An Anthology of Bilingual Child Phonology 9781788928427

Citation preview

An Anthology of Bilingual Child Phonology

SECOND LANGUAGE ACQUISITION Series Editors: Professor David Singleton, University of Pannonia, Hungary and Fellow Emeritus, Trinity College, Dublin, Ireland and Associate Professor Simone E. Pfenninger, University of Salzburg, Austria This series brings together titles dealing with a variety of aspects of language acquisition and processing in situations where a language or languages other than the native language is involved. Second language is thus interpreted in its broadest possible sense. The volumes included in the series all offer in their different ways, on the one hand, exposition and discussion of empirical fi ndings and, on the other, some degree of theoretical reflection. In this latter connection, no particular theoretical stance is privileged in the series; nor is any relevant perspective – sociolinguistic, psycholinguistic, neurolinguistic, etc. – deemed out of place. The intended readership of the series includes fi nal-year undergraduates working on second language acquisition projects, postgraduate students involved in second language acquisition research, and researchers, teachers and policy makers in general whose interests include a second language acquisition component. All books in this series are externally peer-reviewed. Full details of all the books in this series and of all our other publications can be found on http://www.multilingual-matters.com, or by writing to Multilingual Matters, St Nicholas House, 31–34 High Street, Bristol BS1 2AW, UK.

SECOND LANGUAGE ACQUISITION: 142

An Anthology of Bilingual Child Phonology Edited by

Elena Babatsouli and Martin J. Ball

MULTILINGUAL MATTERS Bristol • Blue Ridge Summit

DOI https://doi.org/10.21832/BABATS8410 Library of Congress Cataloging in Publication Data A catalog record for this book is available from the Library of Congress. Names: Babatsouli, Elena, editor. | Ball, Martin J. (Martin John), editor. Title: An Anthology of Bilingual Child Phonology /Edited by Elena Babatsouli and Martin J. Ball. Description: Blue Ridge Summit: Multilingual Matters, 2020. | Series: Second Language Acquisition: 142 | Includes bibliographical references and index. | Summary: “This book is an edited collection of phonological development studies that pertain to themes in child bilingualism. It comprises studies on protolanguage phonology, referring to the development of children’s autonomous linguistic systems from their fi rst meaningful forms to complete cognitive and articulatory acquisition of language”— Provided by publisher. Identifiers: LCCN 2020002476 (print) | LCCN 2020002477 (ebook) | ISBN 9781788928410 (hardback) | ISBN 9781788928427 (adobe pdf) | ISBN 9781788928434 (epub) | ISBN 9781788928441 (kindle edition) Subjects: LCSH: Bilingualism in children. | Children—Language. | Language acquisition. | Grammar, Comparative and general—Phonology. Classification: LCC P115.2 .A58 2020 (print) | LCC P115.2 (ebook) | DDC 404/.2083—dc23 LC record available at https://lccn.loc.gov/2020002476 LC ebook record available at https://lccn.loc.gov/2020002477 British Library Cataloguing in Publication Data A catalogue entry for this book is available from the British Library. ISBN-13: 978-1-78892-841-0 (hbk) Multilingual Matters UK: St Nicholas House, 31–34 High Street, Bristol BS1 2AW, UK. USA: NBN, Blue Ridge Summit, PA, USA. Website: www.multilingual-matters.com Twitter: Multi_Ling_Mat Facebook: https://www.facebook.com/multilingualmatters Blog: www.channelviewpublications.wordpress.com Copyright © 2020 Elena Babatsouli, Martin J. Ball and the authors of individual chapters. All rights reserved. No part of this work may be reproduced in any form or by any means without permission in writing from the publisher. The policy of Multilingual Matters/Channel View Publications is to use papers that are natural, renewable and recyclable products, made from wood grown in sustainable forests. In the manufacturing process of our books, and to further support our policy, preference is given to printers that have FSC and PEFC Chain of Custody certification. The FSC and/or PEFC logos will appear on those books where full certification has been granted to the printer concerned. Typeset by Nova Techset Private Limited, Bengaluru and Chennai, India.

