Sociolinguistic Approaches to Sibilant Variation in Spanish (Routledge Studies in Hispanic and Lusophone Linguistics) [1 ed.] 0367722208, 9780367722203

Citation preview

Sociolinguistic Approaches to Sibilant Variation in Spanish

Social processes and the nature of language variation have driven sibilant variation across the Spanish-​speaking world. This book explores the current state of Spanish sibilants and their dialectal variations. Focusing on different processes undergone by sibilants in Spanish (e.g. voicing, devoicing, weakening, aspiration, elision) in various geographical areas and language contact situations, each chapter offers an analysis on a unique sociolinguistic case from different formal, experimental and data-​ based approaches. The opening chapter orients the reader with an overview of sibilant system’s evolution, which serves as an anchor to the other chapters and facilitates understanding for readers new to the topic. The volume is organized around three thematic parts: Part I, Spain; Part II, United States; and Part III, Central and South America. The collection includes research on dialects in both Peninsular and Trans-​Atlantic Spanish such as Jerezano, Caribbean Spanish in Boston and New York City, Cuban Spanish in Miami, Colombia-​Barranquilla Spanish, northern Buenos Aires Argentine Spanish, and United States heritage Spanish, among other case studies. This volume offers an original and concise approach to one of the most studied variables in Spanish phonetics, taking into account geographically based phonetic variation, sociolinguistic factors and various Spanish language contact situations. Written in English, this detailed synthesis of the wide-​ranging geolinguistic features of Spanish sibilants provides a valuable resource for scholars in Hispanic studies, linguistics, Spanish dialectology and sociolinguistics. Eva Núñez-​Méndez is Professor of Spanish Linguistics at Portland State University, USA.

Routledge Studies in Hispanic and Lusophone Linguistics Series Editor: Dale Koike, University of Texas at Austin

The Routledge Studies in Hispanic and Lusophone Linguistics series provides a showcase for the latest research on Spanish and Portuguese Linguistics. It publishes select research monographs on various topics in the field, reflecting strands of current interest. Titles in the series: Spanish in the United States Attitudes and Variation Edited by Scott M. Alvord and Gregory L. Thompson Spanish in Health Care Policy, Practice and Pedagogy in Latino Health Glenn A. Martínez Los castellanos del Perú historia, variación y contacto lingüístico Luis Andrade Ciudad y Sandro Sessarego (eds.) Language Patterns in Spanish and Beyond Structure, Context and Development Edited by Juan J. Colomina-​Almiñana and Sandro Sessarego The Evolution of Spanish Past Forms Gibran Delgado-​Díaz Sociolinguistic Approaches to Sibilant Variation in Spanish Edited by Eva Núñez-​Méndez Heritage Speakers of Spanish and Study Abroad Edited by Rebecca Pozzi, Tracy Quan and Chelsea Escalante For more information about this series please visit: www.routledge.com/​Routledge-​Studies-​in-​Hispanic-​and-​Lusophone-​ Linguistics/​book-​series/​RSHLL

Sociolinguistic Approaches to Sibilant Variation in Spanish Edited by Eva Núñez-​Méndez Series Editor: Dale A. Koike Spanish List Advisor: Javier Muñoz-​Basols

First published 2021 by Routledge 2 Park Square, Milton Park, Abingdon, Oxon OX14 4RN and by Routledge 52 Vanderbilt Avenue, New York, NY 10017 Routledge is an imprint of the Taylor & Francis Group, an informa business © 2021 selection and editorial matter, Eva Núñez-​Méndez; individual chapters, the contributors The right of Eva Núñez-​Méndez to be identified as the author of the editorial material, and of the authors for their individual chapters, has been asserted in accordance with sections 77 and 78 of the Copyright, Designs and Patents Act 1988. All rights reserved. No part of this book may be reprinted or reproduced or utilised in any form or by any electronic, mechanical, or other means, now known or hereafter invented, including photocopying and recording, or in any information storage or retrieval system, without permission in writing from the publishers. Trademark notice: Product or corporate names may be trademarks or registered trademarks, and are used only for identification and explanation without intent to infringe. British Library Cataloguing-​in-​Publication Data A catalogue record for this book is available from the British Library Library of Congress Cataloging-​in-​Publication Data A catalog record has been requested for this book ISBN: 978-​0-​367-​72220-​3  (hbk) ISBN: 978-0-367-72221-0 (pbk) ISBN: 978-​1-​003-​15394-​8  (ebk) Typeset in Times New Roman by Newgen Publishing UK

Contents

List of figures and maps  List of tables  List of contributors  Acknowledgments  List of abbreviations  Introduction 

vii x xiii xix xx 1

E VA N Ú Ñ E Z -​MÉN D EZ

PART I

Spain 

7

1 An overview of the sibilant merger and its development in Spanish 

9

E VA N Ú Ñ E Z -​MÉN D EZ

2 Sibilants in western Andalusian Spanish: the lack of a Sevillian norm in the Jerezano speech community 

73

JAN N I S  H ARJ US

3 Intervocalic /​s/​-​voicing in Spanish in contact with Catalan  95 J U S T I N DAV I D SO N

PART II

United States 

129

4 Describing and analyzing variability in Spanish /​s/​: a case study of Caribbeans in Boston and New York City 

131

DAN I E L G E RAR D ER K ER A N D MA D ELI N E R E F F E L

vi Contents

5 Variable realization of final /​s/​in Miami Cuban Spanish: the reversal of diachronic language change 

164

AN D RE W LYN C H A N D A N TO N I FER NÁ N D EZ  PARERA

6 Variable /​s/​-​voicing by heritage Spanish speakers in the United States 

192

AMAN DA B O O MER SH I N E A N D JO H N STEVE NS

PART III

Central and South America 

215

7 /​s/ weakening in Nicaragua 

217

WH I T N E Y CH A PPELL

8 A sociophonetic approach to /​s/​-r​ ealization in the Colombian Spanish of Barranquilla 

246

RI C H ARD F I LE- ​M U R I EL, EA R L BROWN A ND MI C H AE L G R A D OV I LLE

9 Sibilants in Ecuadoran Spanish 

262

J OH N   L I P SK I

10 Syllable-​final /​s/​-​variation in a Uruguayan Spanish-​ Portuguese contact variety 

279

MARK WALTER MI R E

11 Variable voicing in Argentine Spanish /​ʒ/​ 

301

MI C H AE L G R A D OV I LLE

Index 

335

Figures and maps

Figures 1 .1 Devoicing of fricatives and affricates  1.2 Boyd-​Bowman’s data: regional origins of the earliest Spanish colonists between 1493–​1519 out of 5,481 settlers  1.3 Boyd-​Bowman’s analysis: origins of population between 1520–​1539 out of 13,262 emigrants  1.4 Settler mixing and koineization  1.5 History of the sound [θ]  1.6 Old Spanish and Judeo-​Spanish sibilants  1.7 Andalusian seseo and Judeo-​Spanish [s]‌and [z]  3.1 Group C, younger female rendition of tras años “after years” (5% voiced)  3.2 Group A, younger male rendition of tras años “after years” (100% voiced)  3.3 Effects of language profile group and age on Spanish intervocalic /​s/​-​voicing  3.4 Effect of gender on Spanish intervocalic /​s/​-​voicing  3.5 Effect of style on Spanish intervocalic /​s/​-v​ oicing  3.6 Effect of stress on Spanish intervocalic /​s/​-​voicing  3.7 Effect of word position on Spanish intervocalic /​s/​-​voicing  3.8 Percentages of 100% voiced [z]‌productions by word position in Barcelonan Spanish  3.9 Percentages of 100% voiced word-​final [z]‌productions by style in Barcelonan Spanish  4.1 Variation across descriptors. 1a, n = 4,325 (Boston data only). 1b, n = 9,125 (Boston and NYC data combined). 1c-​d, n = 7,099 (combined data, non-​deleted tokens only)  4.2 More reduction occurs among codas compared to onsets across all descriptors. 1a, n = 4,325 (Boston data only). 1b, n = 9,125 (Boston and NYC data combined). 1c-​d, n = 7,099 (combined data, non-​deleted tokens only) 

19 24 25 28 32 51 52 107 108 110 111 112 113 114 114 115 147

149

viii  List of figures and maps 4.3 More reduction is observed among Caribbean compared to Central and Andean groups. 1a, n = 4,325 (Boston data only). 1b, n = 9,125 (Boston and NYC data combined). 1c-​d, n = 7,099 (combined data, non-​deleted tokens only)  150 4.4 Reduction of coda /​s/​among Caribbeans diminishes with increased U.S. life experience. 1a, n = 1,054 (Boston data only). 1b, n = 3,052 (Boston and NYC data combined). 1c-​d, n = 1,749 (combined data, non-​deleted tokens only)  153 6.1 Sample spectrogram illustrating voice bar and pulses  201 6.2 Realization of /​s/​by following consonant class and participant group  203 6.3 Realization of /​s/​by word position and following context  203 6.4 Role of stress position on the realization of coda /​s/​  204 6.5 Location of voicing by following context and speaker group  205 6.6 Individual realization rates for heritage speakers before voiced consonant  208 7.1 % sibilance and deletion of word-​final, intervocalic /​s/​in the early 1980s (from Lipski’s 1984 data)  224 7.2 % sibilance and deletion of word-​final, intervocalic /​s/​in the early 2010s (from Chappell 2013’s data)  224 7.3 Coda sibilance insertion in observo from Observo un par de islas “I see a couple of islands”  226 7.4 Spectrogram and waveform of a glottal stop in Levantan las olas “the waves rise up”  228 7.5 Spectrogram and waveform of creaky voice production in las casas entre las montañas “the houses between the mountains”  228 7.6 Percentage glottal constriction by age and task  230 7.7 Percentage glottal constriction by education and task  231 7.8 Percentage sibilance by age and task  231 7.9 Percentage sibilance by education and task  232 7.10 Production of [sʔ] in museos altos “tall museums”  233 7.11 Example of mental representations of word-​final, intervocalic /​s/​for a more educated or younger speaker (left) and less educated or older speaker (right) within an exemplar model  236 8.1 Boxplots of physiological variables by gender  252 8.2 Individual boxplots of COG by F2 of /​i/​(i.e. oral cavity length)  253 8.3 Individual boxplots of voicelessness by F0 of /​i/​(i.e. laryngeal cavity size)  254 11.1 Representation of the lexical connections forming category /​ʒ/​ 307 11.2 Representation of the exemplar clusters of absolute initial yo at the beginning (left), middle (center), and end (right) of the devoicing process  308 11.3 Waveform and spectrogram of a llevar “to take” (/​ʒ/​ fully voiced)  311

List of figures and maps  ix 11.4 Waveform and spectrogram of mayoría “majority” (/​ʒ/​ fully voiceless)  312 11.5 Waveform and spectrogram of ella “she” (/​ʒ/​partially voiced)  313 11.6 Waveform and spectrogram of ellos “they” (/​ʒ/​fully voiced)  314 11.7 Range of voicelessness values of each speaker for /​s/​and /​ʒ/​  319 11.8 Relation between bigram frequency and voicelessness of /​ʒ/​  323 11.9 Range of voicelessness values of /​ʒ/​for each previous context by speaker type  324

Maps 1.1 1 .2 2.1 2.2 2.3 3.1 9.1

Expansion of Castilian (11th–​15th cent.) and the spread of sibilant devoicing  Spread of modern distinción, seseo and ceceo  Western and Eastern parts of Andalusia, and Jerez  Ceceante and seseante zones in Andalusia  Heheo distribution in Andalusia  Regional territories of Catalonia (adapted from Generalitat 2009, 2011)  Map of Ecuador and Afro-​Choteño area 

21 46 74 76 90 97 263

Tables

1 .1 1.2 1.3 1.4 1.5 1.6 1.7 1 .8 1.9 1 .10 1.11 1.12 1.13 1.14 1 .15 1.16 1.17 1.18 1.19 1.20 1.21 1.22 1.23 3.1 3.2 3.3

Voiceless and voiced sibilant phonemes  Examples of medieval sibilant phonemes  Evolution of Old Spanish sibilants  Examples of sibilant phonemes  Castilian versus Andalusian Spanish: distinción and seseo  Boyd-​Bowman’s data: regional origin of the earliest Spanish colonists of America between 1520–​1539  Linguistically significant cities in Spain during the 15th and 16th centuries  Chronology of devoicing and velarization  Chronology of sibilant devoicing according to grammarians’ reports  Grammarians’ opinions about sibilant change  Examples of orthographic confusion from 1270 to 1680  Sibilant merger chronology based on graphemes  Chronology of sibilant devoicing according to rhymes  Chronology of sibilant merger based on grammarians, rhymes and graphemes  Chronology of interdentalization, velarization and devoicing  Sibilant devoicing development according to researchers  Conclusions regarding the sibilant merger and development  Judeo-​Spanish sibilants  Examples of [s]‌and [z]  Examples of the origins of [θ]  Examples of the origins of [x]‌  Examples of palatalization  Examples of distinción, ceceo and seseo  2013 population (%) ages 15+ with Catalan and Spanish as native or habitual language (Institut d’Estadística 2014)  Subject population according to language profile group (2013 Census data: Institut d’Estadística 2014)  Summary of mixed-​effects linear regression model fitted to Barcelonan and Madrid Spanish intervocalic alveolar

11 12 15 17 22 27 29 31 36 37 39 40 44 45 46 48 49 50 53 54 55 56 57 98 104

List of tables  xi

3.4 4 .1 4.2 4.3 5.1 5 .2 5.3 5.4 5.5 5.6 5.7 5 .8 5.9 5 .10 5.11 5.12 5 .13 5.14 5.15 5.16 6 .1 6.2 6.3 6.4 6 .5 7.1

fricatives. Intercept: Group A, younger female casual speech, word-​final unstressed tokens  Frequencies of fully voiced intervocalic [z]‌in casual styles of Spanish varieties  Mixed effects logistic regression results. Frication present  Mixed effects linear regression results. Duration ms.  Mixed effects linear regression results. COG  Results of the binary logistic regression analysis for realization of final /​s/​  Realization of final /​s/​according to following segment  Realization of final /​s/​before pauses according to time of arrival  Realization of final /​s/​before vowels according to time of arrival  Realization of final /​s/​before consonants according to time of arrival  Realization of final /​s/​according to syllable stress and time of arrival  Realization of final /​s/​according to position in word and time of arrival  Realization of final /​s/​according to morphological value  Realization of final /​s/​according to morphological value and time of arrival  Realization of final /​s/​according to time of arrival  Realization of final /​s/​according to self-​ascribed gender  Realization of final /​s/​according to time of arrival and self-​ ascribed gender  Realization of final /​s/​according to task  Center of Gravity of [s]‌among non-​newcomer Mexican women in New York City  Duration and Center of Gravity of [s]‌among third-​ generation PreM Cubans in Miami  Duration and Center of Gravity of [s]‌in the speech of M3_​A (PreM male) in Spanish and in English  Participants’ demographic information  Position and following context of coda /​s/​tokens  Frequency of variants by speaker group and linguistic context  Degree of voicing (percent voiced) by speaker group and following context  Comparative realization of voiced variant  Word-​final, intervocalic /​s/​production in the early 1980s among three social classes before a stressed or unstressed vowel (Lipski 1984a: 175) 

109 117 154 154 155 174 175 176 176 176 177 178 179 179 180 181 181 181 184 184 185 199 200 202 205 206 223

xii  List of tables 7.2 Word-final, intervocalic /s/ production in the early 2010s among three social classes before a stressed or unstressed vowel  223 7.3 Simplified distribution of coda /​s/​variants identified in Chappell (2013)  227 7.4 Average production rate for each variant  229 8.1 Stepwise linear regression: COG  255 8.2 Stepwise linear regression: voicelessness  255 9.1 Realization of coda /​s/​in Ecuadoran Spanish (%)  264 9.2 Breakdown and VARBRUL factor weights for Afro-​Choteño missing plurals  267 10.1 Development of sibilants from Vulgar Latin to Old Spanish (examples from Penny 2002)  282 10.2 Realizations of /​s/​(in percentages) for multiple modern dialects of American Spanish, including that of Rivera, Uruguay  286 10.3 Comparison of probabilities for /​s/​deletion for plural NP constituents in Rivera Spanish with probabilities for /​s/​ retention in Brazilian Portuguese (Scherre 1998) and Rivera Spanish (Carvalho 2006a) according to relative and linear positions within the NP (Waltermire 2011)  288 10.4 Multivariate analysis of the probabilities of co-​occurrence of the aspiration of /​s/​and social factors in the Spanish of Rivera 289 10.5 Multivariate analysis of the probabilities of co-​occurrence of the deletion of /​s/​and social factors in the Spanish of Rivera  290 10.6 Cross-​tabulation of the deletion of /​s/​according to the occupation and sex of consultants  291 11.1 Fricative inventories of Spanish varieties  302 11.2 Demographic information of the nine participants  316 11.3 Distribution of /​ʒ/​ and syllable-​initial, non-​word-​final /​s/​ according to previous context in the data  316 11.4 Distribution of /​ʒ/​ and syllable-​initial, non-​word-​final /​s/​ according to following context in the data  317 11.5 Predictor variables accounted for in the data  318 11.6 Significance and direction of estimates of fixed effects for mixed-​effects models of individual speakers  321 11.7 Fixed effects of mixed-​effects regression predicting voicelessness of word-​initial /​ʒ/​  322 11.8 Fixed effects of mixed-​effects regression predicting voicelessness of word-​medial /​ʒ/​  326

Contributors

Amanda Boomershine University of North Carolina Wilmington, USA Amanda Boomershine is an Associate Professor of Spanish at UNC Wilmington where she teaches courses in Hispanic linguistics, service-​ learning and seminars on topics such as immigration and linguistic variation. In addition, she coordinates the department’s internship program. She earned her doctorate in Hispanic Linguistics from Ohio State University. Her current research topics include teaching pronunciation to heritage speakers of Spanish, the perception of English and Spanish sounds by bilingual, monolingual and heritage speakers of Spanish and English, and a sociolinguistic analysis of the Spanish spoken in North Carolina. Her recent research in socio-​phonetics and language variation includes work/​chapters published in Studies in Hispanic and Lusophone Linguistics, Cambridge Handbook of Hispanic Linguistics, and Advances in Spanish Language Teaching: The Teaching of Spanish Pronunciation; and The Influence of Allophony vs. Contrast on Perception: The Case of Spanish and English. Earl Brown Brigham Young University, Utah, USA Earl Brown works as an Associate Professor of Linguistics at Brigham Young University. He received a PhD in Hispanic Linguistics at the University of New Mexico in 2008, and his doctoral thesis was published in the LINCOM Studies in Romance Linguistics series in 2009. His research agenda centers on the quantification of language variation, especially in Spanish. In most of this research, he employs the acoustic software Praat and uses the corpus linguistics techniques of text searching, data manipulation and statistical analysis with the programming languages Python and R. Some of his research has appeared in the journals Language Variation and Change, International Journal of Bilingualism, Linguistic Geography,

xiv  List of contributors Research Design and Statistics in Linguistics and Communication Science, and Corpus Linguistics and Linguistic Theory. Whitney Chappell University of Texas, San Antonio, USA Whitney Chappell is an Assistant Professor of Hispanic linguistics at the University of Texas at San Antonio. She earned her doctorate from The Ohio State University, and her research focuses on socio-​phonetic variation in the Spanish-​speaking world, or how Spanish speakers construct social meaning through their use of contextualized linguistic variants. Her most recent projects focus on the socio-​phonetic perception of nonstandard variants among monolingual and bilingual Spanish speakers, and her work has been published in prestigious publications such as Language Variation and Change, The Journal of Voice, Hispania, and Studies in Hispanic and Lusophone Linguistics, among many others. Justin Davidson University of California, Berkeley, USA Justin Davidson is an Assistant Professor of Hispanic and Romance Linguistics at the University of California, Berkeley. His research agenda is guided by questions that address language variation and language change in contact situations, specifically as linked to the empirical assessment of contact-​induced linguistic change. His principal research project explores bi-​directional effects of language contact between Spanish and Catalan manifested phonetically in the speech of the diverse community of Catalan-​Spanish bilingual speakers, which is complemented by parallel investigations of Spanish in contact with Quechua in Ecuador and Peru, and Spanish in contact with English in the California Bay Area.   These projects and their associated publications focus on the dynamics of language use in bilingual speech communities, insomuch as why, as well as by what processes, certain linguistic features propagate throughout the wider speech community. He has additionally published on the diachronic development of diaspora varieties of Catalan from a framework of sociohistorical linguistics, as well as the variable acquisition of Spanish inflectional morphology by US heritage speakers and L2-​learners using empirical methodologies informed by the fields of second language acquisition and psycholinguistics. Daniel Gerard Erker Boston University, USA Daniel Gerard Erker is an Assistant Professor of Spanish and Linguistics in the Department of Linguistics at Boston University. He earned his PhD

List of contributors  xv in Linguistics from New  York University. His research interests include sociocultural linguistics, language change, the linguistic practices of immigrant communities, the acquisition and representation of probabilistic linguistic knowledge, and the biological evolution of the human language faculty. He is the director of the Spanish in Boston Project, an NSF-​funded variationist sociolinguistic study of Spanish-​ speaking Bostonians. His work has appeared in a range of journals, including Language, Lingua, Language and Linguistics Compass, Language Variation and Change, and the International Journal of the Sociology of Language. Antoni Fernández Parera University of Miami, USA Antoni Fernández Parera holds the position of lecturer of Spanish at Barnard College. Since completing his PhD in Romance studies at the University of Miami, he has been actively engaged in teaching second and heritage language learners of Spanish in the United States and Spain. His areas of research and interest are sociolinguistics, sociocultural theory and second language acquisition, and pedagogy with special emphasis on comparative second and heritage language research. Richard File-​Muriel University of New Mexico, USA Richard File-​Muriel is an Associate Professor of Hispanic Linguistics at the University of New Mexico. His research focuses on how frequency of use impacts the production and perception of language. He is also interested in research methodology, such as the quantification of lexical frequency and how the composition of corpora can influence lexical frequency counts. His current research attempts to model individual physiology in sound variation, which is routinely ignored in sociolinguistics research. Michael Gradoville Arizona State University, USA Michael Gradoville is a Lecturer of Spanish and Linguistics in the School of International Letters and Cultures at Arizona State University, which he joined in 2016 after having taught for three years at Spelman College. He received his PhD in Hispanic Linguistics and general Linguistics from Indiana University. His main fields of interest are usage-​based linguistics, socio-​phonetics, phonology, language variation and quantitative methods. His research has examined Latin American varieties of Spanish from the US Southwest to Argentina and many points in between as well as multiple varieties of Brazilian Portuguese. His work has appeared in journals such as Lingua, Corpus Linguistics and Linguistic Theory, Sociolinguistic

xvi  List of contributors Studies, the Italian Journal of Linguistics, and the Journal of Research Design and Statistics in Linguistics and Communication Science as well as several edited volumes and conference proceedings. Jannis Harjus University of Innsbruck, Austria Since 2016, Jannis Harjus has been a post-​doc and teaching staff member at the Institute of Romance Languages at the University of Innsbruck (Austria). After studying history, pedagogy and Hispanic linguistics at the Universities of Bremen (Germany) and Córdoba (Spain), he received his PhD from the University of Mainz (Germany) with a work on perceptual variety linguistics in Western Andalusia. He has taught linguistics courses at various universities in Germany and Spain, including the University of Seville, the University Complutense of Madrid, the University of Córdoba and the University of Bremen. His main field of research focuses primarily on variety linguistics (Andalusian and Mexican Spanish; socio-​phonetic and perceptual approaches) and on discourse analysis. His most recent publication is Sociofonética andaluza y lingüística perceptiva de la variación: el español hablado en Jerez de la Frontera (Iberoamericana Vervuert 2018). He has also published articles in peer-​reviewed international journals and anthologies about subjects related to variety linguistics, multimodal discourse analysis, intonation, translation for dubbing and studies on linguistic landscapes. John Lipski Pennsylvania State University, USA John Lipski is Edwin Erle Sparks Professor of Spanish and Linguistics at Pennsylvania State University. His research encompasses language contact, bilingualism and the contributions of the African diaspora to the diversification of Spanish. His recent work combines ethnographic and psycholinguistic approaches to the study of bilingual speech communities in San Basilio de Palenque, Colombia (Palenquero-​Spanish), northern Ecuador (Quichua-​ Media Lengua) and northeastern Argentina (Portuguese-​ Spanish), with particular emphasis on code-​ switching constraints and morphosyntactic processing costs. Andrew Lynch University of Miami, USA Andrew Lynch is an Associate Professor of Spanish and Latin American Studies at the University of Miami, where he serves as Director of the Spanish Heritage Language Program. He also serves as Editor-​in-​Chief of Heritage Language Journal. His research and teaching focus broadly

List of contributors  xvii on sociolinguistics, societal language contact, and critical theories of language. He has published extensively on the topics of Spanish in the United States, Spanish language variation, and heritage language theory and research, and is a co-​author of El español en contacto con otras lenguas (Georgetown University Press, 2009) and editor of Spanish in the Global City (Routledge, forthcoming). Eva Núñez-​Méndez Portland State University, OR, USA Eva Núñez-​Méndez is Professor of Spanish Linguistics at Portland State University (Oregon, USA), where she has been part of the Department of World Languages and Literatures since 2002. She received her PhD from University of Salamanca (Spain). She has taught linguistics courses at various universities in Europe and the USA, including the University of California at Berkeley, Middlebury College (Vermont), the University of Puerto Rico (Río Piedras), the University of Houston (Texas), the National University of Ireland (Galway), and the University of Portsmouth (England). Her main field of research focuses primarily on linguistics (phonetics, syntax and history of Spanish language) and on applied linguistics (language variation, text analysis and translation). Her recent publications include Biculturalism and Spanish in Contact: Sociolinguistic Case Studies (Routledge, 2018), Diachronic Applications of Hispanic Linguistics (Cambridge Scholars Publishing, 2016)  as editor; and Fundamentals of the History of the Spanish Language (Yale University Press, 2012)  as single author. She has also published numerous articles in peer-​reviewed international journals such as Hispanófila, Linguistica, Hispanic Research Journal, Romance Studies, Languages for Specific Purposes, IJLassso and Estudios Hispánicos. Madeline Reffel Boston University, USA Madeline Reffel is a doctoral student in the Department of Linguistics at Boston University. She studies language variation and change, sociophonetics, and the outcomes of dialectal contact. Her research explores these themes as they relate to Spanish-​speaking communities in the United States. John Stevens University North Carolina Wilmington, USA John Stevens is an Associate Professor of Spanish at UNC Wilmington where he teaches courses in Spanish linguistics and language. Originally from La Jolla, California, Professor Stevens holds a bachelor’s degree in Spanish and international relations from the University of Southern California,

xviii  List of contributors a master’s degree in Linguistics from San Diego State University, and a doctorate in Hispanic Linguistics, also from the University of Southern California. His research interests include the acquisition of Spanish as a second language, phonology and Spanish dialectal variation. He is particularly interested in the connection between study abroad experiences and improvement in Spanish language proficiency. Mark Waltermire New Mexico State University Mark Waltermire is an Associate Professor of Linguistics at New Mexico State University. His main research areas are sociolinguistic variation and language contact, particularly regarding the linguistic results of Spanish in contact with Portuguese along the Uruguayan-​Brazilian border and Spanish in contact with English in the United States. He has published original research findings in these areas in journals as well as in the edited collections Laboratory Approaches to Spanish Phonology, Language, Borders and Identity, and The Influence of English on US Spanish.

Acknowledgments

I would like to express my gratitude to the contributors for their support in the development of this book. Without their collaboration, this project would never have been possible. Their remarks and suggestions have contributed to the present outcome of this volume. The journey to complete it has incurred other debts, especially to those attentive readers who kindly reviewed and advised us on sections of the manuscripts, providing feedback to the authors. We would like to thank the anonymous reviewers for their very thorough readings of the manuscript and their helpful comments and feedback. In particular, we would like to thank all of our colleagues in the field that graciously offered their professional opinion in the process of reviewing the manuscript:  Emilia Alonso (Ohio University), Robert Blake (UC Davis), Paul Chandler (University of Hawaii), Melvin González (University or Puerto Rico), Manuel Gutiérrez (University of Houston), Cynthia Kauffeld (Macalester College), Carol Klee (University of Minnesota), Maria Irene Moyna (Texas A&M University), Haris Symeonidis (University of Kentucky), Damián Vergara-​Wilson (University of New Mexico) and Mark Waltermire (New Mexico State University). We are also grateful to Robert Daly for reading a draft of this book, spotting inconsistencies, and offering numerous valuable comments. We thank Routledge and the entire publishing team, for their professionalism and help with the publication of this collection. Thank you! My thanks are also due to my students for their interest in sociolinguistics, their enthusiasm, and their input. Finally, and most importantly, very special thanks go to Portland State University for financially supporting this and earlier research projects with a Faculty Enhancement Grant.

Abbreviations

And. Andalusian BOC Boston Spanish Corpus Braz. Brazilian or Brazilian Portuguese C or Cons. consonant ca. abbreviation for Latin circa, approximately Cast. Castilian Cat. Catalan COG Center of Gravity Cont. Freq. (CF) contextual frequency F0 fundamental frequency F2 second formant Fls. St. false start Hes. hesitation HL heritage language HSS heritage Spanish speaker Lat. Latin L1 first language, mother tongue L2 second language med. Medieval or medial position mod. modern NP Noun Phrase O. Fr. Old French O. Sp. Old Spanish OZC Otheguy-​Zentella Corpus of Spanish in New York PC or Prev. Cntx. previous context Phon. phoneme Port. Portuguese Post-​M Post-​Mariel Pre-​M Pre-​Mariel [s]‌ unvoiced alveolar fricative sibilant, IPA [s̺] unvoiced apico-​alveolar fricative sibilant, IPA; [ś] in RFE [s̪ ] /​[z̪ ] unvoiced and voiced dental fricative sibilants in IPA alphabet; [ş] /​[z̧ ] in RFE

newgenprepdf

List of abbreviations  xxi [ş] /​[z̧]

unvoiced and voiced dental fricative sibilants in RFE alphabet; [s̪ ] /​[z̪ ] in IPA sC coda /​s/​before consonant; also word-​medial, pre-​ consonantal (implosive) position, i.e. s in asma ‘asthma’ s#C coda /​s/​at word boundary before consonant; also word-​ final, pre-​consonantal position, i.e. s in ríos ‘rivers’ s#V coda /​s/​at word boundary before stressed vowel s#v coda /​s/​at word boundary before unstressed vowel s## coda /​s/​at phrase boundary Sp. or Span. Spanish, modern varieties Std. Standard Unkn. unknown v. unstressed vowel V. stressed vowel VhV stressed vowel + aspiration of /​s/​+ stressed vowel VOT voice onset time vs versus Wd. Pos. (WP) word position * Vulgar Latin, not recorded or documented < > letters, graphemes [:] semicolon indicates a long sound [ ] sounds, allophones: variations of a phoneme with no contrastive value. E.g. [s]‌[z] in modern Spanish are allophones of the phoneme /​s/​, e.g. hasta [ásta] and asma [ázma]. E.g. [ʒ] and [ʃ] do not generally contrast in the Rioplatense Spanish as they are allophones of the phoneme /​ʒ/​ [s̺ ] IPA. Apicoalveolar fricative voiceless consonant sound; dialectal in north of Spain /​ /​ indicates phonemic transcription, a broader transcription than phonetic transcription. Symbols contained within have contrastive value, e.g. /​s/​, /​z/​as sip /​sip/​ and zip /​zip/​ in English [ʔ] glottal stop Ø null, no results

Introduction Eva Núñez-​Méndez

This volume presents an edited collection of original contributions, all of which focus on the current state of Spanish sibilants and their variations among Spanish dialects. Social processes and the nature of language change have driven sibilant variation across the Spanish-​speaking world. To reflect the wide range of variables and approaches to this topic, we have followed an overall geographical orientation. In so doing, the book is organized into three main thematic sections according to the type of socio-​geographical scenario that characterizes those particular regions: Spain, United States, and Central and South America. Part I, Spain, includes three chapters: an overview of the sibilant merger and its development in Spanish; a study on the Western Andalusian dialect of Jerez in Southwestern Spain; and an analysis of sibilant cases of voicing in Spanish in contact with Catalan. Part II concentrates on the United States, also with three chapters. This section explores speech varieties in the United States with close examination of Caribbean Spanish speakers in Boston and New  York City, variable realizations of final /​s/​in Miami Cuban Spanish, and variations of voicing among heritage speakers of Mexican Spanish. Part III brings together five chapters on Central and South American dialect zones: Western Nicaragua; Barranquilla, Colombia; Ecuador; Uruguay, a zone in contact with Portuguese; and, finally, northern Buenos Aires, Argentina. To help navigate through the volume, we offer the following brief discussion of the chapters included. Opening Part I, Chapter 1, “An overview of the sibilant merger and its development in Spanish,” offers the reader a summary of the evidence, chronology, dialectal divergence and phonetic background of the sibilant merger to understand current variations. In doing so, it helps to contextualize this particular phonetic variation, beginning with its origins in the Iberia Peninsula. Placing the phenomena within a historical framework is essential for following and exploring its expansion to modern cross-​dialectal realizations. Modern regional variation stems from the historical lack of stability in these sounds. Condensing what prior scholarship has established, this chapter seeks to present an overview of the sibilant merger, while addressing

2 Introduction the most recent research on its complex evolution. This chapter is useful for those interested in the history of the sibilant merger and its later expansion across dialects, as well as those with specific questions about the documentation of the sibilant merger development. Chapter  1 sets the foundations for tracing and framing the sibilants’ geolinguistic features discussed in the following chapters. In Chapter  2, Jannis Harjus analyzes socio-​phonetic data on Jerezano Spanish with regard to sibilants. He investigates the metalinguistic knowledge about a possible phonic norm in western Andalusia that diverges from the national standard of Peninsular Spanish. His results show a clear division between a ceceante (Jerez and rural western Andalusia) and a seseante zone (Seville capital and the other provincial capitals of western Andalusia). He argues that the speakers themselves believe in the non-​existence of a Sevillian norm based on seseo for the speech community. In Chapter 3, Justin Davidson provides an examination of select linguistic and social factors to explain Barcelonan Spanish intervocalic /​s/​-​voicing. Through interviews and phrase-​list readings, his findings are consistent in favoring greater voicing led by Catalan-​dominant younger females in casual speech. Alongside a near-​categorical absence of voiced Spanish [z]‌production in word-​initial contexts, he states that intervocalic /​s/​-v​ oicing is a vital contact feature of Catalonian Spanish. In Part II, centered on sibilant variations in the United States, Chapter 4 authors, Daniel Gerard Erker and Madeline Reffel, focus on syllable-​final /​s/​ among speakers of Caribbean origin who reside in Boston and New  York City. Considering acoustically based descriptions, spectrographic and waveform determination, along with other quantitative measures, both authors conclude that patterns of variation in syllable-​final /​s/​are largely conserved across generations among Caribbeans in these two cities. They first explore and ultimately reject the hypothesis that increased life experience in the United States promotes intergenerational shifts in /​s/​variation. In Chapter 5, Andrew Lynch and Antoni Fernández Parera question the hypothesis that the Miami-​born grandchildren of early exile Cubans tend to produce higher rates of the syllable-​and word-​final /​s/​to differentiate themselves from more recent Cuban immigrant arrivals on social and ideological grounds. They compare the realization of final /​s/​among Miami Cuban heritage speakers whose grandparents were early exiles in the 1960s and 1970s (pre-​Mariel) with that of same-​age speakers whose families left Cuba in the 1980s and 1990s (post-​Mariel). Their results show a reversal of the evolution of [s]‌> [h] > [Ø], widely spread reduction since the sixteenth century, conditioned by sociopolitical and ideological factors, as well as the possible influence of English. In Chapter 6, Amanda Boomershine and John Stevens present data from acoustic analysis of variable voicing of coda /​s/​in the speech of heritage speakers of Spanish living in the United States. They compare the production of syllable-​final and word-​final /​s/​in different voicing contexts by both

Introduction  3 heritage speakers and first language speakers of Mexican Spanish. Their research shows how heritage speakers produce coda /​s/​in a way that is similar to native speakers, unlike second language speakers who rarely voice /​s/​before voiced consonants. Part III, targeting Central and South America, opens with Chapter  7, where Whitney Chappell documents the reduction of coda /​s/​in Western Nicaragua. She points out the emergence of sibilant hypercorrections as the result of this growing reduction in pronunciation, i.e. observos for observo “I observe.” This happens among older speakers of Nicaraguan Spanish with lower levels of education. This group has less experience with coda sibilance than more educated (and younger) speakers, whose exposure to formal Spanish and external varieties tends to be greater. The author accounts for it as an individual variation and also as an age-​graded change. If speakers are exposed to coda sibilance in school but later, being removed from the workforce, are more exposed to glottal variants, speaker’s exemplar representations may shift over time. In Chapter  8, Richard File-​Muriel, Earl Brown and Michael Gradoville look at s-​realization in the Colombian Spanish of Barranquilla, taking into account individual physiology and gender features. They claim that although linguistic behavior has been regarded as purely socially learned, it may also be partially explained by simple physiological differences between females and males. Measures of oral cavity length and vocal fold size can significantly condition s-​realizations. Their results support the oft-​reported gender effect that females tend to produce more canonical s-​realizations overall. Also, individuals with smaller vocal fold size exhibit higher levels of voicelessness, and speakers with shorter vocal tract produce higher centroid measures, independently of the gender. They concur that physiological variables and gender influence linguistic behavior. In Chapter 9, John Lipski takes us to Ecuador. His study shows that the behavior of sibilants varies considerably among Ecuadoran dialects as well as among social strata and in ethnolinguistic enclaves. Coda /​s/​-w ​ eakening characterizes the coastal region, while voicing of word-​final prevocalic /​s/​is found to varying degrees in the central highlands. In the Andean highlands, rhotics also receive sibilant realizations. He explains that the voicing of intervocalic /​s/​does not appear to be a carryover from early colonial Spanish, although remnants of Spanish /​s/​realized as [z]‌are found in the Quechua-​ Spanish mixed language known as Media Lengua. In Chapter 10, Mark Waltermire sheds light on the linguistic and extralinguistic variables that condition the use of syllable and word-​final /​s/​variants in the Spanish spoken along the border between Uruguay and Brazil, mainly in the city of Rivera. He describes that the rate of /​s/​aspiration in Rivera is much lower than that in surrounding national varieties, most likely due to the influence of Portuguese (which lacks aspiration). Aspiration is favored among females and students, those who prefer Spanish and speak it often, and those who have negative attitudes toward Border Uruguayan Portuguese.

