mixing between sub-dialects of the obsolescent language is an increase in variability. Studies of language death routinely distinguish between language change that is attributable to contact with the dominant language and language change which is not a result of contact with the dominant language (see Sect. 2.4). The outcome of both types of language change is however an increase in variability in the dying language and thus an increase in systemic complexity. This is due not only due to the presence of new variable elements in the system, but the existence of new rules that govern the choice of one variable element over another. The overall outcome of the combined effect of transfer and dialect mixing is not the wholesale loss of historically appropriate or minority forms, nor it is systemic simplification. Rather, the outcome is orderly heterogeneity and the non-localisation of minority forms, resulting in more optionality and variability in the dying language.
8.3
Principles of Linguistic Change During Language Revitalisation
The results of the dialect death study (Chapter 7) showed newspeaker phonological systems to have specific characteristics, which are formalised in the data-driven Principle 4 to 7. These principles account for new-speaker acquisition of historically appropriate forms, and the systemic constraints associated with them, for the frequency of dominant language transfer, for the retention or loss of highly localised forms, and for hypercorrect linguistic behaviour due to analogical levelling. • Principle 4: Even when intergenerational transmission is absent, new speakers can faithfully replicate native-speaker phonological systems, along with phonetic realisation rules, and systemic constraints.
8 Towards a New Theory of Language Death
187
o Principle 4.1: New-speaker varieties of the obsolescent language are characterised not by the incomplete acquisition of historically appropriate forms, but by the generalisation and simplification of constraints on transfer and variability. • Principle 5: The speech of new speakers evidences a higher frequency of transfer from the dominant language than the speech of native speakers. • Principle 6 : New speakers retain variably historically appropriate dialectal forms, but levelling also leads to a decrease in variability and to the loss of highly localised sub-dialectal forms. • Principle 7 : Analogical levelling results in the variable realisation of non-local or standard forms as instances of phonetically similar phonemes with which they are not structurally cognate. This section circumscribes, explains, and justifies each of these empirically motivated theoretical principles in turn, with the aim of providing a holistic overview of the impact of language revitalisation on phonetic and phonological change. Principle 4: Even when intergenerational transmission is absent, new speakers can faithfully replicate native-speaker phonological systems, along with phonetic realisation rules, and systemic constraints. The revitalisation of Gascon in the form of Occitan is not language revitalisation in the strictest sense, as it is not a process that has reinstated an unbroken chain of intergenerational transmission in the Gascon speech community (see Sect. 7.4). When native speakers have ceased to pass the language on to their children, as is the case in this study, Sasse is of the opinion that a specific type of revitalisation is possible in extreme cases and that this type of revitalisation involves a ‘generation skip’ (1992: 21): the oldest generation still speaks the language fluently; the middle generation displays a proficiency continuum from ‘semi speaker’ to zero; some of the youngest generation acquire the language anew. Sasse (1992: 21) defines language death as the total interruption of language transmission and states that any revitalisation that takes place after total interruption of language transmission involves the creation of a new language. On that basis, Gascon is a dead language,
188
D. Mooney
and the speech of the new speakers is a new, or reborn language, but the results of the dialect death study do not align well with these definitions of death and revitalisation. New speakers of Gascon were shown to faithfully replicate the stressed and unstressed oral vowel systems of the native speakers and their actual pronunciations of the vowels were not, on the whole, significantly different from those of the native speakers. For the /ʒ/ variable, new speakers were shown to reproduce the system of the native speakers, with comparable variants used and a relatively comparable frequency distribution. Even in the absence of unbroken intergenerational transmission, the new speakers of Gascon demonstrate an ability to learn the language of the older generation and to faithfully replicate the systems of the native speakers. Principle 4.1: New-speaker varieties of the obsolescent language are characterised not by the incomplete acquisition of historically appropriate forms, but by the generalisation and simplification of constraints on transfer and variability. New speakers of Gascon were shown to faithfully replicate the systems and systemic constraints of the native speakers, but the variable rules or constraints on variation acquired tend to show some evidence for focusing, in the form of simplification or generalisation. Simplification in these cases involves the regularisation of variable paradigms, rather than ‘reduction’, which would involve the loss of essential systemic elements and defectivity (Sasse 1992: 15). For example, the variable rules governing transfer of French [ʁ] into Gascon and the distribution of [r] and [ɾ] were only partial replications of the native-speaker variable rules; these rules were similar for both new and native speakers, but not exactly the same. For the transfer of French [j] to replace Gascon /ʎ/, new speakers produced a more regular variable rule than native speakers: /ʎ/ > g[ʎ] / #_V; g[ʎ] = (syllable type). This simplified rule states that /ʎ/ is realised as [ʎ] with a higher probability in word-initial position, and as [j] in all other syllable types. For the /-s/ variable, the variability present between [-s] and [-j] in the speech of the native speakers was completely lost, with variable elements becoming lexicalised in the words mes and james, respectively. Principle 5: The speech of new speakers evidences a higher frequency of transfer from the dominant language than the speech of native speakers.