Contents

Acknowledgements Contributors

vii ix

1

Introduction Elena Babatsouli and Martin J. Ball

2

Investigating the Linguistic World of the Bilingual Child: Transfer and Code-switching Conxita Lleó

27

Seeking Crosslinguistic Interaction in French Bilingual Phonological Development Margaret Kehoe

58

Case Studies of Phonological Development in Six Preschool-aged Russian-Finnish Bilingual Children Olga Nenonen

85

3

4

5

6

7

8

9

1

Enhanced Phonology in a Child’s Weaker Language in Bilingualism: A Portrait Elena Babatsouli

117

Sensitivity to Morphophonological Cues in Monolingual and Bilingual Children: Evidence from a Nonword Task Luca Cilibrasi and Ianthi Tsimpli

140

Lexical-semantic Organization in Monolingual and Bilingual Hebrew Speaking Children: Evidence from a Word Association Task Atalia Hai Weiss The Production of Marked Arabic Consonants by Arabic-English Bilingual Children Living in Canada Anwar Alkhudidi, Yasmeen Hakooz, Madeline Walker, Ryan A. Stevenson and Yasaman Rafat On Heritage Accents: Insights from Voice Onset Time Production by Trilingual Heritage Speakers of Spanish Raquel Llama and Luz Patricia López-Morelos

v

163

187

209

vi

An Anthology of Bilingual Child Phonology

10 Linguistic Outcomes and the Role of Phonology in Typically Developing and Late Talking Toddlers Kakia Petinou and Loukia Taxitari 11 Stylistic Patterns in the Speech of Young Children and their Caregivers: A Study of Variable /s/ Lenition in Dominican Spanish Karen Miller and Rodrigo Cárdenas

232

251

12 Identification of Protracted Phonological Development across Languages: The Whole Word Match and Basic Mismatch Measures 274 Barbara May Bernhardt, Joseph Paul Stemberger, Daniel Bérubé, Valter Ciocca, Maria João Freitas, Diana Ignatova, Damjana Kogošek, Inger Lundeborg Hammarström, Thóra Másdóttir, Martina Ozbič, Denisse Perez and A. Margarida Ramalho 13 Phonological Processing and Nonword Repetition: A Critical Tool for the Identification of Dyslexia in Bilingualism Chiara Melloni and Maria Vender Index

309 334

Acknowledgements

We are thankful to the authors for their contributions and to the reviewers for their thorough and timely reviews. This volume would not have been possible without their willingness to share their work which advances knowledge on themes pertaining to bilingual child phonology. We also wish to extend our gratitude to the book series editors, David Singleton and Simone E. Pfenninger, for hosting this project in the Second Language Acquisition book series at Multilingual Matters.

vii

Contributors

Editors

Elena Babatsouli is an Assistant Professor at the University of Louisiana at Lafayette, co-editor of the Journal of Monolingual and Bilingual Speech and founder of the International Symposium of Monolingual and Bilingual Speech. She has a BA in English (Royal Holloway, University of London), an MA in languages and business (London South Bank University) and a PhD in linguistics (University of Crete). Elena’s research interests are in child monolingual and bilingual acquisition, atypical speech (SSDs), SLA, phonetics/phonology, psycholinguistics, measures and quantitative methods. She has 30 publications, five edited books, two edited conference proceedings and two edited journal special issues. Martin J. Ball is Emeritus Professor of Linguistics at Bangor University, Wales, having previously held positions in Wales, Ireland, the US and Sweden. He co-edits the journals Clinical Linguistics and Phonetics and Journal of Multilingual and Bilingual Speech, as well as book series for Multilingual Matters and Equinox Publishers. Martin has published widely in communication disorders, phonetics, sociolinguistics, bilingualism and Welsh linguistics. Recently he completed co-editing the fourvolume Encyclopedia of Human Communication Sciences and Disorders for Sage. He is an honorary fellow of the Royal College of Speech and Language Therapists and a fellow of the Learned Society of Wales. He currently lives in Cork, Ireland. Authors

Anwar Alkhudidi is a Lecturer in the English Department at UmmAl-Qura University. She holds a BA in English language from Taif University in Saudi Arabia. She obtained an MA in linguistics in 2018 from the University of Western Ontario, Canada and is currently working on her PhD in linguistics at the University of Macquarie in Australia. Anwar’s area of research interests includes phonological and phonetic development in both L1 and L2 speakers. She is also a member of the Canadian Linguist Association, a recipient of the Saudi Ministry of Education Scholarship and has taught a variety of undergraduate courses. ix