4 Introduction Waltermire concludes that Portuguese remains a powerful force with respect to maintenance and deletion of plural /​s/​in this border bilingual region. This collection ends with Chapter  11 by Michael Gradoville, who addresses variable voicing in Argentinian alveopalatal sibilants. His research reconciles recent studies on the devoicing of /​ʒ/​in Argentine Spanish. Taking into account that post-​pausal contexts favor devoicing for word-​initial /​ʒ/​, he maintains that the devoicing process is not as frequent, regardless of whether the speaker is a voicer or devoicer. He argues that the occurrence of voiced variants is not random, but rather a consequence of high-​frequency sequences. As a whole, this volume presents various approaches and analyses that show how sibilant variations across speakers, speech communities, dialects, communicative situations, and time, provide an immense wealth of data to shape new proposals in the field of language variation and change in the Spanish language. Of the 11 chapters, two combine variationist and diachronic analyses (Chapters 1 and 5) while others cover a range of other methodologies including quantitative data analysis (Chapters 3, 7, 8, 10 and 11), along with acoustic assessments (Chapters  4 and 6), and descriptive approaches to sociolinguistic factors (Chapters 2, 4, 9 and 10). Alongside, five chapters address language contact situations:  Spanish in contact with English (4, 5 and 6), Spanish in contact with Quechua (9), and Spanish in contact with Portuguese (10). All chapters are linked together as they present their original research on one particular sound class, Spanish sibilants, within a variationist sociolinguistics framework. The variation of /​s/​pronunciation has attracted more attention than any other aspect of variation in Spanish phonology. There are many articles and research that deal with the pronunciation of /​s/​in Spanish scattered in manuals, encyclopedias and books. This collection innovatively aims to group representative pieces of research under the same criteria, offering an overview of the sibilant development (Chapter 1), along with new insights on sibilants in Spanish dialects from both descriptive and quantitative perspectives. Through a variety of methodological and analytical angles, this project offers a collection of 11 chapters on sibilant variations in 11 Spanish-​speaking geolinguistic communities. Due to the fact that Spanish is widely spoken as the de facto official language in 21 countries and has more than 483 million native speakers worldwide, it is quite challenging to organize a collection of research that is inclusive of all possible variables in the articulation of sibilants. Nevertheless, our solution has been to select and include new significant contributions in the field, following a geographical distribution from a pan-​Hispanic lens. The research selected here, written by well-​known scholars as well as young and promising researchers, has demonstrated the most advanced sociolinguistic approaches regarding cross-​dialectal changes in sibilant pronunciation. Consequently, this project is relevant and innovative because it presents a collection of studies on one particular sound class, Spanish sibilants, focusing on different processes undergone by /​s/​(e.g. voicing, devoicing, weakening, elision), and on different

Introduction  5 contact situations with a wide range of social and theoretical tools. It also presents both diachronic and synchronic commentaries and includes heritage speakers and under-​researched Spanish-​speaking communities (e.g. in Nicaragua, Boston, Miami). Therefore, it examines one sound class from different angles, including a representative variety of Spanish dialects. The volume’s justification fits within the existing literature on sibilants in the areas of phonetics, phonology, sociolinguistics, dialectology, and language variation and change. While there are numerous pieces of research on sibilants in each of these subfields, there is no one volume that ties together this one class of sounds through a variety of methodological and analytical angles. No volume to date has been dedicated specifically to an examination of sibilant change and variation. This compilation represents an attempt to fill this gap. In sum, this project’s scope lies in sociolinguistics and dialectology, revolving around the central theme of language variation and change as evidenced in sibilant variables across Spanish dialects. The sibilant merger is an essential component for understanding and explaining Spanish variation, differences between Transatlantic and Castilian varieties (e.g. phenomena like seseo, distinción, ceceo), dialectal changes, and speech features in American Spanish communities. This volume’s originality is based on its concise approach to one of the most studied variables in Spanish phonetics, taking into account geographically based phonetic variation, sociolinguistic factors, and various Spanish language contact situations. Providing a thorough and updated sociolinguistic analysis of sibilants across Spanish-​ speaking regions has driven us to complete this project, with the hope of contributing to and balancing the current literature on this engaging topic. We hope that these chapters serve as a springboard for discussion and an inspiration for further research. It is our goal that this collection provides students and scholars with new materials and insights into how languages change and the importance of those changes in shaping linguistic and social phenomena.

Part I

Spain

1  An overview of the sibilant merger and its development in Spanish Eva Núñez-​Méndez

The evolution of the medieval sibilant phonetic system is crucial to understanding the origin of Castilian varieties on both sides of the Atlantic. It helps to distinguish varieties such as Andalusian Spanish, Judeo-​Spanish, and trans-​Atlantic Spanish, which constitute proof of many of the diachronic processes happening during and after the late medieval period. This chapter helps the reader understand the linguistic variations of the sibilants in the modern language and explores the origins and different steps in their evolution; furthermore, it evaluates recent research about the timeline of these phonetic changes. Placing the phenomena within a historical framework is essential for following and exploring its expansion into modern cross-​dialectal realizations. Modern regional variation stems from the historical lack of stability in these sounds. Condensing what prior scholarship has established, this chapter seeks to present an overview of the sibilant merger, while addressing the most recent research on its complex development. This chapter sets the foundations for tracing and framing the sibilants’ geolinguistic features discussed in the following chapters. The reorganization in the sibilant paradigm greatly altered the configuration of the Spanish language as we know it today. It also made this language unique compared to other Romance languages that still maintain certain similarities with the old medieval Spanish sibilant system. These transformations happened at different stages and contexts throughout long periods of time, mainly between the fifteenth and seventeenth centuries. The old medieval sibilants presented seven phonemes, distributed in the following pairs, according to their voiceless or voiced, and fricative or affricate features: Fricatives:  /​s/​ –​ /​z/​   /​ʃ/​ –​  /​ʒ/​ Affricates:  /​ts/​ –​ /​dz/​  /​tʃ/​​ The voiceless/​voiced sound pairs [s]‌~[z], [ʃ]~[ʒ], and [ts]~[dz], went through various consonant mutations such as devoicing (loss of the voiced sound), weakening of the affricates (spirantization), and most recently,

10  Eva Núñez-Méndez inter-​ dentalization and velarization, processes unique to Spanish among other peninsular and Latin derived languages. Studying the spoken language in the far past is a difficult task as the recorded manuscripts and data written in Spanish from those centuries are not always reliable. Phonetic changes happening in the spoken language were hardly represented in written form during those early periods, especially before the fifteenth century, and more specifically before the invention of the printing press in 1440. After that date, printing accelerated the spread of culture and the printed word, giving us a record of what was occurring at the oral level. However, many times the orthography did not correspond chronologically to the spoken language. Following a conservative tradition of what was considered to be the scholarly tendency in a given region, the written language did not always match the reality and practicality of the spoken language. Therefore, the study of the evolution of sibilants, based on old texts, rhymes, and records from grammarians and authors from that time presents a challenge and is in many ways restricted to very few documents. The main resources used here come from old poems, written texts from medieval authors, and opinions documented by linguists between the fifteen and seventeenth centuries. Thanks to these resources (useful but limited), some benchmarks have been established to indicate the initial changes, the expansion, and generalization, which resulted in the loss of the phonemic distinction between those pairs. The further we go back in time, the more difficult it is to obtain accurate information about the chronological evolution of a specific phonetic change. The number of written resources decreases, as does their readability and authenticity. Diachronic analysis presents multiple difficulties, especially due to the lack of correspondence between the articulatory phonetic changes happening in the spoken language and their written form. To study the evolution of sibilants, the oldest resources we have are written documents and rhymes, and authors’ opinions from the fifteenth century. The study of this phenomenon has advanced thanks to these written records; however, the accuracy of these documents could prove to be insufficient, as we all know that the written language does not always correspond to the spoken language. Besides, writers’ opinions and style could be influenced by their own perception of what they believed or desired the standard language to be. Analyzing the sibilants’ evolution using old written testimonies presents some methodological obstacles:  first, due to the lack of sufficient graphic documentation showing changes in the spoken language; and second, due to the inaccuracy of these written texts with regards to the chronology of the phonetic change. Nevertheless, and based on the generally accepted hypothesis that the sibilant changes date back to the sixteenth century, this study roots its conclusions in those first written literary manuscripts, including poetic rhymes, and in the later printed testimonies of grammarians and scholars.

An overview: sibilant merger in Spanish  11 Spoken language analysis for distant historical periods presents many challenges for gathering textual evidence. There are many factors to be taken into account, such as the limited number of surviving documents, restricted access to them, unreadable original scripts (many handwritten), the reliability of true originals versus copies, the deciphering of texts and spelling errors, arbitrary punctuation (far from modern norms), and the variability of scribes’ writing customs (mainly based on birth origin and scribal school tradition). Let us add that this lack of standardization in style, including spelling, paragraphs, use of space and capitals, differs greatly from present conventions. Nevertheless, and despite the misleading relationship between writing and speech, these old texts are our only evidence for determining five century-​old language practices. Bearing these challenges in mind, this diachronic approach discusses the timing of the readjustment of sibilants, which certainly brings up some dissension among current academics, and how significant it was to the emerging Spanish varieties, especially those in the south of the peninsula with their trans-​Atlantic projection. There is a substantial corpus of research on the devoicing and development of Spanish sibilants (Espinosa 1935; Catalán & Reinhardt 1957; Dámaso Alonso 1962; Amado Alonso 1967a; Martinet 1955; Malkiel 1971; Parodi 1976; Lloyd 1987; Harris-​Northall 1992; Pensado 1993; Penny 1993; Galmés 1962, among many others). It is not within the scope of the present article to analyze every single aspect of the historical complexity of this phonetic revolution, but to present an updated and straightforward diachronic synopsis that considers the implications of the timing for the trans-​Atlantic expansion of the language.1

1.1  Retrospectives and perspectives: origins According to written evidence, the testimonies of multiple grammarians, and the traditional philologists’ hypotheses, Old Spanish distinguished three sibilant pairs with voicing contrast up until the sixteenth century, as shown in Table 1.1. These six phonemes were represented by the following graphemes in orthography:

Table 1.1 Voiceless and voiced sibilant phonemes Voiceless

Voiced

/​s/​ /​ʃ/​ /​ts/​

/​z/​ /​ʒ/​ /​dz/​

12  Eva Núñez-Méndez a) /​s/​voiceless alveolar fricative, was written as as in passo “step,” passar “to pass”; b) /​z/​voiced alveolar fricative, was represented as , mostly in intervocalic position although also in other word interior positions; e.g. rosa “rose”; c) /​ʃ/​voiceless alveopalatal fricative, was written as as in dixo “s/​he  said”; d) /​ʒ/​voiced alveopalatal fricative was represented by as in fijo “son,” mugier “woman”; e) and finally, dento-​alveolar affricates /​ts/​and /​dz/​were written as and respectively, so decir “descend” and dezir “say” constituted a minimal pair, denoting different meanings. The distinction among these sibilant phonemes could differentiate such words as casa “house,” caxa “box” and caça “hunt”; fixo “fixed” and fijo “son,” as explained in Table 1.2. Even though these orthographic tendencies were in place, the written form often presented many spelling errors, some of which could show the training of the scribes, their different levels of literacy, or the changes occurring in the spoken language. This was a period in which standardization was incomplete, so spelling errors were still common. Frequent errors found across many conventional texts could indicate the historical state of the language; however, isolated errors did not necessarily represent changes in the spoken language. On the other hand, scribes, transcribers, or editors, when writing or copying a manuscript, might have taken some liberties to regularize the spelling, changing the original spelling to what they thought was right at the time. Scribes did not always copy the original manuscripts faithfully. Kauffeld (2016: 184) gives us a very specific example of these transcription practices that affect the analysis of Old Spanish sibilants. She states that there were two variations of the Greek sigma, ς and σ, used extensively and indiscriminately for the graphs , and in the fourteenth and fifteenth-​century manuscripts. This graphic trait may have caused great confusion for transcribing later editions, at a time when the sibilant system was already changing in the spoken language. Words such as poso/​pozo, casa/​ caza, decir/​dezir might have been misspelled due to the interpretation of sigmas used indiscriminately as in poςo or poσo, caςa or caσa, deςir or deσir. Many transcribers chose to follow etymology to represent sigmas; others freely adhered to the visual impression, to whether the graph looked like

Table 1.2 Examples of medieval sibilant phonemes Examples

decir

dezir

passo

casa

fixo

fijo

graphs phonemes

ç, ce,i /​ts/​

z /​dz/​

-​ss-​ /​s/​

-​s-​ /​z/​

x /​ʃ/​

j, ge, i /​ʒ/​

An overview: sibilant merger in Spanish  13 an or like a . The word caça “hunt” could have been transcribed as casa or caza depending on the visual perception of the scribe. These writing practices add more misinterpretations to the study of the sibilant merger in its early stages. We must also consider that most of the manuscripts that survived were related to official, judicial or literary matters. Therefore, they were written in a formal register, far from the spoken language of the time. They were also written by a very select group of the total population, those that could read and write, very few in medieval times. Those scribes did not represent the mass population nor the language spoken by the vast majority. The invention and spread of the printing press during the end of the fifteenth century helped to speed the dissemination of written texts without the laborious, intensive and long process of copying by hand. The faster production of printed texts required certain standardization of the language related to the use of lower and upper-​case letters, punctuation, symbols and spelling. There was a demand for the standardization of the written language among the printing houses, and for having grammar manuals with orthographic guidance at a time when Castilian was far from being standardized. In many ways, printing facilitated the study of language processes in the sixteenth century, where we find more documented sources by authors and contemporary scholars. This was a time when many transitions were happening in the sibilant system. Aside from the Old Spanish graphemes, if we look back at written sources in Latin, these six sibilants occurred in the following environments, examples adapted from Boyd-​Bowman (1980: 11). a) Sound [s]‌in Old Spanish came from Latin S or SS in all positions (except in those for [z]) as in S ALEM > sal “salt,” PASSAT U M > pasado “past,” FA LS UM > falso “false,” MUROS > muros “walls.” b) Sound [z]‌in Old Spanish was the pronunciation of Latin S in intervocalic position, e.g. ROS A M > rosa “rose”; or before a voiced consonant as it is in modern Spanish cosmos (from CO S MŒ) . c) Sound [ts] was the pronunciation in Old Spanish for Latin C E , C I in initial or strong position, and for the combination of consonant plus C E , C I , D E , D I and TE, T I , and for sporadic merging with [s]‌, e.g. C E N T U M > ciento “hundred,” M A RTI UM > marzo “March,” S E RARE > cerrar “to close.” d) [dz] in Old Spanish came from Latin C E, CI, T E , T I between vowels and from GE, GI preceded by n or r, e.g. FAC ER E > hacer “to do,” P U T E U M > pozo “well,” A RGI LLA M > arcilla “clay.” e) Sound [ʃ] was realized in Old Spanish for Latin X, P SE , P SI , SSE , SSI , or for Arabic shin (Boyd-​Bowman 1980: 10). E.g. in AXE M > eje “axis,” C A P S E A M > caxa, caja “box,” BA S S I U > baxo, bajo “low.” There are some cases of merger between Old Spanish [s]‌and [ʃ] as in SAP ON E M > xabón, jabón “soap,” S EPI A M > xibia, jibia, sepia “cuttlefish,” wa sa Allah > oxalá, ojalá “God willing.”

14  Eva Núñez-Méndez f) [ʒ] in Old Spanish came from Latin C’ L, G ’ L , L I and from J before o or u. E.g. O C ( U)LUM > ojo “eye,” FO LI A M > hoja “leaf,” JU V E N E M > joven “young.” Also, there are some cases of initial Latin I in strong position as in IA M MAGI S > jamás “never ever,” I U DAE U M > judío “Jew” (Boyd-​ Bowman 1980: 71). From late Vulgar Latin until the early stages of Castilian, these sound groupings underwent many changes, with the resulting drastic reduction of the sibilant system in the sixteenth century. By no means were these changes carried out uniformly in the peninsula; they went through considerable vacillation before they were standardized depending on the social and geographical proximity of the regions to the Castilian variety. The only sibilants not affected were the alveolar /​s/​and the alveopalatal /​tʃ/​as in modern oso “bear” and chico “small” respectively. 1.1.1  Possible causes for the devoicing and the unstable sibilant patterns The loss of voiced sibilants in medieval Spanish exemplifies one of the most important sound changes in the evolution of the language, with extraordinary repercussions in its trans-​ Atlantic journey. Besides the prepalatal sound [tʃ], which did not undergo any modifications, along with its voiced correspondent [dʒ] (kept as a context variable sound allophone of the current palatal phoneme /​ʝ/​), Old Spanish presented six other sibilant consonants, contrasting with each other by the feature of voicing. In the Castilian territories, four of these six consonant sounds [s]‌, [z], [ts] and [dz], were reduced to two, [s] and [θ], as in coser/​cozer. In the south, in what is known today as Andalusian Spanish, the dominant and most widespread result was [s] for the standard seseo (versus the stigmatized dental sibilant [s̪] for the ceceo variety, similar but not identical to the standard Castilian [θ]). The other pair [ʃ] and [ʒ] merged into a non-​sibilant velar sound [x] as in modern gente. Both resulting sounds [θ] and [x] are unique to standard peninsular Spanish in the family of Romance languages. While with interdental sound [θ] there was an articulatory advance in the point of contact to between the teeth, from [ts], [dz] to [s̪] to [θ]; with velar sound [x] there was a retreat all the way back to the velar area from the alveopalatal points of sounds [ʃ] and [ʒ]. The resulting modern sound [s]‌from the medieval sibilant inventory (of [s]~[z] in the north, together with [ts]~[dz] in the south) is articulated differently according to its origins in northern or southern peninsular dialects. Among the many resulting variations, apicoalveolar, laminodental, laminoalveolar (Quilis 1993:  283), the pronunciation of tends to be (apico)alveolar in Northern Spain [s̺], and (dorso)alveolar in most of Andalusia and Latin America [s].2 See Table 1.3. There is no clear explanation as to why these changes happened only in this variety of Latin in the northern part of the peninsula and not in other Romance languages. Many experts attribute this merger to a very general

An overview: sibilant merger in Spanish  15 Table 1.3 Evolution of Old Spanish sibilants Evolution of Old Spanish sibilants /​s/​~/​z/​ /​ts/​~/​dz/​

> /​s/​ in Andalusian & Latin American Spanish

> alveolar [s]‌

/​s/​~/​z/​ /​ts/​~/​dz/​ /​tʃ/​ /​ʃ/​~/​ʒ/​

> /​s/​ > /​θ/​in Castilian as in masa/​maza, minimal pair > /​tʃ/​as in chico > /​x/​as in general

> (apico)alveolar [s̺]

tendency of phonetic weakening in the first stages, and eventually, to creating a greater distinctiveness among consonants (Penny 2002:  101). The loss of sibilant voicing contrast that developed in the varieties of Late Latin spoken on the Iberian Peninsula is widely accepted as a consequence of the intra-​ linguistic consonant assimilation, weakening and articulatory relaxation occurring in the Western Romance territories. The process of weakening, or reduction in articulatory effort, is considered “lenition.” This term, coined by Thurneysen, describes consonant conditioning which is caused by a reduction of the energy employed in their articulation, and mostly affects consonants in intervocalic position (1946: 74). In this phonetic process, every consonant assumes a new (weaker) articulation, especially when it is surrounded by open articulations, mostly vowels and w, but also, when followed by 1, r, n, (Martinet 1952: 192). It corresponds to a generic pattern that applies not only to sibilants but other consonants as well in many other non-​Romance languages. One of the explanations is that this trend was due to the influence of Celtic articulatory habits or a Pre-​Celtic substratum. Although the phenomenon extends beyond the original Celtic-​speaking domains (as in Catalonia, south of Spain, and Portugal), the geographical distribution of the phenomenon largely coincides with at least some of the sections of Western Europe where Celtic languages must have been spoken around 300 BC . In Celtic, practically any consonant is affected by lenition (Martinet 1952: 203, 214). Indeed, the linguistic term lenition was used exclusively in reference to Celtic languages until 1950 (Lass, Szigetvári, & Zuraw 2008: 91). The lenition tendency of Celtic seems to be a good hypothesis to explain these articulatory weakening processes, although its diachrony still remains undetermined. Other assumptions, far from the Celtic lenition theory, assume that each sound followed its own path or that the Western Romance consonantal development resulted from parallel evolution determined by structural analogy, assimilation, aerodynamic instability and effort reduction. It still remains controversial to consider these processes as lenition. Among linguists, there is little agreement on the criteria for grouping a phonetic change or process under “lenition.” Degemination, deaspiration, voicing/​devoicing, spirantization, flapping, debuccalization, gliding and the loss of sounds are commonly considered leniting processes (Gurevich 2011: 10).

16  Eva Núñez-Méndez The phenomenon of sibilant devoicing was frequent and became regular in intervocalic position in the evolution of Castilian. In the Middle Ages, intervocalic voiced sibilants began to merge with their voiceless counterparts, while other Romance languages (both in and out of the peninsula) such as Portuguese, Catalan, French, Italian and Judeo-​Spanish kept this voiced-​ voiceless contrast, as can be seen with a few examples in Table 1.4. Within the general context of consonantal lenition, the devoicing of medieval Spanish sibilants is unusual, setting Spanish apart from other sister languages. The medieval voiced sibilants occurred almost exclusively in intervocalic position (Bradley & Delforge 2006: 43); therefore, the devoicing also happened mostly in that articulatory context. Nevertheless, devoicing can also occur in a plosive context, as is still the tendency in modern words such as ciudad “city” where the final voiced –​d is weakened, disappearing entirely, or is devoiced to [θ] or even to [h]‌in some peninsular dialects (Ariza 2004: 18). Different theories have been proposed to answer the question of why devoicing happened in Castilian versus other Latin languages. Briefly, it is worth noting some of them: a) The low functional and lexical contrast of the sound, mainly in the case of sound [z]‌, which did not appear often in the vocabulary. b) There is also the need to go forward with a phonetic system that is easier to articulate and has a higher value of functionality and balance (Ariza 2004: 19). c) Following the tendency of functionality, Alarcos (1951: 32) explains that it is due to a process of articulatory economy and simplification; the voiced/​voiceless opposition was scarce and reduced to the intervocalic position, becoming almost redundant. d) Lenition could also be caused by a readjustment in the phonetic system due to the laxing of the tension or intensity of the articulation (Veiga 1988b: 59). The aerodynamic instability of voiced fricatives and affricates is explained by their limited duration and the buildup of intraoral pressure that prevent vocal-​ fold vibrations from continuing (Zygis, Fuchs, & Koenig 2012: 399). In sync with this hypothesis, lenition is also explained by the effect of aerodynamic constraints on changes in the fricative’s place of articulation. Voiced fricatives are relatively difficult to produce due to the antagonistic aerodynamics requirements for voicing (low oral pressure) and for frication at the supraglottal constriction (high oral pressure). Thus, they require a pressure difference sufficient to generate turbulence (Ohala & Solé 2010: 53). e) Similar to d), a shorter duration, in terms of acoustic and articulatory features, is another possible reason for devoicing as it has been proven with laboratory experiments (Martínez Celdrán 1992: 630). f) Most recent research indicates that frequency, probability, entropy and surprisal shed light upon the mechanisms of uncertainty and ambiguity in

newgenrtpdf

Table 1.4 Examples of sibilant phonemes PLATEAM

PUTEUM

CAPTIA v. CAPTARE *CAPTIARE

CASAM

CAUSAM

CAPSAM

FIXUM

FILIUM

Spanish (Madrid norm)

plaza ‘square’ /​θ/​ plaça /​ts/​ plaça /​s/​ praza /​θ/​ praça /​s/​ piazza /​tts/​ place /​s/​ praça /​s/​

pozo ‘well’ /​θ/​ pozo /​dz/​ pou –​ pozo /​θ/​ poço /​s/​ pozzo /​tts/​ puits –​ podzo, pozo /​z/​

caza ‘hunt’ /​θ/​ caça /​dz/​ caça /​s/​ caza /​θ/​~/​s/​ caça /​s/​ caccia /​ttʃ/​ chasse /​s/​ kaça /​z/​

casa ‘house’ /​s/​ casa /​z/​ casa /​z/​ casa /​s/​ casa /​z/​ casa /​z/​ chez –​ kaza, casa /​z/​

cosa, causa ‘thing, cause’ /​s/​ cosa (cosa) /​z/​ cosa, causa /​z/​ cousa, causa /​s/​ coisa, causa /​z/​ cosa, causa /​z/​ chose, cause /​z/​ cosa /​s/​

caja ‘box’ /​x/​ caxa /​ʃ/​ caixa /​ʃ/​ caixa /​ʃ/​ caixa /​ʃ/​ cassa /​ss/​ caisse /​s/​ caxa /​ʃ/​

fijo ‘fixed’ /​x/​ fixo /​ʃ/​ fix /​ks/​ fixo /​ʃ/​ fixo /​tʃ/​ fisso /​sso/​ fixe /​ks/​ fixo /​ʃ/​

hijo ‘son’ /​x/​ fijo /​ʒ/​ fill /​l/​ fillo /​ʎ/​ filho /​ʎ/​ figlio /​ʎ/​ fils /​s/​* fijo /​ʒ/​

Old Castilian Catalan Galician Portuguese Italian French Judeo-​Spanish

Note: */​s/​ in French fils is not etymological.

An overview: sibilant merger in Spanish  17

LATIN

18  Eva Núñez-Méndez

g)

h)

i)

j)

k)

the sibilant system prior to dissimilation, following Information Theory basic principles (Zampaulo 2013: 172).3 In accordance with this approach, high and low frequency elements tend to lead to language change: low frequency and articulatory complexity contribute to instability thus promoting change; furthermore, elements with extremely distinctive cues will also be unstable. Sounds and sequences at the extreme ends of the noticeability pole are less stable. Common sound sequences would be situated away from surprisal extremes, and therefore, less prone to change (Hume & Mailhot 2013: 43). Amado Alonso explains that the devoicing was an internal process due to the vibration or rehilamiento of the sibilants within the general trend of consonantal weakening. A consonant with less vibration has less sonority (Amado Alonso 1955: 379). His opinion follows Rousselot’s compensation law (no longer accepted as valid) in experimental phonetics, which states that in the articulation of a sound when two organs are involved if one organ gets stronger, the other weakens; therefore, if the vocal cords are very active, there will be less energy in other articulators (such as the lips or tongue). This means that the voiceless sounds are stronger, thicker and more intense than their voiced counterparts (Serrano 1982:  212). Like A. Alonso and Alarcos, Pensado agrees with intralinguistic evolution explanations for this change; however, she adds that it is not due to the weakening of the consonant but to intense friction. When there is stronger friction, there is less vibration in the vocal cords. The friction level is much more intense in the sibilants than in the rest of the fricatives. Therefore, an intense friction articulation is incompatible with the voicing (1993: 199, 214).4 The main factor to cause this merger was the loss of the occlusive element in sounds [ts] and [dz] according to experts like Alvar (1983:  132), Lapesa (1957:  86), Catalán & Reinhardt (1957:  309) and Cock (1969: 13). In line with internal triggers for the devoicing, Catalán & Reinhardt argue that it was not due to a phonetic process, but rather a shift in the phonemic system (1957: 320). Phonological systems tend towards symmetry, such that contrastive features will tend to be used in parallel across the inventory (Zygis et al. 2012: 300). This is observed, for example, in classical distinctive feature theory (Jacobson and Halle 1956; Chomsky and Halle 1968; Clements 1985; Hall 2001). The unstable sibilant patterns are a result of the influence of the Basque language, which does not have voiced sibilants (Martinet 1955:  316). However, other experts such as Dámaso Alonso and Alarcos disagree as Galician, Aragonés, and Leonés do not have voiced sibilants and have not been in contact with Basque (Cabrera Morales 1992: 6). Another less plausible possibility is based on the influence of the Leonés dialect, a theory widely rejected as the devoicing also happens in another close dialect such as the Aragonés (Ariza 2004: 19).

An overview: sibilant merger in Spanish  19

Figure 1.1 Devoicing of fricatives and affricates.

l) Devoicing could have also happened partially due to the influence of Mozárabe, as proposed by Dámaso Alonso (1972: 138), or even Arabic, as is suggested by Galmés (1965: 92). m) The influence of the substratum explains the devoicing, following the Celtic or pre-​Celtic approach. Basically, explanations for the unstable medieval sibilant system and its results can be summarized under three main viewpoints: first, the influence of other languages (including the substratic hypothesis) thanks to external dynamics; second, intra-​linguistic articulatory factors due to internal processes; and finally, a combination of these two causes. In other words, it was due to substratum by external causes, internal evolution, or a mixture of both. Without a consensus, scholars in the field vary their theories based on internal or external factors to explain the devoicing of fricative sounds ([z]‌and [ʒ]) and affricates ([dz], [dʒ]), which eventually created very unstable sibilant patterns in medieval Spanish and the resulting foundations for southern varieties and trans-​Atlantic Spanish. See Figure 1.1. 1.1.2  Expansion of the merger: geographical implications The traditional approach to the original localization and further expansion of this phenomenon is based on the idea that it went from the north to the south of the peninsula, following closely the increasing reach of Castilian in other non-​Castilian territories. Most experts have agreed on this tendency;5 nevertheless, the regularity and intensity of its completion differ according to different geographical areas. The consolidation of use was slow and intermixed in many regions. The change of cultural and political centers in Castile in 1561, when Madrid replaced Toledo as the Court center under Philippe II’s reign, influenced the speech norm dynamics. Madrid became a model site for the northern pronunciation, as the new Court speakers were associated with the innovative speech form. Meanwhile in Toledo there was still a traditional sense of the pronunciation. Many testimonies have recorded the differences between the pronunciations from Old Castile, with Madrid as its center, from those of Toledo in New Castile. In 1578, Juan de Córdoba (from Andalucia) writes:

20  Eva Núñez-Méndez Those from Castile say haçer and in Toledo they say hazer. And they say xugar and in Toledo they say jugar. And they say yerro, and in Toledo they say hierro. And they say alagar and the others say halagar, and many other words that I will skip here to avoid prolixity.6 This illustrative quote shows that there were pronunciation differences within Castile itself, depending on the areas; therefore, the confusion was not at all uniformly widespread. It is strongly believed that the merger started in Old Castile, in the north, and expanded from there.7 It spread southward towards Toledo and from there it expanded towards the east and west: Extremadura, Murcia, Andalucia and America. This sudden phonetic revolution was born from a dialect area in Castile, where devoicing had existed for some time already; it was not born out of the Court speech. Speakers of the northern meseta, from Benavente to Burgos, and speakers in the Basque region followed this old Castilian practice. It is remarkable how this dialectal practice from the far north of the Guadarrama mountains attained rapid prestige among the educated classes in Philip II’s Madrid and easily overthrew the traditional Toledan Court speech. This dissident practice without voiced sibilants sprang forth in Castile, and with little opposition, became the norm in the speech of the Spanish-​speaking community. Parallel to the socio-​political transformation of the sixteenth century in the Madrid of the Counter-​reformation, this dialectal innovation succeeded in opposing the communal norm and imposing itself over the prestigious norm of the imperial Court of Toledo.8 Due to the fact that the phenomenon started in the north, some linguists have defended the Basque influence in the confusion (there is almost an exclusive predominance of voiceless consonants in Basque); however, this theory of the Basque role has been discarded as the devoicing was not only characteristic of Castilian, it also affected Galician and the Leonese and Aragonese dialects, even the Valencia Apitxat variety (Cabrera 1992: 6), which were not in direct contact with Basque. The theory suggests that the extension of the phenomenon was early in the north and from there it spread much later to the south; nevertheless, it was a very slow process with multiple intersections of contradictions supported by the textual evidence. The south was resistant to implement this new tendency from the north. This delayed feature is proven by some instances in a few dialects from provinces south of Salamanca and north of Cáceres where speakers still use some of the voiced sibilants in words such as vecino “neighbor.”9 Based on textual examples, we see that the graphic confusion was not distinctively separating the northern and southern speech practices. In the sixteenth century, there are many written records by Andalucian writers such as Nebrija, Guillén de Segovia, A. de Palencia, Fernando de Herrera and Juan Sánchez that maintained the graphic distinction, indicating that differences were still noticeable between the voiced and voiceless sounds in oral speech. For instance, in Nebrija’s Reglas de Orthographia (1517), he

An overview: sibilant merger in Spanish  21 explains that there was a different level of articulatory tension, in his own words apretado “strong” for /​s/​and floxo “weak” for /​z/​. Nevertheless, a century later, in the seventeenth century, another Andalucian, Mateo Alemán in his Ortografía castellana (1609), severely criticized those that wrote with such a graphic distinction that was nonexistent in the spoken language. On the other hand, if we consider authors from the north-​center such as Teresa de Jesús (from Avila) and Juan de Valdés (from Cuenca), some contradictions exist. De Jesús does not make graphic distinctions and she writes tuviese, matasen, açer, deçir, dijera, teoloxia por tuviesse, matassen, hazer, dezir, dixera, teologia; nevertheless, Juan de Valdés, also from Castile, explains in his Diálogo that he writes when the pronunciation is stronger and also writes with words that some Spaniards neither pronounce nor use in writing. His examples could indicate that authors were either still trying to preserve the traditional written norm or that the devoicing had not yet spread to that area. Also, we can find early devoicing in some southern territories as some examples from Cancionero de Baena (copied in Andalucia) and Fernando de Rojas’s testament from Toledo in 1541 demonstrate (Cabrera 1992: 7). See Map 1.1. We have seen that this phonetic phenomenon of devoicing spread southward from the north of the peninsula over a long period of several hundred years during the reconquest. The expansion followed the same tendencies as the spread of the Castilian language: from the center-​north towards the south, almost in a triangular shape, leaving the marginal territories for the development of sister languages: Catalan, Galician and Portuguese. Uniformity was not characteristic of this change, and as a result, two different tendencies grew from the loss of phonemic distinction: one in the north, the Castilian norm; and another in the south, the Andalusian norm. These two phonetic trends set the foundations for major dialectal variations and for the Spanish transatlantic phase. According to the geographic peninsular areas, and considering the articulatory complexity of the affricates /​ts/​and /​dz/​, the neutralization of these two

Map 1.1 Expansion of Castilian (11th–​15th cent.) and the spread of sibilant devoicing.