8 Towards a New Theory of Language Death
189
The types of change that we observe in the structure of a dying language are not notably different from those well established in the study of language change in general, ‘but the time span for change seems compressed and the amount of change seems relatively large’ (Dorian 1981: 154). It is clear from the analysis of the speech of the native speakers in Chapter 7 that this is absolutely the case for Gascon. New-speaker varieties of Gascon are characterised by an increase in frequency of transfer from French: they exhibit a dramatic increase in the realisation of historically inappropriate plosive variants of /b d ɡ/ in intervocalic position and significantly more incorrect grammatical gender marking for nouns which differ in gender between French and Gascon. In apparent-time, new speakers appear to be leading an externally motivated change that involves the replacement of intervocalic lenited forms with plosive at the same place of articulation. Increased transfer from French also involves a concomitant reduction in the numbers of variants that can be used for a given variable, as the transferred French forms replace minority variants. Principle 6: New speakers retain variably historically appropriate dialectal forms, but levelling also leads to a decrease in variability and to the loss of highly localised sub-dialectal forms. Jones (1998: 154) notes that for highly competent new speakers, deviance from the conservative norms of native speakers is most often perceived in terms of their use of standard linguistic forms and consequent lack of dialectal forms. New speakers of Gascon are in contact, at least to some extent, with native-speaker varieties and with standardised Occitan varieties. As noted previously, standard Occitan is to some extent a pluricentric language in that it is diasystemic; localised variation in the primary dialect areas is accommodated within the standard model. As such, in the Gascon context, ‘standard’ forms could refer to majority Gascon forms, as opposed to highly localised Gascon forms, or to Lengadocian forms from the dialect on which the Occitan standard is based. Overall, the speech of new speakers is not characterised by the complete loss of Gascon features, but by a reduction in the number of variants available for a given variable, and by the loss of the highly localised forms of the Gascon sub-dialects. For example, the new-speaker
190
D. Mooney
realisations of /d͡ʝ/ are predominantly localised Gascon forms, not ‘standard’ Lengadocian [t͡ʃ], with a preference for the most common form across sub-dialects of Gascon in Béarn, [d͡ʝ]. There is, however, some evidence in the speech of the new speakers for levelling ‘proper’: for /b/ and /d/ in particular, the new speakers demonstrate markedly reduced variability or optionality for the variables considered; focusing, paradigm simplification, the reduction in exponents of phonological categories are the hallmarks of levelling ‘proper’. On the whole, the Gascon new speakers were shown, in Sect. 7.3, to use localised historically appropriate Gascon forms for the majority of the variables considered; they had not completely abandoned Gascon forms in favour of ‘standard’ Occitan, or Lengadocian, forms, as is often suggested anecdotally. Principle 7: Analogical levelling results in the variable realisation of non-local or standard forms as instances of phonetically similar phonemes with which they are not structurally cognate. Weinreich et al. (1968: 142) note that ‘reinterpretation of phonemes’, also known as ‘rephonologisation’ can result in language variation and change. During dialect contact, for example between Gascon and ‘standard’ Occitan or Lengadocian varieties, phonetic realisations of the external system may be incorrectly reinterpreted as realisations of phonemes in the local system, with which they are not historically cognate. For example, Gascon /ʒ/ is phonemically cognate with /d͡ʒ/ in standard Occitan and there is evidence in the Gascon data of /ʒ/ being realised as [d͡ʒ]. There are also instances, however, where the borrowed afficate is palatalised to [d͡ʝ], effectively realising borrowed [d͡ʒ] as an instance of Gascon /d͡ʝ/, not Gascon /ʒ/, with Gascon /d͡ʝ/ being phonemically cognate with standard Occitan /t͡ʃ/, not /d͡ʒ/. This rephonologisation results from the analogical association of borrowed [d͡ʒ] with the most phonetically similar phoneme in Gascon, /d͡ʝ/. Analogical change of this type occurs when a phonetic realisation is changed to reflect the phonetic specification of another phoneme in the system on the basis of perceived similarity. This kind of analogical change is evidenced in the speech of both native and new speakers; in Arzacq, for example, Gascon /ʒ/ is realised as [d͡ʝ] categorically (see Table 6.22). For
8 Towards a New Theory of Language Death
191
these new speakers, the use of [d͡ʒ] is most likely the result of convergence in the direction of standard Occitan and the use of [d͡ʝ] is most likely a hypercorrect or analogical form of standard Occitan [d͡ʒ]. This type of analogical levelling can lead to the potential collapse of phonemic contrast, for example, when Gascon /ʒ/ and /d͡ʝ/ are both realised as [d͡ʝ]. Gascon /d͡ʝ/ is phomemically cognate with standard Occitan /t͡ʃ/ and there is evidence in both the speech of native and new speakers for /d͡ʝ/ being realised as [t͡ʃ]. There is also evidence in the speech of both speaker types for /d͡ʝ/ being realised as [t͡ç]. Again, the use of [t͡ç] for /d͡ʝ/ may be due to the analogical association of standard Occitan /t͡ʃ/ with the phonetically similar Gascon phoneme, /t͡ç/, producing /d͡ʝ/ as a hypercorrect palatalised variant of [t͡ʃ], or [t͡ç], using the phonetic realisation rules for Gascon /t͡ç/.
8.4
Internal, External, and Extralinguistic Factors
The results of the language and dialect death studies revealed very little evidence for internally motivated linguistic changes from below, or for independent innovations in Gascon; the vowel system, for example, was shown to be remarkably stable over time. There were very few instances of syntagmatic or phonetic change in the data, and paradigmatic changes, such as the merger of /-e/ and /-O/ in Arzacq, were already well established in the literature. When Gascon, however, was shown to adopt linguistic forms transferred from French, it appears to do so with quantifiable regard to its own pre-existing internal structure. The existence of French sounds in Gascon is largely constrained by the internal structure of Gascon and, while these changes are externally motivated, they exhibit the characteristics of conditioned phonetic or syntagmatic changes post-adoption, in that they are favoured in particular (internal) linguistic contexts. The internal constraints on the distribution of transferred features are to some extent determined by the (initial) interlingual identification (Flege 1995: 98) of the phonologies of the two languages in contact with each other (see Principle 1.1).