x

An Anthology of Bilingual Child Phonology

Barbara May Bernhardt was on the faculty of the School of Audiology and Speech Sciences at the University of British Columbia, Vancouver, Canada, from 1990 to 2017. She has been a speech-language pathologist since 1972. Her primary focus is phonological development, assessment and intervention across languages. In collaboration with co-investigator Joseph Paul Stemberger and colleagues in over 15 countries, Barbara has been conducting an international crosslinguistic project in children’s phonological acquisition (phonodevelopment.sites.olt.ubc.ca). Other areas of expertise include the utilization of ultrasound in speech therapy, language development, assessment and intervention, and approaches to service delivery to Indigenous people in Canada. Daniel Bérubé has been a faculty member in the Audiology and SpeechLanguage Pathology Department at the University of Ottawa, Canada, since 2017. He has been a speech-language pathologist since 2005. His primary teaching and research focus is on the language and literacy development of bilingual and multilingual children. Daniel has collaborated on international and national projects, including an international crosslinguistic project in children’s phonological acquisition (phonodevelopment. sites.olt.ubc.ca) and a Health Canada funded project assessing the oral language and pre-literacy skills of bilingual French-English children. Rodrigo Cárdenas (PhD in psychology, Michigan State University) is an Assistant Teaching Professor at Penn State University. His research focuses on how parental care and mate selection has affected the evolution of human cognition. This work examines how people use facial and vocal cues to make decisions about how to interact with conspecifics, and how these perceptual and decisional processes are regulated by neuroendocrine systems. Rodrigo’s work on parental care studies how cognitive mechanisms are attuned to the detection, encoding/storing and processing of infant-related information, and how they vary as a function of the individual and cultural differences that affect interest in infants. Luca Cilibrasi is Lecturer in Psycholinguistics and Language Acquisition at Charles University (Department of English and ELT Methodology and Department of General Linguistics). His main research interests are acquisition in bilinguals, as well as acquisition in children with language impairment and in children with dyslexia. Before his appointment in Prague, Luca completed a De Vincenzi Postdoctoral Fellowship at the University of Cambridge, a PhD at the University of Reading (clinical language sciences) and BA and MA studies at the University of Siena (communication and linguistics). Valter Ciocca is a Professor in the School of Audiology and Speech Sciences at the University of British Columbia. Between 1992 and 2006 he

Contributors xi

was a faculty member in the Department of Speech and Hearing Sciences at the University of Hong Kong. Valter’s research expertise includes the perceptual and acoustic analyses of speech produced by speakers with and without communication disorders, the perception and production of pitch in speech and other sounds, and auditory scene analysis. Maria João Freitas is an Associate Professor of Linguistics in the Faculty of Humanities, University of Lisbon (FLUL), and is a member of the CLUL research center. She is currently the director of the PhD program in linguistics at FLUL. She has a BA in Romance languages and literatures, an MA in phonetics, a PhD in humanities (phonological acquisition) and an Agregação in Portuguese linguistics (phonological acquisition). Maria’s research interests are in phonological and morphological acquisition, in both typical and atypical language development. She has published in the fields of phonological acquisition, clinical linguistics and educational linguistics. Yasmeen Hakooz is currently a second year PhD student in the Faculty of Education at Western University, Canada, where she specializes in the field of applied linguistics. Her doctoral research examines the identity formation of Canadian-born Muslim students in Canadian post-secondary schools. Yasmeen holds a BSc in psychology from the University of Toronto. She also holds an MA in linguistics from Western University, where she examined the Circassian language using UNESCO’s language vitality and endangerment scales. Previously, Yasmeen has chaired WISSLR, a student symposium held at Western University, and she is part of the Congress 2020 Program Advisory Committee. Diana Ignatova has a PhD in speech-language pathology. She is a Lecturer at Sofia University ‘St. Kliment Ohridski’. She gives academic courses on several disciplines: introduction to clinical linguistics; diagnostic assessment of communicative disorders and team work; learning disabilities; communicative disorders in craniofacial malformations; and non-verbal learning disabilities. Diana is the author of the two-part monograph, Reading and Specifi c Reading Disabilities: Theoretical Aspects and Social Perspectives (I, 2009) and Specific Reading Disabilities: Diagnostic Assessment (II, 2010). In the past she has worked as a speech therapist for children with a wide range of disabilities. Margaret Kehoe is a Senior Lecturer in the Psycholinguistics Department at the University of Geneva. She received a Master’s degree in speech and hearing sciences from the University of Arizona and a PhD from the University of Washington. Her research focuses on the phonetic and phonological development of young children within a crosslinguistic context. Margaret has studied linguistic features such as stress, rhythm, syllable