22  Eva Núñez-Méndez phonemes had different outcomes in the Castilian and Andalusian norms. /ts/ and /​dz/​were weakened to dental fricative sounds [s̪] and [z̪] in the south, which finally devoiced in [s̪] and merged in [s]‌; while in the north /​ts/​and /​dz/​ weakened directly to a dental sibilant sound [s̪], which became interdental [θ] to distance itself from the resulting alveolar [s] from the pair /​s/​and /​z/. Therefore, these two phoneme pairs /​ts/​~/​dz/​and /​s/​~/​z/​evolved differently in the north-​central and the southern parts of the peninsula: two solutions /​θ/​and /​s/​resulted in the north, with contrastive value as in coser/​cozer, casar/​ cazar, poso/​pozo etc.; and only one in the south, /​s/​, which created leveling and homonyms, the precedent for seseo. See Table 1.5. Table 1.5 Castilian versus Andalusian Spanish: distinción and seseo

Evolution of voiceless interdental fricative /θ/ calça

examples

‘trousers’

graphemes

ç, ce,i

sounds

[ts]

fizo

cossa ‘thing’

‘house’

z

-ss-

-s-

[dz]

[s]

[z]

‘did (3rd person sing.’

casa

16th cent.

[s]

[s]

1650

[θ]

[s]

Castilian Spanish: North-Center Distinción

Evolution of Andalusian seseo calça

examples

‘trousers’

graphemes

ç, ce,i

sounds

[ts]

fizo

cossa ‘thing’

‘house’

z

-ss-

-s-

[dz]

[s]

[z]

‘did (3rd person sing.’

casa

14th cent. 16th cent.

[s]

17th cent. present

[z] [s]

[s] seseo

[s] ≈ [θ] ceceo

Andalusian Spanish Seseo

An overview: sibilant merger in Spanish  23 On the other hand, the resulting sound [s]‌both in the north and the south had slightly different articulations within the same alveolar point. In the north, it was pronounced with the very tip of the tongue, as apicoalveolar [s̺]. In the south, the dorso-​alveolar [s] was instead pronounced with a flatter tongue. The latter is the common resulting [s], shared with Portuguese, Italian, French and English. In contemporary Spanish, the seseo indicates that speakers use the phoneme /​s/​, the sole sibilant survivor of those medieval changes in the south, for graphemes ,, and . In all dialects of standard Spanish, with or without distinción, modern [s]‌presents phonetically gradient and variable voicing in syllable final position when it is preceding a voiced consonant.10 Therefore, the phoneme /​s/​can be realized as voiceless [s] and as voiced [z] sounds according to the context.

H

   [s]‌ e.g. sí “yes,” casi “almost,” más “more,” hasta “until” /​s/​    [z]‌+ voiced cons. e.g. asma “asthma,” desde “from,” Israel It is crucial to point out that, both in the northern and southern variations, positional markedness played an important role in tracing the development of Spanish sibilants from the medieval period to the modern variations. The interior-​intervocalic position is the most prone to articulatory changes compared to the strong initial word position. Syllable initial sibilants showed standard articulatory faithfulness to be voiceless, while voiced sibilants rarely occurred outside of the intervocalic pattern. The original phenomenon of seseo that started as a differential Romance dialectal characteristic between the north and the south in the peninsula became a point of departure for transatlantic Spanish varieties. Consequently, the evolution of Andalusian Spanish or andaluz, with seseo as one of its main features, plays a decisive role in understanding linguistic variations in American Spanish. 1.1.3  Spanish emigration to the New World: andalucismo in the precolonial period The development of seseo and distinción marked a very significant phase of transition in the history of Spanish. From there on, the Castilian variety would distinguish itself from other trans-​peninsular variations spread to the Americas, the Philippines and Africa, in addition to other strategic Mediterranean sites occupied by Judeo-​Spanish. The Andalusian Spanish speech would become the foundation for many modern Spanish varieties. The practice of seseo, mainly, and other characteristics (such as the weakening of syllable-​final /​s/​ and yeísmo) from the southern dialect are commonalities for the Spanish spoken outside of the peninsula. “There are many features of American Spanish which demonstrate that southern Peninsular tendencies

24  Eva Núñez-Méndez have successfully gained the upper hand in all or most of Spanish America” (Penny 2002: 25). The spread of seseo in the new continent arrived with the first waves of Spanish colonists in 1492 and the continuous demographic settlements at the end of the fifteenth century. The initial stage started in the Antillean region, in Santo Domingo (Dominican Republic), and from the Caribbean islands began the conquest of the mainland, Mexico; from there, it continued over Central and South America. Every year there were expeditions with new settlers, mostly from Andalusia, especially coming from Seville and Huelva. Considering data on demographic distribution, gender, maritime jobs, manual-​labor occupations and economic class, the Andalusian presence was always well represented in the early phase of colonization, as Boyd-​Bowman’s figures demonstrate. His findings confirm that of the 5,481 settlers studied in the early colonization period from 1493 to 1519, one in three colonists was Andalusian, one in five was from the province of Seville, and one in six was from the city of Seville itself. Sixty percent were from Andalusia, while Extremadura, the two Castiles, Leon and the Basques contributed roughly 6% each, and all other sources combined to 11% (1973: 3), as shown in Figure 1.2. In the second stage of the colonization from 1520 until 1539, Andalusians still remained as the major immigrant group with 32% of the 13,262 colonists, followed by Castilians at 18%, and Extremadurans at 17% approximately (Boyd-​Bowman 1973: 17), as explained in Figure 1.3. The Spanish cities that provided more immigrants to the new world were: Seville with 58% and Huelva with 20% of all the Andalusians migrating in the initial period. In the second period, half of all emigrants were from Seville, Badajoz, Caceres, Toledo, Salamanca and Valladolid, with Seville still furnishing one out of every six men and half of all the women. Between 1560 and 1579, roughly three out of every four emigrants came from the Southern half of the Peninsula and 28.5% were women. In later periods, Andalusian settlers lost ground principally to those of Extremadura and New Castile. Old Castile, Leon and the Basque provinces show negligible changes (Boyd-​ Bowman 1973: 17).

Figure 1.2 Boyd-​Bowman’s data:  regional origins of the earliest Spanish colonists between 1493–​1519 out of 5,481 settlers.

An overview: sibilant merger in Spanish  25

Figure 1.3 Boyd-​Bowman’s analysis: origins of population between 1520–​1539 out of 13,262 emigrants.

From the sociolinguistic perspective, if we consider gender and social status in the second period of colonization, Boyd-​Bowman indicates that just over 6% of settlers were women. Of those, over 58% were Andalusians (34% from Seville), followed by women from Extremadura and Old Castile with over 10% each. From 1493 to 1519, one of every two women was from Seville city or the province; later on, still during the early period of colonization, it was one of every three Spanish women arriving in America. The presence of Andalusian women had a great impact on language development as they were linguistic models for the next generation, in addition to setting the precedent for the prestige of the Seville speech norm among white women in the colonia. Regarding social standing and occupations, almost 48% of the sailors were from Andalusia; nevertheless, positions of leadership, power and nobility, such as governors, captains, bookkeepers, bankers, merchants, accountants, scriveners, councils, priests, artisans, and so on, came from Old Castile. Boyd-​ Bowman’s data identifies a higher proportion of population in leadership from Castile, almost 20%, although that population made up just 18% of the colonists; while Andalusia provided 24% of the leaders but represented 32% of all the migrants. These numbers resemble a correspondence with modern Spanish-​American dialectal distribution:  those in the coastal regions show similar features to Andalusian trends, while those in the highlands, specifically in the old colonial capitals and cultural centers close to the viceregal courts, have kept a conservative linguistic tendency, mainly due to closer connections to the Peninsular hierarchy, the Court and culture of the Spanish Crown.11 The majority of Spanish settlers came from popular classes, which permeated the type of language that would spread in the new continent. “For every noble man that made it to the New World in the first period, 10 unrestrained men of low and dark origins would come along” (Oviedo 1959: 36).12 American Spanish has been recognized to be more popularly oriented than European Spanish, with a tendency to maintain archaisms, colloquialisms and vulgar usage when compared to the literary and Court Spanish of Madrid. In many ways, in its first stages, it reflected what was happening with the common people of southern Spain in the fifteenth century, who were mostly illiterate, uneducated, and of low status, with a popular rustic speech.13

26  Eva Núñez-Méndez The colonial demographics between 1520–​1539 studied by Boyd-​Bowman show that the southern predominance was always present in the earliest dialect mixing in America except for Nicaragua, Venezuela and the New Granada Kingdom (now Colombia) where Castilians were the largest group. In other continental territories, the most important colonist groups were approximately as follows in percentage and numbers (Boyd-​Bowman 1973: 27–​32), as shown in Table 1.6. Thanks to Boyd-​Bowman’s extensive geo-​demographic records proving that Andalusians outnumbered other groups in the early stages of the colonization, the impact of the southern peninsular dialects in the configuration of Spanish in its early days in the New World is undeniable. Andalusian was indeed the most important influence in precolonial American Spanish; an assumption that initiated the andalucista theory supported by modern scholars.14 During the fifteenth and sixteenth centuries, two main cities in Andalusia had control of all the commercial traffic to the Indies: Seville and Cadiz. Since the House of Trade, la Casa de Contratación, was in Seville, many settlers came from the city itself. If they came from other areas in the peninsula, they lived in Seville, sometimes many years, before embarking for the Americas. Natives or not, most emigrants had to stay lengthy periods in Seville. Furthermore, they spent weeks aboard ships on their transatlantic journey, providing more opportunities to permeate other settlers’ pronunciation habits before arrival. Among Sevillian innovative speech patterns at the end of the fifteenth century, the seseo was the predominant one and was directly transferred to the Canary Islands and to American soil. Even those linguists that reject the andalucista theory of American Spanish accept the andalucismo of the seseo, as the most frequent feature in all colonial documents, including those from South America. They concur that seseo could not have been developed independently from peninsular Spanish. Furthermore, they accept this phonetic trend as a sevillanismo proper (Moreno de Alba 2007: 48).15 In the sixteenth and seventeenth centuries, two southern cities, Seville and Cadiz, monopolized all the maritime traffic between Spain and the Indies. This was a time of profound changes; not only in socio-​politics but also in the way people were speaking. The pronunciation was changing on both sides of the Atlantic with Seville as the meeting point for departure to the colonies. Seville became the only hub for travelling to America, the link between the Old and the New Worlds. The first creoles to speak Spanish were mainly exposed to the Seville and Andalusia varieties as a consequence of demographics and commercial sailing logistics.16 Based on Boyd-​Bowman’s conclusions that in the initial Antillean period by far the largest single group, in every year, and on all major expeditions, was the Andalusians, the seseante speech has been recognized as the most significant linguistic item of the Andalusian influence in American Spanish. Once in America, in the early American-​Spanish koiné, the seseo was adopted by colonists from all parts of the Peninsula as a result of dialect contact and

newgenrtpdf

Table 1.6 Boyd-​Bowman’s data: regional origin of the earliest Spanish colonists of America between 1520–​1539 Andalusians

Old Castilians

New Castilians

Extremadurans

Basques

1,372 195 108 4,022 958 467 1,340 1,088 145 906

45.6% 626 41% 80 26.9% 29 35% 1,412 33% 316 25% 119 22.2% 297 41.3% 449 33.9% 49 18% 163

13.4% 184 17.4% 34 30.6% 33 17.3% 693 14.8% 142 21.7% 101 22.2% 298 14.7% 160 13% 19 20.5% 186

10.8% 146 7.2% 14 1.9% 2 12.6% 507 11.3% 109 7.5% 35 13.9% 186 9.9% 107 9% 13 13.8% 125

12.8% 175 15.9% 31 14.8% 16 14.8% 598 22% 211 22.7% 106 20.4% 274 6.3% 69 1.4% 2 12.7% 115

3.4% 46 3.1% 6 5.6% 6 4.4% 177 6% 57 3.9% 18 5.5% 74 4.9% 53 9.7% 14 5.7% 52

An overview: sibilant merger in Spanish  27

Santo Domingo (Dominican Republic) Cuba Puerto Rico México Panama area Guatemala and Chiapas Perú La Plata River Asunción (Paraguay) New Granada (Colombia)

Colonists

28  Eva Núñez-Méndez mixtures. The northern norm opposition of /​θ/​and /​s/​came to have little to no impact in areas of mixed speech during precolonial and colonial times, even though there were continuous waves of settlers coming from Old and New Castile and other regions during the sixteenth century.17 In this environment of dialect mixing, most speakers would simply adopt one feature to facilitate communication, making it the norm, instead of making a new but unpredictable distinction.18 Mergers are more frequent in the case of dialect mixture. Consequently, the pair [θ]/​[s]‌, pronounced by these early Castilian colonists, became less resistant to merger, and the result was the continuation and spread of seseo in the Americas. The linguistic background of early colonial times was a dynamic one including the following aspects: the predominantly Andalusian mix of settlers from all the regions of Spain, migratory shifts, dialect mixing, languages in contact (including Amerindian ones), and the growing social prestige of Spanish. The language varieties of those first colonists in long-​term contact went through adaptive processes and periods of accommodation. Colonists adjusted their speech, sometimes by eliminating minor variants, reproducing salient features in the speech of others, or mixing variants. In speech variety contact situations, it is common to encounter a gradual tendency to reduce, level, simplify and regularize one variant from a group of competing variants, a process known as koineization. The koine becomes the common or standard variety of a larger area where there are opportune conditions for dialect mixing and leveling. See Figure 1.4. Demographics and migration patterns must be carefully considered to describe the spread of Spanish in America in precolonial times. The impact of the Andalusian influence in the linguistic foundations of the New World has been highlighted; however, what was Andalusia like at the end of the fifteenth century? Being the most southern part of the peninsula, it was the latest to be reconquered from the Moors, and as a result, it was also the last to be resettled by Christians from the northern territories, which resulted in a repopulation shift with a consequent linguistic environment of speech and dialect leveling. The Andalusian speech of this period was far from standardized; it developed from a koine of mainly Castilian with Arabic and Mozarabic elements, from

Figure 1.4 Settler mixing and koineization.

An overview: sibilant merger in Spanish  29 Table 1.7 Linguistically significant cities in Spain during the 15th and 16th centuries

15th cent.

Linguistic Norms

16th cent.

Toledo (Court center until 1560)

Traditional Castilian

Madrid

Innovative Castilian

Madrid (capital) new prestige norm

Seville (trade with America)

Andalusian dialect

Andalusian dialect

which it borrowed multiple lexical voices that are still present in modern southern Spanish. Due to sociopolitical circumstances, transatlantic enterprises, maritime economics and trade, human mobility and demographic convergence, Andalusian became the pioneer dialect. It was the language variety spoken in Seville, not the one from Toledo or Madrid, that set the first linguistic norms in the colonization of the New World; furthermore, in the first stage of colonization, it was the Spanish koine from the peninsula with many antillanismos that was taken to the mainland by the first conquerors (Boyd Bowman 1964: 24–​25).19 A century after the discovery, the growing political importance of Seville was discarded by the preeminence of Madrid as the new capital. Old Castile centralization seriously impacted the competing linguistic norms. The reputation of the Toledo norm as the traditional stronghold of “good Castilian” (as Menéndez Pidal describes it)20 was replaced by the emergent innovative Madrid norm, which became the prestige norm from 1560. From this background the Andalusian dialect arose, grew and expanded. See Table 1.7.

1.2  Diachronic review and alternative accounts: phenomenon propagation and completion By the end of the sixteenth century, most areas under Castilian influence had finally eliminated voiced sibilants from the spoken language. This chronologic framework is generally accepted and in accordance with the opinions of grammarians and treatise writers, as well as attestations and graphic misspellings from documented records. Nevertheless, polemics arise when dating the origin of the phenomenon in its phase of transatlantic displacement. Was seseo a well-​established trait in the Andalusian koine at the end of the fifteenth century? Did the sibilants suffer from phonetic devoicing around the same time? Was this revolutionary phonetic shift the result of diachronic processes such a “falling domino” effect or push/​drag phenomena in Martinet’s sense of push chain (chaîne de propulsion) and drag chain (chaîne de traction)?21 The devoicing and evolution of these sibilants did not happen at the same time and were not homogeneous. Experts do not unanimously agree on the chronology of the changes; however, some general consensus can be extracted from their hypotheses. The departure point rests on the assumption that it was the Old-​Castilian “system,” the “innovative” norm of a language without

30  Eva Núñez-Méndez voiced sibilants, imposing itself on Madrid and Toledo and displacing the old Toledo Court system in the last third of the sixteenth century at the latest. It is not a question of transformation or evolution in speech but of the imposition of a foreign practice (Catalán & Reinhardt 1957: 287). The chronology of the devoicing goes hand in hand with the spread of the phenomenon southward, expanding from a focal point towards surrounding peninsular territories. It is important to remember the following brief steps in its history: a) During the medieval ages, there was a clear distinction between voiced and voiceless sounds as has been recorded in written documents from the times of Alfonso X in the thirteenth century. Before that time, such graphic difference cannot be attested due to the lack of records. b) From the end of the thirteenth century until the orthographic reform instated by Nebrija in the fifteenth century, there were some instances of devoicing, especially affecting the pair /​s/​~/​z/​; however, it is difficult to prove due to the graphic instability throughout the Middle Ages that occurred, despite Alfonso X’s orthography standardizing efforts. Nebrija consolidates the grapheme for the voiceless sound [s]‌. There are also a few examples of devoicing of the pairs /​ts/​~/​dz/​and /​ʃ/​~/​ʒ/​ studied by Menéndez Pidal (1919: 27–​29) such as façer (alternating with fazer), raçon, deçir, rayçes in a Mountain document from 1410, together with usso~uso. Similar cases appear in the fifteenth-​century manuscript 64 of the National Library of Madrid, such as raçones, diçesse, desfiçiesse, diçen, pobreça, façian, façienda, tristeça, bajas, baja, linguaxe. In the Oraçional de Cartagena, from the mid-​fifteenth century, many examples of graphic confusion were recorded for the sounds [s] and [z] as in casa~cassa, causa~caussa, esposa~espossa, guisa~guissa, inposible~inpossible, invisibles~invessibles, nescesarias~neçessarias, pasiones~passiones, pasar~passar, paso~passo, progreso~progresso, vasos~vassos, etc. In Arnalde and Griselda by Diego de San Pedro, published in 1492, there are similar examples: eso, necesidad, diese, sobrase, pasión en lugar de esso, necessidad, diesse, sobrasse and passión, and so on (Cabrera 1992: 4). c) At the end of the fifteenth century and beginning of the sixteenth, in the pre-​classic period, many more illustrative cases of the devoicing occur in written texts. However, it is the beginning of the spread of the phenomenon as many grammarians, treatise writers and rhymes of that period still confirm this distinction. Furthermore, we have the Judeo-​Spanish linguistic petrification from the 1492 expulsion as a testimony to the existing opposition between voiced and voiceless sibilants at the end of this century. d) At the end of the sixteenth century the phenomenon spread to more linguistically conservative peninsular areas and became consolidated as a general speech practice. It was in the south where its completion was most delayed.

An overview: sibilant merger in Spanish  31 e) By the beginning of the seventeenth century there were no voiced sibilants in the standard speech. f) The devoicing happened first in the north from early periods; while in Toledo it was a widespread phenomenon from the mid-​sixteenth century and was not generalized until the seventeenth century. g) The merger of the prepalatal/​palatoalveolar fricative pair /​ʃ/​~/​ʒ/​gives way to a new velar sound [x]‌. The [ʃ] articulation retreated to the velar area to become the velar fricative [x]. Some treatise writers and grammarians attest this at the end of the sixteenth century. Torquemada, López de Velasco and Oudin report that the sounds represented by the letters , and were articulated close to the throat, which implies that both the palatal and velar pronunciation coexisted at this time. In the seventeenth century, there are a few descriptions of this velar pronunciation. Jiménez Patón and Robles confirmed it; the latter writes that the sound for the letter was pronounced with the tongue far back in the mouth, almost twisted towards the throat, which indicates the modern velar realization.22 Correas still explains the articulation of this grapheme as palatal; however, being from Extremadura, his description could denote that the spread of velarization was not yet general in his southern region (Blanco 2006: 85). It is difficult to determine the exact date for the triumph of the velar sound [x]‌in the standard speech, as there is not much written documentation about this process. The general conclusion is that the palatal articulation was dominant during the sixteenth century and started to change at the end of it. After a long coexistence period of palatal and velar pronunciations, well advanced into the seventeenth century, the change from [ʃ] to [x] took place and the velar [x] became the norm. Nevertheless, the realization of this velar sound in the south became aspirated as a [h], merging with the aspiration descending from the initial Latin F -​, while in the north, the descendent of the prepalatal fricative pair favored a more intense articulation as an uvular fricative sound [χ]. This continues as a modern contrast between northern and southern dialects. h) Most linguists agree that the velarization of sounds [ʃ]~[ʒ] also has its origin in the north of the peninsula from where it spread southwards; however, from the chronological point of view, it was the devoicing (of [z]‌) that preceded the velarization in this area. In the center-​southern regions both phenomena probably extended simultaneously (Alarcos 1988: 56).23 See Table 1.8. Table 1.8 Chronology of devoicing and velarization North

1st 2nd

/​s/​~/​z/​  >  /​s/​ /​ʃ/​~/​ʒ/​ > /​x/​

Center-​southern areas

Simultaneously

/​s/​~/​z/​  >  /​s/​ /​ʃ/​~/​ʒ/​ > /​x/​

ej. casa [káza] > [kása] ej. dixo [díʃo] > [díxo] dijo

32  Eva Núñez-Méndez i) The confusion of the affricate sounds [ts]~[dz] evolved through various steps of spirantization (loss of their initial occlusive elements [t]‌~[d]), devoicing (loss of [z]) and interdentalization (creation of a voiceless interdental [θ]) with different results in the north (distinción) and the south (seseo and ceceo). The spirantization happened first with the voiced [dz] around the mid-​sixteenth century, while it affected the voiceless [ts] at the end of the sixteenth century. At the beginning of the seventeenth century both the spirantization and devoicing were complete, although there were some educated minority groups that made the distinction between sounds [ts] and [s̪] (based on their articulation as affricate/​fricative, not voicing) until the twenties of the seventeenth century. The interdentalization was happening at the end of the sixteenth century as is mentioned briefly by some writers; however, it took a long time to spread, and in the seventeenth century was still a minor phenomenon (Blanco 2006: 86). See summary in Figure 1.5. Venegas and Valdés’s testimonies confirm the distinction between sounds [ts] and [dz] until the mid-​sixteenth century. After that date, other writers such as Torquemada and López de Velasco describe the sound represented by the letter ([z]‌< [dz]) as being fricative. There are no more references to the spirantization of [ts] until the end of the sixteenth century when López de Velasco writes about it. Nevertheless, in the seventeenth century Correas, Bravo Grájera, Jiménez Patón, Luna, Pérez de Nájera and Villar inform us of the merger of the sounds of both letters and . Still other grammarians like Alemán and Bonet attest the distinction (at that point of sounds [ts] and [s̪]), criticizing the confusion as a common practice; in the same line, Sebastián, Dávila and Cascales add that the pronunciation of and is different but very close. These late testimonies of distinction in the seventeenth century contradict the opinion of most writers; the experts are inclined to believe that it was a conservative trend to resist innovation (Blanco 2006: 81). A. Alonso supports the idea that the opposition of voicing lasted longer, until the end of the sixteenth century.24 In the same line of thought, Lapesa points out that in Toledo the distinction between sounds [ts] and [s̪] did not survive the first 30 years of the seventeenth century.25 Catalán discards the idea that the opposition was functioning after having lost

Figure 1.5 History of the sound [θ].

An overview: sibilant merger in Spanish  33 the voicing; likewise, Cano Aguilar defends that, without the voicing contrast, there was no distinction in the manner of articulation (Blanco 2006: 83). According to treatise writers, the evolution of the sounds [ts]~[dz] had its key period between the middle of the sixteenth century and the beginning of the seventeenth century. At that time, the distinction and the merger coexisted, that is to say, in some regions the sound [ts] kept the old affricate pronunciation while in others it was fricative [s̪]; in some areas, the [z̪] was still voiced while in others it was already voiceless [s̪]. The distinction was articulated by a few and taught by a limited group of grammarians, while the confusion was practiced among most people and recorded by the vast majority of writers. 1.2.1  Chronological accounts: grammarians and treatise writers As a point of departure, it is very important to consider grammarians’ and treatise writers’ opinions about pronunciation habits and practices in the past in order to date phonetic changes. A few of them have been selected according to their time period and relevance. During the fifteenth and sixteenth century, the modern notion of voiced/​voiceless did not exist so the scholars used other ways to describe the articulatory contrast: for voiceless they used fuertes strong, apretadas tight, espesas thick, densas dens, recias stout, or of sonido entero whole sound; and for voiced, blandas soft, suaves soft, or of medio sonido half sound. The distinction was based on the energy used in the articulation of the sound. 1.2.1.1  Alveolar pair /​s/​~/​z/​ The first attestation of the voiced sound [z]‌comes from Nebrija’s Gramática (1492) where he differentiates the soft from the strong . He states that the only consonants that are doubled in Spanish are the and the ; which can be written as single letters according to their pronunciation.26 During the sixteenth century, scholars followed Nebrija’s description, terminology and distribution for both graphs and . Valdés, in 1535, also defends the different articulation between the thick (written as a geminated ss) from the non-​thick (written as a single letter); nevertheless, he criticizes the graphic confusion of these two letters.27 In 1558, Villalón also distinguishes between and , and in his Gramática castellana, he emphasizes several orthographic rules, clarifying the matter, which implies that the sounds were already merging in speech. According to A.  Alonso (1955:  348), Villalón, being from Valladolid, was defending the Old Castile practice of distinction, the Toledo norm, although the new devoicing had already spread among speakers. It is very likely that he himself was not speaking with this distinction. In 1589, Cuesta writes about the need to differentiate and in writing and insists in not mistaking and

34  Eva Núñez-Méndez misspelling them, which indicates that the mix-​up between the sounds [s] and [z] was common.28 In 1625, Correas does not write about this pair; he only writes about one type of , which proves that the voiced [z] was no longer differentiated. In 1609, Alemán, who does not distinguish them either, gets rid of and only uses the simple with two typographic variations: a long s at the beginning of the word or syllable and a short s and the end of the word or syllable.29 Like Alemán, both Sebastián, in 1619, and Bonet, in 1620, do not differentiate the letters and . However, around that time, Jiménez Patón, in 1614, still writes about the distinction and the that sounds soft between vowels,30 which is contradictory to what his contemporaries were describing. From sixteenth-​century grammarians’ statements, it can be concluded that the graphemes and had different articulations; however, from the seventeenth-​century reports, excluding Jiménez Patón’s, the inference is that there was not a voiced [z]‌pronunciation with graphic correspondence since the beginning of this century. 1.2.1.2  Dento-​alveolar pair /​ts/​~/​dz/​ Going back to the first grammar in Spanish written by Nebrija in 1492, there is no clear description for the sounds represented by and . Nebrija separates the sounds from Latin origins, which he names as propios “proper,” from sounds from Arabic, named as prestados “borrowed” or ajenos “foreign.” He considers that has a borrowed sound and relates its pronunciation to the Hebrew letter Samekh. On the other hand, he writes that the letter has been kept from Latin.31 Nevertheless, Nebrija does not define the way these letters are pronounced. His contemporary writers do the same; they acknowledge the difference in pronunciation without explaining it. In 1531, Venegas says that it is a compound letter, which suggests an affricate articulation.32 In 1535, Valdés adds that Castilian has added three letters to the Latin alphabet:  , and and that the sound of letter is espesso “thick” as in çapato, coraçón, açucar.33 In 1558, Villalón explains that has the same pronunciation as but pronounced twice; his definition is not clear; however, he seems to denote different pronunciations for both graphemes.34 In 1552, Torquemada writes that the sounds written with and have different but very similar pronunciations, and consequently, many people interchange them. He underlines that the is less strong, referring perhaps to the manner of articulation or its voicing.35 In 1582, López de Velasco states that the sounds for and are different; nevertheless, he confirms that the confusion was quite general among all people.36 In 1589, Cuesta, following López de Velasco, differentiates the letters , and and adds that the confusion was a common practice among children.37 In 1578, Fray Juan de Córdoba describes that the distinction was not practiced in Old Castile where speakers pronounced the same sound for both letters and .38

An overview: sibilant merger in Spanish  35 It seems that during the sixteenth century, although grammarians criticize the confusion, the sounds were perceived as different. On the other hand, during the seventeenth century, all authors attest the neutralization of these two sounds as common in speech; those that still write about the distinction (such as Alemán, Sebastián, Bonet and Dávila) point out that it is very weak and difficult to maintain, only practiced by a few, admitting that the in-​distinction has become a habit and widespread phenomenon. In 1625, Correas states that the distinction was lost in common speech and criticizes those authors that still defend different articulations for and .39 He also does not maintain the graphic distinction between , and , using only .40 In 1604, Pérez de Nájera censures the use of letter instead of , which means that the phonetic distinction was not practiced at that time.41 In 1614, Jiménez Patón writes that both sounds for letters and are limited to only , without any further reference to the letter .42 In 1620, Luna claims that the sounds for both letters and are the same.43 In 1634, Bravo Grájera admits the same sound for the two letters; however, he insists in writing them differently according to their etymological origin.44 In 1651, Villar acknowledges the same sound for both graphemes and leaves the distinction in writing to the author’s free will and guessing.45 1.2.1.3  Alveopalatal pair /​ʃ/​~/​ʒ/​ In 1492, Nebrija defines the sound of the grapheme as an Arabic borrowing, confirming that while it existed in Latin, the pronunciation was borrowed from the Arabs.46 Specifically, A.  Alonso identifies it with the Arabic sound shin (palatal voiceless fricative) (1949:  73). Likewise, Nebrija describes the sound of as belonging to Arabic.47 A. Alonso associates it with the Arabic sound gim (palatal voiced affricate) (1949:  73). In 1535, Valdés writes about the difference in sounds for the letters and or ; he clarifies the use of the grapheme for Latin X [ks] and of for the Arab sound [ʃ]. He also compares the sound of ~ with the Italian gi, which could denote a fricative [ʒ] or affricate pronunciation [dʒ].48 In 1552, Torquemada explains that the sounds of and or are very similar and many writers mistake them.49 In 1558, Villalón reports the indistinction; both letters and have the same sound.50 He, being an Old Castilian, may not have pronounced them differently; however, he also mentions that there is a slight differentiation.51 In 1578, Fray Juan de Córdoba tells us that Old Castilians could not distinguish between the voiced and voiceless palatals.52 In 1582, López de Velasco, although claiming their distinction, declares the confusion both in writing and in pronunciation.53 In 1589, Cuesta reports that many people (both children and elders) mistake these two sounds written as and .54 In 1597, Oudin, a French grammarian, explains that the sounds for letters and or were different but very close; he also adds that the sound for was articulated very close to the throat.55

36  Eva Núñez-Méndez Table 1.9 Chronology of sibilant devoicing according to grammarians’ reports Sibilant merger according to Grammarians End 15th cent. Mid 16th cent. End 16th cent. 17th cent.

Clear distinction of sounds [s]‌~[z], [ʃ]~[ʒ] and [ts]~[dz] The merger (voiced/​voiceless) began spreading Confusion was widespread and common; still some reports about distinction No distinction

At the end of the sixteenth century we find the first reports about the velar or glottal pronunciation of [ʃ] distancing itself from its palatal articulation. It is then that we can assume a regression in its point of articulation to velar [x]‌. In 1552, Torquemada already mentions this articulation in the throat.56 In 1582, López de Velasco observes the pronunciation of as articulated from inside the throat.57 In 1597, Oudin defines the sound of as articulated in the throat.58 In 1630, Correas still describes as palatal; nevertheless, he supplies us with the details that in Andalusia the sound for this letter is mistaken with that of the letter .59 This implies that if the aspiration was happening in Andalusian, then the velarization of [ʃ] to [x] was also already occurring in standard Spanish, though it was still in its expansion stage. In 1614, Jiménez Patón criticizes the Spanish pronunciation of letters and , as being different from other Romance languages, which suggests that they were already pronounced velar as [x] at this time.60 In 1619, Sebastián advises not to mispronounce the letters and , which have very similar articulations.61 In 1631, Dávila attests how difficult it was to keep the distinction between and .62 Treatise writers from the sixteenth century (except Nebrija and Valdés) agree that the confusion between [ʃ] and [ʒ] was quite common after the middle of the century. Grammarians from the seventeenth century confirm that the indistinction was general in this century, as shown in Tables 1.9 and 1.10. Grammarians’ opinions show important details about the chronology and geography of medieval sibilant changes. See below a summary with names, work titles (first printing), origins, and whether they comment on the sibilant merger (indicated by “+”) and where it happens. 1.2.2  Chronological accounts: graphemes If we consider the graphemes, the equivalency will be letters for sound [ts], for [dz]; double , initial , final for [s]‌, intervocalic for [z]; for [ʃ], or for [ʒ]. There is confusion when the writing does not regularly reflect this correspondence, that is to say, when words such as fuerça, fazer, esse, casa, dexar, ceja are written as fuerza, facer, ese, cassa, dejar and cexa. Only then can it be sustained that the writing was showing the lack of distinction in speech. Some illustrative examples follow from medieval Spanish (Blanco 2006) in Table 1.11.

newgenrtpdf

Table 1.10 Grammarians’ opinions about sibilant change Authors

Book Title

1st edition

Author’s Origin

Merger [s]‌~[z]

Merger [ts]~[dz]

Merger [ʃ]~[ʒ]

Alemán, Mateo (1547–​1615)

Ortografía castellana

1609, México

Seville, Andalusia

+ general [s]‌

+ general [ʃ]

Bonet, Juan Pablo (c. 1573–​1633)

1620, Madrid 1634, Madrid

Torres de Berrellen, Aragon Coria, Cáceres?