192
D. Mooney
At first glance, the external motivations of linguistic change in Gascon are not particularly novel: it involves the transfer of phonetic features from French, at a high frequency, and over a relatively short period of time (see Dorian 1981: 154). The discovery, however, of linguistic constraints on transfer, formalised as variable rules, has reframed this process of transfer in the context of language obsolescence. Contact with French does not involve the wholesale loss or abandonment of historically appropriate Gascon forms (see Principle 1); instead, the adoption of linguistic material from French increases the amount of orderly heterogeneity in Gascon (see Principle 1.2). The establishment of these variable rules that govern transfer from French to Gascon demonstrates that there exist internal constraints on the outcomes of language contact and that transfer is not of the simplistic form ‘X replaces Y’. Dialect contact between sub-dialects of Gascon was also shown to produce externally motivated phonetic changes in the language death study. The outcomes of dialect contact between the sub-dialects of Gascon were not aligned with the traditional outcomes of the regional dialect levelling (RDL) process, whereby levelling ‘proper’ and geographical diffusion results in a reduction in variability over geographical space or to a ‘reduction in exponents of phonological […] categories’ between the varieties in contact (Kerswill and Williams 2005: 1024). Instead, dialect contact appeared to promote the appearance of highly localised variants in subdialects with which they are not traditionally associated (see Principle 2) and a concomitant increase in the amount of variability over geographical space (see Principle 2.1), at least for native speakers. The combined effect of external factors on language change in Gascon, as a result of both language and dialect contact, is a marked increase in exponents of phonological categories or in the number of variant choices for linguistic variables (see Principle 3). The influence of extralinguistic factors on sound change in Gascon was relatively limited. There were isolated instances of sex differential behaviour, such as the potential inversion of the sociolinguistic gender pattern in Sect. 6.3.2, and of change in apparent-time, such as the low-magnitude /a/-raising for new speakers when compared with native speakers. The overall findings align, however, with previous analyses, such as Dorian (1981: 154), that note the potential reduction of the
8 Towards a New Theory of Language Death
193
effect of extralinguistic factors in situations of language contraction and death. Language death theory has previously focused on the distinction between fluent native speakers, younger fluent speakers, rusty speakers, and semi-speakers when examining variability in speech behaviour during language obsolescence. Dorian (1981: 117) has noted that the ‘community norm’ is defined with reference to the speech of fluent native speakers and that there is a proficiency continuum across and within other speaker categories that is most frequently age-graded, with younger speakers using the most historically inappropriate forms. These various speaker types were not the object of study in this book, which has instead drawn as distinction between fluent native speakers and fluent new speakers, with near-native competence in Gascon. Fluent new speakers are distinguished from younger fluent speakers in that they have acquired Gascon through the education system and their existence is bound up with questions of authenticity, legitimacy, and power relations, relative to the native speakers (Kasstan 2017: 2). Up until this point, the speech of new speakers has more frequently been analysed within this powerstructure framework; the dialect death study presented in this book has shown, for new speakers at least, that the traditional theorisation of dialect death, or dedialectalisation, does not seem to adequately account for their speech behaviour. For example, the speech of new speakers has been traditionally perceived as ‘a potential “xenolect”, or as a standardized non-local variety in which dialectal variation is largely absent’ (Jones 1998: 257), or as ‘far removed from community norms’ (Kasstan 2017: 2). As such, the speech of new speakers has been traditionally viewed through the lens of unidirectional dialect convergence towards the standard language, which is often held in opposition to the authentic local varieties of the fluent native speakers. The new speakers of Gascon demonstrate, however, a remarkably faithful replication of the fluent native-speaker phonological systems, along with phonetic realisation rules, and systemic constraints (see Principle 4 ). Their speech is not characterised by a departure from ‘community norms’, by the incomplete acquisition of historically inappropriate form, or by large levels of convergence in the direction of standardised Occitan (see Principle 4.1). While the new speakers in
194
D. Mooney
this study do appear have more phonological transfer from French in their variety of Gascon, they also largely replicate the constraints on this transfer present in the speech of the native speakers. Where they differ from native speakers is in the regularisation and simplification of these constraints and of the associated variable rules (see Principles 4.1 and 5 ). Jones (1998: 289) has noted that dialect convergence towards the standard in the speech of new speakers is often characterised by simplification (see Trudgill 2009: 100) and by a reduction in irregularities but, in the case of the new Gascon speakers, these processes do not involve categorical changes in linguistic form. The simplification and regularisation identified involves systematic refinement of the internal constraints on the distribution of transferred features, rather than a departure from community norms in the more general sense. We might be tempted to interpret these systemic changes as genuine changes in apparent-time (see Dorian 1981: 117, for discussion), with the caveat that there is no intergenerational transmission involved, at least not in the strictest sense (see Sect. 8.3). In the context of koinéization, Kerswill and Williams (2000: 67–71) have shown that younger generations are instrumental in filtering out variation from the speech of older generations. While the dialect contact scenarios between sub-dialect in Béarn, and between new and native speakers, do not meet the criteria for a situation in which koinéization would occur, it is worth noting that the speech of new speakers also involves systemic focusing of the type seen in situations of in-migration. New speakers have been shown to retain historically appropriate Gascon forms, while also reducing variability and abandoning the use of highly localised sub-dialectal forms (see Principle 6 ). In this chapter, the results of the language and dialect death studies have been formalised into data-driven principles; these principles take the form of testable hypotheses and it is hoped that future research on language death and revitalisation will use these hypotheses as a basis for analyses of linguistic change during the obsolescence process. The scope of the theorisation presented is necessarily limited to the domains of phonetic and phonological analysis, but the integration of variationist tools into these analyses has permitted at least some theoretical refinement of the processual outcomes of language and dialect contact in the context of obsolescence. The concept of the variable rule is essential to
8 Towards a New Theory of Language Death
195
understanding not only the transfer of linguistic features from the dominant language, but also the constraints on transfer between dialects in contact. Indeed, the principal outcomes of language and dialect contact, and of language and dialect death, do not appear to be particularly different. The overall picture of new-speaker phonology is that, if Gascon is to live on only in its revitalised form, it will not do so as an illegitimate ‘xenolect’. It will in fact constitute a historically accurate representation of the language of mother-tongue Gascon speakers of generations past.
References Dorian, Nancy C. 1981. Language death: The life cycle of a Scottish Gaelic dialect. Philadelphia: University of Pennsylvania Press. Flege, James E. 1995. Second-language speech learning: Theory, findings, and problems. In Speech perception and linguistic experience: Issues in crosslanguage research, ed. Winifred Strange, 229–273. Timonium, MD: York Press. Jones, Mari C. 1998. Language obsolescence and revitalisation: Linguistic change in two sociolinguistically contrasting Welsh communities. Oxford: Clarendon Press. Kasstan, Jonathan. 2017. New speakers: Challenges and opportunities for variationist sociolinguistics. Language and Linguistics Compass 11: 1–16. Kerswill, Paul, and Anne Williams. 2000. Creating a new town koiné: Children and language change in Milton Keynes. Language in Society 29: 69–115. Kerswill, Paul, and Anne Williams. 2005. New towns and koinéization: Linguistic and social correlates. Linguistics 43: 1023–1048. Labov, William. 2007. Transmission and diffusion. Language 83: 344–387. Sasse, Hans-Jürgen. 1992. Theory of language death. In Language Death, ed. Brenzinger, Matthias, 7–30. Berlin & Boston: De Gruyter Mouton. Trudgill, Peter. 1986. Dialects in contact. Oxford: Blackwell. Trudgill, Peter. 2009. Sociolinguistic typology and complexification. In Language Complexity as an Evolving Variable, ed. Geoffrey Sampson, David Gil, and Peter Trudgill, 99–109. Oxford: Oxford University Press. Weinreich, Uriel, William Labov, and Marvin I. Herzog. 1968. Empirical Foundations for a Theory of Language Change. In Directions of Historical Linguistics: A Symposium, ed. Lehmann, W. P., and Yakov Malkiel, 96–195. Austin & London: University of Texas Press.