xii

An Anthology of Bilingual Child Phonology

structure, phonological vowel length and voice onset time in English, German, Spanish and French speaking children as well as in bilingual children. She teaches classes in speech sound disorders and bilingualism. Damjana Kogovšek is an Assistant Professor at the University of Ljubljana. She is a speech and language pathologist and a teacher of the deaf and hard of hearing. She previously worked as a speech and language pathologist with preschool children and adolescents with communication disorders in public school (Institution of the Deaf and Hard of Hearing Ljubljana) and kindergarten. She received her PhD from the University of Ljubljana, Slovenia in 2007, and is currently working as an Assistant Professor in the Faculty of Education Ljubljana. Damjana teaches and conducts research in several areas, including the rehabilitation of speech and language disorders, communication disorders, education of the deaf and hard of hearing and the inclusion of students with special educational needs. Raquel Llama is a Postdoctoral Fellow in the Department of Language Education at Stockholm University as well as a member of the Language Acquisition Research Lab (LAR Lab) at the University of Ottawa, and associate editor of the Spanish Journal of Applied Linguistics (Revista Española de Lingüística Aplicada, RESLA). Her main area of expertise is third language acquisition. Raquel’s most recent publications and conference presentations include research on lexical access and heritage language acquisition in trilingual speakers, with a focus on the sub-field of (heritage) phonetics and phonology. Conxita Lleó, now retired, was Professor of Romance Linguistics at the University of Hamburg, Germany. She studied Romance languages at the University of Barcelona and general linguistics at the University Washington (Seattle), and received two PhD degrees, one from each university. Conxita’s focus of research lies in bilingualism, child language, phonological acquisition and sound change, where she has published about 100 studies and several books. Over a period of about two decades she obtained research grants from the German Science Foundation. Luz Patricia López-Morelos is an instructor in the School of Linguistics and Language Studies (SLaLS) at Carleton University and a member of the Language Acquisition Research Lab (LAR Lab) at the University of Ottawa. Her main area of research interest is heritage language acquisition. More specifically, she explores diverse aspects of the acquisition and use of Spanish as a heritage language by trilingual speakers. Her current research aims at investigating the status of grammatical gender in children and adolescents who speak Spanish as a heritage language, speak English as a dominant language and are learning French as a third language.

Contributors

xiii

Inger Lundeborg Hammarström is an Assistant Professor at the University of Linköping, Sweden, and co-editor of Logopedics Phoniatrics Vocology. Inger is a certified speech-language pathologist and has worked clinically with children with speech and language impairments for many years before entering the research community. Inger has a PhD in otorhino-laryngology (Linköping University). Her research interest is in children with atypical speech and language, both children with CLP and in children with no known causation behind their difficulties. Inger has 25 publications and is the co-author of two book chapters. She has also, together with colleagues, developed three Swedish assessment materials. Thóra (Þóra) Másdóttir is an Assistant Professor and a Program Director at the University of Iceland, School of Health Sciences. She has an MA in speech and language pathology (Indiana University, USA) and a PhD in the same field (Newcastle University, UK). Thóra has worked as a speech and language pathologist since 1990 and holds a certificate of clinical competence (CCC) from the American Speech-Language-Hearing Association. Her primary research focus is clinical child phonology as well as speech and language test development. Thóra collaborated with Barbara May Bernhardt and Joseph P. Stemberger in an international crosslinguistic study in children’s phonological acquisition. Chiara Melloni is currently Associate Professor of General Linguistics at the University of Verona, Italy. Her research falls within the lexical semantics/morphosyntax interface and focuses on the formal and interpretative properties of the morphologically complex lexicon. She also conducts psycholinguistic research aimed at exploring morphological and phonological processing and awareness skills in typical and atypical language acquisition, in both monolingual and bilingual settings. Chiara is the author of a monograph (Event and Result Nominals: A Morphosemantic Approach, 2011, Peter Lang) and of numerous articles in international journals, and has given invited lectures both nationally and internationally. Karen Miller (PhD in linguistics, Michigan State University) is an Associate Professor of Spanish Linguistics and the co-director of the Center for Language Science at Penn State University. Her research focuses on (developmental) sociolinguistics; she is especially interested in how children and adults acquire variation in L1 and L2 acquisition. Karen’s work also investigates the impact of input type (e.g. variable input, inconsistent input, non-variable input) on the acquisition process, focusing on whether different input types impact acquisition outcomes (i.e. acquisition paths and time course). Her work on language acquisition and variation has been funded by the National Science Foundation.