+ general [s]‌

Correas, Gonzalo (1571?-​1631)

Reduction de las letras y arte para enseñar ablar los mudos Breve discurso, en que se modera la nueva orthographía de España Arte de la lengua española castellana

+ people from Andalusia, Toledo and Old Castile [ts]~[s̪] +

1625 pub. [1954]

Jaraiz de la Vera, Extremadura

Cuesta, Juan de (?) Davila, Nicolás (?) Córdoba, Fray Juan de (1503–​1595)

Libro y tratado para enseñar a leer y escribir Compendio de la ortografia castellana Arte de la lengua zapoteca

1589, Alcalá de Henares 1631, Madrid

Guadalajara?

+ general [s]‌ +

Cartagena, Castile

+ children [ts]~[θ] +

1578, México

Cordoba, Andalusia

+ Old Castile

Jiménez Patón, Bartolomé (1569–​1640) López de Velasco, Juan (c. 1530–​1598)

Ortografía

1614, Baeza

Ciudad Real, Castile

 

[s]‌~[z]

+ general [ts]

Orthographia y pronunciación catellana

1582, Burgos

Soria, Castile

+ [s]‌~[z]/​ [s]

+ many [s̪]~[z̪]

+ general + general [s̪]

+ general [ʃ]/​ [x]‌? + many + general + [ʃ]~[ʒ] Old Castile + [x]‌ general + [ʃ]~[ʒ] (continued)

An overview: sibilant merger in Spanish  37

Bravo Grájera, Gonzalo (?-​1672)

+ general [ʃ]

newgenrtpdf

Authors

Book Title

1st edition

Author’s Origin

Luna, Alejandro de (?)

Ramillete de flores poéticas y notables jeroglíficos Gramática de la lengua castellana Grammaire espagnolle expliquee en françois Ortografía castellana

1620, Toulouse

Toledo, Castile

1492, Salamanca 1597, Paris

Lebrija, Sevilla, Andalusia

1604, Valladolid

Valladolid, Castile?

Othographia y Orthologia

1619, Zaragoza

?

Manual de escribientes

c. 1552–​59, Madrid pub. 1970 1535 pub. 1736 1531, Toledo

Astorga, Leon

1558, Amberes 1651, Valencia

Valladolid, Castile? Arjonilla, Jaen, Andalusia

Nebrija, Antonio (1441–​1522) Oudin, César (1560?-​1625) Pérez de Nájera, Francisco (1530–​1619) Sebastián, Miguel (?) Torquemada, Antonio de (c. 1507–​1569) Valdés, Juan de (c. 1500–​1541) Venegas, Alejo (c. 1497–​1562) Villalón, Cristobal (c. 1510–​1562?) Villar, Juan P. (1585–​1660)

Diálogo de la lengua Tractado de orthographía y accentos en las tres lenguas principales Gramática castellana Arte de la lengua española

Cuenca, Castile Toledo, Castile

Merger [s]‌~[z]

Merger [ts]~[dz]

Merger [ʃ]~[ʒ]

+ general _​_​ [s]‌~[z]

_​_​

_​_​ +

+ general + general [s]‌

+ [ts]~[dz]

+ general

+ many [ts]~[z̪]

+ many

+ [s]‌~[z]

_​_​

_​_​

+ [s]‌~[z]/​[s]

_​_​

_​_​

+ general

+ [ʃ]~[ʒ]

38  Eva Núñez-Méndez

Table 1.10 Cont.

An overview: sibilant merger in Spanish  39 Table 1.11 Examples of orthographic confusion from 1270 to 1680 [ts] written as

[dz] written as

[s]‌ written as

[z]‌ written as

[ʃ] written as

pezes cozes lanza dulzes cozes carzel laço perdiçes feliçes nariçes amenaça paces saçon racimo escaso paso escasas gruesa huesos promesas mjeses masa pasos hueso paso masa promesas escaso mieses pessar susso messes sesso besso prossa rrossa usso pessar pusse sesso quesso tejo lejos madeja queja lejos

ca. 1430 1600 1610 1635 1656 1674 1330? 1330? ca. 1430 ca. 1430 1605 1619 1652 1670 1270 1270 1300 1330? 1331 1330? 1330? 1344 1400 1528 1554 1612 1622 1653 1680 1270 1300 1330? 1330? ca. 1430 ca. 1430 ca. 1430 ca. 1430 1458 1559 1623 1660 1344 1528 1612 1620 1652

Cancionero de Baena El conde Alarcos Peribáñez y el comendador de Ocaña La vida es sueño Un bobo haze ciento La nave del mercader Libro de Buen Amor Libro de Buen Amor Cancionero de Baena Cancionero de Baena El curioso impertinente La prudencia en la muger El desden, con el desden Sueños ay, que uerdad son Primera Crónica General de España Primera Crónica General de España El libro de Caballero Zifar Libro de los estados El Conde Lucanor Libro de Buen Amor Libro de Buen Amor Crónica General de España Danza de la muerte Retrato de la lozana andaluza La vida del Lazarillo de Tormes Fuente ovejuna La fingida Arcadia El lindo Don Diego El indulto general Primera Crónica General de España El Libro de Caballero Zifar Libro de Buen Amor Libro de Buen Amor Cancionero de Baena Cancionero de Baena Cancionero de Baena Cancionero de Baena Tratado de las armas Los siete libros de Diana El mejor alcalde el rey La reina Maria Estuarda Crónica General de España Diálogo de Mercurio y Carón Fuente ovejuna El cavallero de Olmedo El desden con el desden

40  Eva Núñez-Méndez Table 1.12 Sibilant merger chronology based on graphemes Grapheme analysis of sibilant merger by 17th cent. [ts]~[dz] [s]‌~[z]

affricates alveolar fricatives

end of 16th cent. 17th cent.

[ʃ]~[ʒ]

alveo-​palatal fricatives

all periods

clear distinction confusion was a minor phenomenon constant distinction

From the grapheme corpus analyzed (Blanco 2006), the results confirm that the graphic distinction among the affricates [ts]~[dz] and fricatives [s]‌~[z] lasted until the end of the sixteenth century and the confusion was still minor in the seventeenth century. The graphic distinction between the palatal pair [ʃ]~[ʒ] was kept almost constantly, which could imply a phonological distinction in speech or just orthographic conservatism. However, the few graphic confusions for this pair jeopardize the study of its phonetic evolution. This chronological conclusion from graphemes diverges with the traditional hypothesis that the confusion of sibilants was a general phenomenon at the end of the sixteenth century and beginning of the seventeenth century. According to Blanco (2006:  100) the generalization was later and it did not finish until the beginning of the seventeenth century. See Table 1.12. Spelling in books corroborates the impression we get from the treatise writers:  in the first half of the sixteenth century all printers accepted the Toledan norm. On the other hand, from a little after 1580, Madrid printing-​houses published books with very abundant cacographies (Catalán & Reinhardt 1957: 295). In the first quarter of the seventeenth century, the Court speech from Madrid became the new standard: it is the triumph of the new Madrid norm. 1.2.3  Chronological accounts: rhymes Based on rhyme analysis of a corpus from the twelfth until the sixteenth century, gathered together by Blanco, some outcomes have been extrapolated to clarify the controversy presented by graphemes: whether the level of graphic distinction was determined by phonological changes, and when the confusion started. If we look at the words that rhyme, independently of the letters, most texts from old medieval and classic Spanish (up to the sixteenth century) denote the distinction as is illustrated with the following selection of examples.

An overview: sibilant merger in Spanish  41 Rhymes [ts]~[ts] 2.1. Esta enbidia es por la priuança   … que su daño faze aquel que la alcança   que preça vos vistes andar en balança   de grandes señores venjr a proueza   por ende tenet que faze proeza   … el que se qujta desta tribulança   Por ende señores con grant acordança   … a qujen sera dada la tal buen andança  (Cancionero de Baena, ca. 1430)

2.2. Que si el justo dolor mueve a vengança   alguna vez el español corage,   despedaçada con aguda lança   compensarás, muriendo, el hecho ultrage  (Herrera, Poesía Castellana 379, ca.1580) 2.3. Dichoso el hombre que de Dios alcança   ser corregido aquí por esto amigo   sufre su disciplina con templança   (Fray Luis de León, Poesía Completa 579, ca. 1590)

Rhymes [dz]~[dz] 2.4. Con gran desgradecimiento   Me pagas y satisfazes   quien deshizo nuestras pazes   quien turbo tu sufrimiento  (Cancionero de las obras, 1496)

2.5. Este Lauro, que tiene’n su corteza   … i en él el manso zéfiro resuena   mi mal, su resplandor i su belleza,   cuando el sol elevado en más alteza   … i entonces m’alegrava l’aspereza  (Herrera, Poesía Castellana 530, ca.1580)

Rhymes [s]‌~[s] 2.6. Después a las noches venciendo los dias   y estando las tierras hinchadas y gruessas   cobrando ya sombras los montes espessas   las silvas y bosques con hojas sombrias  (Cancionero de las obras, 1496)

2.7. El sueño diste al coraçon umano,   para qu’al despertar mas s’alegrasse   del estado gozoso, alegre i sano;   Que como si de nuevo le hallasse,   haze aquel interval, qu’a passado,   qu’el nuevo gusto nunca’l bien se passe.   (Garcilaso de la Vega, Obras 451, ca. 1530)

Rhymes [z]~[z]‌ 2.8. Y porque eres tan hermosa   te quiero mira veras   quiere me quiere me mas   pues por ti dexo a mi esposa   y toma toma esta rosa   … ni por mi se te de la cosa.  (Cancionero de las obras, 1496)

2.9. Confieso qu’es ansi, que nadie es parte   si Dios, respondió a Job, al hombre acusa   a con justa razon guarder su parte.   Que quien con la baraja, si ya vsa   de todo su saber, dara turbado   por mill acusaciones vna escusa.   (Fray Luis de León, Poesía completa 593, ca. 1590)

42  Eva Núñez-Méndez Rhymes [ʃ]~[ʃ] 2.10. Porque a los miembros y anexos   De san Juan el de letran   Todas sus gracias se dan   Aunque esten aca muy lexos.  (Cancionero de las obras, 1496)

2.11. Que, como en el amor le fuiste junto,   justo es qu’en tal estrecho no t’alexes   d’aquel divino i celestial trassunto.   I antes qu’el peso inútil veloz dexes,   lleva d’el muerto amante la memoria,   aunque tardando con razón te quexes…  (Herrera, Poesía castellana 720, ca. 1580)

Rhymes [ʒ]~[ʒ] 2.12. Es un enxambre de abejas   que todas van tras su rey   es un pastor y una grey   pastor de cien mil ovejas:   desvelar quemar las cejas   por privar y desprivar   cien mil consejos consejas   cien mil cosas nuevas viejas   un usar y desusar.  (Cancionero de las obras, 1496)

2.13. ¿Es éste’l fruto, Amor, qu’al fin recojo   d’el continuo servicio de mis años?;   ¿está es la cierta fe de tus engaños?;   ¿de tus promesas, éste es el despojo?   !Ai, que bien yo meresco el mal qu’escojo,   pues que cierto los ojos en mis daños,   i huyo de tus claros desengaños,   i contra mí tan sin razón m’enojo!  (Herrera, Poesía castellana 615, ca. 1580)

As a result of the rhyme studies, the sibilant merger must have started in the sixteenth century and the loss of [ts]~[dz], [s]‌~[z] and [ʃ]~[ʒ] oppositions was not general until the seventeenth century. Rhymes help to explain that the graphic differences between [ts] () and [dz] (); [s] () and [z] (); [ʃ] () and [ʒ] () suggest a phonological distinction, minor but still present, in the sixteenth century. However, once the devoicing spread in the seventeenth century, the graphemes , and (of voiced [dz], [z] and [ʒ] respectively) stopped resembling pronunciation practices and simply reflected inherited etymological patterns. See examples below of rhymes with confusion from the seventeenth century. Rhymes with no distinction [ts]~[dz] 2.14. Con la sombra del jarro y de las nueces   la sed bien inclinada se alborota;   todo graznate esté con mal de gota,   hasta dejar las cubas en las heces.   Los brindis repetidos y las veces   crezcan el alarido y la chacota,   y el agua aguachirle que las peñas trota   buen provecho les haga a rana y peces.  (Quevedo, Un Heráclito 378, 1613)

2.16. Espíritus sanguineos vaporosos   suben del corazón a la cabeza,   y saliendo a los ojos su pureza   pasan a los que miran amorosos,   … rayos sintiendo en la sutil belleza   como de ajena son naturaleza…  (Lope, Rimas de Tomé de Burguillos 790, 1634)

An overview: sibilant merger in Spanish  43 2.15. Pasos otro dio al aire, al suelo coces.   Y premïados gradüadamente,   advocaron a sí toda la gente,   … mancebos tan veloces…  (Góngora, Soledad Segunda 407, 1613)

These cases of rhyme confusion between [ts] and [dz] from the first half of the seventeenth century show that the distinction was not commonly maintained, and it was the voiceless variant that leveled the contrast. Therefore, those words that should have been pronounced with voiced [dz] were pronounced as voiceless and in rhyme with [ts] words: nueces (Lat. N U X-​ N UC IS ) , heces (Lat. FEX -​F ECI S ), veces (Lat. V I X-​V I C I S ) with [dz] and peces (Lat. P E S C E-​P IS C I S ) with [ts] in 2.14; coces (Lat. C AL X-​C ALC I S ) with [ts] and veloces (Lat. V E LOX -​V ELO C I S ) with [dz] in 2.15; cabeza (Lat C AP I T I A ) with [ts].63 pureza, belleza, naturaleza (Lat. suffix –​I TI A > -​eza) with [dz] in 2.16. Rhymes with no distinction [s]‌~[z] 2.17. Que no me quieren bien todas confieso,   que yo no soy doblón pra dudallo.   … Si me aborrecen, no será por eso.   Con quien tiene codicia tengo seso,   en pagar soy discípulo de el gallo,   … en estas retenciones que profeso.  (Quevedo, Un Heráclito 354, 1613) 2.18. Las redes califica menos gruesas,   Sin romper hilo alguno,   Pompa el salmon de las reales mesas,   Cuando no de los campos de Neptuno…  (Góngora, Soledad Segunda 435, 1613)

2.19. Esta cabeza, cuando viva, tuvo   Sobre la arquitectura destos huesos   carne y cabellos, por quien fueron presos   los ojos que mirándola detuvo.   Aquí la rosa de la boca estuvo   marchita ua con tan helados besos,   aquí los ojos de esmeralda impresos,   color que tanta almas entretuvo.  (Lope, Rimas Sacras 633, 1614)

The above rhymes from the beginning of the seventeenth century show illustrative examples of the lack of distinction between the voiceless [s]‌and the voiced [z] as confieso (med. Lat. CO NFES SARE ) , eso (Lat. I P SU M ) and profeso (Lat. P ROFES S US ) are pronounced with [s] rhyming with seso (Lat. SE NS UM ) with [z] in 2.17; gruesas (Lat. GROS S US ) with [s] rhymes with mesas (Lat. M ENS A M ) with [z] in 2.18; huesos (Lat. O SSU M ) with [s] rhymes with presos (Lat. P R E NS US ) with [z], besos (Lat. BA SI U M ) with [z] rhymes with impresos (Lat. IMPRES S US ) with [s] in 2.19.

44  Eva Núñez-Méndez Rhymes with no distinction [ʃ]~[ʒ] 2.20. Vase en el mundo dilatando el día   en cercos de oro y arreboles rojos,   y en las hojas las perlas del rocío;   mas cuando tan hermoso el sol salía   anocheció para mis tristes ojos…  (Lope, Rimas 149, 1602) 2.21. Bien sé que al viento doy quejas baldías,   pues antes de llegar a tus orejas,   con ir ardiendo en fuego, vuelven frías.   Pero veo también que si me dejas   el alma, el cuerpo y el honor perdido,   no importa que se pierdan estas quejas.  (Lope, Rimas 435, 1602)

2.22. Las aves que leyeren mis tristezas   luego pondrán en tono mis congojas   y cantarán mi mal en las cortezas   al son que hiciere el aire con las hojas.   Cualquier viento, templado a mis ternezas,   de las cuerdas, Amor, que no me aflojas…  (Quevedo, Un Heráclito 267, 1613) 2.23. De Alcides lo llevó luego a las plantas   que estaban, no muy lejos,   trenzándose el cabello verde a cuantas   da el fuego luces y el arroyo espejos.  (Góngora, Soledad Primera 327, 1613)

These verses from the beginning of the seventeenth century indicate that the merger of the prepalatal pair [ʃ]~[ʒ] favored the voiceless [ʃ]: rojos (Lat. RU S S U S ) with [ʃ] rhymes with ojos (Lat. O C U LU M ) with [ʒ] in 2.20; orejas (Lat. AU R IC U L AM ) with [ʒ] rhymes with both dejas (Lat. L AXARE /​DARE ) and quejas (Lat.*QUA S S I ARE/​Q UA S S ARE ) with [ʃ] in 2.21; both congojas (Cat. congoixa) and aflojas (Lat. FLUXUS ) with [ʃ] rhyme with hojas (Lat. F OL I AM ) with [ʒ] in 2.22; and lejos (Lat. LAXI US ) with [ʃ] rhymes with espejos (Lat. S P E C ULUM ) with [ʒ] in 2.23. Consequently, the results from the rhyme corpus analysis delay the generalization of the confusion to the seventeenth century, while the opinions of grammarians and scholars place it at the end of the sixteenth century. See summary in Table 1.13. Considering the opinions of grammarians, treatise writers, and scholars, the analysis of graphemes and rhymes, and the traditional hypothesis of the history of the language, the chronological changes can be approximately summarized as follow in Table 1.14.

Table 1.13 Chronology of sibilant devoicing according to rhymes Rhyme analysis (following pronunciation) 15th cent. 16th cent. 17th cent.

Clear distinction between [s]‌~[z], [ʃ]~[ʒ], and [ts]~[dz] Sibilant merger began. Minor phenomenon at the end of 16th cent. Sibilant merger was general. No distinction between voiced/​voiceless

An overview: sibilant merger in Spanish  45 Table 1.14 Chronology of sibilant merger based on grammarians, rhymes and graphemes 15th cent.

End of 16th cent.

17th cent.

Grammarians

Distinction

Rhymes

Distinction [s]‌~[z], [ts]~[dz] and [ʃ]~[ʒ] Distinction ~   ~  ~ < j, ge, i >

The phenomenon is extended and common The merger is minor

No distinction: articulatory levelling General phenomenon Beginning 17th cent. The merger is general

The merger started

Beginning 17th cent. Confusion is a minor phenomenon

Graphemes

1.2.4  Chronological accounts: creation of new sounds The sibilant merger in the sixteenth century resulted in two new sounds in Spanish, non-​existent in other Romance languages: the interdental fricative voiceless [θ] (from the pair [ts]~[dz]) and the velar fricative voiceless [x]‌(from the pair [ʃ]~[ʒ]). These processes are named interdentalization and velarization, respectively. Sources on the process of interdentalization are few as grammarians did not elaborate on this change. The first records are from the end of the sixteenth century. Venegas, in 1531, refers to this Greek sound [θ] as interdental; however, he does not relate it to the Spanish . López de Velasco, in 1582, compares the pronunciation of letter with the Greek ; therefore, he was noticing some similarities. Cuesta, in 1589, is the first one to confirm an interdental fricative consonant very close to our modern [θ]; nevertheless, it was in its first stages and was not common. There are few records from the seventeenth century about this interdental sound; only Bonet, in 1620, relates an interdental pronunciation for [s̪], which was perhaps close or equal to modern [θ]. The conclusion is that the interdental [θ], although it seemed to appear around the end of the sixteenth century, was not a common practice during the seventeenth century; it was still very far from the modern distinción or ceceo. See Map 1.2 for modern distribution. The velarization, according to written documentation, started to take place at the end of the sixteenth century while coexisting with the predominant palatal pronunciation of [ʃ]; in the seventeenth century, after a long period of concurrence, the shift from [ʃ] to [x]‌took place and succeeded in imposing itself as the norm. Grammarians such as Torquemada, López de Velasco and Oudin, from the sixteenth century, describe the letters , and with a back-​throat pronunciation, denoting both palatal and almost velar pronunciation at that time. Writers such as Jiménez Patón and Correas, from the seventeenth century, report the velar pronunciation for [ʃ] and [ʒ] (Patón) and

46  Eva Núñez-Méndez

Map 1.2 Spread of modern distinción, seseo and ceceo.

Table 1.15 Chronology of interdentalization, velarization and devoicing Chronology of interdentalization, velarization and devoicing Creation of new phonemes and sounds /​θ/​ and /​x/​

Loss of voiced sibilants

Rhyme analysis Grapheme analysis Grammarians’ sources: • [s̪] > [θ] end 16th cent. 17th cent.

no conclusions

confusion was early in the north

phenomenon started still a minor phenomenon late distribution and generalization starting phenomenon phenomenon spread

17th cent. confusion was common and wide spread

• [ʃ] > [x]‌

16th cent. mid 17th cent.

as an extension of the in Andalusian (Correas). It is difficult to determine exactly when the velar [x] became standard, displacing the palatal articulation, as there are not that many documented indications. See Table 1.15. 1.2.5  Updated modern chronological accounts Some discrepancies emerge concerning when the devoicing spread and consequently prompted other derivative phonetic changes. In general terms, the sibilant merger originated early in the north and its generalization was completed in the seventeenth century. Most of the earlier scholars maintain that it was a

An overview: sibilant merger in Spanish  47 general phenomenon before the end of the sixteenth century. Menéndez Pidal (1940: 35) places it with the so-​called “phonetic revolution” at the end of the sixteenth century, based on literary works. Martinet believes its completion was in the sixteenth century. Alarcos (1950: 270) states that the lack of distinction between voiced/​voiceless differentiated the Old Castile and Toledo norms during the first half of the sixteenth century, and that eventually the Castile norm was imposed on Toledo and the rest of the Spanish-​speaking territories during the sixteenth century. Catalán and Reinhardt (1957:  314) think that the devoicing of the affricates (and consequently the alveolars and palatals) was done in the last third of the sixteenth century. Penny (2000: 42) defends that the devoicing of the three pairs happened before the end of the Medieval Ages in some Castilian areas (in Cantabria and north of Burgos), and that from there it spread southwards until it became the norm in the second half of the sixteenth century. Cano Aguilar (1988: 238) describes the devoicing as a general phenomenon in the fifteenth century (or even before) and asserts that its regularization was not until the second half of the sixteenth century Other philologists delay the phenomenon to the seventeenth century. Among them, Cuervo (1987: 344) proposes that the merger between voiceless and voiced sibilants was not completed until the first half of the seventeenth century. A.  Alonso (1951a:  38) also confirms that, although the devoicing was an old phenomenon in certain areas, its generalization and completion was much later; the devoicing was still in process at the end of the sixteenth century, and the voiceless sibilants were only consolidated in the seventeenth century. On the other hand, based on more recent and comprehensive documentary studies, other authors suggest that the devoicing and deaffrication occurred much earlier than prior studies have suggested. Frago brings forth many textual examples of graphic confusion as early as the last quarter of the thirteenth century concluding that seseo was part of the Andalusia speech at the end of the fifteenth century (1992:  118). Kauffeld, gathering evidence from examples of confusion of or for as early as 1398, indicates a considerable presence of seseo in both Seville and Cordoba from the fourteenth century onward and dates seseo as well-​established throughout the south by the end of the fifteenth century (2016: 187). In addition, this author supplies us with early attestations of seseo, dating to 1544 and 1545, in the colonial New Spain corpus. Also, Parodi provides even earlier documentation of seseo in the Spanish of New Spain as early as 1523 (1976: 124).64 D. Alonso (1962: 141) explains that in the north the devoicing of sounds [z]‌, [dz] and [ʒ] was advanced in the fourteenth and fifteenth centuries, but he does not say anything about its spread to the south. The sibilant confusion was also evident in early colonial Spanish texts. The documentary record shows seseo as well-​rooted in early American Spanish speech, appearing in the writings of people of all social classes and ethnic groups and indicating a complete graphic confusion in all positions within the word.65 Furthermore, this confusion applies to both Indian and Creole

48  Eva Núñez-Méndez Table 1.16 Sibilant devoicing development according to researchers Sibilant Devoicing Development: Chronological Interpretations Frago Gracia (1992-​3) Kauffeld (2011, 2016) D. Alonso (1962) Penny (2000)

Cano Aguilar (2004) Martinet (1951-​52) Alarcos (1950)

Catalán (1957) Bradley et al. (2006) Menéndez Pidal (1940) Lloyd (1987) Cuervo (1987) A. Alonso (1951a, 1955)

There are graphic confusions in the last quarter of the 13th cent. End of the 15th cent. seseo in Seville, Cordoba and the south. 14th cent. onward seseo in Seville and Cordoba. End of the 15th cent. seseo was general in Andalusian dialect. Devoicing was normal in the north during 14th and 15th cent. • Devoicing happened before the end of Middle Ages in the north (Cantabria and Burgos). • 2nd half of the 16th cent. devoicing was the norm. It was a regular phenomenon in the 15th cent.; however, it was widespread in the second half of the 16th cent. 16th cent. it is a widespread phenomenon. • 1st half of the 16th cent. there was opposition in the Toledo norm • 1st half of the 16th cent. there was no opposition in the Castile norm. • Throughout the 16th cent. the Castile norm became widespread. Devoicing was general in the last third of the 16th cent. Elimination of voiced sibilants by the 1580s. By the end of 16th cent. devoicing is generalized.

13th cent.

14th cent

15th cent.

16th cent.

End of 16th cent. It was not common until the first half of the 17th cent. Devoicing was happening at the end of the 16th cent. but it was widespread in all the territories in the 17th cent.

17th cent.

writers, and to those Spaniards not originally from seseo-​areas, which further confirms its strong hold on the speech practices of the area (Kauffeld 2016: 189). See summary in Table 1.16. However, when it comes to date the development of interdentalization ([θ]) and velarization ([x]‌), most authors agree. Corpus analysis, grammarians’ attestations and traditional studies postulate that they are late phenomena, which did not generalize until the second half of the seventeenth century or later, with their origins at the end of the sixteenth century. Alarcos (1950: 272) suggests that the interdentalization of [s̪] was complete in the second part of the seventeenth century approximately. Lapesa (1981:  373) posits that the interdental articulation was practiced in the north in the second half of

An overview: sibilant merger in Spanish  49 Table 1.17 Conclusions regarding the sibilant merger and development General conclusions regarding the sibilant merger and development 14th cent. Traditional-​ Prior Studies

–​

End 15th cent. –​

End 16th cent.

Beginning 17th cent.

General phenomenon

Recent Some graphic Devoicing It was a Documentary confusions was normal widespread Studies phenomenon Devoicing in Seseo general but was not the north in Seville, general in Cordoba and all areas of the south the Spanish-​

Devoicing was generalized. No one made the distinction in speech.

speaking peninsula.

the sixteenth century, while in Toledo and other areas the leveling was not completed until the first third of the seventeenth century. Lloyd (1987: 531) indicates that, although there is evidence of an early interdental pronunciation for the letter from the first years of the sixteenth century, its generalization did not consolidate until the beginning of the eighteenth century. Likewise, Cano Aguilar (1988:  240) places this phenomenon at the end of the seventeenth century or even at the beginning of the eighteenth century. A.  Alonso (1955:  336) proposes that the ceceo and the distinción were not common until the middle of the eighteenth century, although there were some early traces in the mid-​sixteenth century. Turning to velarization, Lapesa (1981: 379) considers that the imposition of the velar [x]‌was not effective until the first third of the seventeenth century after a long period of coexistence with the palatal [ʃ]. Alarcos (1950: 271) dates this late phenomenon to the seventeenth century. Kiddle (1975: 74) specifies that the modern sound [x] becomes general after 1660. Lloyd (1987: 544) indicates that it was during the middle and end of the sixteenth century when [x] was spread, and it was in the middle of the seventeenth century when it became a general practice. Cano Aguilar (1988: 239) implies that this occurred in the first decades of the seventeenth century. Bearing in mind that the first signs of devoicing and its expansion did not all develop simultaneously, at the same rate or with the same origin, we can make some conclusions about the sibilant merger and devoicing (in Table 1.17.).

1.3  Tracking Old Spanish sibilants: Judeo-​Spanish It is always of great interest to philologists to analyze Judeo-​Spanish as a record of many medieval phonetic changes that took place in Old Spanish.

50  Eva Núñez-Méndez While those changes transitioned to what is today modern Spanish, some of them were preserved in Judeo-​Spanish as it was the language spoken by those Jews expelled from Castile-​Leon in 1492. Cut off from the Iberian Peninsula, they carried various peninsular dialects from their regions of origin (Aragon, Castile, Catalonia, etc.) to their new surroundings. Most Sephardis took the road east towards the land of the Ottoman Empire and settled in major cities: Istanbul, Salonica, Izmir, Sofia, Monastir, Sarajevo and others, some of which became Sephardic centers. In contact with other languages, such as Turkish, Serbo-​Croatian, Bulgarian and Greek, their language underwent a process of koineization in which Castilian turned out to be dominant, and this koine became the lingua franca of the Mediterranean Jewish world, used for business, at the synagogue, in education and in general community life (Bürki 2013: 336). Their language, well-​kept by minority groups, is still used, mostly in songs, prayers and selective media sources. The survival of Judeo-​ Spanish shows unprecedented historical value to study language variation in the fifteenth century. Unlike modern Spanish, Judeo-​ Spanish has kept three pairs of sibilants from medieval Spanish in the pronunciation of both fricatives and affricates. The voiced sound [z]‌, pronounced differently from the voiceless [s], is used when is in intervocalic position as in casa /​káza/​ “house,” beso /​bézo/​“kiss.” This is the case in modern Portuguese, e.g. casa /​z/​“house” versus cassa /​s/​“gauze”; and in French, e.g. poisson /​s/​ “fish” versus poison /​z/​“poison.” Some scholars argue that the neutralization between sounds [s] and [z] is regular in some varieties of JudeoSpanish (Bürki & Romero 2014). Judeo-​Spanish /​s/​and /​z/​are phonemes, meaning they have a contrastive value and can differentiate word meaning. See examples in Table 1.18. Judeo-​Spanish also kept the distinction between the Old Spanish prepalatal sibilant sounds [ʃ] and [ʒ] in words such as jabón [ʃaβón] “soap,” deshó [deʃó] “to leave,” trusheron [truʃéron] mod. trajeron “to bring, brought,” and muzer [muʒér] “woman,” fijo [fíʒo] “son.” In Spanish, these two sibilants converged in the voiceless [ʃ], which shifted its point of articulation further back becoming a velar fricative sound [x]‌, or an uvular fricative [χ] in the north-​central variety and an aspiration [h] in the south, merging with the aspiration descending

Table 1.18 Judeo-​Spanish sibilants voiceless

voiced

alveolar fricative affricate

/​s/​ e.g. abrasso ‘hug’

voiceless

voiced

(pre)palatal /​z/​ e.g. beso ‘kiss’

/​ʃ/​e.g. debasho ‘under’ /​tʃ/​e.g. chiko ‘small’

/​ʒ/​e.g. ojos ‘eyes’ /​dʒ/​e.g. yente ‘people’

An overview: sibilant merger in Spanish  51 from Latin initial F-​. Due to the geographical proximity, Moroccan Judeo-​ Spanish closely follows Spanish evolution and they also speak pronouncing the velar sound [x]. In Judeo-​Spanish, the prepalatal affricates /​tʃ/​ and /​dʒ/​maintained their phonological value, and the voiced sound [dʒ] survived in words such as yente mod. gente “people,” yuntos mod. juntos “together,” while it was lost in Spanish. As in the Castilian variety, Judeo-​Spanish lost the dento-​alveolar affricate pair [ts] and [dz]. This pair changed to the alveolar fricatives [s]‌and [z], respectively, in words such as sinco, cinco [síŋko] “five,” senar, cenar [senár] “to have dinner” (following the same tendency of the Andalusian seseo of the south) and casa [káza] “house.” Although Judeo-​Spanish kept six of the seven medieval Spanish sibilants, their distribution in the spoken language was slightly different than that of medieval Castilian of the fifteenth century. These six sounds [s], [z], [ʃ], [ʒ], [tʃ] and [dʒ] were kept and spoken in the Judeo-​ Spanish variety, but not in the same phonetic contexts as Old Spanish. See summary in Figure 1.6. Following the Andalusian variety, Judeo-​ Spanish maintained two of the four dento-​alveolar and alveolar medieval sibilant sounds dropping the affricates [ts] and [dz]; but keeping the voiced distinction between [s]‌and [z], without any further merger into just a single phoneme /​s/​as in Andalusian seseo, and without evolving into the two resulting fricative phonemes of the Castilian distinción between /​s/​and /​θ/​. Castilian speakers distinguish words such as casa/​caza “house/​hunting,” caso/​cazo “case/​pan” and laso/​lazo “straight/​bow” as minimal pairs. Compare examples in Figure 1.7. Although Judeo-​Spanish underwent important linguistic changes in recent centuries, linked to geopolitical circumstances, Sephardis succeeded in keeping their vernacular language, with their first major signs of literary creativity in the eighteenth century. Today, it is an almost obsolete minority language, but with great sociolinguistic value and a key role in diachronic philology. Far from sustaining a chauvinistic Iberianism, the Sephardis developed cross-​ethnic language relations in their Mediterranean settlements; nevertheless, they surprisingly preserved their tight traditions, identity, and old

Figure 1.6 Old Spanish and Judeo-​Spanish sibilants.

52  Eva Núñez-Méndez

Figure 1.7 Andalusian seseo and Judeo-​Spanish [s]‌and [z].

language in complex and finely graded forms of hybridism. Their ability to cross cultural borders while still maintaining cohesion as a minority group stands out as particularly remarkable. In decline, the Sephardis share the fate of other diasporic minorities. The seduction of assimilation and the advent of modernity have contributed to the erosion of the bonds of the Sephardic collective identity and, consequently, their language.