9 Conclusions
This monograph took inspiration from Dorian (1981) and Jones’ (1998) seminal studies of language obsolescence and death; both studies prioritised data-driven theorisation of the myriad processes involved in language death instead of simply documenting the language varieties involved. The present study, on the variety of Gascon spoken in Béarn, France, has attempted to go further than Dorian (1981) and Jones (1998) by firmly integrating into the analysis well-established theories of language variation and change. This study also had its own theory-building ambitions and aimed to advance data-driven principles that account for linguistic variation and change during obsolescence and revitalisation, at least in the domains of phonetics and phonology. It also aimed to build on the existing literature on France’s regional and minority languages by incorporating more methodological rigour and depth of study, in the form of instrumental analysis and statistical modelling, as is required in modern variationist sociolinguistics. From the outset, the Gascon study made a formal distinction between language death and dialect death, and between their potential linguistic outcomes. This distinction was framed, in Chapter 2, as part of the more general theoretical distinction between language contact and dialect © The Author(s), under exclusive license to Springer Nature Switzerland AG 2023 D. Mooney, Language and Dialect Death, https://doi.org/10.1007/978-3-031-51101-1_9
197
198
D. Mooney
contact. At least in the initial stages of the study, these distinctions have been useful for the examination of language variation and change in Gascon: Gascon finds itself under pressure from both the dominant national language, French, and, to some extent, from the standardised southern Gallo-Romance variety, Occitan. Contact with French would traditionally be considered language contact whereas contact with Occitan would be considered dialect contact. Additionally, within the region of Béarn, contact between speakers of different non-standard sub-dialects of Gascon would, of course, be traditionally conceptualised within a dialect contact framework. The Gascon case study therefore provided an excellent opportunity to examine the processes of language and dialect death and to examine the implications of the relationship between the processes for variationist theories of language and dialect contact. The research questions advanced in Chapter 1 framed the overall analysis of language variation and change in Gascon and informed the specific methodological approach adopted, as outlined in Chapter 3: two sub-studies, notably the ‘language death’ study and the ‘dialect death’ study mapped directly onto the research questions. To recapitulate, the research questions were as follows: 1. What phonetic and phonological changes are occurring in obsolescent Gascon as a result of contact with French? 2. To what extent is phonetic and phonological change in Gascon the result of levelling between its localised sub-dialects? 3. What effect has revitalisation in the form of Occitan had on the phonological structure and phonetic realisation of Gascon? The language death study addressed specifically research questions 1 and 2. This sub-study examined data for nine continuous vocalic variables and ten categorical variables in the speech of thirty native Gascon speakers from five fieldwork sites in Béarn. The selection of linguistic variables was also informed by research questions 1 and 2: the language death study examined variables evidencing contact-induced transfer from French as well as variables whose realisation traditionally differs between the five fieldwork sites in Béarn. In essence, research question 1 is about
9 Conclusions
199
the linguistic outcomes of language contact and research question 2 is concerned with changes that are not the result of contact with the dominant language, but with contact between the remaining dialects of the dying language. The dialect death study addressed research question 3 by considering the same linguistic variables in the speech of ten new speakers or néo-locuteurs, and by comparing phonetic and phonological variation and change in their speech to that of the older native speakers. It had the additional aim of establishing the extent to which there existed a structural cleavage between the speech of the native and new speakers, as is the predominant folk perception. The language death study in Chapter 6 identified the phonetic and phonological changes occurring in obsolescent Gascon as a result of contact with French and the results of these analyses were formalised as data-driven principles in Chapter 8. The Gascon stressed oral vowel system was shown to be remarkably stable in the speech of the native speakers from Béarn, and there was little or no evidence for transfer from French. For the consonantal variables considered, transfer from French was shown to be variable, not categorical, and the analysis identified the existence of language-internal constraints on transfer. These languageinternal constraints were formalised as variable rules which predict the linguistic environments in which transfer is more likely to occur. The language death study also found that the language-internal constraints identified are at least partially the result of the equivalence classification of the phonological systems of the two languages in contact: the identification of cognate phonemes in Gascon and French, as well as their surface realisations, had some bearing on the variable outcomes of language contact. The language death study also established that the introduction of variable rules into the obsolescent language results in increased systemic complexification, not simplification, because the level of variability in the dying language, and the systemic constraints on variability, are seen to increase. The language death study also considered the extent to which phonetic and phonological change in the Gascon of the native speakers is the result of contact between localised sub-dialects in Béarn. The outcomes of dialect contact were not wholly characteristic of levelling ‘proper’ in
200
D. Mooney
that that patterns of change did not always involve a reduction in differences between the sub-dialects for the variables analysed. The outcomes of dialect contact between the sub-dialects were better characterised by dialect mixing or by the appearance of highly localised variants in subdialects with which they were not traditionally associated. Dialect mixing was shown to result in localised variants becoming non-local and the amount of variant choices was shown to increase across geographical space. This effectively meant that all localised variants from the subdialects were at least an option, however infrequent, in all of the other sub-dialects. Therefore, the combined effect of transfer from the dominant language, French, and of dialect mixing between the sub-dialects of Gascon in Béarn is a marked increase in variability in the dying language, not the wholesale loss of traditional Gascon linguistic forms. The dialect death study in Chapter 7 examined the effect of revitalisation on the phonological structure and phonetic realisation of Gascon in the speech of ten new Gascon speakers who had acquired the language within the Occitan educational framework. The perceived cleavage between the speech of native and new speakers was not identified: even in the absence of an unbroken chain of intergenerational transmission, new Gascon speakers were shown to faithfully replicate the phonological system of the native speakers, including phonetic realisation rules, and systemic constraints. There was little or no evidence for incomplete acquisition in the speech of the new speakers; they were shown to largely maintain historically appropriate Gascon forms in their speech. The new speakers differed from the native speakers in their tendency to use more regular variable rules or by generalising and simplifying the constraints on transfer and variability in the speech of the native speakers. New-speaker language was also characterised by increased transfer from French, but never by the wholesale loss of traditional Gascon speakers; transfer remained variable and systemically constrained. The use of highly localised sub-dialect forms was less common for new speakers and the outcome of contact with traditional sub-dialects is characterised by the focusing mechanism of levelling ‘proper’: new-speaker phonologies showed less variability in variant choice for these features and evidenced a reduction in the use of highly localised sub-dialectal forms.