xiv

An Anthology of Bilingual Child Phonology

Olga Nenonen is University Instructor of the Russian Language and Culture in the Faculty of Information Technology and Communication Sciences at Tampere University, Finland. Her research focuses on phonology/phonetics, child language acquisition and bilingualism. Her research interests include typical and atypical language acquisition (children with SLI) with a crosslinguistic focus. Martina Ozbič PhD, has been a speech and language therapist and special pedagogist, assistant (1997–2010) and Assistant Professor (2010–2017) at the University of Ljubljana, Slovenia. Since 2017 she has had her own bilingual private clinical practice in Slovenia, collaborating since 2018 as a researcher with Associazione La Nostra Famiglia, as a Lecturer with the University of Trieste, Italy, in courses for speech and language therapy, and further with the Faculty for Applied Social Sciences in Nova Gorica, Slovenia. Martina’s current area of research and professional expertise is protracted phonological development and language impairment, especially in bilingual children. She has (co)authored several papers about phonological development, the prevention of specific learning impairment and adaptations of instruments in Slovenian language. Denisse Pérez Herrera is a Titular Professor at the University of Valparaíso, Chile. She is the Director of the Center for Development Research in Cognition and Language (CIDCL) of the same university. Denisse is a speech therapist (University of Chile), and has a Master’s in applied linguistics from the Pontifical Catholic University of Valparaíso and a PhD in psychology from the University of Granada, Spain. Her research interests are in child monolingual acquisition, communication disorders, and phonetics in Down Syndrome. Kakia Petinou is an Associate Professor and Vice Dean of the Department of Rehabilitation Sciences, School of Health Sciences, Program of Speech and Language Pathology at the Technological University Cyprus. She is also the founder and director of ‘TheraLab’ at the Cyprus University of Technology. She holds a BSc and a Master’s degree in communication sciences and disorders from the University of Georgia, USA as well as a PhD in the field of speech and hearing sciences and developmental psycholinguistics from City University of New York, Graduate Center, USA. Kakia’s research interests focus on typical and atypical speech and language development in paediatric populations including phonological disorders, early intervention, intensive phonological treatment, early speech correlates of developmental apraxia of speech, phonological and semantic interface, and clinical practices in ASD. Her research interests also lie in the development of valid and reliable assessment tools for the evaluation, diagnosis and intervention of speech and language disorders in children.

Contributors xv

Yasaman Rafat is an Associate Professor of Hispanic Studies (Linguistics) in the Department of Language and Cultures and an Associate Member of the Brain and Mind Institute at Western University. She holds a PhD in Hispanic linguistics from the University of Toronto. Yasaman works on second language speech learning, bilingualism and language change and attrition. Ana Margarida Ramalho is a speech language therapist and a member of the CLUL Research Center (University of Lisboa). She graduated in speech and language therapy, has a postgraduate degree in neuropsychology and rehabilitation, and holds a Master’s in language and communication sciences and a PhD in linguistics. Margarida’s research and clinical interests are in typical and atypical child acquisition and development (DDL, SSD and dyslexia), phonology, psycholinguistics, and measures and quantitative methods for assessment and intervention. She is also an investigator on the Cross-linguistic Child Phonology Project, a worldwide project on children’s phonological acquisition (phonodevelopment.sites.olt.ubc.ca). Joseph Stemberger is a Professor Emeritus in the Department of Linguistics at the University of British Columbia. His research addresses language processing (especially for morphology, phonology, phonetics and interactions between them, for adult language production and fi rst-language acquisition), and intersects linguistics, cognitive psychology and speechlanguage pathology. One goal is to compare across languages and explain what is similar and what is different. Current projects focus on typical and protracted phonological development in many languages: the Valley Zapotec project and the Cross-linguistic project. Joseph also does traditional dancing (English and Slovenian), is in two choirs and likes to go hiking and cycling. Ryan Stevenson is an Assistant Professor at the University of Western Ontario in the Psychology Department and the Brain and Mind Institute. His research focuses on multisensory integration, the development of sensory processing and perception, and how sensory perception influences cognitive processes. Loukia Taxitari is a psycholinguist, interested in language acquisition. She has a BA in Greek philology from the University of Athens, an MPhil in general linguistics and a DPhil from the University of Oxford. Her research interests focus on lexical development, how words acquire their reference and how infants use general knowledge about the world to help in this task. Finally, in Cyprus Loukia is studying lexical and grammatical development in a bidialectal community, and how the coexistence of a regional dialect along with a standard language influences language development.