1.4  Tracking sibilant development: overview of other Romance languages It is worth noting here that Romance languages do not derive from classical Latin, but from everyday colloquial “vulgar” (or popular) Latin, carried by Roman soldiers, merchants and slaves into the conquered provinces of the Empire. This version became the common tongue in the Mediterranean with striking resemblances to modern Italian. The political decentralization of the later Empire allowed regional variations and the formation of dialects that eventually drifted apart, becoming mutually unintelligible and resulting in new languages. Nevertheless, these languages kept strong similarities in grammar, vocabulary and pronunciation. One of these commonalities can be reconstructed by the evolution of sibilant consonants. Geographically close Romance languages like Portuguese, French and Italian differ from Spanish in maintaining voiced sibilants alongside the unvoiced. The drastic change affecting Old Spanish sibilants during the course of the sixteenth century did not occur in other sister languages. Besides, the emerging interdental sound [θ] (from pair [ts]~[dz] with its advancement of the point of contact to between the teeth), and the velar [x]‌(from pair [ʃ]~[ʒ] with its retreat to the velar area) developed only in Spanish. Taking these pairs into account, the main contrasts of sibilant consonants can be summarized in general terms with the following representative examples: a) The pair of voiced and unvoiced fricative alveolar phonemes /​s/​~/​z/​ was kept in most Romance languages, mainly represented graphically by ;

An overview: sibilant merger in Spanish  53 /​s/​originated as Latin S or S S in all positions and /​z/​as intervocalic -​S -​ or S preceded by a voiced consonant. French evolved to a silent sound in word final position . Portuguese developed an unvoiced prepalatal [ʃ] in coda and in word final position, and a voiced prepalatal [ʒ] in intervocalic position. The changes NS > S [z]‌and PS, RS > S S [s] occurred very early in Latin as is attested in numerous Roman inscriptions and by the concordance with Modern Romance. In the case of NS , the loss of the N was balanced by a closing of the preceding vowel (as in Portuguese mês). Compare examples in Table 1.19. b) The affricate pair /​ts/​~/​dz/​ presents more variations. Romance languages did not develop the interdental Spanish sound [θ], but they frequently resulted in a voiced fricative sound [s]‌as in the Andalusian seseo variation. These two phonemes essentially originated from Latin combinations C E , C I , T E, TI , and sporadically from DE, DI , GE, GI . The voiced sound [dz] occurred mainly in intervocalic position or preceded by a voiced consonant like n or r. In the initial position or after consonants, the letter from C E , C I survives with different pronunciations: palatal [tʃ] in Italian and alveolar [s]‌in Portuguese and French. In the intervocalic position, Portuguese and French voice it to [z], spelling it and respectively. In the final position, we find pronounced as prepalatal [ʒ] in Portuguese; in French or with a silent consonant. Latin C was pronounced with [k]‌; however, over the course of time, and in combination with [e] and [i], the tongue made contact further forward than before other vowels, resulting in a more palatal pronunciation, coming to palatal [tʃ], which is the modern Italian pronunciation as in noce [nótʃe] “nut” (Latin NUC EM ). The other three languages carried the process further:  in Portuguese and French, the pronunciation went from [tʃ] to [ts] to the alveolars [s] or [z]; similarly, in Andalusian Spanish it went to [s]. Old Spanish shared dento-​alveolar [ts] with Old French and Old Portuguese (Bowman 1980: 33). Table 1.19 Examples of [s]‌and [z] Latin

Spanish

Portuguese

French

Italian

English Cognate

sanum [s]‌

sano [s]‌ pasado [s]‌ besar [s]‌ pesar [s]‌

são [s]‌ passado [s]‌ beijar [ʒ] pesar [z]‌

sain [s]‌ passé [s]‌ baiser [z]‌ peser [z]‌

sano [s]‌ passato [s]‌ baciare [tʃ] pesare [z]‌

sane ‘healthy’ past [s]‌ –​‘to kiss’ pensive

mes [s]‌ prosa [s]‌

mês [ʃ] prosa [z]‌

mois prose [z]‌

mese [z]‌ prosa [z]‌

menstrual [s]‌ prose

passatum [s]‌ basiare [s]‌

pensare [s]‌

(pesare)

mensem [s]‌ prorsa [s]‌

54  Eva Núñez-Méndez In modern Spanish, the Latin combination C E , C I and T E , T I resulted in the interdental sound [θ] (and in [s]‌in seseo), with the graphemes and . The regular development of TE, TI was [z] in Portuguese represented by and sometimes with , pronounced [s] (confused with C E , C I ); [ts] or [dz] in Italian with the grapheme ; and [s] in French (from Old French [ts]) with various spellings , , , and (which could be [z] between vowels as poison or silent at the end of a word as puits, palais) (Bowman 1980: 141). See Table 1.20. c) The fricative alveopalatal pair /​ʃ/​~/​ʒ/​, lost in Spanish, was kept in other Romance languages with a palatal or alveolar articulation of some sort. After the devoicing, it become a velar sound [x]‌in Spanish, written with or . The pair sources were Latin X, P SE , P SI , SSE , SSI and C ’L , G’L , LI , I O, I U , respectively; nevertheless, in some Spanish popular words, there is an anti-​etymological /​x/​spelt with from the fusion with Arabic shin as in jabón “soap.” The prepalatal sound [ʃ] derived from intervocalic Latin X gave Italian double [ss] written as (but [ʃ] for , ); Portuguese [ʃ] for or [s]‌ for ; and French [s] for . The Latin groups P SE , P SI , SSE , SSI had similar solutions; in French, we have a voiced prepalatal [ʒ] for . The Latin combination C’ L , resulting from the fall of an unstressed vowel, has a regular development in Romance, giving palatal solutions. It becomes [ʎ] in Portuguese, [k:j] in Italian, and [i:j] in French. Spanish and Portuguese also palatalized the Latin group G ’ L as in teja “tile”; the exceptions are mainly related to ecclesiastical words which show semi-​ developments such as regla “rule” or siglo “century.”

Table 1.20 Examples of the origins of [θ] Latin

Spanish

Portuguese French

Italian

English Cognate

centum [k]‌

ciento [θ], [s]‌ ciudad [θ], [s]‌ cocina [θ], [s]‌ decir [θ], [s]‌ placer [θ], [s]‌ marzo [θ], [s]‌ O. Sp. março tristeza [θ], [s]‌

cento [s]‌ cidade [s]‌ cozinha [z]‌ dezir [z]‌ prazer [z]‌ março [s]‌

cento [tʃ] città [tʃ] cucina [tʃ] dire piacere [tʃ] marzo [dz]

cent city kitchen diction ‘to say’ pleasure March

civitatem [k]‌ cocinam [k]‌ dicere [k]‌

placere [k]‌

martium [ti] tristitia [ti] pescem, piscis

pez [θ], [s]‌ *piscionem [k]‌ crucem [k]‌ cruz [θ], [s]‌ radicem [k]‌ raíz [θ], [s]‌

cent [s]‌ cité [s]‌ cuisine [z]‌ dire plaisir [z]‌ mars [s]‌

tristeza [z]‌ tristesse [s]‌ tristezza [ddz] pez [ʒ] poisson [s]‌ pece [tʃ] poix cruz [ʒ] croix croce [tʃ] raiz [ʒ] racine [s]‌ radice [tʃ]

–​ ‘sadness’ fish cross radish ‘root’

An overview: sibilant merger in Spanish  55 The Latin group LI developed into [ʎ] in Portuguese, [ʎ] in Italian, and [:j] in French. In dialectal Spanish, for words written with , there is a lateral pronunciation [ʎ] as in Portuguese and Italian, for instance in batalla “battle.” In other words, LI gives the same results as C ’ L in all but Italian, where it becomes [ki] (Bowman 1980: 84). Finally, the Latin group I O, I U , survives in Italian as affricate [dʒ] spelled , and in Portuguese and French as fricative [ʒ] spelled or . In Spanish the Latin initial I -​(as well as initial G E -​, G I -​, and D E -​, D I -​) shows three developments: before e and a the result is palatal [ʝ], spelled as in ya (< IA M ) “already,” yema ( ese, tuviesse “had” > tuviese. In 1815, the letter was allocated to the sound combinations [ks] and [gs], as in Latin, (i.e. examen, exención); the new velar sound [x]‌is written with as in caxa “box” > caja, lexos “far” > lejos, together with the letters for those cases with etymological Latin GE, GI, as gente “people,” género “gender,” and so on. Words such as México, Oaxaca, versus Méjico, Oajaca, and so on, accept both spellings for complex historic-​political reasons.67 This was the last graphic change to end the distinction between voiced and unvoiced sibilants, which stopped being a common practice in speech two centuries before (Lapesa 1981: 423). This spelling reform in the eighteenth century eliminated and (for [x]‌); however, the orthography has never differentiated between the Castilian system and the seseo and ceceo variants. Because the graphic distinction between and does not have a phonemic contrast in seseo dialects, misspellings are common. On the other hand, writers who want to parody rural ceceo speakers will use an orthographic z instead of s to indicate the interdental articulation as in zi, zeñor [θí θeɲór] instead of sí, señor [sí seɲór] “yes, sir.” See Table 1.23. By 1815 the orthography became fixed as in modern Spanish. Later changes have been reduced to accentuation and minor instances. Throughout its history, RAE has mostly kept a conservative approach with a faithful respect Table 1.23 Examples of distinción, ceceo and seseo orthography

distinción [θ]~[s] ceceo [θ]

seseo [s]

cocer ‘to boil’ coser ‘to sew’ censura ‘censure’ sincero ‘sincere’

[koθér] [kosér] [θensúra] [sinθéro]

[kosér] [kosér] [sensúra] [sinséro]

[koθér] [koθér] [θenθúra] [θinθéro]

58  Eva Núñez-Méndez for etymology and unity. Still, since 2012, RAE has shown some incipient flexibility in keeping up with the times and accepting estadounidismos as part of the Spanish lexicon. Including the Spanish of the United States as one of the Spanish-​speaking varieties and giving the country its own Language Academy in 1973 came after a long-​fought battle against RAE traditionalism. This accommodation is an important recognition for the country that will soon have the most Spanish speakers. Although its origins were based on Castile Spanish, today the RAE works to guarantee a common standard across many countries in accordance with its founding goal:  making sure that the changes do not break the essential unity it enjoys throughout the Spanish-​speaking world. Since 1992, the RAE and the other 21 language Academies have collaborated in producing the Dictionary of the Spanish Language, adding, deleting or modifying words according to language changes in all geographical varieties of Spanish.

1.6  Conclusions The evolution of the Old Spanish sibilants differentiates the two principal linguistic varieties of Spanish, Peninsular and American, and characterizes Spanish versus other Romance languages. After considerable vacillation, the sibilant merger that started in the north of the peninsula extended towards the south and was completed by the end of the seventeenth century. Nevertheless, the Madrid prestige-​court variant imposed in the south overlapped some other tendencies that today define the Andalusian dialects. Whereas a few parts of Andalusia use only [θ] and say zí, zeñor [θí θeɲór], most parts, together with the whole of Spanish America (and other peninsular Romance languages), only use [s]‌(i.e. sine [síne] for cine “cinema”). Castilian Spanish with its apicoalveolar [s̺] clearly distinguishes the consonants [θ] and [s] as in cien “hundred” and sien “temple.” In this respect, the southern region offers a close parallel with Portuguese, Catalan, Italian and French as these languages never innovated [θ], although they kept a voiced [z] /​voiceless [s] distinction. We have concluded that the similarities in pronunciation between the sibilant pairs threatened their distinction, with devoicing being the first step in their neutralization. As a result, three solutions emerged to continue that opposition: the Castilian opposition of [θ]~[s]‌; the Andalusian leveling into [s] (for seseo); and the later creation of the velar [x]; with both [θ] and [x] being unique in Romance. The new Castilian system attained rapid prestige and overthrew the Seville seseo speech; however, it was the latter that made its way to transatlantic territories. It is still an enigma for scholars to justify how and why this non-​conformist Castilian practice, distant from communal and Court norms, flourished and imposed itself. Without indulging in philological abstractions, my purpose here has been to clearly illustrate the primary causes for the sibilant system’s transformation, development and ultimate stabilization under a comprehensive chronological

An overview: sibilant merger in Spanish  59 framework backed by recent research. The historical complexity of this phenomenon raises as many questions as it solves. Nevertheless, the goal here has been to facilitate the reader’s exploration of the topic from a descriptive perspective. The sibilant reorganization in late medieval and early modern Spanish is a fundamental topic for understanding the famous northern/​ southern dialect split in Spain and its implications for the development of cross-​ dialectal American Spanish variations. In sum, this chapter offers a summary of the evidence, chronology, dialectal divergence and phonetic background of the sibilant merger to understand current variations. In doing so, it helps to contextualize this particular phonetic variation within the wide range of dialects across the Hispanic-​speaking world, which we will see in the following chapters.

Notes 1 The main scope is to frame the modern Spanish system of sibilants within an updated perspective in diachronic linguistics without engaging in the false myth of ceceo based on a king who had a lisp, which society tried to emulate as a prestige marker. See Guitarte (1992 and 1987) and González Ollé (1987) to read more about this fabled Spanish “lisp.” 2 See Aleza (2010:  63) for the multiple variations of the pronunciation of sound [s]‌in América, e.g. it is apicoalveolar in Guatemala, rural parts of Honduras, Bolivian plains, inland Colombia, etc. 3 For further information about Information Theory and Information-​theoretic terms please see Shannon 1948, Cover and Thomas 2006, Goldsmith 1998, Hume and Mailhot 2013. 4 “Las fricativas sonoras adolecen de una contradicción intrínseca: un ruido intenso de fricación es incompatible con la sonoridad” (Pensado 1993: 214). 5 Frago is one of the few linguists that disagree with the well-​grounded idea that the gestation of the phonetic changes comes from the north (even when he himself defended this idea in his previous research) after studying southern manuscripts: “a la vista de los conocimientos que he ido adquiriendo en textos antiguos del sur de España y de los territorios aragonés y leonés, me he limitado a defender […] la palmaria imposibilidad de atribuir su origen al llamado ‘sistema cantábrico’, el inicio de la cadena de alteraciones fonéticas” (1989: 132). 6 “Los de Castilla la Vieja dizen haçer, y en Toledo hazer. Y dicen xugar, y en Toledo jugar. Y dizen yerro, y en Toledo hierro. Y dizen alagar, y dizen halagar, y otros muchos vocablos que dexo por evitar prolixidad” (in A. Alonso 1967a: 23). 7 See A.  Alonso 1967; Alarcos 1988; Cuervo 1987; Lapesa 1981; Martinet 1974; Pascual 1988 and Gavel 1929. 8 For further information see Diego Catalán (1957: 321). 9 See Espinosa’s “Arcaísmos dialectales” (1935). 10 Campos-​Astorkiza states that there is a scarcity of data supporting the gradient and variable voicing assimilation of /​s/​before a voiced consonant. She analyzes voicing assimilation in Spanish as an instance of gestural blending (2014: 18). 11 Menéndez Pidal explains that marine traffic frequented the coasts with more direct, close, and persistent waves of colloquial speech from the metropolis, in contrast with those regions in the mainland interior (1962: 142).

60  Eva Núñez-Méndez 12 “Por cada hombre noble y de clara sangre que pasaba al Nuevo Mundo en los primeros tiempos, venían diez descomedidos y de otros linajes oscuros y bajos” (Oviedo 1959:  36). Opina Frago (1999:  12) que “en la emigración a Indias predominó con mucho el elemento popular”; y escribe Zamora Vicente (1967:  378) que “el fondo patrimonial idiomático aparece vivamente coloreado por el arcaísmo y por la tendencia a la acentuación de los rasgos populares.” Añade Álvarez (1987: 35) que “el grueso de la población española que llegó en un principio a nuestras playas, y en términos más amplios a las de América en general, pertenecía a las clases populares.” 13 “El pueblo que se desgajó de España para poblar América […] estaba compuesto de rústicos, villanos, artesanos, clérigos, hidalgos, caballeros y nobles, aproximadamente en la misma proporción que el pueblo que se quedó en España” (A. Alonso 1967b: 15). 14 Most linguists agree on the andalucista theory; however, Henríquez Ureña first opposed it based on the idea of multiple causation, or polygenetic theory, claiming that similarities on both sides of the Atlantic are due to parallel independent developments. Henríquez Ureña, in line with Amado Alonso, does not accept the andalucismo of American Spanish; they believe that the Andalusian presence was not clearly predominant over all colonial America. See Noll (2005). 15 “Ese importante rasgo fonológico no puede explicarse como algo autóctono ni de Canarias ni de América, sino como un verdadero sevillanismo” (Moreno de Alba 2007: 48). 16 “Sevilla y Cádiz monopolizaron durante los siglos XVI y XVII el comercio y relaciones con Indias. En un momento en que la pronunciación estaba cambiando rápidamente a ambos lados del Atlántico, Sevilla fue el paso obligado entre las colonias y la metropoli, de modo que para muchos criollos la pronunciación metropolitana con que tuvieron contacto fue la andaluza” (Lapesa 1981: 586). 17 Danesi’s data points out that Boyd-​Bowman’s studies on 5,481 settlers in the early colonial period only represent 2.74% of the 200,000 immigrants that arrived in the sixteenth century (1977: 192–​193). 18 Tuten points out that “the regularization of seseo […] occurred as part of a particularly far-​flung process of roughly simultaneous koineization(s) […] The inherited tendencies or structural features were the weakly-​marked phonemic distinction and incipient neutralization. In the koineizing context(s), speaker-​learners everywhere could easily have generalized the merger and thereby established it as a new norm” (2003: 264). 19 “En cuanto a la colonización del Nuevo Mundo fue el lenguaje de Sevilla, no el de Toledo o de Madrid, el que estableció las primeras normas […] La época inicial o antillana está claramente dominada […] por las provincias andaluzas Sevilla y Huelva y que fue precisamente la koiné española insular desarrollada en aquel tiempo, con su caudal de antillanismos la que llevaron consigo desde las islas los primeros conquistadores de Tierra Firme” (Boyd Bowman 1964: 24–​25). 20 “La revolución de fines del siglo XVI no fue […] sino la última y decisiva batalla librada por una norma dialectal castellana vieja contra el prototipo lingüístico cortesano toledano” (Menéndez Pidal 1962: 101). 21 Martinet’s push/​drag chains explain a rationale for phonological changes in a language over time. The structure of a language requires relationships between its units in the system; any change may potentially initiate a chain of derivative changes, in the end giving rise to drag chains and the like. He claims that phonemes

An overview: sibilant merger in Spanish  61 (as autonomous units) move around in response to internal pressures generated by the interaction of function, structure, economy and the natural asymmetries of the speech organs. The inception of sound change is gradual and imperceptible, a continuous process in which the allophonic norm of a phoneme assumes a new position by infinitesimal steps (1955: 48–​49). 22 Robles writes: “se pronuncia entrándose la lengua tan adentro, que casi se dobla hacia la garganta” (in Blanco 2006: 85). 23 “La desonorización debió de preceder en el norte al reajuste del punto de articulación; en las zonas centro-​ meridionales, donde el ensordecimiento se propagó desde el norte, lo mismo que la velarización, es posible que los dos fenómenos fueran simultáneos” (Alarcos 1988: 56). 24 “La sonoridad se perdió en el ultimo tercio del siglo XVI […] La oposición ç-​z siguió siendo funcionante después de haber perdido su marca única de oposición, la sonoridad” (A. Alonso 1955: 317). 25 “Durante algún tiempo se mantuvo un resto de oposición entre la /​ts/​(escrita c o ç) y la fricativa, sorda ya también, prodecedente de /​dz/​, y transcrita con z; pero esta diferencia no sobrevivió al primer tercio del siglo XVII, y la igualación en /​θ/​ fue completa” (Lapesa 1981: 374). 26 “La en que comiença la sílaba siguiente suena poco […] pero si suenan apretadas, doblarse han en medio de la palabra, como diziendo:  amassa, passa […] de donde se puede coger, quándo estas dos letras se han de escreuir senzillas, y quándo dobladas, mirando a la pronunciación, si es apretada, o si es floxa” (Nebrija in Blanco 2006: 61). 27 “En muchos vocablos he mirado que scrivís dos eses adonde otros se contentan con una, y una donde otros ponen dos; ¿tenéis alguna regla para esto? […] Generalmente pongo dos eses quando la pronunciación ha de ser espessa, y donde no lo es pongo una sola” (Valdés in Blanco 2006: 63). 28 “La pronunciacion de la.s. […] tiene sonido doblado y sencillo como si dixessemos en algunas partes esse, dandole fuerça. Y en otras, ese, dandole sonido floxo […] para bien pronunciarlas y escreuirlas” (Cuesta in Blanco 2006: 65). 29 “En todo principio de dición o silaba usamos de la ʃ larga, i a fines dellas, de la s pequeña” (Alemán in Blanco 2006: 65). 30 “S. en principio y medio de parte, quando tienen ante sí otro consonante, suenan recias, aunque estén sencillas, como quando las ponemos dobladas […] Mas quando se ponen entre dos vocales suenan floxas y con poca fuerça” (Jiménez Patón in Blanco 2006: 66). 31 “Que ni los griegos ni los latinos […] la sienten ni conocen por suia; de manera que, pues la c, puesta debaxo aquella señal, muda la substancia de la pronunciación, ia no es c, sino otra letra, como la tienen distinta los judíos y moros, de los cuales nos otros la recibimos […] “ “De letras que tenemos prestadas del latín para escribir castellano, sola mente nos sirven por sí mesmas estas doze: a, b, d, e, f, m, o, p, r, s, t, z” (Nebrija in Blanco 2006: 48). 32 “Ai a este syluo aplicamos una.d. de suerte que precede al syluo: como la culebra le haze […] quedara formado el verdadero sonido de la.z. porque esta y la.r. son letras dobladas” (Venegas in Blanco 2006: 49). 33 “La lengua castellana, de más de a.b.c. latino, tiene una j larga, que vale lo que al toscano gi; y una cerilla que, puesta debajo de la c, la haze sonar casi como z”. “La cerilla se ha de poner quando, juntándose con a, con o y con u, el sonido ha de ser espesso, diziendo çapato, coraçón, açucar” (Valdés in Blanco 2006: 50).

62  Eva Núñez-Méndez 34 “z tiene la mesma pronunçiación que la c con çedilla dos vezes pronunçiada” (Villalón in Blanco 2006: 51). 35 “La c y la z se pareçen casi tanto en el sonido de la pronunçiaçión como la b y la v, de manera que muchas personas no saben diferenciarlas ni escreuirlas, y muchas vezes hallaréis puesta la vna por la otra, avnque en las pronunçiaçiones son tan diferentes […] çaço que ambas vezes se pronunçia la c con la fuerça que he dicho y vazío […] con la mitad de la fuerça menos que aueis pronunçiado la c” (Torquemada in Blanco 2006: 51). 36 “Parecido el sonido de la.c. cedilla al de la.z. viene a no percibirse por todos la differencia que ay del vno al otro; y el error y confussion que hay en la pronunciacion dellas, no sólo entre gente sin letras, pero entre curiosos y obligados a saberlo” (López de Velasco in Blanco 2006: 52). 37 “Sepan los niños distinguir el sonido de la c á la z porque ordinariamente […] muy pocos niños hazen diferencia en pronunciar estas dos letras, es la causa no mirar los que enseñan al principio de ello” (Cuesta in Blanco 2006: 53). 38 “Que los de Castilla la Vieja dizen haçer y en Toledo hazer” (F. Juan de Córdoba in Blanco 2006: 53). 39 “Muchos […] quieren dezir que la çedilla es blanda, i la zeda mas fuerte i rrezia. I es error imaxinar que tenemos mas de un sonido de ze en Castellano” (Correas in Blanco 2006: 54). 40 “I hallando tanta confusion en que aia tres letras de un sonido, devemos desechar las otras dos c […] i usar sola esta z […] i ansi evitaremos la anbiguedad que deseamos quitar” (Correas in Blanco 2006: 54). 41 “En qué se falta en la orthographia castellana? Faltase […] de tres maneras […] La tercera mudando como hacemos por hazemos, goçamos por gozamos […] y assí de otros poniendo en lugar de z, ç con cedilla” (Pérez de Nájera in Blanco 2006: 58). 42 “C. es letra muda, pronúnciase Ce, tiene tres oficios. El segundo es prestado […] cuando se pronuncia como zeta griega, poniéndole debajo vna zerilla, como en estas diciones, zapato, çapato, zedaço, zarça, çeniça, çieno. La qual pronunciación hallo que se pegó a los castellanos de los moros andaluzes” (Jiménez Patón in Blanco 2006: 59). 43 “Esta letra ç con aquella coma abajo, y esta z tienen en todo el mismo valor, ansí çarça, zerda” (Luna in Blanco 2006: 59). 44 “Dos letras Griegas […] que de la lengua Griega se auian de comunicar a la nuestra, como son y, z, […] que aunque las latinas i, c, lo podían suplir, no era bien usar dellas, sino que las Griegas señalassen en el vocablo el origen” (Bravo Grájera in Blanco 2006: 59). 45 “La poca, o ninguna diferencia que ay en la pronunciación de ç y z […] da licencia para escrivir con la que mas presto a la memoria venga, y a lo sumo, solamente señalara yo al uso de la z, al principio […] y en el medio […] y también quando se halla entre dos vocales como dezir, hazer, induzir, etc.; esto es inteligible […] lo demas es adivinar (Villar in Blanco 2006: 60). 46 “La x en el latín no es otra cosa que la breuiatura de cs o de gs; nosotros dámosle tal […] pronunciación propia de la lengua aráuiga, de donde parece que nosotros la recebimos, porque otra lengua ninguna la reconoce por suia, y los moros siempre la ponen en lugar de nuestra s, y por lo que nosotros dezimos señor, san, […] por s, ellos dizen xenor, xan […] por x” (Nebrija in Blanco 2006: 68). 47 “La g […] cuando después de ella se siguen e, i, es propria de nuestra lengua, que ni judíos ni griegos, ni latinos, la sienten, ni pueden conocer por suia, salvo

An overview: sibilant merger in Spanish  63 el morisco, de la cual lengua io pienso que nos otros la recebimos” (Nebrija in Blanco 2006: 68). 48 “De manera que podremos usar la s en los vocablos que viéremos tener origen del latín, y la x en los que nos pareciere tienen origen del arávigo.” “Quanto a la j larga […] suena al castellano lo que al toscano gi” (Valdés in Blanco 2006: 69). 49 “Estas tres letras traen en gran baraja y discordia la buena ortographía, porque, con pareçer tan diferentes en si, tienen tanta semejança en la pronunçiaçión, que muchas vezes se ponen la vna por la otra […] Donde han de dezir dixo ponen dijo, y por poner lexos ponen lejos; […] poniendo la x por j, deziendo enoxos por enojos, mensaxero por mensajero” (Torquemada in Blanco 2006: 70). 50 “La x en el castellano tiene la mesma pronunçiaçión en el vocablo que tiene la j larga […] porque poca differençia haze dezir jarro o xarro, jornada o xornada” (Villalón in Blanco 2006: 71). 51 “Verdad es que algo mas aspera se pronuncia la x, que la j, consonante […] El sonido de la pronunciacion le enseñara con que letra deua escreuir. Dira jarro y no xarro, dira xara y no jara. Dira xabón y no jabón” (Villalón in Blanco 2006: 71). 52 “[Los de Castilla la Vieja] dizen xugar, y en Toledo dizen jugar” (F. Juan de Córdoba in Blanco 2006: 71). 53 “En muchas palabras apenas percibe la oreja la differencia que ay entre ellos, como trabajo y abaxo […] No puede el oydo sacar la verdadera escriptura dellas” (López de Velasco in Blanco 2006: 71). 54 “Es menester que los que enseñan leer y escriuir adviertan en que sus discipulos tengan entendido como han de diferenciar de la .x. a la.i. jota, porque muchas vezes he visto descuydarse en esto, no digo los niños solamente, sino los de mayor edad, que por escriuir Guadalajara dizen con .x. Guadalaxara. Y otras vezes por el contrario por decir con .x. Xaramillo dizen con la.i. Iaramillo” (Cuesta in Blanco 2006: 72). 55 “Faut noter qu’il y a grand affinité de pronunciation entre le g mis deuant e ou i, le j […] and l’x […] Car i’ay remarqué des mots escrits indifferemment par ces trois lettres, comme tixeras, tigeras & tijeras qui signifie des ciseaux” (Oudin in Blanco 2006: 72). 56 “Estas letras se pronunçien en lo vltimo del paladar cerca de la garganta […] la de la j sale blanda y amorosamente, y la de la x con mayor fuerça, diferenciándose […] viejos, lexos; hijo, dixo; juez, Xuárez” (Torquemada in Blanco 2006: 72). 57 “La .x. […] es como la.g., pero más denso y metido a la garganta” (López de Velasco in Blanco 2006: 73). 58 “j consone que les Espanols appelent jota & le prononcent […] au dedans de la gorge” (Oudin in Blanco 2006: 73). 59 “Avia aki de rreprehender al vulgo de Andaluzia la baxa, ke la truekan feamente por xe, i la xe por h. Mas los kuerdos, ó no lo hazen, o se korrixen dello en Kastilla, o komunikando kastellanos viexos, I se ofenden de tan torpe vizio” (Correas in Blanco 2006: 73). 60 “La G.  tiene dos oficios:  vno propio, como suena en […] nego, legas, garça, gordo; otro, prestado, como suena en […] Gerónimo, Gil, neges, legis. Esta vltima pronunciación es de sólo el romance castellano, y así las otras naciones no pueden sufrir que se la demos al latín” (Jiménez Patón in Blanco 2006: 74). 61 “Devese llevar cuydado, en no confundir la.j. y la .x. por lo mucho que en el son son parecidas” (Sebastián in Blanco 2006: 75). 62 “La j tiene tanta semejança con la x, que solo el cuidado las puede diferenciar” (Dávila in Blanco 2006: 76).

64  Eva Núñez-Méndez 63 Before the end of the sixteenth century, there were more restrictions to consonantal rhymes than in modern Spanish poetry, and words such as passa and casa, cabeça and belleza, baraja and baxa, suave and sabe could not rhyme (Menéndez Pidal 1962:  115). All sources studied indicate that cabeça was pronounced with [ts], nevertheless Pharies (2015: 133) describes it with [dz]. For more information about rhymes see Domínguez Caparrós (2014). 64 “Me parece que no solo puede adelantarse […] en medio siglo la fecha del ensordecimiento, sino que además es posible sostener que éste y el seseo coexistieron desde los primeros años en que se introdujo el castellano en América” (Parodi 1976: 124). 65 See Cock (1969) and his studies on seseo in New Granada from 1550–​1650. 66 See Vázquez Balonga (2015: 201–​207). 67 Véase Junco, La jota de Méjico y otras danzas (1967).

Glossary Andalucismo:  in linguistics, this theory classifies Latin American Spanish varieties as an extension of Andalusian Spanish, based on similar dialectal characteristics like seseo. Andalusian Spanish:  a Spanish dialect from the southern part of the peninsula. It includes southern variants which differ from both northern varieties and standard Spanish. Seseo and ceceo are current sociolinguistic phenomena in this region. Ceceo:  is the reduction of the phonemes /​θ/​and /​s/​(for graphemes , ; and , respectively) into only one dental sibilant /​s̪/​, similar to /​θ/. This lisping articulation is predominant in southern parts of Andalusia (Cádiz, southern Huelva, Granada and Seville  –​except the city  –​, and most of Málaga). E.g. pozo/​poso are pronounced the same, [pós̪o] almost [póθo]. Devoicing:  is a sound change where a voiced consonant becomes voiceless due to the influence of its phonological environment. Shift in the opposite direction is referred to as voicing or sonorization. Distinción:  describes the opposition between the phonemes /​θ/​ and /​s/​, e.g. pozo/​poso; as opposed to seseo/​ceceo dialects which have no opposition between and . It characterizes mainland Spain dialects. Koineization:  a tendency to reduce, level, simplify and regularize one variant from a group of competing language varieties in contact. Lenition:  in phonetics, it refers to a sound change that weakens consonant articulation making it become voiced, spirantized or deleted. “Lenition” means “softening” or “weakening.” The word comes from Latin L E N I S , “weak.” The opposite is fortition. Merger:  in historical linguistics, it is defined as the collapse of a phonemic distinction as one sound becomes identical with another. The affected phonemes only merge when they appear in certain phonological contexts. E.g. voiced [z]‌and voiceless [s] sibilants merged into [s] in old Spanish.

An overview: sibilant merger in Spanish  65 Seseo:  is the reduction of the phonemes /​θ/​and /​s/​(for graphemes , ; and , respectively) into only /​s/​. Seseo is rare in mainland Spain. It occurs in Andalusia (mainly coastal regions), in the Canary Islands, and Latin American variants of Spanish. E.g. pozo/​poso are pronounced the same [póso]. Sibilant:  in phonetics, it is a consonant sound, in which the tip of the tongue is brought near the roof of the mouth and air is pushed making a whistling or hissing sound, e.g. [s]‌, [z], [ʃ] and [ʒ].

References Alarcos Llorach, Emilio (1950). Fonología española. Madrid: Gredos. Alarcos Llorach, Emilio (1951). Gramática estructural. Madrid: Gredos. Alarcos Llorach, Emilio (1988). De nuevo sobre los cambios fonéticos del siglo XVI. In M. Ariza, A. Salvador & A. Vindas (Eds.), Actas del I Congreso Internacional de Historia de la Lengua Española (pp. 47–​59). Madrid: Arco Libros. Alarcos Llorach, Emilio (1991). Fonología diacrónica del español. In Fonología española. Madrid: Gredos. Aleza Izquierdo, Milagros, & Enguita Utrilla, José (Eds.) (2010). La lengua española en América: norma y usos actuales. Valencia: Universitat de Valencia. Alonso, Amado (1949). Examen de las noticias de Nebrija sobre Antigua pronunciación española. Nueva Revista de Filología Hispánica, 3, 1–​82. Alonso, Amado. (1951a). Cronología de la igualación c-​z en Española. Hispanic Review, 19, 37–​58, 143–​164. Alonso, Amado (1951b). Cómo no se pronunciaban las ç y z antiguas. Hispania, 34(1),  51–​53. Alonso, Amado (1955). De la pronunciación medieval a la moderna en español. Vol 1. Madrid: Gredos. Alonso, Amado (1967a). De la pronunciación medieval a la moderna en español I y II. Madrid: Gredos. Alonso, Amado (1967b) La base lingüística del español americano. ELTH Estudios Lingüísticos. Temas hispanoamericanos, pp. 7–​60. Alonso, Dámaso (1962). Ensordecimiento en el norte peninsular de alveolares y palatales fricativas. In M. Alvar et  al. (Eds.), Enciclopedia lingüística hispánica I (suplemento) (pp. 85–​103). Madrid: CSIC. Alonso, Dámaso (1972). Temas y problemas de fragmentación fonética peninsular. Obras completas 1.  Estudios lingüísticos peninsulares (pp. 13–​ 290). Madrid: Gredos. Alvar, Manuel (1983). A vueltas con el seseo y ceceo. In F. Marcos Martín (Ed.), Introducción plural a la gramática histórica (pp. 130–​144). Madrid: Cincel. Álvarez Nazario, Manuel (1987). Orígenes del español en Puerto Rico. In H. López Morales et al. (Eds.), Actas del I Congreso Internacional sobre el español de América I (pp. 33–​46). Madrid: Arco Libros. Ariza Viguera, Manuel (2004). Revisión del cambio fonético y fonológico. Lexis, 28,  7–​27. Blanco, Marta (2006). Aproximación a la cronología de las transformaciones funcionales de labiales y sibilantes del español. Santiago de Compostela: Universidad Santiago de Compostela.