9 Conclusions
201
The legitimacy of new-speaker language is not compatible with the ideologies of more localised or grassroots associations in Béarn, whose mission is centred around the preservation and celebration of the language and culture of generations past. Essentially, the Béarnist and Gasconist movements take a postvernacular approach: ‘postvernacular dynamics do not aim at reversing language shift, nor to a return to the everyday use of the language in public and private spheres’ (Pivot and Bert 2017: 277–278). The efforts of the grassroots movements in Béarn focus on keeping the language alive in its current form, not on re-establishing language practices that are lost or soon to be lost. The postvernacular approach emphasises the difference between native and new speakers by devaluing the function of new speakers: when reversing language shift is not the goal, new speakers become insignificant and delegitimised. The existence of new speakers is furthermore seen to compromise the legitimacy of the language as a whole, by attempting to reproduce a language variety that is perceived, in this case, to be dying with dignity. Revitalisation in the form of Occitan, even when it is essentially a carbon copy of the native-speaker variety of Gascon, is seen to be completely incompatible with the celebration of the language and culture of days gone by. New speakers are less concerned with what they perceive to be ideological in-fighting: ‘quoan sera exactament la medisha, en hèit, mes bon, ne cau pas possar los vielhs au tujar’ (Speaker AN) (‘when of course, it’s exactly the same [language]. But you know, there’s no need to force old people to think the way we do’). The overall outcome of revitalisation for Gascon is generally a positive one, at least where maintenance of the native-speaker norm is the end goal. The perceived separation between the speech of the native speakers and the new speakers appears to be completely ideological in nature and, at least where phonology and phonetics are concerned, this folk perception has little basis in scientific reality. This does not mean that the ideology cannot in and of itself create an alternate reality in the language obsolescence context; indeed, we have seen that discussion of new speakers’ language is more often concentrated on these issues of authenticity. However, in acknowledging the role that folk perceptions have to play in the revitalisation context, what we can say is that revitalisation in the form of Occitan appears to produce the desired outcome:
202
D. Mooney
new speakers acquiring the language in the Occitan educational framework are, in Béarn at least, acquiring a historically appropriate and strikingly accurate language variety that respects local Gascon norms. The new speakers are learning a slightly modified version of the nativespeaker language and, as such, we can conclude that Occitan educational approach works well, not for revitalising the language per se, because the speakers’ numbers remain low, but for promoting and supporting the faithful replication of the community norm. In the absence of intergenerational transmission, and in the absence of any real effort on the part of grassroots movements in Bearn to support and encourage new speakers, the Occitan approach constitutes the only viable means of surviving death for Gascon.
References Dorian, Nancy C. 1981. Language death: The life cycle of a Scottish Gaelic dialect. Philadelphia: University of Pennsylvania Press. Jones, Mari C. 1998. Language obsolescence and revitalisation: Linguistic change in two sociolinguistically contrasting Welsh communities. Oxford: Clarendon Press. Pivot, Bénédicte., and Michel Bert. 2017. Orthography creation for postvernacular languages: Case studies of Rama and Francoprovençal revitalization. In Creating Orthographies for Endangered Languages, ed. Mari C. Jones and Damien Mooney, 276–290. Cambridge: Cambridge University Press.
Appendix 1: Wordlist translation task
Variable
Context
Gascon Target
French Stimulus
(i)
C_C C_C C_C C_C C_C C_C C_C C_C C_C C_C C_C C_C C_C C_C C_C C_C C_C C_C C_C C_C C_C C_C
pic guit petita vita díser punt puish dus putz sus pesca qu’ei véder péder dever pè qu’èi pèc tè bèth par cantar
amer le canard petite la vie dire le point puis deux puits sur la pêche il est voir croire devoir le pied j’ai stupide le thé beau la paire chanter
(y)
(e)
(E)
(a)
(continued) © The Editor(s) (if applicable) and The Author(s), under exclusive licence to Springer Nature Switzerland AG 2023 D. Mooney, Language and Dialect Death, https://doi.org/10.1007/978-3-031-51101-1
203
204
Appendix 1: Wordlist translation task
(continued) Variable
(u)
(O)
(-e)
(-O)
(r)
Context
Gascon Target
French Stimulus
C_C C_C C_C C_C C_C C_C C_C C_C C_C C_C C_C C_C C_C C_# C_# C_# C_# C_# C_# C_# C_# C_# C_# V_V V_V V_V V_V V_V #_V #_V #_V #_V #_V C_V C_V C_V C_V C_V V_# V_# V_# V_#
portar estar crompar poma hlor nebot tots boca pòrc lòc còth còp tròp càder díser brave arbe aute pòrta cada gojata vita mongeta poret porret beròja sorélh haria hrèita regent hromatge hred repèish crampa trin tribalhar craba praube hèr horn dur per