xvi

An Anthology of Bilingual Child Phonology

Ianthi Maria Tsimpli is Professor of English and Applied Linguistics at the University of Cambridge and a Fellow of Fitzwilliam College. She works on first and second language development in children and adults, language impairment, attrition, bilingualism, language processing and the interaction between language, cognitive abilities and print exposure. She holds a BA in modern Greek philology from the University of Athens and a PhD in linguistics from University College London (1992). Before moving to Cambridge, Ianthi taught linguistics at University College London, the University of Newcastle-upon-Tyne, Aristotle University of Thessaloniki and the University of Reading. Maria Vender is currently a Research Fellow in Psycholinguistics and Adjunct Professor at the University of Verona, Italy. Her research interests lie in the area of language acquisition in typical and atypical contexts, focusing in particular on the development of linguistic competence in children suffering from specific language and learning disabilities, in both monolingual and bi-/multilingual settings. Maria is the author of a monograph (Disentangling Dyslexia – Phonological and Processing Impairment in Developmental Dyslexia, 2017, Peter Lang) and of a number of articles published in international journals. She is member of Bilinguismo Conta, the Italian branch of the organization Bilingualism Matters involved in the promotion of bilingualism, and she participates in research projects funded at a national and European level. Madeline Walker is a graduate of the Masters in linguistics and the Honours Specialization Program at the University of Western Ontario. Her research focuses on morphosyntactic second language acquisition. Currently, Madeline is completing a BEd at Wilfrid Laurier University, where she continues to invest time into child language and literacy acquisition. Madeline is presently involved in research at Wilfrid Laurier University focusing on promoting a culturally and linguistically diverse teaching profession. Atalia Hai Weiss is a speech-language pathologist who has been working mainly with children with language impairment. She is a lecturer in the Communication Disorders Department at the Hadassah Academic College of Jerusalem. Her PhD, conducted at the Hebrew University, Jerusalem, focused on a population of musicians with dyslexia. Atalia’s current research is concerned, mainly but not exclusively, with lexical development among bilingual children in Israel, both typically developing and with developmental language disorder.

1 Introduction Elena Babatsouli and Martin J. Ball

An overarching term, protolanguage refers to a child’s early linguistic system as an autonomous intermediate linguistic system in development that spans the acquisition path from the first meaningful forms to complete cognitive and articulatory acquisition of language introduced. Consequently, protolanguage covers all different contexts in typical and atypical language acquisition in childhood, i.e. monolingual, bilingual/ multilingual and dialectal/multilectal, as well as speech delay and disorder (e.g. Babatsouli & Ingram, 2018a). The present volume resumes this theme on protolanguage, compiling original research studies relating to phonology in childhood which is, by default, developmental phonology. Our specific goal in this anthology was to compile phonological development studies that pertain to themes in bilingualism, meaning the acquisition/use of more linguistic codes than one, whether it is actual languages, dialects or communication modes, in an array of contexts such as endogenous and exogenous bilingualism, heritage language, bilectalism, trilingualism, typical and atypical use, etc. An interdisciplinary field, bilingualism has shown unprecedented research growth over the past halfcentury, especially on topics concerning measures intended to help decipher the representations and language processing mechanisms of bilingual speakers. Advancements in this field have led to some reevaluation of known studies in protolanguage phonology. For example, Smith’s (1973) seminal study on English as a fi rst (single) language actually involved a child being raised in England (and visits abroad) by a multilingual mother whose English was a second language (Babatsouli, 2013; Ingram, pers. commun.; Lleó, 2018). Work done on bilingual phonology is now seen in the light of heritage language (e.g. Kupisch & Rothman, 2018; Lleó, this volume: Chapter 2), and it is argued that the children involved in some bilingual German-Spanish studies were raised with ‘some bilingualism’, ‘some trilingualism’ and some ‘bidialectalism’ being practiced at home (Lleó, this volume: Chapter 2). There is, thus, non-conclusiveness in the semantics involved in some terms in language acquisition research (e.g. Babatsouli & Ball, 2019; Babatsouli & Ingram, 2018a; Treffers-Daller, 2009), what Kupisch and Rothman (2018) have very accurately phrased as ‘Terminology matters!’. 1