66  Eva Núñez-Méndez Boyd-​Bowman, Peter (1964). Índice geobiográfico de cuarenta mil pobladores españoles de América en el siglo XVI. Tomo I. Bogotá: Instituto Caro y Cuervo. Boyd-​Bowman, Peter (1968). Índice geobiográfico de cuarenta mil pobladores españoles de América en el siglo XVI. Tomo II. México: Jus. Boyd-​Bowman, Peter (1973). Patterns of Spanish Emigration to the New World (1493–​ 1580). Buffalo, NY: Council of International Studies. Boyd-​Bowman, Peter (1980). From Latin to Romance in Sound Charts. Washington D.C.: Georgetown University Press. Bradley, Travis, & Delforge, Ann Marie (2006). In R. Gess & D. Arteaga (Eds.), Historical Romance Linguistics (pp. 19–​52). Amsterdam: John Benjamins. Bürki, Yvette, & Romero, Elena (Eds) (2014). La lengua sefardí: aspectos lingüísticos, literarios y culturales. Berlin: Frank & Timme. Bürki, Yvette (2013). The status of Judeo-​Spanish in the Ottoman Empire. In J. del Valle (Ed.), A Political History of Spanish: The Making of a Language (pp. 335–​ 349). Cambridge: Cambridge University Press. Cabrera Morales, Carlos (1992). Las sibilantes medievales:  reflexiones sobre los problemas de desonorización. Medievalia, 11, 1–​18. Campos-​Astorkiza, Rebeka (2014). Sibilant voicing assimilation in Peninsular Spanish as gestural blending. In M. Côté & E. Mathieu (Eds.), Variation within and across Romance Languages (pp. 17–​38). Amsterdam: John Benjamins. Cano Aguilar, Rafael (1988). El español a través de los tiempos. Madrid: Arco. Cano Aguilar, Rafael (2004). Cambios en la fonología del español durante los siglos XVI y XVII. In Cano Aguilar (Ed.), Historia de la lengua española (pp. 825–​916). Barcelona: Ariel. Catalán, Diego (1956). El ceceo-​zezeo al comenzar la expansión atlántica de Castilla. Boletim de Filología, 16, 305–​34. Catalán, Diego, & Reinhardt, Karl J., trans. (1957). The end of the phoneme /​z/​in Spanish. Word, 13(2), 283–​322. Chomsky, Noam, & Morris, Halle (1968). The Sound Pattern of English. New York: Harper. Clements, George N. (1985). The geometry of phonological features. Phonology Yearbook, 2, 225–​252. Cock, Olga (1969). El seseo en el Nuevo Reino de Granada 1550–​1650. Bogotá: Instituto Caro y Cuervo. Cover, Thomas, & Thomas, Joy (2006). Elements of Information Theory. Hoboken, NJ: Wiley-​Interscience. Cuervo, Rufino José (1987). Disquisiciones sobre antigua ortografía y pronunciación castellanas. In Obras III (pp. 344–​476). Bogotá: Instituto Caro y Cuervo. Danesi, Marcel (1977). The case of andalucismo re-​examined. HR XLV, 181–​193. Domínguez Caparrós, José (2014). Métrica española. Madrid: UNED. Espinosa, Aurelio M. (1935). Arcaísmos dialectales:  la conservación de s y z sonoras en Cáceres y Salamanca. Madrid: Anejo XIX de la Revista de Filología Española. Frago Gracias, Juan Antonio (1989). ¿Sólo gritas en el edificio del reajuste fonológico? LEA, 11, 125–​143. Frago Gracias, Juan Antonio (1992). El seseo: orígenes y difusión Americana. In C. Hernández Alonso (Ed.), Historia y presente del español de América (pp. 113–​142). Valladolid: Junta de Castilla y León.

An overview: sibilant merger in Spanish  67 Frago Gracias, Juan Antonio (1993). Historia de las hablas andaluzas. Madrid: Arco Libros. Frago Gracias, Juan Antonio (1999). Historia del español de América. Madrid: Gredos. Galmés de Fuentes, Álvaro (1962). Las sibilantes de la Romania. Madrid: Gredos. Galmés de Fuentes, Álvaro (1965). Resultados de -​LL-​y -​LY-​, -​C’L-​en los dialectos mozárabes. RLIR, 29, 60–​97. Gavel, Henry (1929). Grammaire Basque I. Bayonnne: Imprimerie du Courrier. Gess, Randall, & Arteaga, Deborah (Eds.) (2006). Historical Romance Linguistics. Amsterdam: John Benjamins. Gil Arrando, M. Consuelo (1986). Contribución al estudio lingüístico del español del siglo XVII. Cauce Revista de Filología, 9, 39–​60. Goldsmith, John A. (1998). On Information Theory, Entropy, and Phonology in the 20th Century. Paper presented at the Royaumont CTIP II on Phonology in the 20th century, University of Chicago. González Ollé, Fernando (1987). Primeras noticias y valoraciones del andaluz. Boletín Real Academia Española, 67, 347–​387. Guitarte, Guillermo (1987). Cecear por gracia. In Actas del VII Congreso de la Asociación de Lingüística y filología de America Latina. Tomo I (pp. 127–​141). Santo Domingo, R.D.: UNPHU. Guitarte, Guillermo (1992). Cecear y palabras afines. In Actas del Congreso Internacional de Historia de la Lengua Española. Tomo I (pp. 127–​164). Madrid:  Pabellón de España. Gurevich, Naomi (2011). Lenition. Faculty Research and Creative Activity 10. Retrieved from http://​thekeep.eiu.edu/​commdis_​fac/​10 (accessed 1 December 2020). Hall, T. A. (2001). Distinctive Feature Theory. Berlin: De Gruyter. Harris-​Northall, Ray (1992). Devoicing, affrication and word-​final  –​z in medieval Spanish. Hispanic Linguistics, 4, 245–​274. Hume, Elizabeth, & Mailhot, Frédéric (2013). The role of entropy and surprisal in phonologization and language change. In Alan Yu (Ed.), Origins of Sound Patterns:  Approaches to Phonologization (pp. 29–​49). Oxford:  Oxford University Press. Jakobson, Roman, & Morris, Halle (1956). Fundamentals of Language. The Hague: Mouton. Junco, Alfonso (1967). La jota de Méjico y otras danzas. México: Jus. Kania, Sonia (2016). Diachronic perspectives on varieties of Spanish pronunciation:  seseo and yeísmo. In E. Núñez-​Méndez (Ed.), Diachronic Applications of Hispanic Linguistics (pp. 200–​238). Newcastle, UK: Cambridge Scholars. Kauffeld, Cynthia (2011). Andalusian Spanish:  A linguistic study of 14th-​and 15th-​ century Texts from Sevilla and Córdoba. New York: Hispanic Seminary of Medieval Studies. Kauffeld, Cynthia (2016). Andalusian Spanish: A diachronic survey of its origins and footprints in the Americas. In E. Núñez-​Méndez (Ed.), Diachronic Applications of Hispanic Linguistics (pp. 167–​199). Newcastle, UK: Cambridge Scholars. Kiddle, L. B. (1975). The chronology of the Spanish sound change š >x. In Studies in honor of Lloyd A.  Kasten (pp. 73–​100). Madison, WI:  Hispanic Seminary of Medieval Studies.

68  Eva Núñez-Méndez Kunz, Marco (2014). El judeoespañol polaco. In Y. Bürki et  al. (Eds.), La lengua sefardí:  aspectos lingüísticos, literarios y culturales (pp. 109–​123). Berlin:  Frank & Timme. Lapesa, Rafael (1957). Sobre el seseo y el ceceo andaluces. In Estructuralismo e Historia. Miscelánea Homenaje a A. Martinet Vol. 1 (pp. 67–​94). Laguna: Universidad de la Laguna. Lapesa, Rafael (1981). Historia de la lengua española. Madrid: Gredos. Lass, T. S., Szigetvári, P., & Zuraw, K. (2008). Lenition, weakening and consonantal strength:  Tracing concepts through the history of phonology. In J. Brandão de Carvalho et al. (Eds.), Lenition and Fortition (pp. 9–​92). Berlin: De Gruyter. Lloyd, Paul M. (1987). From Latin to Spanish. Philadelphia, PA:  American Philosophical Society. Lloyd, Paul M. (1993). Del latín al español. I. Fonología y morfología históricas de la lengua española. Madrid: Gredos. Malkiel, Yakov (1971). Derivational transparency as an occasional co-​determinant of sound change: A new causal ingredient in the distribution of “-​ç-​” and “-​z-​” in ancient Hispano-​Romance. Romance Philology, 25(1), 1–​52. Martinet, André (1951). The unvoicing of Old Spanish sibilants. Romance Philology, 5, 133–​156. Martinet, André (1952). Celtic lenition and Western Romance consonants. Language 28(2), 192–​217. Martinet, André (1955). Économie des changements phonétiques. Berne: A. Francke. Martinet, André (1974). Economía de los cambios fonéticos. Madrid: Gredos. Martínez Celdrán, Eugenio (1992). Un mismo parámetro fonético en el fondo de la lenición protorromance: la duración. In Bartol Hernández et al. (Eds.), Estudios filológicos en homenaje a Eugenio de Bustos Tovar II (pp. 621–​640). Salamanca: Universidad de Salamanca. Menéndez Pidal, Ramón (1919). Documentos lingüísticos de España I, Reino de Castilla. Madrid: Estudios Históricos. Menéndez Pidal, Ramón (1940). Manual elemental de gramática histórica española. Madrid: Espasa Calpe. Menéndez Pidal, Ramón (1962). Sevilla frente a Madrid. Algunas precisiones sobre el español de América. In Diego Catalán (Ed.), Miscelánea homenaje a André Martinet: Estructuralismo e historia III (pp. 99–​165). Canarias: Universidad de La laguna. Moreno de Alba, José (2007). Introducción al español americano. Madrid: Arco Libros. Noll, Volker (2005). Reflexiones sobre el llamado andalucismo del español de América. In V. Noll et al. (Eds.), El español en América: aspectos teóricos, particularidades, contactos (pp. 95–​111). Madrid: Iberoamericana-​Vervuert. Ohala, J. J., & Solé, M. J. (2010). Turbulence and phonology. In S. Fuchs et al. (Eds.), Turbulent Sounds: An Interdisciplinary Guide (pp. 37–​97). Berlin: De Gruyter. Oviedo, Gonzalo Fernández de (1959 [1532]). Historia general y natural de las Indias. Madrid: Atlas. Parodi, Claudia (1976). Para el conocimiento de la fonética castellana en la Nueva España:  1523. Las sibilantes. In Actas III Congreso:  Asociación de Lingüística y Filología de América Latina (pp. 115–​125). San Juan: Universidad de Puerto Rico. Parodi, Claudia (1995). Orígenes del español americano. México City: UNAM. Parodi, Claudia (2017). Spanish loanwords in Amerindian languages. In K. Dakin et al. (Eds.), Language Contact and Language Change in Mesoamerica and Beyond (pp. 155–​169). Philadelphia, PA: John Benjamins.

An overview: sibilant merger in Spanish  69 Pascual, Jose Antonio (1988). Notas sobre las confusiones medievales de las sibilantes. LEA, 10, 125–​131. Penny, Ralph (1993). Neutralization of voice in Spanish and the outcome of the Old Spanish sibilants:  A case of phonological change rooted in morphology. In D. Mackenzie & I. Michael (Eds.), Hispanic Linguistic Studies in Honor of F. W. Hodcroft (pp. 75–​88). Oxford: Dolphin. Penny, Ralph (2000). Variation and Change in Spanish. Cambridge:  Cambridge University Press. Penny, Ralph (2002). A History of the Spanish Language. Cambridge:  Cambridge University Press. Pensado, Carmen (1993). El ensordecimiento castellano: ¿un fenómeno extraordinario? Anuario de lingüística hispánica, 9, 195–​230. Pharies, David (2015). A Brief History of the Spanish language. Chicago, IL: University of Chicago Press. Quilis, Antonio (1993). Tratado de fonética y fonología españolas. Madrid: Gredos. Rosenblat, Angel, & Tejera, M. J. (2002). El español de América. Caracas: Biblioteca Ayacucho. Rousselot, Jean (1901). Synthese phonetique. La Parole, 11, 641–​647. Serrano, Abraham E. (1982). Estudio de teoría ortográfica del español vol 2. Murcia: Editum. Shannon, Claude (1948). A mathematical theory of communication. The Bell System Technical Journal, 27, 379–​423, 623–​656. Thurneysen, Rudolf (1946). A Grammar of Old Irish. Dublin: Institute for Advanced Studies. Tuten, Donald N. (2003). Koineization in Medieval Spanish. Berlin: De Gruyter. Vázquez Balonga, Delfina, & Sánchez Prieto Borja, Pedro (2015). ¿Seseo en el centro peninsular? Revista de Historia de la lengua española, 10, 201–​207. Veiga Rodríguez, Alexandre (1988a). El rasgo fónico tensión y los procesos protohispánicos de lenición consonántica. In M. Ariza et  al. (Eds.), Actas del I  Congreso Internacional de Historia de la Lengua Española I (pp. 193–​ 208). Madrid: Arco Libros. Veiga Rodríguez, Alexandre (1988b). Reaproximación estructural a la lenición proto-​ romance. Verba, 15, 17–​78. Zamora Vicente, Alonso (1967). Dialectología española. Madrid: Gredos. Zampaulo, André (2013). Sibilant dissimilation in the history of Spanish:  An information-​theoretic approach. In C. Howe et  al. (Eds.), Selected Proceedings of the 15th Hispanic Linguistics Symposium (pp. 172–​ 178). Somerville, MA: Cascadilla Press. Zygis, Marzena, Fuchs, Susanne, & Koenig, Laura L. (2012). Phonetic explanations for the infrequency of voiced sibilant affricates across languages. Laboratory Phonoloy, 3(2), 299–​336.

70  Eva Núñez-Méndez

Topics for discussion (1) Briefly summarize the origin and development of Spanish sibilants. Give lexical examples that show the changes. (2) Provide possible causes and theories for the sibilant devoicing. (3) Explain the epicenter of the sibilant merger and its geographical implications. (4) Describe the andalucista theory. How does Boyd-​Bowman’s data contribute to this theory? (5) What factors affected the propagation and completion of the sibilant merger? (6) Discuss the chronological framework generally accepted by grammarians and treatise writers as well as the attestations and graphic misspellings from documented records of the sibilant merger. (7) Read the following words (from the rhymes in this chapter) and provide the phonetic symbol for the old sibilant. Example: balança [ts]

a) alegrasse b) anexos c) belleza d) corteza e) cosa f) dexes g) enojo h) escusa i) eso



j) espejos k) esposa l) hojas m) huesos n) orejas o) ovejas p) passe q) pazes r) quexes



s) rojos t) rosa u) seso v) templança w) veces x) vengança y) viejas z) vsa

(8) Briefly describe the RAE’s graphic standardization of sibilants. (9) What are the modern phonetic solutions that emerged from the sibilant merger? (10) Provide the cognates of these Latin words in modern Spanish, Portuguese, French, and Italian, and trace the phonetic rules or tendencies for the sibilant. Provide an English cognate if possible. Example:  PASSATUM [s]‌, pasado in Spanish pronounced [s], passado in Portuguese with [s], passé in French with [s], and passato in Italian with [s]; in English past. The double –​S S -​in Latin evolved to the voiceless [s] in these romance languages 

An overview: sibilant merger in Spanish  71

A) AXEM B)

G E NTEM

C)

J U V E NEM

D ) M A RTI UM



E) M E N SE M F) OC U LU M G) RU SSE U M H) SP ON SAM

(11) Read the following poem and explain the rhymes between pesa, cesa and besa (from Lat. PENS ARE, C E SSARE , and BASI ARE respectively). What does the rhyme indicate about the evolution of the original Latin sibilants? Ni puedo encarecer lo que me pesa que tuviese tal crédito contigo. Dichosa aquella dama que no cesa de reírse de Alcina entre tus brazos, cuya risa tu boca adora y besa. (Lope de Vega, Rimas, 437) (12) Consider Parodi’s quotes about colonial American Spanish and provide your own view on the matter. a) “Several Hispanists, such as Boyd-​Bowman (1976), Lapesa (1956), Menéndez Pidal (1962) and others, believed the varieties of Latin American Spanish had an exclusive link with Andalusian Spanish, mainly with the dialect from Seville, since there were no obvious traces of other dialects from Spain, including the most widespread peninsular varieties, such as Old Castilian and Toledan Spanish in the Americas. This has been called the andalucista theory of Latin American Spanish” (2017: 158). b) “By examining the phonetic shape of the Spanish loanwords in Amerindian languages such as Aymara, Guajiro, Huastec Nahuatl and others, it is possible to show that these loanwords show traces of various Spanish dialects, in particular, Old Castilian, Andalusian, Toledan and the newly formed Spanish koines in the Americas. These facts […] disfavor the strong version of the andalucista theory of Latin American Spanish and support the multiple koines hypothesis, advanced […] by myself and others in Parodi (1995, 2009), del Valle (1998) Fontanella de Weinberg (1992), de Granda (1994)” (2017: 158). c) “According to the multiple koines hypothesis, several dialects, mainly Andalusian, Old Castilian and the Toledan norm leveled their differences, forming several koines in Latin American Spanish. Since the early 16th cent. the historian cronistas have declared that speakers from everywhere in the Peninsula came

72  Eva Núñez-Méndez to the Americas. These newcomers spoke different varieties of Spanish and some of them knew, in addition, more than one Peninsular language or dialect” (2017: 159). d) “The Toledan spelling was used beginning in the 13th cent. in Spain, after Alfonso X established it to be employed in legal and in formal written documents. In the Americas, a few highly educated individuals spoke and wrote the prestigious Toledo norm. The majority of settlers spoke Andalusian Spanish, Old Castilian or one of the Latin American koines. The few that knew how to read and write used a watered-​ down variant of the Toledan variety that many times revealed their pronunciation through their misspellings” (2017: 158). e) “El çeçeo-​zezeo, es decir, la confusión de /​ts/​en /​s/​y de /​dz/​ en /​z/​] se ha documentado en textos sevillanos del siglo XV, lo cual ha permitido afirmar que durante el último cuarto de siglo está profundamente arraigado en el habla común de Sevilla. Solo una minoría cultivada debió de seguir la norma toledana distinguidora de esta ciudad. Al iniciarse la expansión transatlántica de España en Canarias y América, se supone que la mayoría de los sevillanos debieron haber sido çeçeantes-​zezeantes […] Ello no solo prueba la monogénesis de esta peculiaridad del español atlántico, sino que favorece la hipótesis de que en las Antillas se originó un español nivelado ‘sevillanizante’, base del ulterior español americano” (1995: 36)

2  Sibilants in western Andalusian Spanish The lack of a Sevillian norm in the Jerezano speech community Jannis Harjus

2.1  Introduction: the Jerezano speech community In recent years, there have been several sociophonetic works on the Jerezano speech community (Carbonero 1992; García-​Amaya 2008; Harjus 2018a). Due to the increased number of studies, differences in local salient phonetic characteristics have also surfaced. The idea of a speech community that shows a predominant realization of ceceo from the traditional dialectological approach (Alvar, Llorente, & Salvador 1973)  was replaced by the first sociolinguistic work by Carbonero and colleagues, who demonstrate a clear tendency to seseo in young and educated speakers (1992:  24). Villena concludes from this, and other Andalusian sociolinguistic data, that the local variety of Seville has the greatest influence on the phonetics of western Andalusian Spanish (2008:  144). According to Villena, a possible regional standard is mainly based on the realization of Sevillian seseo and, in the urban speech communities of western Andalusia, it leads to a reduction of the national standard of European Spanish (reducing the distinction between the phonemes /​s/​and /​θ/​) as well as to a decline of less prestigious phenomena of western Andalusian Spanish, especially ceceo (2008). However, recent socio-​phonetic data on spoken Spanish in Jerez, in particular García-​Amaya (2008) and Harjus (2018a), do not support this hypothesis and present data that show a tendency towards either maintaining ceceo or shifting to sibilant distinction in the speech community. In order to arrive at a more precise answer to the norm question in the Jerezano speech community, here we will be focusing on a brief presentation of the term norma sevillana, an exhaustive sociophonetic analysis of the use of sibilants in Jerez, and how the use of sibilants in western Andalusia is perceived by Jerezano speakers themselves. Consequently, we do not only present a classic sociolinguistic study on sibilants in Jerez, but we also point out what the Jerezano speakers themselves generally know about the sibilant variation and, more precisely, what they think about a possible seseo-​based regional norm in western Andalusian Spanish:  for this second purpose, we

74  Jannis Harjus

Map 2.1 Western and Eastern parts of Andalusia, and Jerez.

use the theoretical framework of Perceptual Dialectology (Preston 1999) and Perceptual Variety Linguistics (Krefeld & Pustka 2010b), which have already been illustrated in Harjus (2018a). See Map 2.1 for approximate location of western and eastern Andalusian Spanish.

2.2  Sibilants in Andalusian Spanish: some data about geographical and social distributions The first chapter already indicates that the sibilant merger in the fifteenth century took another course in Andalusia than it did in Castile. We emphasize that the /​s̹/​and the /​z̹/​resulting from the sibilant merger in Andalusia were also too similar, which is why these sounds neutralized into a single phoneme, the voiceless fricative /​s̹/​, which had two possible realizations:  a postdental [θ] and a predorsal [s]‌. Thus, today, there is the current ceceo for the postdental realization [θ], and the seseo with the predorsal [s] (Frago 1993: 213). The two possible modern realizations of /​s̹/​are both articulated with the lingual apex in the lower incisors and a constriction of the oral cavity. The first realization is predorsal in the upper incisors, which produces the ceceo, while the second realization, the seseo, or, better, Sevillian seseo, is in the alveoli. While the first pronunciation causes a weak sound, the seseo tone is jarring. In sum, both articulations have two tongue contacts:  both have tongue tip contact at the lower incisors, but tongue body contact occurs upper incisor for ceceo, and in the alveolar area for seseo. Both realizations approximate the timbres of the standard Castilian /​s/​and /​θ/​([s]‌apical and [θ] interdental) but maintain their own character. The sound of seseo can be distinguished from the ceceo by divergent articulatory intensity, less intense in the pronunciation of the [θ] and with maximum intensity in the [s] sound. The difference of intensity between [s] and [θ] in Andalusia is less clear than between [s] and [θ] in Castilian, whose average is close to 20 dB (Lasarte Cervantes 2010: 500).

Sibilants in western Andalusia: Jerezano  75 So there are two divergent types of pronunciation of /​ s/​among the Andalusian seseo speakers, with a clear preference for the Sevillian realization, already described above as predorsal. The Cordoban type of seseo, with a flatter lingual position, is restricted to the province of Cordoba, while the Sevillian /​s/​has spread not only in the far west of Andalusia, but also in Granada and Malaga. Especially, the Andalusian /​θ/​differs from the interdental Castilian in a more dental than interdental realization, although it is not uncommon to find interdental solutions in the production of ceceo as well (Narbona Jiménez et al. 2011: 175). Even Narbona warns of certain linguistic polymorphism in ceceante and seseante pronunciations: even the same speakers tend to perform /​s/​and /​θ/​in different ways, sometimes closer to the standard distinction, other times closer to the southern sounds. In addition, there are speakers who are not distinct seseantes or ceceantes and they exchange the divergent realizations, for which they are called seceantes or ceseantes, depending on their articulatory tendencies (2011: 175). The seseo also reached Spanish speakers in ultramar, the Canary Islands and Spanish America. Despite the fact that ceceo today is known basically by its numerous speakers in the south of Spain, there are also speech communities with ceceantes in the Americas, especially in Colombia and other Caribbean areas. Today, scholars agree on the fact that the phenomenon of Andalusian dephonologization arose in the valley of the Guadalquivir in the fifteenth century through the coexistence of reconquistadores from different origins (Frago 1993: 307). Other linguists, such as Mondéjar (1985), share this point of view. Although the first linguists denied it (Navarro et al. 1933), it seems that the seseante realization was slightly before the ceceante and spread from the city of Seville to bordering areas and, along with the advance of the Christian reconquista of the Iberian Peninsula, towards the Kingdom of Granada. Especially in rural areas and among less educated speakers, seseo seems to have become ceceo, although the evidence is still scarce (Narbona Jiménez et al. 2011: 83). Nowadays, the Andalusian territory is divided into two large zones: that of the distinction, more or less the entire northern and eastern strip of the autonomous community, and a southern one of dephonologization. The latter is divided in two parts: the speakers in the center of the Andalusian territory, especially in the south of Cordoba, the north of Malaga, the center of Huelva and the north of Seville, as well as the great urban centers of western Andalusia, such as the Cities of Cadiz and Seville, are mostly seseantes; those from the other major part, especially the rural areas, the Andalusian Atlantic and Mediterranean coast, from Huelva to Almeria, are ceceantes, including the urban spaces in the south of Seville and of the provinces of Malaga and Granada (Alvar et al. 1973: 1580, 1705). See Map 2.2. Nowadays, however, it is known that, rather than being geographical, the division between ceceo and seseo zones is sociolinguistic:  while the ceceo area seems to be more rural, the seseo one includes the larger urban speech

76  Jannis Harjus

Map 2.2 Ceceante and seseante zones in Andalusia.

communities, especially, Seville, Cordoba, Malaga and Cadiz: “the normative prestige of the seseo variety is greater than that of ceceo and as such tends to be found among speakers of elevated sociocultural status, even though they belong to a traditionally ceceante area” (Jiménez 1999: 33). In eastern Andalusia, the situation seems to be quite different. Melguizo, in a recent study on the variety of rural Granada, emphasizes the dominance of ceceo among rural speakers. These speakers, however, make the distinction as soon as they move to the capital of Granada (Melguizo 2007). Similar data is provided for the speech community of urban Malaga (Ávila 1994). However, the high number of distinctions in the study of the Malaga speech community is striking. Whereas the urban Sevillian speech community shows a tendency towards the seseo, the rural one tends to the ceceo (Carbonero 1982), although recently there has been a certain increase in the distinction between young people, as González (1993) points out. The data for Cordoba is quite clearer:  Uruburu (1990) concludes that the younger speakers only realize seseo or distinction. In Huelva, Heras (1996) does not find seseo, but rather ceceo and distinction. With this background, it can be seen that the phenomena of seseo and ceceo share the Andalusian dialect area with the Castilian distinction of /​s/​and /​θ/​. The latter one is increasingly present among the most educated classes and among the youngest Andalusian speakers.

2.3  The Sevillian norm: historical and actual approaches 2.3.1  Descriptive linguistic norms The term “linguistic norm,” which is not defined coherently, derives from the Latin word norma, translated “rule,” which is often used in human sciences as a precondition for central human actions which, in turn, are based on

Sibilants in western Andalusia: Jerezano  77 conventions and which are performed on a daily basis. Since human linguistic behavior is part of social behavior, there are also linguistic norms that are simultaneously part of the social norms that regulate social life (Dittmar 1997: 163). In modern variationist linguistics, the use of the term “linguistic norm” has been imposed in its plural form:  linguistic norms (Bartsch 1987:  5). The term linguistic norms suggests that there is not one fixed standard, but rather multiple possible variants of diatopic, diastratic, or diaphasic norms. In addition, the term itself is often divided into two parts:  while a prescriptive norm is equal to the codified and prestigious standardized language in formal communicative situations, the descriptive norm refers to linguistic uses of certain social groups that can be observed empirically (Polzin-​Haumann 2013:  45). Carbonero subdivides the descriptive norms into linguistic norms and sociolinguistic norms:  while the former is a set of linguistic habits considered as normal and used by most speakers of a speech community respectively (2003: 81), a sociolinguistic norm refers to a process of selectivity of linguistic uses, since not all speakers of a given speech community share the same variables. Furthermore, this means that a descriptive norm covers usual linguistic realizations in a certain speech community, but it always exists in correlation with other descriptive norms (Harjus 2018a: 27). We define linguistic norms as social aspects of linguistic communication; linguistic norms are abstractions of linguistic awareness that are based on expectations regarding adequate linguistic use within a speech community in specific situations (Gloy 2005: 121). This linguistic awareness and, consequently, the linguistic norms grow with communicative experiences (Dittmar 1997: 165). Therefore, it is important to analyze the speaker’s metalinguistic knowledge, since the linguistic norms imply evaluative processes that can lead to conflicts in communicative situations. This is why in modern variationist linguistics we take into consideration the speakers’ own linguistic awareness (Harjus 2018a: 29). However, it ought not to be forgotten that, due to the normative processes initiated by linguistic policies, there is a certain tension between the descriptive and prescriptive norms (Polzin-​Haumann 2013:  46). This tension appears in the Andalusian variety area, since the Andalusian language policy putatively protects the heterogeneity of Andalusian varieties (Parliament of Andalusia 2007, article 12.3.2), but, in fact, often intervenes both in the media and in secondary education in a prescriptive manner. These prescriptive norms supposedly lead to a loss of overt and covert prestige of linguistic variants, such as ceceo (Harjus 2018b). These norms imposed upon Andalusian speakers, apart from relying on northern Spanish as a national standard (Carrascosa 1997: 429), are frequently based on the linguistic variants of the urban variety of Seville. Some linguists consider this urban variety to act as a descriptive linguistic norm for a large part of western Andalusia.

78  Jannis Harjus 2.3.2  Sevillian norm The term Sevillian norm has come to light in early Hispanic dialectology. Both Menéndez (1962) and Alvar (1990) mention the normative function of the Sevillian speech variety for the expansion of Spanish in the Americas: “From the city came a linguistic norm that was a canon in the great expansion of Spanish” (Alvar 1990: 37). Carrasco (2007) also supports the Sevillian norm thesis for the late middle ages. However, this theory does not hold true for the nearby western Andalusian area. In the same way, the emergence of Canarian and American Spanish varieties fuels the theory of Andalucismo proposed by Wagner (1927), which has gained more credibility by the data provided by Boyd-​Bowman (1956). Far from these historical visions of the Sevillian norm and Andalucismo theories, Carbonero attempts to create a norm for Andalusian speakers based on supposedly generalized and prestigious phonic forms throughout the western part of Andalusia. On the one hand, phonic phenomena, such as the seseo, used by western Andalusian speakers, according to Carbonero, have an overt prestige. On the other hand, phonic features, such as ceceo, are abandoned by speakers in formal communicative situations (2003: 114). Unlike Carbonero, Mondéjar concludes that no divergent norm, neither based on the Sevillian variety nor on the Granadian one, exists in Andalusian Spanish (1995:  39). Recently, however, Villena (2008) aimed to demonstrate a phonetic-​phonological norm for the western part of Andalusia as well and collected data from several studies on Andalusian speech communities to underline a division between western and eastern areas. For this reason, Villena sees two divergences within the Andalusian area: while the eastern area of Andalusia is lacking a regional center that can act as a normative model for speakers, the speech community of Seville has an impact on the nearby area, which is essential to form the basis of a regional norm (2006: 240). The varieties of eastern Andalusia are closer to the Spanish national standard. In other words, the varieties of eastern Andalusia are in the process of “vertical convergence” (Villena 2008: 144) towards the peninsular standard of European Spanish and, therefore, are similar to what Coseriu calls tertiary dialects (1980: 113). The western area, however, may follow Seville and normative European Spanish as described below. Partially based on the theoretical scheme of the diglossia of Auer (2005), Villena (2006) focuses on the question of a supposed western norm as a regional standard located between the local varieties of western Andalusia and standard European Spanish. Therefore, in western Andalusian there are three distinct varieties: while the national standard acts as a reference model and the different vernacular varieties form the basis of the variational continuum, the regional standard, also defined as norma sevillana (Villena 2008: 144), is between the two opposite poles. Although it has an overt prestige similar to the national standard of European Spanish in its area of ​​communication, it

Sibilants in western Andalusia: Jerezano  79 is based on the phonic feature of seseo in western Andalusia: “Seseo and lenition [ʃ] of /​tʃ/​would easily form part of the western regional standard due to their social prestige” (Villena 2008: 156). However, Harjus already warns that the database taken by Villena in consideration for its analysis of the variational situation of western Andalusia is quite obsolete (2018a: 34). Removing Villena’s most recent findings on the malagueño speech community, the figures regarding the dephonologization of /​s/​ and /​θ/​in Cordoba, Huelva, Cadiz, Granada, Jaen and Jerez de la Frontera derive from studies of the 80s and 90s of the last century. In the case of Jerez, which is central in this chapter, we already have more recent data on seseo and ceceo that must be considered (García-​Amaya 2008; Harjus 2018a). Nevertheless, Samper already points to a possible rejection of a regional western norm due to the data: “This shift implies a move towards the canon of Madrid rather than that of Seville” (2011: 116).