porter être acheter la pomme la fleur le neveu tous la bouche le porc le lieu le cou la fois trop tomber dire courageux un arbre un autre la porte chaque la jeune fille la vie le haricot le poulet le poireau belle le soleil la farine le besoin un instituteur la fromage froid le repas la chambre le train travailler la chèvre le pauvre le fer le four dur par (continued)
Appendix 1: Wordlist translation task
205
(continued) Variable
(ʎ)
(t͡ç)
(d͡ʝ)
(b)
Context
Gascon Target
French Stimulus
V_# _C# _C# _C# _C# _C# [n]_ #_V #_V #_V #_V V_V V_V V_V V_V V_V V_# V_# V_# V_# V_# V_[s]# V_[s]# V_[s]# V_[s]# V_[s]# V_# V_# V_# V_# V_# V_V V_V V_V V_V V_V [R]_V [R]_V [z]_V [z]_V V_[a] V_[a]
renart mort pòrc obèrt sèrp verd genre lheit lhevar lheta lhèu huelha botelha vielha familha aulha sorelh hilh uelh vielh alh uelhs hilhs fesilhs miralhs recuelhs castèth cavath navèth martèth còth vilatge viatge maridatge vitatge arrelòtge corbaish garba desvelhar desbrombar arribar trovar
renard mort porc ouvert serpent vert genre lit lever liseron peut-être feuille bouteille vieille famille brebis soleil fils oeil vieux ail les yeux les fils les fusils les miroirs les recueils château cheval nouveau marteau cou village voyage mariage vignoble horloge corbeau gerbe réveiller oublier arriver trouver (continued)
206
Appendix 1: Wordlist translation task
(continued) Variable
(d)
(ɡ)
(ʒ)
Context
Gascon Target
French Stimulus
V_[y] V_[y] V_[u] V_[u] V_[O] V_[O] V_V V_V [R]_V [R]_V V_[a] V_[a] V_[y] V_[y] V_[u] V_[u] V_[O] V_[O] V_V V_V [R]_V [R]_V [l]_V [l]_V V_[a] V_[a] V_[u] V_[u] V_[y] V_[y] V_[O] V_[O] V_V #_V #_V #_V #_V #_V medial medial medial medial
recebut decebut neboda bohet loba craba béver déver pèrder ordonar nadar agradar credut sedut pedolh hredor seda moneda créder véder carga cargar belga promulgar pagar ligar arrigor pregon conegut paregut ahromiga ahraga seguir jaune ièr jòc junh juntar minjar iranje gojat ploja
reçu deçu nièce soufflet la louve la chèvre boire devoir perdre ordonner nager plaire j’ai cru assis pou froideur la soie argent croire voir la charge/le fret charger belge promulguer payer lier la rigueur profond connu paru fourmis fraise suivre jaune hier jeu juin joindre manger orange jeune homme pluie (continued)
Appendix 1: Wordlist translation task
207
(continued) Variable (s)
(gender)
Context
Gascon Target
French Stimulus
medial lexical lexical lexical article article article article article article article article article article article article article article article article
gaujor qu’ei jamei mei un ahar un arrelòtge un cremalh lo deute lo grèish lo hum un limac un par la bilheta la lèbe la léit la mensonja la sang la sau la sèrp ua ungla
gaiété il est jamais mais une affaire une horloge une crémaillère la dette la graisse la fumée une limace une paire le billet le lièvre le lait le mensonge le sang le sel le serpent un ongle
Appendix 2: Language Death Study—Supplementary Data
See Fig. A2.1, A2.2, A2.3, A2.4, A2.5, A2.6, A2.7 and Table A2.1, A2.2, A2.3, A2.4, A2.5, A2.6, A2.7, A2.8, A2.9, A2.10, A2.11, A2.12, A2.13, A2.14, A215
© The Editor(s) (if applicable) and The Author(s), under exclusive licence to Springer Nature Switzerland AG 2023 D. Mooney, Language and Dialect Death, https://doi.org/10.1007/978-3-031-51101-1
209
210
Appendix 2: Language Death Study—Supplementary …
Fig. A2.1 Stressed oral vowel system for (male and female) Arzacq speakers, normalised data
Fig. A2.2 Stressed oral vowel system for (male and female) Lembeye speakers, normalised data
Appendix 2: Language Death Study—Supplementary …
211
Fig. A2.3 Stressed oral vowel system for (male and female) Nay speakers, normalised data
Fig. A2.4 Stressed oral vowel system for (male and female) Ouest speakers, normalised data
212
Appendix 2: Language Death Study—Supplementary …
Fig. A2.5 Unstressed oral vowels for (male and female) Lembeye speakers, normalised data
Fig. A2.6 Unstressed oral vowels for (male and female) Nay speakers, normalised data
213
Appendix 2: Language Death Study—Supplementary …
Fig. A2.7 Unstressed oral vowels for (male and female) Ouest speakers, normalised data Table A2.1 Arzacq speakers (normalised data): Regression analysis of F1 values for /e/ and /E/, with ‘speaker’ and ‘word’ as random effects Dependent variable = F1 N = 51 Grand mean = –0.102
R2 : 0.747 Degrees of freedom: 8
Factor group
Factor
Coefficient
N
p-value
Phoneme
/E/ /e/
+0.489 –0.489
22 29
0.00
Non-significant factor groups were: syllable type (p = 0.185) and speaker sex (p = 0.776) Table A2.2 Arzacq speakers (normalised data): Regression analysis of F2 values for /e/ and /E/, with ‘speaker’ and ‘word’ as random effects Dependent variable = F2 N = 51 Grand mean = 0.898
R2 : 0.136 Degrees of freedom: 8
Factor group
Factor
Coefficient
N
p-value
Phoneme
/e/ /E/
+0.202 –0.202
29 22
0.04
Non-significant factor groups were: syllable type (p = 0.49) and speaker sex (p = 0.763)
214
Appendix 2: Language Death Study—Supplementary …
Table A2.3 Arzacq speakers (normalised data): Regression analysis of F3 values for /e/ and /E/, with ‘speaker’ and ‘word’ as random effects Dependent variable = F3 N = 51 Grand mean = 0.114
R2 : 0.253 Degrees of freedom: 8
Factor group
Factor
Coefficient
N
p-value
Phoneme
/e/ /E/
+0.511 –0.511
29 22
0.00
Non-significant factor groups were: syllable type (p = 0.183) and speaker sex (p = 0.2) Table A2.4 Lembeye speakers (normalised data): Regression analysis of F1 values for /e/ and /E/, with ‘speaker’ and ‘word’ as random effects Dependent variable = F1 N = 56 Grand mean = –0.036