2

An Anthology of Bilingual Child Phonology

In this light, there is also what appears to be an unspoken covenant in the literature that the term bilingualism may interchangeably be used with multilingualism, bi(multi)lingualism, in the sense that bi- ‘two’ (i.e. more than one) also suggests multi- ‘several’ (i.e. more than one/two), and vice versa. To give some examples, the International Expert Panel on Multilingual Children’s Speech (www.csu.edu.au/research/multilingualspeech/iepmcs) was formed to ‘provide direction and practical strategies for SLPs [speech-language pathologists] and related professionals working with children who are multilingual and/or multicultural’ (International Expert Panel on Multilingual Children’s Speech, 2012: 1). So, ‘[c]hildren who are multilingual are able to comprehend and/or produce two [our italics] or more languages in oral, manual, or written form with at least a basic level of functional proficiency or use, regardless of the age at which the languages were learned’ (International Expert Panel on Multilingual Children’s Speech, 2012: 1, adapted from Grech & McLeod, 2012: 121). On similar lines, the Phonetics, Acquisition and Multilingualism Lab (PAMLab) in Boston, MA, USA, broadly construes multilingualism as the ‘command/use of more than one [our italics] language’ (http://sites.bu.edu/ pamlab/). The Multilingual Phonology Lab in Chicago, IL, USA (https:// hip.uic.edu/research-groups/multiphon/), however, shows a clear-cut preference in the investigation of second and third language adult phonological acquisition. Additionally, multilingual may sometimes be loosely used to mean ‘across languages’ (e.g. McLeod, 2012; McLeod & Verdon , 2017; and as in the inclusion of Bernhardt et al., this volume: Chapter 12). On the other hand, an overwhelming pattern of preference for using bilingualism is exemplified by well-established or more recent individual efforts on national and academic institution levels to increase our understanding of child bilingualism (focusing on protolanguage) as by, for instance: the Centre for Research on Bilingualism in Bangor, Wales (https:// www.bangor.ac.uk/bilingualism/); the Childhood Bilingualism Research Centre in Hong Kong (http://cbrchk.org/); the Bilingual Phonology Lab in Arizona, USA (https://sites.google.com/site/bilingualphonologylab/); the THALES project (Bilingual Acquisition & Bilingual Education: The Development of Linguistic & Cognitive Abilities in Different Types of Bilingualism) in Cambridge, UK; the Irish Network in Childhood Bilingualism (https://childbilingualismresearch.com/); the Bilingualism Matters research and information center in Edinburgh with 26 international branches (http://www.bilingualism-matters.ppls.ed.ac.uk/); and also research done in Geneva (e.g. Kehoe, this volume: Chapter 3), among others. Formative work on bilingualism has been shaped within the premises of the Hamburg Studies on Multilingualism (https://www.uni-hamburg.de/en/ sfb538/publikationen-sfb538/studies-on-multilingualism.html), which has incorporated child bilingual projects like PEDSES and PhonBla (see Lleó, this volume: Chapter 2) and research on Catalan in the bilingual CatalanSpanish context (Benet et al., 2012).

Introduction 3

This arrangement of fusion/separation of what is basically a single although broad-spectrum concept is reiterated in the titles of related conferences (e.g. the International Symposium on Monolingual and Bilingual Speech, the International Symposium of Bilingualism, the International Conference on Multilingualism and Multilingual Education, etc.) and academic journals (Linguistic Approaches to Bilingualism, the International Journal of Bilingualism, the Journal of Monolingual and Bilingual Speech, the International Journal of Bilingual Education and Bilingualism, the International Journal of Multilingualism, the Journal of Multilingual and Multicultural Development, etc.). Finding Room for this Anthology in Published Works