2.4  Analysis: descriptive data on sibilants in Jerezano Spanish 2.4.1  State of the art Traditionally, seseo does not form part of Jerezano Spanish, since it is presented as exclusively ceceante (Alvar et al. 1973: 1580). This classic description has been modified by the first sociolinguistic work on the speech community, in which Carbonero assures that 44% of the speakers are seseante and 47% ceceante (1992: 24). This conclusion has been repeated in the large monographs on Andalusian Spanish (Jiménez 1999: 33) as well as in several articles by Villena (2006, 2008). According to the latter, the data offered by Carbonero (1992), especially the increase of seseo by more educated speakers, underscore the possible extension of a Sevillian norm, which is mainly based on the use of seseo, within Jerezano Spanish. However, the most recent study on sibilants in Jerez shows that seseo does not seem to have penetrated the local speech community (García-​Amaya 2008: 65). This analysis emphasizes a high index of ceceo, but, at the same time, it indicates high numbers of standard Spanish distinction (García-​Amaya 2008). In the subsequent section, we will analyze data from a new project. 2.4.2  Actual data from an oral corpus: global visions The present analysis is based on data from an oral corpus. One-​hour interviews were conducted both informally and formally with 18 respondents that we knew personally and nine Jerezano speakers we did not know. In total, we conducted 27 interviews with Jerezano speakers, who differ in gender, age, and socio-​cultural background (see Harjus 2018a: 101). Realizations of ceceo are quite frequent in the corpus:  more than two-​ thirds (72%) of the realizations of phoneme /​s/​are [θ]. This analysis hints at a very striking increase in ceceo in the Jerez speech community with respect to

80  Jannis Harjus the data of Carbonero, who find only 44% of ceceo realizations in his corpus (1992: 25). These results are close to those obtained by García-​Amaya, who shows 73% of ceceo realizations in his corpus (2008:  61). In total, 21 out of 27 respondents (78%) tend to ceceo; five speakers differ in almost all the realizations (19%) and one respondent is not decisively ceceante (4%), in spite of performing many /​s/​as /​θ/​. These data are strikingly different from those published by Carbonero (1992) and much higher compared to other urban-​ speaking communities of western Andalusia. While at least a quarter of the speakers are ceceante in Malaga (Ávila 1994) and Huelva (Heras 1996), only 6% are in the capital city Seville (Carbonero 1982). The corpus only shows a few realizations of the phoneme /​θ/​ like [s]‌: the Sevillian seseo still has not reached the speech community of Jerez, since only 3% of the realizations of the phoneme /​θ/​in the corpus are [s]. These data contradict those obtained by Carbonero (1992), but it comes close to the data offered by Alvar (1973), in which seseo is not mentioned for Jerezano Spanish, and agree with García-​Amaya’s assertion that the speech community is not seseante, since only 6.5% of the realizations of /​θ/​are [s] (2008:  63). In total, only one of the 27 speakers is distinctively seseante, a number that would be equivalent to 4%. So, the corpus contradicts the conclusions outlined by Carbonero (1992), who presents 44% of seseo in his corpus. In addition, seseo realizations by five more speakers have been found in the corpus, but in the great majority of their pronunciations the /​θ/​ is produced as [θ], so they are not seseantes. Apart from the speech community of Huelva with seseo numbers very similar to those of our corpus (Heras 1996), the data provided with regard to other cities in western Andalusia is completely different:  87% of the population in Seville (Carbonero 1982), 78% of the speakers in Cadiz (Payán 1988:  35), and 31% of the speakers in Malaga tend to use seseo (Ávila 1994). As there are some sociolinguistic differences in our corpus as well, these will be addressed in the following section. 2.4.3  Actual data from an oral corpus: sociolinguistic variation The use of ceceo varies depending on the age of the Jerezano speakers. While middle-​aged (82%) and elderly (74%) respondents perform a [θ] instead of /​s/​in most cases, less than half of the realizations of the /​s/​done by the third generation (48%) is a [θ]. These data clearly indicate that the phenomenon is linked to the speaker’s age, the second generation being the one that uses the ceceo more frequently, and the third that does it less frequently. In Carbonero (1992), it is the second generation of speakers who uses ceceo less frequently (37%) and those who are younger (50%) and older (53%) are those who use it the most (1992: 25). In comparison to the extralinguistic variable of age, the diasexual variable does not seem to be important: the female speakers (73%) use ceceo a little more than the male ones (64%). So, the gender variable does not need to be

Sibilants in western Andalusia: Jerezano  81 investigated more profoundly regarding the use of ceceo in the speech community of Jerez. This conclusion corresponds with Carbonero’s data on Jerez (1992: 42), and with Ávila’s data on Malaga (1994: 362). In addition to age, the most interesting factor regarding the use of ceceo in Jerezano Spanish seems to be the variable of educational level. The more educated the Jerezano speakers are, the less they use ceceo: lower educated speakers pronounce a [θ] instead of /​s/​in 90% of the cases, middle-​class speakers reach 78%, and higher educated respondents reach 38%. However, the number of ceceantes with university degrees is still high if we compare these data with the 12% obtained by Carbonero. In spite of his affirmation that lower educated people use the ceceo more frequently (1992: 24), the figures of that study are still quite low compared to those shown in this contribution. The realization of /​s/​as ceceo is 46% among speakers of average education and 76% among speakers with less education (Harjus 2018a: 146). The few seseo realizations of the corpus are associated with younger speakers, especially the youngest generations (7%), and middle-​aged speakers (1%) (Harjus 2018a: 150). Carbonero considers the middle-​aged group (58%) to perform the seseo more frequently (1992:  25). Our data cannot confirm this point. The diasexual variation in the distribution of the phenomenon of seseo is almost zero: while the female speakers realize seseo up to 4%, the males produce it up to 1%. Similar to Carbonero (1992: 25), where women reach 49% and men 40%, in our corpus with respect to ceceo, the gender differences are minimal. The same applies to the data referring to Malaga (Ávila 1994: 362) and Cadiz (Payán 1988: 35). None of the lower educated Jerezano speakers use seseo (Harjus 2018a: 147). The number of seseo realizations increases with the respondent’s level of education. But this minimal increase in seseo pronunciations in our corpus –​1% among averagely educated speakers and 7% among the highest educated –​is not comparable with the numbers of the Jerez speech community concluded by Carbonero (1992), which rise from 11% to 76%. The numbers in our corpus resemble those offered by Melguizo for a rural speech community in Granada, where the use of seseo increases from the lowest sociocultural levels (0%) to the highest (10%) (2007: 8). In this way, we can conclude that seseo is a rare accomplishment among the youngest and most educated Jerezano speakers. However, the low figures of this specific form of the neutralization of the phonemes /​θ/​ and /​s/​ indicate that seseo is far from being a favorite realization among the highest social strata, nor is it part of the Jerezano speech community, as Garcia-​Amaya has already indicated (2008). With the data obtained here, it is clear that the speech community of Jerez seems to follow a different path from that of other urban areas of Andalusia: while the Granadian speech community usually leans towards the national standard of European Spanish (Moyá Corral, & García Wiedemann 1995), among speakers of western capitals, for example, in Cadiz, the use of

82  Jannis Harjus seseo increases (Payán 1988). Our contribution, combined with data from García-​Amaya (2008) and Harjus (2018a), shows that the Jerez speech community maintains a strong use of traditional ceceo, including among the most educated speakers. The majority of the highest educated speakers also make the distinction between /​s/​and /​θ/​, but nearly nobody uses seseo. Accordingly, the Jerezano speakers prefer the distinction over the seseo whenever they distance themselves from the ceceo. This conclusion brings us closer to a comparison of Jerez with the rural areas of eastern Andalusia, in which education is a decisive factor in the linguistic performance of speakers: the higher educated the respondents are, the higher the possibility for distinction and the less educated they are, the higher the possibilities of conservation of ceceo (Melguizo 2007: 8). Next, we will focus on the question of whether there is a norma sevillana in Jerezano Spanish from the speakers’ own points of view.

2.5  Theoretical basis: perceptual variety linguistics Previous metalinguistic works on Andalusian Spanish narrow the concept of speakers’ metalinguistic consciousness down to the emotional component, by analyzing primarily the speakers’ attitudes towards Andalusian Spanish. Both current Perceptive Dialectology and Perceptual Variety Linguistics have been used as frameworks to study Andalusian Spanish. Perceptual Variety Linguistics extends Perceptual Dialectology, traditionally focused on diatopic aspects, to the analysis of social and situative variation in spoken language. In doing so, Perceptual Variety Linguistics also defines the term perception more profoundly theoretically (Krefeld & Pustka 2010b: 10). This sub-​discipline of variety linguistics attempts to analyze the entire variational system of a certain language ​​from the speakers’ own point of view. It is necessary to emphasize that the speakers’ metalinguistic knowledge and their concrete speech influence each other in a circular process between linguistic production and linguistic perception (Postlep 2010): therefore, speakers change their linguistic production permanently due to constant perceptions (Pustka 2007: 18). This framework follows the concept of modern sociolinguistics, claiming that language is not a homogenous system. Here, like in Pustka (2007:  9), speakers’ concepts of linguistic variation are located in their mental concepts of metalinguistic knowledge. These linguistic representations of language variation are again subdivided into abstract knowledge about linguistic variation, which a speaker can retrieve without external stimulus, and concrete perceptions about linguistic variation, which can be expressed on the basis of direct perceptions (Krefeld & Pustka 2010b: 11). Although both metalinguistic representations are mutually interdependent and in constant exchange with each other, it is essential for the methodology to differentiate between these two aspects in Perceptual Variety Linguistics (see Harjus 2018a:  50, 2018c: 3).

Sibilants in western Andalusia: Jerezano  83

2.6  Methodology: corpus-​construction for perceptual variety linguistics Methodologically, this contribution follows the programmatic division of metalinguistic consciousness about variation into abstract, permanent representations and concrete perceptions. We combined various approaches of Perceptual Dialectology and Perceptual Variety Linguistics in a questionnaire developed for our purposes (Harjus 2018a: 265). First, we developed a mechanism to get access to the abstract representations of Jerezano speakers. Once again, we distinguish between the metalinguistic knowledge of the participants of the Andalusian varieties, which is particularly evident in mental maps and associated linguistic features, and their language attitudes towards proper Andalusian variety areas they have identified. This part of the survey primarily follows the methodological guidelines of Perceptual Dialectology, in which speakers were asked to draw dialect zones on an empty Andalusian map and to characterize the corresponding regions with linguistic features (Harjus 2018c: 7). In particular, this last aspect is interesting, since the content-​related dimension of metalinguistic knowledge can be analyzed here. Although references to linguistic variation and its phonetic features may be complex and difficult to interpret, strictly emic data were collected and rendered interpretable by a linguistic categorization similar to Anders (2010: 268). This scheme consists of five categories: phonetic, morpho-​syntactic, and lexical aspects as well as variety-​describing and -​valuing elements, which have been subdivided into 24 subgroups. In this chapter we merely analyze the data on the sibilants. The last two aspects of the first part of the survey are the “degree-​of-​difference,” namely the subjective distance of idiolect and exposed dialect zones, as well as the attitudes-​evoking “correct and pleasant” (Preston 1999: 34). On the one hand, participants were able to choose between an extreme proximity (1) and a large difference between idiolect and certain dialect zones (5). On the other hand, language attitudes were queried only on a binary scale, as positive or negative. The second part of the questionnaire is devoted to concrete perceptions, in which the participants are given a total of 13 different recordings –​nearly 30 seconds each –​by Andalusian speakers. The participants must assign the samples, representing various places and social milieus in western Andalusia, to a certain dialect or sociolect on the basis of salient features. In total, 30 speakers from different social provenances in Jerez participated in this fieldwork (Harjus 2018a: 287). The number of participants therefore only offers a tendency within the speech community and does not aim at providing a comprehensive or complete representation. While 13 speakers are university graduates and 11 have at least a secondary school diploma, six speakers are barely educated or even illiterate. The inclusion of less educated speakers in particular is an innovation in the analysis of Perceptual Variety Linguistics, since most of the work relies exclusively on data from young university students, for example Pustka (2007).

84  Jannis Harjus

2.7  Analysis of perceptual variety linguistics 2.7.1  Metalinguistic concepts of Jerezano speakers On average, the speakers divide the empty maps of Andalusia in the draw-​ a-​map task into seven different dialect zones and always include Seville and Jerez. The mental conception of the Andalusian variety area goes from a minimum of three zones to a maximum of 16 different regions. The biggest sociolinguistic difference in the Jerezanos’ mental maps is age:  while the youngest (6.0) and oldest generation (6.4) show similar numbers of dialect zones, the middle generation (8.7) draws significantly more varieties within Andalusian Spanish. Almost all speakers draw their map of the Andalusian dialect zone in a “selective” way (Anders 2010:  197):  in the mental maps of the participants, the Andalusian variety does not represent a continuum, but rather a conglomeration of individually, selectively perceived varieties. This statement highlights data found for other languages (Anders 2010). A  single speaker structures the blank card exhaustively, namely without blind spots in it. The mapping of Andalusian provincial capitals, such as Seville, Cordoba or Granada, emphasizes the extraordinary weight of the administrative centers in the minds of the Andalusians. In addition to the importance of cities, provincial boundaries are also represented as subjective isoglosses, which in turn suggests the strength of these administrative boundaries in the cognitive representation of the Andalusian space of Jerezano speakers (Harjus 2018a: 300). With regard to the areas of the southwest Andalusian area, the speakers name the speech communities of Jerez, Seville City, Cadiz City and some rural areas of the province of Cadiz. The urban centers Seville and Jerez that have been mentioned elsewhere are most frequently identified in the corpus (Harjus 2018c: 7). The linguistic features listed by Jerezano speakers to characterize Andalusian varieties highlight their metalinguistic knowledge. Partially they name even correct scientific terms, such as “ceceo” or “seseo.” In total, 121 different linguistic phenomena are listed, of which the phonetic elements are most numerous, and among these, the sibilants are most abundant, with ceceo as the most salient feature in the characterization of their own speech community. The variety of Seville city is characterized differently by the Jerezano speakers than their own speech community: nearly two-​thirds of all speakers emphasize the dephonologization in favor of /​s/​for the variety of Seville. Therefore, for the Jerezano speakers, the seseo is salient for the speech community of Seville. Generally, we rarely find valorative remarks for the variety of Seville: here, however, we not only find consistently positive attributes, such as muy fino “very refined” or self-​referential remarks as in el habla es más fina “their accent is more refined,” but also negative remarks as in chulita “hoity-​ toity” (Harjus 2018a: 335).

Sibilants in western Andalusia: Jerezano  85 In contrast to its own speech community, and the urban variety of Seville, the speakers often add negative comments, such as demasiado vulgar “too vulgar,” or with ironic utterances, such as refinado, jejeje “snobby, jejeje,” when referring to the rural dialects of the province of Cadiz, especially the varieties of Sierra de Cadiz. These negative ratings are consistently from university graduates. In addition to prosodic features, such as hablan cantando “they sing when they speak,” the use of [h]‌instead of /​s/​in the rural areas in Cadiz is emphasized, as the example pronuncian s como j (Puerto Jerrano) “they pronounce [s] as [h] (Puerto [h]errano)” stands out. Furthermore, the Jerezano speakers highlight the ceceo with respect to Sierra de Cadiz, although the same characteristic is salient for their own speech community, too. The rural areas of Cadiz and Seville as well as Jerez, although separated by speakers’ subjective isoglosses, become characterized by the same salient feature of ceceo (Harjus 2018c: 8). The speakers divide southwestern Andalusia into a seseo zone, which is made up of the provincial capitals Malaga, Seville and Cadiz, and into a clearly rural ceceo zone, in which, however, their own speech community of Jerez is included. Similar conclusions can be drawn from the Degree of Difference task. The participants subdivide the western part of Andalusia into two parts: while the urban speech communities of Malaga and Seville are characterized by an average distance of more than 2.9 and thus quite different, the speakers perceive the rural parts of Seville and Cadiz as well as Cadiz city similar to (1.8–​2.4) their own idiolect. It should be mentioned that middle-​ class speakers see the varieties of rural zones as more similar to their own while establishing greater distance between the speech of the urban centers (Cadiz, Seville, Malaga) and their idiolect. Young and university-​educated speakers, on the other hand, perceive similarity to the urban variety of Cadiz, but not that of Seville. The gender differences are quite similar:  while the women surveyed show a greater subjective proximity to rural ceceo areas, especially the Sierra de Cadiz, and greater distance to urban seseo zones (Cadiz, Seville), men are more likely to sketch proximity to the seseo regions they have identified (Harjus 2018a: 349). With regard to the language attitudes of the Jerezano speaker, we have to conclude that they have a higher sympathy towards the nearby western Andalusian areas. Together with its own speech community, it is the rural area of the region that received the greatest sympathy. Nevertheless, the urban speech community of Seville is also rated more sympathetically than unsympathetically, especially by men and younger speakers. Women, the middle generation, and, above all, the less educated find the urban variety of Seville less pleasant, although, with the exception of women, the sympathy values are ​​ still slightly higher than antipathy. With regard to the correctness of western Andalusian speech communities, it should be noted that the rural areas of the provinces of Cadiz and Seville have the lowest values. This result highlights a possible stigma of the rural varieties of Andalusian Spanish (Ropero 2001). Their own speech community

86  Jannis Harjus of Jerez, and also of Cadiz City, is rated slightly more correct than incorrect. Similarly, almost two-​thirds of all participants rate the urban variety of Seville as correct: on the one hand, this result demonstrates a relative prestige for the urban variety of Seville; on the other hand, the more than a third of negative reviews shows the limits of this prestige from the point of view of the Jerezano speakers, who seem to rate the adequacy of a local variety from an economic and political point of view rather than from a bare linguistic one (Harjus 2018a: 377). 2.7.2  Perceptions of Jerezano speakers The Jerezano speakers have also demonstrated their metalinguistic knowledge in the section on concrete perceptions. Although the respondents of the second generation, and those of average education, seem to be the respondents who know the regional background of the perceived persons on the voice tests somewhat more than the rest of the respondents, we must refute the hypothesis that older speakers locate other speakers better than young ones:  for the speech community of Jerez it should be emphasized that young people also show a high linguistic consciousness. It is noteworthy that there are no major differences in the accuracy of the perception regarding the level of education of the respondents of the corpus, although it seems certain that the least educated speakers use the appropriate linguistic terms far less than respondents obtaining university degrees (Harjus 2018c: 9). The linguistic features perceived in greater numbers have been those of the sibilants, first and foremost, of the dephonologization of /​s/​and /​θ/​, associated with quite divergent dialectal zones in western Andalusia. Dephonologization in favor of ceceo is always located in a self-​referential way, either within the Jerezano speech community itself or within rural speech communities in the province of Cadiz: therefore, ceceo is mostly associated with their own speech community, or in the closest communities of the rural area, highlighted in the degree-​of-​difference by its proximity. The separation of both areas in the concrete perception seems to be more difficult for the respondents, specifically in comparison to the section on abstract representations where they still separated two different zones. If they perceive ceceo in a stimulus they cannot differentiate whether the speaker comes from their own speech community or from the rural areas of Cadiz and Seville provinces. On the one hand, this result underscores Ropero’s findings on the linguistic attitudes of Sevillian speakers in the 1980s, illustrating that ceceo is more frequently associated with the speech communities of rural Andalusia (2001: 39). Notwithstanding, the people of Jerez associate the perceived ceceo not only with the modalities of the rural areas of Cadiz and Seville, but also with their own modality. On the other hand, this fact emphasizes an outstanding

Sibilants in western Andalusia: Jerezano  87 rurbanity with the speech community of Jerez (Harjus 2018a: 438), since, contrary to the abstract concepts, the respondents do not perceive any difference between the two variational areas. On the contrary, the dephonologization in favor of seseo is never associated with their own speech community, but always with the modalities of the provincial capitals of western Andalusia, with Seville, Cadiz and, although rarely, with the speech community of Malaga or the area of Cordoba in eastern Andalusia. Therefore, the respondents divide the western Andalusian space into two zones: the modalities of the large urban conglomerates, Seville, Cadiz and Malaga, are associated with the phonetic feature of the seseo, while the rural areas of the provinces of Cadiz and Seville along with the urban variety of Jerez are linked to the phenomenon of ceceo (Harjus 2018c: 12). Therefore, according to the metalinguistic consciousness of the Jerezano speakers, there is a multidimensionality of linguistic communication in the variational space of ​​western Andalusia, since they distinguish between an urban area, a rural one, and another rurban. Piredda analyzes the same phenomenon in a study on the perceptions of Sardinian speakers, in which she summarizes that her respondents cannot exactly locate the stimulus but can always distinguish perfectly between the Sardinian countryside and the Sardinian urban area (2014: 75–​79). Unlike Sardinian speakers who do not mention specific linguistic markers, the Jerezano speakers justify their locations in the urban area (Seville, Cadiz, Malaga or Cordoba) or the rural areas of Cadiz or Seville, including its own speech community, with the distinctive features of seseo or ceceo respectively. Above all, the absolute and total rejection of the seseo for their own speech community indicates, as in the part of the abstract representations, that this phonic feature is perceived as a shift away from Jerezano Spanish. Regarding the perception of diastratic factors, it must be stated that the dephonologization in favor of the seseo, as a salient phenomenon of a stimulus, leads to a social categorization of the speaker. A higher educational level is associated more frequently with seseo than ceceo. A perceived ceceo in a voice sample is often associated with less educated speakers. This result seems to confirm Ropero’s ideas of the Sevillian speech community, that seseo has more prestige than ceceo throughout western Andalusia (2001: 39). Within the local speech community of Jerez, traits of seseo in a voice sample seem to be salient for more educated speakers. Thus, it can be concluded that it is also very prestigious, since there is a correlation between the evaluation of certain social characters and the acceptance of a perceived linguistic variable (Krefeld 2010a: 171). However, it must be reiterated that, despite the better social signaling of seseo versus ceceo, this phonic feature is never associated with the local speech community itself. Seseo, however prestigious it may be, forms no part of the local linguistic variety for Jerezano speakers and, consequently, a perceived stimulus with the salient phonic phenomenon of seseo is never located within their own speech community.

88  Jannis Harjus In addition, seseo is not always associated with more educated speakers. While only three respondents believe that the seseo speaker of the stimulus (d) has a university degree, almost three times as many respondents (9) perceive a speaker with little instruction. And even among the social locations of the speakers of the stimuli (b), (e) and (k), which are mostly characterized as people of higher education, up to 35% of perceptions are of a speaker without any education. Not only does this fact show that the individual qualification of phonic features can vary considerably (Anders 2010: 354), but also that the use of seseo is not always prestigious. Hence, we must reject a supposed loss of prestige for the Jerez speech community: the social locations in the corpus clearly show that the ceceo is realized by educated speakers in Jerez, since the Jerezano speakers recorded on voice samples are perceived as highly educated speakers. However, it is noteworthy that among the lower educated respondents almost no one identifies a highly educated speaker: the less educated respondents do not associate the feature of ceceo with a high level of education. On the contrary, the rest of the respondents do not hesitate to locate the ceceantes among the higher educational levels. Although the possibility that certain high social categorizations of the ceceo speakers (c; f; m) and the lower categorization of the seseante (d) could be motivated by protection of the local variety cannot be ruled out, this result emphasizes the persistence of the old stereotype of the uncultivated ceceante among less educated speakers. Among the more educated Jerezano speakers, this stereotype seems to be overcome and the identifications of speakers with high educational levels such as university degrees show the overt prestige which ceceo enjoys even among the most educated speakers of the local speech community. As a result, “the ceceante realization is not at all a phonological feature linked to the lower social strata within the Jerezano speech community” (Harjus 2018c: 11).

2.8  Conclusions: the lack of a Sevillian norm in Jerezano Spanish This chapter has shown that we do not find many realizations of seseo in Jerezano Spanish, not even among young, well-​educated Jerezano speakers. Therefore, we can refute the older data provided by Carbonero (1992) and a seseo-​based regional norm that deviates from the national standard of European Spanish for the Jerezano speech community: the data either demonstrate a tendency towards the standard variety distinction between /​s/​and /​θ/​ or a retention of traditional dialectological ceceo in Jerez, even in formal communicative situations. Consequently, both ceceo and distinction have overt prestige in Jerez speech community. The results of perceptual variety linguistics’ data lean towards the same direction: the use of actual scientific terms underlines that metalinguistic communication constitutes a part of everyday life in the Jerezano speech community. In particular, the use of linguistic terms, for example

Sibilants in western Andalusia: Jerezano  89 for the use of sibilant terms such as seseo, ceceo or even heheo, refutes Spiekermann’s thesis on the Jerezano speech community, maintaining that linguistic laymen were not able to put their metalinguistic thoughts into words (2010: 222). The most important result of the analysis of speaker’s concepts is the division of a ceceante and a seseante zone in western Andalusia. The speakers’ abstract, permanent representations clearly distinguish between Seville and Cadiz City as seseo-​communities and their own speech community plus rural Cadiz, which are characterized by ceceo. Results do not support former studies (Alvar et  al. 1973). Akin to the Huelva speech community (Regan 2017), the analysis of speakers’ concepts clearly demonstrate that seseo does not play a role for the Jerezano speakers in their own speech community. This conclusion clearly supports the descriptive sociolinguistic part of this chapter, and the latest sociophonetic studies (García-​Amaya 2008, Harjus 2018a). This tends to prove that a regional standard based on the urban variety of Seville has not arrived in the Jerez community, and is not applicable to all speech communities of western Andalusia (Harjus 2018a). The fairly positive attitudes towards the use of ceceo demonstrate the overt prestige of the phenomenon in the local speech community. The results of the perception test also support this conclusion, indicating the increasing overt prestige of ceceo and distinction. The results also demonstrate that the salience of phonetic features does not necessarily lead them to be withdrawn (Lenz 2010: 106). The high salience of ceceo and the increased use even in formal communicative situations show that the sound characteristic is an identification marker of the speech community (Lenz 2010: 108). With regard to the speakers’ language attitudes, it is noticeable that these become more positive as the population of the Andalusian urban centers increase. The fewer speakers in a community, the worse their variety is rated by the Jerezano speakers. Linguistic phenomena, such as seseo or ceceo, seem to play a rather inferior role as the examples show:  Jerez and Cadiz with different phonetic characteristics but the same correctness values; Cadiz City and Seville City with the same salient characteristics but different evaluations. The metalinguistic consciousness of Jerezano speakers manifests itself in a center-​periphery model in which linguistic correctness increases with increasing cultural, political and demographic power of the urban centers. Therefore, while the speakers often identify the Sevillian seseo as correct, they do not place it in their own speech community. Narbona has already pointed out that Andalusians recognize other Andalusian varieties very quickly, and classify them as artificial, or foreign (2011: 327). Thus, a perceived seseo in the Jerezano speech community is first and foremost a departure from the local variety, and secondarily a prestigious form of pronunciation, which still does not function as a western Andalusian regional standard for sibilants.

90  Jannis Harjus

Glossary Ceceo:  ceceo refers to the lack of distinction between phonemes /​s/​and /​θ/​ in favor of /​θ/​; i.e. sí “yes” is pronounced as [θí]. Concepts: Abstract concepts or representations are part of a speaker’s metalinguistic ideas and, as part of the langue, they are always present, also without stimuli. Heheo or jejeo:  initial /​s/​is pronounced as [h]‌or dropped at the beginning of the syllable; i.e. pasa “raisin” is pronounced [páha]. See Map 2.3. Jerezano: Spanish variety spoken in the speech community of Jerez de la Frontera in Southwestern Andalusia, province of Cadiz. Jerez has more than 215,000 habitants. Norma sevillana:  a possible influence of the linguistic variety of urban Seville on the western zone of Andalusia. It is based on the salient features from Sevillian Spanish, basically seseo. Perceptual variety linguistics (PVL): a theoretical and methodological approach to analyze speakers’ metalinguistic ideas. In contrast with the Perpetual Dialectology, it focuses on all types of linguistic variation, making a distinction between abstract concepts and perceptions. Perceptions: metalinguistic representations which are based on concrete perceptions and belong to the parole (in contrast to abstract concepts). For the analysis of concrete perceptions, we need stimuli. Salient feature: a distinguishing or iconic feature of a sociolinguistic variety. Seseo:  refers to the lack of distinction between phonemes /​s/​and /​θ/​in favor of /​s/​; i.e. cine “cinema” is pronounced as [síne].

Map 2.3 Heheo distribution in Andalusia. Source: Adapted from Prieto (2014: 132).

Sibilants in western Andalusia: Jerezano  91

References Alvar, Manuel (1990). Norma lingüística sevillana y español de América. Madrid: ICI. Alvar, Manuel, Llorente, Antonio, & Gregorio, Salvador (1973). Atlas Lingüístico y Etnográfico de Andalucía. Vol. 6: Fonética, morfología, sintaxis. Granada: CSIC. Anders, Christina (2010). Wahrnehmungsdialektologie  –​Das Obersächsische im Alltagsverständnis von Laien. New York: De Gruyter. Auer, Peter (2005). Europe’s sociolinguistic unity, or:  A typology of European dialect/​ standard constellations. In N. Delbecque et  al. (Eds.), Perspectives on Variation: Sociolinguistic, Historical, Comparative (pp. 7–​42). New York: Mouton de Gruyter. Ávila Muñoz, Antonio (1994). La variación reticular e individual de s/​z en el vernáculo urbano malagueño. Datos del barrio de Capuchinos. Analecta Malacitana, 7(2), 343–​367. Bartsch, Renate (1987). Sprachnormen:  Theorie und Praxis. Eine normtheoretische Untersuchung von Sprache und Sprachverhalten. Tübingen: Niemeyer. Boyd-​Bowman, Peter (1956). The regional origins of the earliest Spanish colonist of America. Publications of the Modern Language Association of America, 71, 1152–​1172. Carbonero Cano, Pedro (2003). Estudios de sociolingüística andaluza. Sevilla: Universidad de Sevilla. Carbonero Cano, Pedro (1982). El habla de Sevilla. Sevilla: Ayuntamiento. Carbonero Cano, Pedro et al. (Eds.). (1992). El habla de Jerez. Estudio Sociolingüístico. Jerez de la Frontera: BUC. Carrasco Cantos, Pilar (2007). Rasgos fonéticos de la norma sevillana en la época medieval. In J. Moyá Corral (Ed.), Las hablas andaluzas y la enseñanza de la lengua. Actas de las XII Jornadas sobre la Enseñanza de la Lengua Española (pp. 47–​56). Granada: Universidad de Granada. Carrascosa Pulido, José (1997). El habla en Canal Sur Satélite (radio y televisión). In A. Narbona Jiménez et al. (Eds.), El habla andaluza. Actas del Congreso del Habla Andaluza (pp. 421–​432). Sevilla: Seminario Permanente del Habla Andaluza. Coseriu, Eugenio (1980). Historische Sprache und Dialekt. In J. Göschl (Ed.), Dialekt und Dialektologie. Ergebnisse des Internationalen Symposiums ‘Zur Theorie des Dialekts (pp. 106–​122). Marburg/​Lahn. Wiesbaden: Steiner. Diaz Campos, Manuel, & Navarro-​Galisteo, Inmaculada (2009). Perceptual categorization of dialect variation in Spanish. In J. Colletine (Ed.), Selected Proceedings of the 11th Hispanic Linguistic Symposium (pp. 179–​195). Sommerville, MA: Cascadilla. Dittmar, Norbert (1997). Grundlagen der Soziolinguistik  –​ein Arbeitsbuch mit Aufgaben. Tübingen: Niemeyer. Frago Gracia, Juan (1993). Historia de las hablas andaluzas. Madrid: Arco Libros. García-​Amaya, Lorenzo (2008). Variable norms in the production of the /​θ/​in Jerez de la Frontera, Spain. In J. Siegel (Ed.), IUWPL7:  Gender in language:  Classic questions, new contexts (pp. 49–​71). Bloomington, IN: IULC Publications. Gloy, Klaus (2005) Norm. In U. Ammon et al. (Eds.), Sociolinguistics: An International Handbook of the Science of Language and Society (pp. 392–​399). New York: De Gruyter. González Bueno, Manuela (1993). Variaciones en el tratamiento de las sibilantes. Inconsistencia en el seseo sevillano: un enfoque sociolingüístico. Hispania, 76(2), 392–​398.

92  Jannis Harjus Harjus, Jannis (2018a). Sociofonética andaluza y lingüística perceptiva de la variación: el español hablado en Jerez de la Frontera. Madrid: Iberoamericana Vervuert. Harjus, Jannis (2018b). Versuchte Normierung einer komplexen Varietät? Der praktische Umgang mit innerer Mehrsprachigkeit im sekundären Bildungssektor Andalusiens. In M. Dannerer et al. (Eds.), Formen der Mehrsprachigkeit –​Sprachen und Varietäten in sekundären und tertiären Bildungskontexten (pp. 105–​ 122). Tübingen: Stauffenburg Verlag. Harjus, Jannis (2018c). Perceptual Variety Linguistics:  Jerezano speakers’ concepts and perceptions of phonetic variation in western Andalusian Spanish. Loquens, 4(2),  1–​15. Heras, Jerónimo de las et al. (1996). Perfil sociolingüístico del habla culta de la zona periurbana de Huelva. Aestuaria, 4(4), 109–​124. Jiménez Fernández, Rafael (1999). El andaluz. Madrid: Arco Libros. Krefeld, Thomas (2010a). Italiano, ma popolare? Einige nicht standardsprachliche Merkmale im Spiegel des Varietätenbewusstsein. In T. Krefeld et  al. (Eds.), Perzeptive Varietätenlinguistik (pp. 151–​180). Frankfurt: Peter Lang. Krefeld, Thomas, & Pustka, Elissa (2010b). Für eine perzeptive Varietätenlinguistik. In T. Krefeld et  al. (Eds.), Perzeptive Varietätenlinguistik (pp. 9–​28). Frankfurt: Peter Lang. Lasarte Cervantes, María de la Cruz (2010). Datos para la fundamentación empírica de la escisión fonemática prestigiosa de /​θ/​en Andalucía. Nueva Revista de Filología Hispánica, 58(2), 483–​516. Lenz, Alexandra (2010). Zum Salienzbegriff und zum Nachweis salienter Merkmale. In C. Anders et al. (Eds.), “Perceptual Dialectology” Neue Wege der Dialektologie (pp. 89–​110). New York: De Gruyter. Melguizo, Elisabeth (2007). La variación de /​θ/​: estudio comparativo de dos muestras de población granadinas. Estudios de Linguística Universidad de Alicante, 21, 245–​259. Menéndez Pidal, Ramón (1962). Sevilla frente a Madrid: algunas precisiones sobre el español de América. In Miscelánea Homenaje a André Martinet (pp. 99–​165). La Laguna: Universidad de La Laguna. Mondéjar, José (1985). Disquisiciones histórico-​críticas y metodológicas sobre la interpretación de los datos en el estudio del seseo. Revue de Linguistique Romane, 49, 195–​196, 271–​286. Mondéjar, José (1995). La norma lingüística del español y la pretendida norma de las hablas andaluzas. Analecta Malacitana, 8(1), 29–​40. Moyá Corral, Juan, & García Wiedemann, Emilio (1995). El habla de Granada y sus barrios. Granada: Universidad de Granada. Narbona Jiménez, Antonio et  al. (2011). El español hablado en Andalucía. Sevilla: Universidad de Sevilla. Navarro Tomás, Tomás et  al. (1933). La frontera del andaluz. Revista de Filología Española, 20, 225–​227. Payán Sotomayor, Pedro (1988). La pronunciación del español en Cádiz. Cádiz: Universidad de Cádiz. Piredda, Noemi (2014). Perzeption des Italienischen in Sardinien: Stadt und Land im Vergleich. In T. Krefeld et  al. (Eds.), Perzeptive Linguistik:  Phonetik, Semantik, Varietäten (pp. 65–​85). Stuttgart: Franz Steiner Verlag. Polzin-​Haumann, Claudia (2013). Standardsprache, Norm und Normierung. In J. Born et  al. (Eds.), Handbuch Spanisch. Sprache, Literatur, Kultur, Geschichte in

Sibilants in western Andalusia: Jerezano  93 Spanien und Hispanoamerika: Für Studium, Lehre und Praxis (pp. 44–​56). Berlin: Erich Schmidt Verlag. Postlep, Sebastian (2010). Zwischen Huesca und Lérida. Perzeptive Profilierung eines diatopischen Kontinuums. Frankfurt: Peter Lang. Preston, Dennis (1999). Introduction. In D. Preston (Ed.), Handbook of Perceptual Dialectology (pp. 23–​39). Amsterdam: John Benjamins. Prieto, Juan P. (2014). Distribución geográfica del jejeo en español y propuesta de reformulación y extensión del término. Revista española de lingüística, 38(2), 129–​144. Pustka, Elissa (2007). Phonologie et variétés en contact. Aveyronnais et Guadeloupéens à Paris. Tübingen: Narr. Regan, Brendan Patrick (2017). A study of ceceo variation in Western Andalusia (Huelva). Studies in Hispanic and Lusophone Linguistics, 10(1), 119–​160. Ropero Nuñez, Miguel (2001). Sociolingüística andaluza: problemas y perspectivas. In P. Carbonero Cano et al. (Eds.), Identidad lingüística y comportamientos diversos. Sociolingüística andaluza (pp. 21–​48). Sevilla: Universidad de Sevilla. Samper Padilla, José (2011). Sociophonological variation and change in Spain. In M. Díaz-​Campos (Ed.), The Handbook of Hispanic Sociolinguistics (pp. 98–​119). Oxford: Wiley-​ Blackwell. Spiekermann, Helmut (2010). Visualisierungen von Dialekten:  Ein Beitrag zum Nutzen der Laiendialektologie. In C. Anders et al. (Eds.), Perceptual Dialectology. Neue Wege der Dialektologie (pp. 221–​244). New York: De Gruyter. Uruburu Bidaurrázaga, Augustín (1990). Estudios sobre la lengua española en Córdoba. Córdoba: Diputación Provincial. Villena Ponsada, Juan (2001). La continuidad del cambio lingüístico. Granada: Universidad de Granada. Villena Ponsada, Juan (2006). Andaluz oriental y andaluz occidental: estandarización y planificación ¿en una o dos comunidades de habla? In A. Cestero Mancera et al. (Eds.), Estudios sociolingüísticos del español de España y América (pp. 233–​254). Madrid: Arco. Villena Ponsada, Juan (2008). Sociolinguistic patterns of Andalusian Spanish. Sociology of Language, 193, 139–​160. Wagner, Max Leopold (1927). El supuesto andalucismo de América y la teoría climatológica. Revista de Filología Española, 14, 20–​32.

94  Jannis Harjus

Topics for discussion (1) Briefly discuss the difference between “prescriptive” and “descriptive” linguistic norms. (2) Explain the term norma sevillana from a historical linguistic approach. (3) Describe the term norma sevillana for current Andalusian Spanish. (4) Briefly describe the differences between Perceptual Dialectology and Perceptual Variety Linguistics. (5) What are the differences between metalinguistic representations and perceptions? (6) Are seseo and ceceo salient features of western Andalusian Spanish? Explain. (7) Is there a Sevillian norm in western Andalusia? Explain. (8) What are the advantages and disadvantages of examining speakers’ metalinguistic ideas? (9) What is ceceo? Explain and provide examples.