R2 : 0.675 Degrees of freedom: 8
Factor group
Factor
Coefficient
N
p-value
Phoneme
/E/ /e/ C_C C_#
+0.53 –0.53 +0.545 –0.545
27 29 44 12
0.00
Syllable type
0.04
Non-significant factor groups were: speaker sex (p = 0.609) Table A2.5 Lembeye speakers (normalised data): Regression analysis of F2 values for /e/ and /E/, with ‘speaker’ and ‘word’ as random effects Dependent variable = F2 N = 56 Grand mean = 0.763
R2 : 0.329 Degrees of freedom: 8
Factor group
Factor
Coefficient
N
p-value
Phoneme
/e/ /E/
+0.149 –0.149
29 27
0.03
Non-significant factor groups were: speaker sex (p = 0.101) and syllable type (p = 0.411)
215
Appendix 2: Language Death Study—Supplementary …
Table A2.6 Lembeye speakers (normalised data): Regression analysis of F3 values for /e/ and /E/, with ‘speaker’ and ‘word’ as random effects Dependent variable = F3 N = 56 Grand mean = –0.269
R2 : 0.175 Degrees of freedom: 8
Factor group
Factor
Coefficient
N
p-value
Phoneme
/e/ /E/
+0.304 –0.304
29 27
0.04
Non-significant factor groups were: speaker sex (p = 0.347) and syllable type (p = 0.835) Table A2.7 Nay speakers (normalised data): Regression analysis of F1 values for /e/ and /E/, with ‘speaker’ and ‘word’ as random effects Dependent variable = F1 N = 55 Grand mean = –0.027
R2 : 0.496 Degrees of freedom: 8
Factor group
Factor
Coefficient
N
p-value
Phoneme
/E/ /e/ C_C C_#
+0.533 –0.533 +0.403 –0.403
27 28 43 12
0.00
Syllable type
0.03
Non-significant factor groups were: speaker sex (p = 0.829) Table A2.8 Ouest speakers (normalised data): Regression analysis of F1 values for /e/ and /E/, with ‘speaker’ and ‘word’ as random effects Dependent variable = F1 N = 57 Grand mean = –0.089
R2 : 0.64 Degrees of freedom: 8
Factor group
Factor
Coefficient
N
p-value
Phoneme
/E/ /e/
+0.443 –0.443
28 29
0.00
Non-significant factor groups were: speaker sex (p = 0.076) and syllable type (p = 0.277)
216
Appendix 2: Language Death Study—Supplementary …
Table A2.9 Ouest speakers (normalised data): Regression analysis of F3 values for /e/ and /E/, with ‘speaker’ and ‘word’ as random effects Dependent variable = F2 N = 57 Grand mean = 0.754
R2 : 0.365 Degrees of freedom: 8
Factor group
Factor
Coefficient
N
p-value
Phoneme
/e/ /E/
+0.152 –0.152
29 28
0.00
Non-significant factor groups were: speaker sex (p = 0.06) and syllable type (p = 0.107) Table A2.10 Lembeye speakers (normalised data): Regression analysis of F2 values for /-e/ and /-O/, with ‘speaker’ and ‘word’ as random effects Dependent variable = F2 N = 49 Grand mean = –0.019
R2 : 0.727 Degrees of freedom: 9
Factor group
Factor
Coefficient
N
p-value
Phoneme
/-e/ /-O/
+0.476 –0.476
23 26
0.00
Non-significant factor groups were: preceding phoneme (p = 0.154) and speaker sex (p = 0.636) Table A2.11 Nay speakers (normalised data): Regression analysis of F1 values for /-e/ and /-O/, with ‘speaker’ and ‘word’ as random effects Dependent variable = F1 N = 46 Grand mean = –0.039
R2 : 0.312 Degrees of freedom: 9
Factor group
Factor
Coefficient
N
p-value
Phoneme
/-O/ /-e/
+0.199 –0.199
23 23
0.02
Non-significant factor groups were: speaker sex (p = 0.141) and preceding phoneme (p = 0.826)
217
Appendix 2: Language Death Study—Supplementary …
Table A2.12 Nay speakers (normalised data): Regression analysis of F2 values for /-e/ and /-O/, with ‘speaker’ and ‘word’ as random effects Dependent variable = F2 N = 46 Grand mean = –0.009
R2 : 0.764 Degrees of freedom: 9
Factor group
Factor
Coefficient
N
p-value
Phoneme
/-e/ /-O/
+0.591 0.591
23 23
0.00
Non-significant factor groups were: speaker sex (p = 0.375) and preceding phoneme (p = 0.375) Table A2.13 Ouest speakers (normalised data): Regression analysis of F2 values for /-e/ and /-O/, with ‘speaker’ and ‘word’ as random effects Dependent variable = F2 N = 41 Grand mean = –0.107
R2 : 0.561 Degrees of freedom: 9
Factor group
Factor
Coefficient
N
p-value
Phoneme
/-e/ /-O/
+0.369 –0.369
20 21
0.00
Non-significant factor groups were: speaker sex (p = 0.673) and preceding phoneme (p = 0.994) Table A2.14 All native speakers (normalised data): Regression analysis of F2 values for /-e/, with ‘speaker’ and ‘word’ as random effects Dependent variable = F2 of /-e/ N = 110 Grand mean = 0.46
R2 : 0.346 Degrees of freedom: 12
Factor group
Factor
Coefficient
Place of origin
Laruns Nay Lembeye Ouest Arzacq
+0.256 +0.173 +0.105 –0.175 0.359
N
Non-significant factor groups were: speaker sex (p = 0.897)
p-value 25 23 23 20 19
0.00
218
Appendix 2: Language Death Study—Supplementary …
Table A2.15 All native speakers (normalised data): Regression analysis of F2 values for /-O/, with ‘speaker’ and ‘word’ as random effects Dependent variable = F2 of /-O/ N = 125 Grand mean = –0.472
R2 : 0.599 Degrees of freedom: 10
Factor group
Factor
Coefficient
Place of origin
Arzacq Ouest Lembeye Nay Laruns
+0.490 +0.023 –0.039 –0.129 –0.345
N
p-value 28 21 26 23 27
0.00
Non-significant factor groups were: speaker sex (p = ..208) and preceding phoneme (p = 0.621)
Index
A
acquisition 5, 9, 14, 16, 18, 19, 38, 169, 172–174, 186–188, 193, 200 agency 25, 32 allophone 80, 81, 84, 87, 91, 93, 107, 134, 137, 170 analogical levelling 176, 186, 187, 190, 191 analogy 86, 152, 171 apocope 76, 82, 85 apparent time 25, 39, 58, 102 attention paid to speech 24–26 authenticity 3, 37, 193, 201 Auvernhat 48, 49, 51, 62, 73, 74, 85, 86, 91, 92, 96–98
B
Basque 3, 52, 53, 77, 102
betacism 83 Breton 3, 17, 38
C
Calandreta 3, 8, 65–67, 105 CAPES 67 Catalan 3, 50, 51, 72, 73, 77, 78, 90, 91, 138 change from above 22, 24, 145 change from below 24, 101, 150 classical orthography 78, 105 community of practice 26 compensatory lengthening 74, 87 complementary distribution 82, 134 complexification 150, 151, 182, 184, 199 condenser microphones 108 constraints 14, 31, 35, 82, 86, 123, 134, 136–138, 140, 141, 150,
© The Editor(s) (if applicable) and The Author(s), under exclusive licence to Springer Nature Switzerland AG 2023 D. Mooney, Language and Dialect Death, https://doi.org/10.1007/978-3-031-51101-1
219
220
Index
151, 155, 162–166, 172–175, 180–183, 186–188, 191–195, 199, 200 creolisation 29 crossover effect 24
Expanded Graded Intergenerational Disruption Scale (EGIDS) 172 external setting 28, 180
F D
depalatalisation 87 dialect contact 4, 6–8, 10, 13, 15, 19, 20, 22, 23, 35, 40, 102, 143, 152, 153, 179, 181, 182, 185, 190, 192, 194, 195, 198–200 dialect convergence 34, 36, 193, 194 dialect mixing 9, 22, 33, 35, 101, 107, 123, 140, 144, 146, 149, 152, 168–170, 175, 176, 182, 184–186, 200 diasystem 62, 78, 79, 81, 85, 86, 91, 96, 97, 171 diffusion 20, 21, 35, 145, 147, 148, 151, 152, 162, 169, 171, 173, 174, 176, 177, 184, 192 diphthongisation 72, 98 domains of use 28, 60, 68, 180 dominant language 2, 10, 16, 28, 30, 31, 38, 55, 103, 123, 133, 151, 182–184, 186–188, 195, 199, 200
E
epenthesis 73 European Charter for Regional or Minority Languages 64 exogamy 60
fixed effects 119 focusing 8, 13, 20, 22, 23, 47, 107, 109, 123, 166, 170, 175–177, 181, 188, 190, 194, 200 folk perception 199, 201 fortition 139, 140, 147, 171 Francoprovençal 3, 48, 73, 74
G
Gaelic 5, 31, 33, 39 Gallo-romance 2, 7, 8, 34, 47–51, 64, 72, 73, 76, 90, 198 geminate 80, 88, 93 gender marking 108, 142, 143, 151, 167, 175, 189 gender paradox 25 generalisation 30, 151, 187, 188 generation skip 187 glide 78, 79, 84, 96–98, 112, 113, 138 gradual death 28, 29, 58, 180 grammatical gender 107, 142, 143, 167, 175, 189
H
hypercorrection 176
Index
I
Lois Jules Ferry 57
imperfect learning 16, 17, 19, 162, 173, 174 incrementation 14, 15, 39 interlingual identification 19, 170, 174, 191
M
221
merger 14, 79, 129, 130, 132, 143, 144, 150, 151, 168, 191 mixed-effects regression 119 multilayer formulation 185
K
koinéization 22, 23, 34, 35, 151, 194
L
language contact 2, 4, 6, 8, 9, 15, 18, 19, 23, 30, 31, 34, 40, 47, 57, 68, 170, 179, 182, 183, 192, 197–199 language shift 1, 2, 17, 28, 29, 33, 34, 36, 37, 40, 55, 56, 58, 60, 63, 68, 201 langue d’oïl 48, 52 langues de France 3 legitimacy 37, 193, 201 Lemosin 48–51, 73, 74, 83, 85, 87, 90, 92, 94–96 Lengadocian 48–52, 61, 75–78, 81, 83–88, 92, 94, 95, 97, 175, 189, 190 lenition 14, 138, 140, 147 levelling 2, 7, 9, 10, 20–22, 31, 34, 35, 107, 123, 143–145, 147, 148, 152, 166, 169, 170, 175–177, 184, 185, 187, 189–192, 198–200 lexical domain 143 linear predictive coding (LPC) 114 listing effect 109 loanword 86, 90, 91
N
nasal vowel 81, 95, 96 non-mobile older rural males speakers (NORMs) 148, 152 normalisation 115–117
O
Occitano-Romance 50, 51, 77, 137 orderly heterogeneity 31, 183, 186, 192 Ordonnances de Villers-Cotterêts 54
P
palatalisation 72–75, 78, 80, 87, 148, 149 paradigmatic change 14, 191 paradigm simplification 166, 190 parlar negre 91, 93, 94, 128, 144, 150 patois 65 phonemic cognate 106, 137, 138, 164, 169, 173, 184 phonetic realisation rules 169, 186, 187, 191, 193, 200 pidginisation 29 pluricentric language 189 postvernacular 201
222
Index
proficiency continuum 29, 38, 39, 58, 187, 193 Provençal 3, 48–52, 60–63, 74–78, 80–95, 97, 98
Q
quota sampling 103
R
random effects 119, 120, 125, 157, 158, 160, 162–164, 167, 213–218 reduction 20, 22, 23, 28, 30, 34, 152, 166, 175–177, 184, 185, 188–190, 192, 194, 200 regional dialect levelling (RDL) 20, 22, 31, 35, 151, 192 regularisation 23, 188, 194 regularity 23, 30, 151, 165, 174, 184 rephonologisation 190 revitalisation 3, 7–9, 13, 36–40, 47, 63, 65–68, 155, 179, 180, 186–188, 194, 197, 198, 200, 201 rusty speaker 28, 29, 37, 193
social network 20, 22, 25–27, 32, 104 socio-political context 180 sound change 6, 7, 13, 14, 129, 180, 192 spectrogram 112–115 speech accommodation 20–22 speech behaviour 28, 180, 193 Speech Learning Model (SLM) 18, 19, 173 split 14 stance 27 standardisation 9, 33, 35, 36, 47, 61, 62, 68, 171 structured obsolescence 5, 29 style-shifting 26, 27 supralocalisation 10, 20, 22 syncope 72, 93 syntagmatic change 14, 191
T
transfer of literacy 61, 62 transmission 14, 21, 39, 58, 60, 65, 158, 161, 162, 172, 173, 186–188, 194, 200, 202 triphthong 75, 98 typological distance 17, 18
S
U
semi-speaker 28, 29, 37–39, 58, 59, 193 shift-induced transfer 16, 17 simplification 22, 23, 30, 34, 35, 144, 151, 153, 172, 174, 177, 184–188, 194, 199
urban hierarchical diffusion 21
V
variable rule 30, 31, 35, 119, 120, 136, 138, 140, 150, 151, 162,
Index
165, 173, 174, 177, 181–184, 188, 192, 194, 199, 200 variable rule programme 119, 120 variationist sociolinguistics 2, 4, 6, 9, 13, 14, 23, 25, 27, 40, 197 velarisation 75 Vivaro-Alpin 48–51, 73–75, 78, 80–90, 92–94, 96–98 vocalisation 74, 86, 87, 96, 98
223
Vulgar Latin 48, 72
W
Welsh 1, 4, 30, 33, 34, 38, 40, 185
X
xenolect 38, 39, 193, 195