3  Intervocalic /​s/​-​voicing in Spanish in contact with Catalan Justin Davidson

3.1  Introduction Though the highly variable production of the Spanish voiceless alveolar fricative /​s/​constitutes the most widely researched consonantal phenomenon in Hispanic Linguistics, the majority of studies on Spanish /​s/​focus on its aspiration or deletion (Chappell 2016), a trend appropriately echoed in this very volume on Spanish sibilants.1 Considerably fewer studies investigate the variable voiced or voiceless quality of /​s/​, and of those that do, the majority are limited to contexts of a following consonant, constituting treatments of variability in regressive voicing assimilation (see for example Campos-​Astorkiza 2014, as well as Boomershine’s chapter in this volume). The voicing of Spanish /​s/​outside of the context of a following (voiced) consonant, or more precisely in syllable-​initial or intervocalic contexts, is accordingly perhaps the least studied phenomenon pertaining to /​s/​production, and constitutes the focus of the present chapter (though refer to File-​Muriel’s and Lipski’s chapters in this volume for additional treatments). Ascribed to first-​language (L1) transfer from Catalan and considered a minority variant characteristic of the “Catalanized” Spanish of Catalan-​L1 bilinguals (Arnal 2011; Sinner 2002), the intervocalic voicing of /​s/​to [z]‌(e.g. los años [lo.sá.ɲos] > [lo.zá.ɲos] “the years”) has most often been treated in impressionistic and/​or dialectological work (e.g. Casanovas 1995; Hualde 2014; Payrató 1985; Serrano 1996; Vann 2001; Wesch 1997). In order to assess its modern vitality and use in Catalonian Spanish, especially given recent claims that phonetic features of Catalonian Spanish, such as intervocalic /​s/​-​voicing, are no longer present in the Spanish of the newest generation of Barcelonan bilinguals (Arnal 2011), we adopt a framework of variationist sociolinguistics (Labov 2001; Tagliamonte 2012) and offer empirical evidence in support of intervocalic /​s/​-​voicing as a regional marker of the Spanish of Catalonia. This chapter is organized as follows: Section 3.2 consists of a brief overview of linguistic history and language trends in Catalonia in order to better contextualize the sociolinguistic phenomenon in question. Section 3.3 offers an account of alveolar fricative production in Spanish and Catalan,

96  Justin Davidson detailing select prior findings that motivate the current study’s research questions. Section  3.4 describes the experimental methodology and test instruments. Section 3.5 discusses data analysis techniques and results from data collection. Section 3.6 offers a discussion of the results of the present study. Lastly, we conclude in Section 3.7 by offering directions for future study.

3.2  Spanish and Catalan in Catalonia Though contact between Spanish and Catalan is centuries old, practical and widespread societal bilingualism is a relatively recent linguistic outcome in Catalonia’s history. Until the formation of the Spanish Crown in 1469, a product of the union between the Arago-​Catalan and Castile Kingdoms with the marriage of King Ferdinand II (Aragon) to Queen Isabella (Castile), Catalan was effectively the national and only language of Catalonia (Turell 2000; Vallverdú 1984). This alliance between the kingdoms marked the beginnings of Catalan-​Spanish bilingualism in Catalonia, initially restricted to the ruling classes and noblemen, whereas the rest (and majority) of Catalonian society remained monolingual in Catalan. The status of Catalan as the national language of Catalonia was not definitively challenged until the Spanish War of Succession in the early eighteenth century, once Castilian King Phillip V signed the Ordinance of New Plant, a legislation that suppressed all institutions and privileges (e.g. media, administration, courts, etc.) of territories formerly part of the Arago-​Catalan Kingdom (Àngel Pradilla 2001; Blas-​Arroyo 2007; Vila-​Pujol 2007). Bilingualism via the acquisition of Spanish by a Catalan-​speaking populace began to extend from the upper classes down through the middle and lower classes, significantly propelled by the Moyano Law of 1857 that imposed compulsory public education taught in Spanish (Mas 1993; Vallverdú 1984; Vila-​Pujol 2007). The ultimate rise of Spanish hegemony over Catalan came with Spain’s fascist dictatorship under General Francisco Franco from 1939 until his death in 1975, throughout which he actively passed legislation to eliminate –​or completely Castilianize –​all Catalan (and other non-​Spanish) public institutions, as well as outlaw Catalan in the public sphere (Àngel Pradilla 2001; Arnal 2011; Newman, Trenchs-​Parera, & Shukhan 2008; Turell 2000; Vallverdú 1984; Vila-​ Pujol 2007). Three years after Franco’s death, Spain’s democratic period was inaugurated with the ratification of a new constitution that divided the country into quasi-​federal Autonomous Communities (Newman et  al. 2008). The Spanish Constitution declares Spanish as the only official national language of Spain, and that all Spaniards have an obligation to know Spanish. Three additional languages of Spain –​namely Catalan, Basque and Galician –​are permitted to be co-​official alongside Spanish in the particular Autonomous Communities that choose to grant them co-​official status (Vila-​ Pujol 2007:  68). Accordingly, Turell remarks that the Spanish Constitution reflects a conception of Spain in which Catalan, Basque and Galician are in fact not a question of the State, consigning them as minority languages

Intervocalic /s/-voicing: Spanish, Catalan  97 to asymmetric bilingualism in their respective Autonomous Communities (Turell 2000; Vila i Moreno, Vial, & Galindo 2005; Newman et al. 2008). With respect to the modern usage of Catalan relative to Spanish in Catalonia, though the aforementioned asymmetric bilingualism unsurprisingly favors Spanish over Catalan, reported competences in Catalan are notably high. Approximately 94% of L1-​Spanish speakers in Catalonia (ages 15+) report understanding Catalan and nearly 75% claim speaking competence, whereas the parallel figures for both Spanish competences by L1-​Catalan speakers are approximately 100%. The disparity is more evident for immigrant populations in Catalonia for whom neither Spanish nor Catalan is a native language: whereas roughly 77% report understanding Catalan and 51% report speaking competence, the parallel figures for Spanish competences are respectively 99% and 98% (Institut d’Estadística 2014). With regard to language use and the size of the Catalan-​L1 population, the Barcelona Metropolitan Area (henceforth BMA, visualized in Map 3.1) stands as somewhat of an outlier in comparison to the Autonomous Community of Catalonia as a whole. Renowned for having the lowest presence of Catalan and fewest number of L1-​Catalan speakers in all of Catalonia (Lleó, Cortés, & Benet 2008), approximately 23% of the current (2013) BMA population are L1-​Catalan speakers, whereas approximately 64% are L1-​Spanish speakers2 (Institut d’Estadística 2014). Moreover, although there are currently over 600,000 (8% of total population of Catalonia [of 7,553,650]) non-​native speakers of Catalan who use Catalan as their habitual language in Catalonia (i.e. native speakers of either Spanish or another language who learned Catalan as a second language),

Map 3.1 Regional territories of Catalonia. Source: Adapted from Generalitat (2009, 2011).

98  Justin Davidson Table 3.1 2013 population (%)  ages 15+ with Catalan and Spanish as native or habitual language

Native Language Habitual Language

Catalan Spanish Catalan Spanish

Barcelona Metropolitan Area

Catalonia

Catalonia Excluding BMA

23.3 64.3 27.8 60

31 55.1 36.3 50.7

44.5 39.16 51.21 34.52

Source: Institut d’Estadística (2014).

the BMA continues to rank the lowest in all of Catalonia for the number of speakers habitually using Catalan, at approximately 28% of its population, as compared to 60% for habitual Spanish use (Generalitat de Catalunya 2014; Institut d’Estadística 2014). A fuller comparison of native and habitual language patterns between the BMA and Catalonia appears in Table 3.1, in addition to the calculations for Catalonia excluding the BMA, which crucially indicate that were it not for the BMA, Catalan in fact would constitute the majority language in this Autonomous Community. The predominance of Spanish in the BMA (and the city of Barcelona in particular)3 comes as a direct result of immigration trends consistent since the mid-​twentieth century, led by L1-​Spanish speakers both from other territories of Spain and more recently in the twenty-​first century from Latin America. Approximately 21% of the BMA population is comprised by L1-​ Spanish immigrants from Spain, which is roughly ten times as many as found in the region with the next to highest population of Spanish immigrants, Tarragona. Still, despite this territorial pocket of Spanish predominance, the aforementioned Catalan competence census data show that Catalonian society is characterized by a widespread and high degree of Catalan-​Spanish bilingualism, which has come about only gradually since the end of the Franco dictatorship with sweeping educational and governmental reforms to revitalize Catalan. Notably, it has been claimed (see for example Vila-​Pujol 2007) that Catalan holds more political power and social prestige than Spanish in Catalonia: The social stratification of the two languages in Catalonia is unique because “even if Spanish is, in principle, the stronger language, from certain perspectives it is also the less prestigious” [Siguan 1988: 454]. This could be so because Catalan “is the language of a large part of the economic and intellectual middle class, as well as the language of local political power,” [Siguan 1988:  454] while Spanish is the language of the immigrants and the lower social strata. (Sinner 2002: 161)

Intervocalic /s/-voicing: Spanish, Catalan  99 Thus, though Catalan is a stateless language insomuch as the lack of official status throughout the country as a whole, it nonetheless has a very high degree of use and linguistic capital (or afforded prestige and power) throughout Catalonia, giving it arguably more administrative power than Spanish.

3.3  Prior treatments of (intervocalic) fricative production 3.3.1  Prescriptive characterizations of alveolar fricative production in Spanish and Catalan Northern-​Central Peninsular Spanish features an apical-​alveolar voiceless /​s/​, articulated with a gesture of the tongue tip toward the alveolar ridge (Dalbor 1997; Hualde 2014; Martínez Celdrán & Fernández Planas 2007; Quilis 1981). In monolingual Spanish varieties that do not exhibit aspiration or deletion of /​s/​in pre-​consonantal positions, such as Mexican Spanish and North-​Central Peninsular Spanish, two allophones of /​s/​, namely voiceless [s]‌and voiced [z], are found in complimentary distribution via regressive assimilation of voicing to the following consonantal segment. Before voiced (semi)consonants, /​s/​is realized as [z] (e.g. rasgo [ráz.ɣo] “feature”; mis hierbas [miz.ʝéɾ.βas] “my herbs”), whereas in all other contexts, /​s/​is produced as [s] (e.g. rasco [ráz.ko] “I scratch”; casa [ká.sa] “house”; patos [pá.t̪os] “ducks”) (Azevedo 1992; Dalbor 1997; Hualde 2014, 2010; Navarro Tomás 1918; Morgan 2010; Pieras 1999; Schwegler, Kempff, & Ameal-​ Guerra 2010; Teschner 1996). Accordingly, monolingual Spanish productions of [z] outside of the context of a following voiced (semi)consonant (e.g. the intervocalic context in particular) are prescriptively disallowed: La s sonora aparece únicamente, en nuestra lengua, en posición final de sílaba, precediendo inmediatamente a otra consonante sonora; en cualquier otra posición su presencia es anormal y esporádica. [The voiced /​s/​in our language appears solely in syllable-​final position immediately preceding another voiced consonant; in any other position its presence is abnormal and sporadic.] (Navarro Tomás 1918: 83) In contrast to Spanish, Central Catalan features two apical-​alveolar fricative phonemes, voiceless /​s/​and voiced /​z/​. This phonemic voicing contrast is active word-​initially and word-​medially, producing minimal pairs such as zel “zeal” [zέɫ] (Spanish celo [θé.lo]) /​ cel “sky” [sέɫ] (Spanish cielo [θjé.lo]) and pesar “to weigh” [pə.zá] (Spanish pesar [pe.sáɾ]) /​ passar “to pass” [pə.sá] (Spanish pasar [pa.sáɾ]). Critically, this phonemic voicing contrast is neutralized word-​finally, resulting in [s]‌or [z] depending on the voicing feature of the following segment (that is, the voicing neutralization of word-​final Catalan alveolar fricatives [and in fact all Catalan sibilants] resolves by means

100  Justin Davidson of anticipatory assimilation). When followed by a voiced segment, such as a vowel, the word-​final fricative is systematically voiced (e.g. gos [s] “dog”; gos estrany [z] “strange dog”) (Hualde 1992, 2014; Julià 2008; Pieras 1999; Prieto 2004; Recasens 2014; Wheeler 2005). In summary, voiced intervocalic fricatives are systematically present in Catalan, resultant from word-​initial /​z/​, word-​medial /​z/​, and as a product of voicing assimilation of word-​final prevocalic /​s/​and /​z/​[or archiphoneme /​S/​]). This accordingly sets up an interesting pair of opportunities for bidirectional contact influence (i.e. L1-​transfer) contingent on syllable position. With respect to syllable-​initial contexts, productions of Spanish pesar “to weigh” or casa “house” as [pe.záɾ] and [ká.za] on the part of an L1-​Catalan speaker could evidence the transfer of a Catalan phoneme (/​z/​) into Spanish, whereas productions of Catalan pesar “to weigh” or casa “house” as [pə.sá] and [ká. sə] on the part of an L1-​Spansh speaker could evidence the substitution of Spanish /​s/​for Catalan /​z/​, potentially eliminating the phonemic voicing contrast in Catalan. With respect to word-​final contexts, the production of Spanish las albas “the dawns” as [la.zál.βas] by an L1-​Catalan speaker, or the production of Catalan les albes “the dawns” as [lə.sál.βəs] by an L1-​Spanish speaker, would constitute a case of largely phonetic, rather than phonemic, transfer (i.e. the respective transfer of a Catalan or Spanish phonotactic voicing rule, which wouldn’t create nor eliminate any phonological contrasts). 3.3.2  Empirical characterizations of alveolar fricative production The aforementioned prescriptive characterizations of Spanish and Catalan alveolar fricative production as absolute (i.e. Spanish [s]‌and [z] allophony as perfectly categorical, and Catalan /​s/​, /​z/​, and /​S/​production as non-​variable) are plainly refuted by recent empirical phonetic research, which demonstrates wide degrees of variability in voicing and non-​voicing in both languages. Focusing on Spanish (for Catalan, see Ballart 2013; Benet 2012; Bonet 1998), instances of [s], partially voiced [s̬], and [z] were all found to occur, in varying frequency, in each of the contexts of a following vowel, a following voiceless consonant, and a following voiced consonant in Northern-​Central Peninsular Spanish (Campos-​Astorkiza 2014), Mexico City Spanish (Schmidt & Willis 2011), Highland Colombian Spanish (García 2013), Quito Spanish (Strycharczuk et al. 2014), and in the speech of a group of (presumably monolingual or Spanish-​dominant) speakers from Madrid, Valencia, and Galicia (Clegg & Strong 1992). Although Clegg declares “there is no question as to the existence of the phenomena of sporadic voicing of /​s/​in all positions and extensive voicing in some speech communities” (1992: 32), the aforementioned studies attesting to a lack of voicing before voiced consonants in several Spanish varieties merits a broader generalization, namely that /​s/​-​voicing assimilation is highly variable, as well as gradient. The treatment of /​s/​-​voicing as a gradient phenomenon is supported by cross-​linguistic research evidencing the gradiently voiced or voiceless quality

Intervocalic /s/-voicing: Spanish, Catalan  101 of alveolar fricatives, prepalatal fricatives, the fricative portion of prepalatal affricates, as well as other /​s/​lenition phenomena (e.g. aspiration, elision), in Portuguese (Jesus 2003), English (Smith 1997; Stevens 1992), Catalan (Carrera-​Sabaté 2009), Italian and Dutch (Rivas 2006), and several varieties of Spanish including Caleño Spanish (File-​Muriel & Brown 2011), Argentinean Spanish (Gradoville 2011; Rohena-​Madrazo 2015), Madrid Spanish (Torreira & Ernestus 2012), Castilian Spanish (Romero 1999), and New  York City Spanish (Erker 2012). In this line of research, fricative voicing assimilation is modeled within gestural phonology (Browman & Goldstein 1989, 1991) as a product of the relative timing and coordination of opposing (or conflicting) laryngeal gestures necessary to restrict or permit voicing during the fricative segment in conjunction with adjacent segments. In particular, the conflicting laryngeal gestures for [s]‌(vocal fold abduction) and a following voiced consonant (vocal fold adduction) can give rise to gestural blending (Browman & Goldstein 1989, 1991), resulting in a single vocal fold adduction gesture that extends fully into the /​s/​segment, yielding [z] (Campos-​Astorkiza 2014; Hualde 2014; Romero 1999). Fricative voicing before a voiced (semi)consonant or vowel can therefore be understood as a lenition phenomenon, in that the conflicting glottal gestures for [s] and the following voiced segment are gradiently reduced to a single glottal gesture that maintains voicing throughout. 3.3.3  Intervocalic /​s/​-​voicing in Barcelonan Spanish Prior treatments of phonetic features of Barcelonan Spanish often discuss intervocalic [z]‌as a product of Catalan-​Spanish transfer that characterizes a Catalanized variety of Spanish of Catalan-​dominant speakers. For example, impressionistic interviews by Sinner (2002) suggest that [z] is a possible linguistic marker (Labov 2001) of Catalonian Spanish. Having interviewed 12 speakers of Barcelonan Spanish and monolingual (Madrid) Spanish aged 27–​41 regarding their awareness of linguistic features of Catalanized Spanish, intervocalic [z] was only named by select Barcelona speakers, who additionally discussed feeling an “obligation to correct or adjust their [Spanish] pronunciation when talking in public” (Sinner 2002:  163). More recently, a matched guise experiment was conducted by Davidson (2019) in order to empirically confirm Barcelonans’ social evaluation of intervocalic [z] vs. [s]. Findings revealed that Spanish intervocalic [z] is positively associated by Barcelonan bilinguals with higher solidarity and a Catalan-​Spanish bilingual identity, though these associations were exclusively covert, as listeners were unable to explicitly identify the phonetic difference between guises with [z] vs. [s]. Wesch (1997) examined possible effects of age and social class on intervocalic [z]‌production in Barcelona Spanish. The variant was impressionistically characterized as frequent, while unlinked to age and social class. Parallel results describing intervocalic [z] production as frequent and unlinked to social

102  Justin Davidson factors (gender, age, social class) have been found for the Spanish spoken in Palma de Mallorca and Sóller, Balearic Islands (Pieras 1999; Serrano 1996). Davidson (2015) empirically assessed the frequency of intervocalic [z]‌in the careful speech (i.e. word-​list reading) of 20 Barcelonan female youths (ages 18–​27) stratified by language dominance, and explored linguistic factors that conditioned the production of intervocalic /​s/​. While [z] production was most strongly favored by speakers with the greatest exposure to and usage of Catalan (32% [z] production), speakers with greater exposure to and use of Spanish nonetheless used [z] over 25% of the time. Both groups additionally produced intervocalic [z] significantly more often in the word-​final prevocalic context than the syllable-​initial context (by roughly 40%), whereas a significant effect of syllable stress favoring [z] across unstressed intervocalic /​s/​sequences over stressed ones was only obtained for speakers with the greater exposure to and use of Catalan. To synthesize, as the majority of research on intervocalic /​s/​-v​ oicing in Catalonian Spanish consists of impressionistic characterizations of [z]‌as a frequent, hallmark feature of Catalan speakers’ Spanish (Casanovas 1995), an exhaustive account of its modern usage has yet to be obtained. Social correlates of Barcelonan [z] production are characterized as absent by Wesch (1997), and though Davidson (2015) found that frequencies of [z] production were sensitive to speakers’ exposure to and usage of Catalan, additional sensitivities to speaker gender, age and speech style, three of the foundational social factors that inform variationist sociolinguistic evaluations of language variation and possible change (Labov 2001; Tagliamonte 2012), were unexamined. Overt awareness of Barcelonan Spanish [z] appears relatively absent, and positive links with solidarity and a shared bilingual identity may facilitate the diffusion of this feature throughout the bilingual speech community (Sinner 2002), though such a hypothesis must be treated as speculative until an empirical investigation of the social and linguistic correlates of intervocalic /​s/​-​voicing is performed. As the only empirical quantification of Barcelonan intervocalic /​s/​-v​ oicing consists of Davidson’s (2015) examination of young females’ careful speech, it is difficult to assess the extent to which [z]‌is present relative to [s] in Barcelona, particularly in more natural, spontaneous speech. Arnal (2011) has claimed (crucially, impressionistically) that due to the aforementioned massive influx of L1-​Spanish speakers to Barcelona since the mid-​twentieth century, the speech of the current youth generation no longer exhibits Catalanized features, and in particular phonetic innovations that would comprise a Catalanized accent of Spanish. While Davidson’s (2015) findings of [z] production at rates above 25% appear inconsistent with this claim, the present study offers a considerably more exhaustive examination of intervocalic /​s/​-​voicing as conditioned by a full array of linguistic and social factors, and will definitively demonstrate the degree to which this feature persists in Barcelonan Spanish, and indeed is actively being further propagated throughout the speech community as a regional marker of Catalonian, bilingual Spanish.

Intervocalic /s/-voicing: Spanish, Catalan  103

3.4  Experimental methodology 3.4.1  Social factors and subject population Four social factors, namely gender, style, age, and language dominance, were employed in the present research on Spanish intervocalic /​s/​-​voicing. Following the Variationist Sociolinguistic framework (Labov 2001; Tagliamonte 2012), gender, style, and age stratification reveal important insights into the current status and trajectory of [z]‌production in Barcelonan Spanish. With respect to gender and style, females are expected to conform more closely than men to overtly prescribed linguistic norms but conform less than men when they are not overtly prescribed (Chambers 2004; Labov 2001) a pattern commonly termed the gender paradox. Given the lack of overt awareness of non-​standard [z], coupled with its positive covert link to solidarity and bilingual identity, we accordingly may expect a gender stratification favored greater voicing by females over males, as well as greater voicing when speakers speak more spontaneously than when they more closely monitor their speech while reading. As for age, we employ the apparent-​time construct (Bailey 2004, 1991; Chambers 2004; Sankoff & Blondeau 2007; Tagliamonte 2012) by establishing two age groups approximately one generation apart (18–​30 year olds vs. 48–​60 year olds) and interpret synchronic variation as indicative of a diachronic trajectory whereby the younger group is considered more advanced than the older group. With regard to the last social factor, language dominance, participants were recruited in two principal testing sites, namely Barcelona and Madrid, mirroring Sinner (2002).4 The former was divided into three groups based on profiles of language dominance and a city/​village divide, whereas the latter served as a smaller control group that permitted a comparison of intervocalic /​s/​production of the Catalan-​Spanish community with those of an outsider, monolingual Spanish community. Bilingual participants were classified into speaker groups based on reported usage of Catalan and Spanish in their daily lives and familial upbringing, though as previously discussed in Section 3.2, all bilingual participants displayed a functional command of both languages. Barcelona participants all hail from the Barcelona Metropolitan Area (BMA), though those from the urban capital (population = 1,573,318 [Institut d’Estadística 2011)] are grouped separately from those from smaller, Catalan-​prevalent villages (average population = 7,419 [Institut d’Estadística 2011]) on the outskirts of the BMA, capturing potential differences in intervocalic /​s/​production reflecting an urban/​rural divide as previously obtained in Sinner (2002) and Davidson (2019). A  total of 54 speakers participated, visualized in Table 3.2. 3.4.2  Linguistic factors The present investigation incorporates two linguistic factors, namely word position and stress, as motivated by prior empirical research by Davidson

104  Justin Davidson Table 3.2 Subject population according to language profile group Listener Group

Younger (18–​30) /​ Older (48–​60)

Home Language /​ Native Language /​ Parent Native Lang.

Daily Spanish Use

2013 Linguistic Census Data for Catalan Competencies (understand /​speak /​read /​write)

A. Catalan-​ dominant, village B. Catalan-​ dominant, city C. Spanish-​ dominant, City D. Madrid (monolingual)

5M 5F /​ 3M 3F

Catalan

7% (SD = 5.8)

98% /​88% /​88% /​ 70%

5M 5F /​ 3M 3F 5M 5F /​ 3M 3F 3M 3F

Catalan

10% (SD = 6.9) 77% (SD = 10.8) 100% (SD = 0)

95% /​72% /​79% /​ 53%

Spanish Spanish

N/​A

Source: 2013 Census data: Institut d’Estadística (2014).

(2015) and Pieras (1999), as well as by descriptions reported in Serrano (1996), Vann (2001), and Wesch (1997). These factors are word position, stress and word class. First, with respect to word position, Vann (2001), Wesch (1997) and Vázquez (1996) only list examples of [z]‌production in word-​final intervocalic position, not mentioning word-​initial and word-​medial intervocalic contexts. Davidson (2015) found a strong effect of word position, such that intervocalic fricative voicing in word-​medial contexts was heavily disfavored over word-​final prevocalic contexts. In the present study, we slightly modify the treatment of word position, namely by differentiating between intervocalic /​s/​segments in word-​final position (e.g. las amigas “the friends”) and in word-​initial position (e.g. estaba salado “it was salty”). The decision to substitute the word-​initial intervocalic context for the word-​medial context was made in order to facilitate the full crossing of linguistic factors, since vowels surrounding word-​ medial /​ s/​are typically oppositely stressed (e.g. masa /​má.sa/​ “mass,” pasado /​pa.sá.do/​ “past”). Moreover, it should be noted that significant differences in the (small) degree of intervocalic fricative voicing conditioned across word-​initial and word-​medial intervocalic contexts were not obtained for Madrid Spanish (Torreira & Ernestus 2012). In accordance with the aforementioned research, we hypothesize that intervocalic /​s/​-v​ oicing will be favored in the word-​final context over the word-​initial context, especially in so much as the word-​initial context is uniquely the site of a phonological voicing contrast in Catalan (as previously discussed in Section 3.1). Additionally, it should be noted that the favoring of greater voicing, and indeed other analogous lenition phenomena like /​s/​-​aspiration in word-​final contexts over word-​initial ones, has been accounted for in non-​contact varieties of Spanish as a general consequence of

Intervocalic /s/-voicing: Spanish, Catalan  105 coda contexts being a prosodically weaker position, thus favoring segmental weakening (File-​Muriel & Brown 2010; Nam, Goldstein, & Saltzman 2009; Torreira & Ernestus 2012). Second, with regard to stress, Davidson found that intervocalic /​s/​-​voicing was favored in contexts of surrounding unstressed vowels for select Barcelonan female youths (2015: 13). In the present study, we have opted to simplify the treatment of syllable stress (which previously was investigated as two separate variables, namely prior vowel stress and proceeding vowel stress) in order to facilitate a fuller crossing of linguistic factors. We shall consider stress effects on /​s/​production in terms of two contexts:  surrounding unstressed vowels (e.g. las amigas “the friends,” carta sin nombre “letter without a name”) and at least one (two whenever possible) stressed vowels (e.g. las águilas “the eagles,” está sin dinero “s/​he is without money,” serás ágil “you will be agile”). Since Spanish has no determiners ending in /​s/​that exhibit final-​syllable stress, these tokens, alongside (unstressed) prepositions (e.g. sin “without,” sobre “on,” tras “after”), are the only tokens belonging to the “stressed” group that feature a single adjacent stressed vowel as opposed to two. The conditioning of stress, on a theoretical ground, invokes the concept of local hyper-​articulation for stressed syllables, or the notion that the speaker may reduce otherwise expected effects of gestural overlap with a neighboring segment across stressed syllables, since these kinds of syllables have longer durations in Spanish (Hualde 2014) and allow the speaker to better time articulatory gestures independently of one another (cf. Browman & Goldstein 1989, 1991). More concretely, this would suggest that tokens with stressed surrounding vowels, such as serás ágil, would be the most resistant to /​s/​-​voicing as an effect of the greater opportunity (across stressed syllables) for the successful coordination of vocal fold abduction for voiceless [s]‌relative to the vocal fold adduction gestures of the surrounding (voiced) vowel segments. 3.4.3  Test instruments Three test instruments were utilized in this study. The first, a socio-​demographic questionnaire, was used to screen participants according to the social criteria outlined in Section 3.4.1 so as to facilitate their binning according to the language profile groups that appear in Table 3.2. The second instrument, a recorded phrase reading in Spanish (see the Appendix for all stimuli), was used to elicit more self-​monitored or careful speech. Subjects were asked to read aloud, using their best pronunciation, a series of 80 target phrases with intervocalic /​s/​, stratified according to the aforementioned two linguistic factors of word position and stress (20 tokens per cell). With the exception of target items with the unstressed prepositions sin “without” and sobre “on,” all intervocalic /​ s/​stimuli consisted of a sequence of /​a/​ –​/​s/​ –​/​a/​. The motivation to minimally vary the quality of vowel surrounding /​s/​tokens stems from a related /​s/​-​lenition study by File-​ Muriel (2011) that found a significant effect of surrounding vowel height on

106  Justin Davidson the lenition of /​s/​to [h]‌and [∅]. The selection of /​a/​as the vowel to most often surround /​s/​follows Quilis, who notes that of all the Spanish vowels, /​a/​is the one that exerts the least influence on the frequency of the energy produced during an /​s/​segment (1981: 235). The third instrument, a 20-​minute sociolinguistic interview (Labov 2001), was used to elicit more casual, spontaneous speech. Participants were asked to discuss casual topics such as food preferences, hobbies and vacation spots, with the goal of eliciting as much casual speech data as possible from each participant. The inclusion of this task alongside the aforementioned phrase reading permits an analysis of speech style of intervocalic /​s/​production, notably with the expectation that non-​standard speech variants, like intervocalic [z]‌, are more likely to be avoided in more formal speech styles (Moreno 2009; Tagliamonte 2012). 3.4.4  Data collection methods Each participant was recorded individually during a single experimental session lasting approximately 45 minutes. Participants were recorded using an SE50 Samson head-​mounted condenser microphone and an H4n Zoom digital recorder (sampling at 44,100hz) in an audiometric booth in the phonetics laboratory at the Universitat Autònoma de Barcelona, in an empty classroom at the Universitat de Barcelona or Universitat Pompeu Fabra, or (for monolinguals) in a quiet room in the Centro de Estudios de Posgrado at the Universidad Autónoma de Madrid. Recruitment was accomplished principally through the use of posted flyers that advertised the study and its compensation for completion of 15 euros.

3.5  Data analysis and results 3.5.1  Acoustic analysis Several methods of analyzing (de)voicing phenomena in fricatives can be found in the phonetics and phonology literature, including the percentage of the fricative segment’s duration that is voiced (which relates to spectral properties of the segment and can be computed manually or by using Praat’s voice report function, which uses a pulse-​based algorithm), harmonicity, intensity and center of gravity (all of which relate to the segment’s energy properties), and duration (which relates to temporal properties of the segment). Gradoville (2011) offers a brief explanation of the theory behind each measurement and sources in the literature for each. As Gradoville addresses, few attempts to weigh or compare the validity of a particular measurement type against another have ever been made, and thus his research aims to objectively determine which method(s) is/​are most valid for the study of fricative (de)voicing phenomena. We have chosen to measure intervocalic /​s/​-​voicing in terms of the percentage of each segment’s voiced duration, which, aside

Intervocalic /s/-voicing: Spanish, Catalan  107 from being corroborated by Gradoville as a valid measure of fricative voicing, has been used to examine /​s/​-​voicing assimilation in Mexican Spanish (Schmidt 2011), Madrid Spanish (Hualde 2014; Torreira & Ernestus 2012), US Spanish (Chapter  6 of this volume), Costa Rican Spanish (Chappell & García 2017), Colombian Spanish (Chapter  8 of this volume), and North-​ Central Peninsular Spanish (Campos-​ Astorkiza 2014), facilitating more transparent comparisons of the present Barcelonan Spanish data with these distinct varieties. In order to calculate voicing durations for each intervocalic fricative segment, fricative boundary segmentation was performed manually by marking left and right boundaries for each segment by using both the waveform and spectrogram to find the zero-​intercept in the waveform closest to the first and last signs of aperiodic noise (Campos-​Astorkiza 2014; Erker 2012; File-​Muriel & Brown 2011; Rohena-​Madrazo 2015; Schmidt 2011). Once intervocalic fricative segments were segmented, exact voicing durations were measured as portions of each fricative segment that exhibited a fundamental frequency (that is, a pitch track), a voice bar at the bottom of the spectrogram, and glottal pulses (Campos-​Astorkiza 2014; Gradoville 2011; Hualde 2014; Rohena-​Madrazo 2015; Schmidt & Willis 2011; Torreira & Ernestus 2012). Note that voicing duration measurements were calculated only after adjusting the Praat spectrogram viewing window to be exactly twice the size of the /​s/​segment and centered on the /​s/​segment, as the F0 contour in the spectrogram is calculated with respect to the segments in the visible window. Example spectrograms illustrating a less voiced and more voiced realization of Spanish intervocalic /​s/​produced by different speakers appear as Figures 3.1 and 3.2. The phrase reading task yielded a total of 4,320 Spanish /​s/​tokens. Those (relatively few) tokens with notable speaker disfluencies (principally pauses

Figure 3.1 Group C, younger female rendition of tras años “after years” (5% voiced).

108  Justin Davidson

Figure 3.2 Group A, younger male rendition of tras años “after years” (100% voiced).

disrupting the word-​final /​asa/​sequence in items like explorarás áreas “you will explore areas”) were discarded from analysis, leaving 4136 Spanish /​s/​ tokens (roughly 77 out of a possible 80 tokens per speaker) for subsequent statistical analysis. Regarding the casual Spanish /​s/​data from the sociolinguistic interview, each participant contributed a total of precisely 20 casual Spanish /​s/​tokens, perfectly balanced across the factors of word position and stress, as this was the highest number of intervocalic /​s/​tokens produced by all speakers that offered the most parallel distribution of linguistic factor contexts as compared with the /​s/​tokens elicited from the careful speech task, permitting a more valid comparison of /​s/​production across the two speech styles. With the addition of these 1,080 casual Spanish intervocalic /​s/​tokens, the total amount of Spanish /​s/​tokens available for statistical analysis was 5,216, which equates to roughly 97 tokens per speaker. 3.5.2  Statistical analysis A single mixed-​effects linear regression model was performed in R using the percentage of voiced segment duration as the dependent variable, testing for fixed effects of two linguistic factors (word position [initial vs. final] and stress [stressed vs. unstressed]) and three social factors (gender [male vs. female], style [careful vs. casual], and the combination of language profile group with age, since the control group of Madrid monolinguals was not stratified with respect to age [Group A, Younger vs. Group B, Younger vs. Group C, Younger vs. Group A, Older vs. Group B, Older vs. Group C, Older vs. Group D, Younger]). Interaction terms between the combined language profile group and age factor and each of all the other independent variables were included in order to assess whether or not any of the effects varied significantly according

Intervocalic /s/-voicing: Spanish, Catalan  109 Table 3.3 Summary of mixed-​effects linear regression model fitted to Barcelonan and Madrid Spanish intervocalic alveolar fricatives. Intercept:  Group A, younger female casual speech, word-​final unstressed tokens

(Intercept) Group A, Older Group B, Older Group C, Older Group C, Younger Group D (Younger) Male Careful Word-​initial Stressed Group D (Younger): Male Group A, Older: Word-​initial Group B, Older: Word-​initial Group C, Older: Word-​initial Group B, Younger: Word-​initial Group C, Younger: Word-​initial Group D (Younger): Word initial

β

t

p

95.79 -​27.2 -​18.66 -​47.99 -​21.69 -​60.56 -​10.71 -​15.28 -​65.09 -​12.4 9.88 28.72 29.45 38.79 8.98 30.01 44.56

20.70 -​4.46 -​3.06 -​7.86 -​4.1 -​9.92 -​3.82 -​4.6 -​41.72 -​7.97 2.95 14.4 14.67 19.23 5.18 17.21 22.26