Bilingual phonological development is at the heart of matters in language research and, also, in this edited compilation. There are numerous textbooks on child phonological acquisition available in the literature, presenting clear and concise introductions to the main concepts, issues and debates (e.g. Bernhardt & Stemberger, 1998; Ingram, 1989; Johnson & Reimers, 2010; Vihman, 1996, 2014; Vihman & Keren-Portnoy, 2013). Similarly, textbooks on bilingual, multilingual and second language acquisition in general are also abundant (e.g. Archibald, 1995; Aronin & Hufeisen, 2009; Auer & Wei, 2007; De Houwer, 2009; Doughty & Long, 2005; Grosjean, 2010; Grosjean & Li, 2012; Martin-Jones et al., 2012; Ortega, 2009; Wei, 2000). The handbook by Simonet (2016) provides an online resource on the phonetics and phonology of bilingualism. Some other published volumes link bilingualism to speech sound disorders (SSDs) and specific language impairment (SLI) (e.g. Ball et  al., 2009; Battle, 2012; Kohnert, 2008; McLeod & Goldstein, 2012; Paradis et al., 2011) or view bilingualism as part of the more general arena of linguistic and cultural diversity to be considered in assessment and intervention methods in atypical development contexts (e.g. Goldstein, 1999; Goldstein & Oller, 2012; Verdon et al., 2016). Furthermore, several publications specialize in specific theoretical or practical aspects related to bilingualism, such as the interaction between language, literacy and cognition (e.g. Bialystok, 2001; Bratt Paulston, 1998; Brisk & Harrington, 2006; Libben et al., 2017), minority languages (e.g. Lauchlan & Parafita Couto, 2017), research on the acquisition of different combinations of languages (e.g. Gut, 2015; Qi, 2011) or case studies (e.g. Deuchar & Quay, 2000; Gut, 2015). Cenoz and Genesee (2001) may be singled out as an edited volume on early bilingual acquisition tackling several linguistic levels, i.e. phonological, lexical, syntactic and pragmatic development. Thomas and Mennen have compiled a volume (2014) on theoretical and empirical aspects of bilingual research which includes some work on phonology. Other recent edited volumes on language acquisition contain solitary studies of

4

An Anthology of Bilingual Child Phonology

bilingual phonological development in typical and atypical contexts (e.g. Babatsouli & Ingram, 2018b; Babatsouli et al., 2017; Ball & Müller, 2016; Yavaș, 2015), but bilingualism is not the central theme of these collections. Cardoso et al. (2017) have compiled a volume on bilingual acquisition of phonology concentrating on German-Romance languages only. Although the majority of L3 acquisition research has engaged in the study of adult sequential bilinguals (Cabrelli Amaro et al., 2012), a current original collection on child phonological acquisition has investigated the profi les of bilingual children and adolescents acquiring an L3 (trilingualism), the majority of which were heritage speakers (early bilinguals) (edited by Wrembel & Cabrelli Amaro, 2017). There is more need for edited volumes presenting a crosslinguistic collection of studies on bilingual child phonology. There is also a shortage of compilations on protolanguage phonology, i.e. typical and atypical developmental language from babyhood to childhood in mono-/bi-/multilingual settings (e.g. Babatsouli, 2020). The present Anthology of Bilingual Child Phonology aims to fill in some of this gap in the literature. Fashioning Definitions Further

As mentioned before, there is opacity with regard to the semantics of terms used in language research (e.g. Babatsouli & Ball, 2019; Babatsouli & Ingram, 2018a; Kupisch & Rothman, 2018; Treffers-Daller, 2009). Defi ning ‘bilingualism’ is another such case in point. ‘Overall, there is no consensus of what constitutes bilingualism and how bilingual competence is represented’ (Auer & Wei, 2007: 246). There are perhaps as many bilingual children in the world as there are monolingual children (Tucker, 1998). With more than half of the world’s population being bi- or multilingual (Crystal, 1995), bilinguals are very diverse and the resulting terminological ambiguity discloses this. Thus, a definition of bilingual acquisition may well start from an acknowledgement of the phenomenon as prevalent and conventional. Roeper (1999) has ventured the argument that ‘the concept of bilingualism has never received a widely acknowledged formal defi nition’. As early as 1967, Mackey states that ‘bilingualism, far from being exceptional, is a problem which affects the majority of the world’s population’ (Mackey, 1967: 11). That the actual nature of bilingualism is very intricate, leading to ‘open ended semantics’ (Baetens Beardsmore, 1982), is also suggested in Grosjean’s (2008) statement that researchers do not yet fully understand who bilinguals really are. Haugen’s (1953) minimalist position states that a bilingual is the speaker of one language who can produce meaningful utterances in another. A determination of bilingual status in children has customarily relied on a range of defi nitions and related constructs in bilingualism based on different variables like age and timing of exposure, sociolinguistic contexts, language status and dominance (e.g. Babatsouli, 2013; Babatsouli

Introduction 5

& Ingram, 2018a). Hamers and Blanc (1983) summarize the taxonomies of factors determining bilingualism as follows: age and context of acquisition, relative status of the two languages, group membership and cultural identity, motivation and context of use. Pienemann and Keßler (2007: 248) propose at least three ‘yardsticks’ for measuring language in bilingualism: developmental trajectories, competence and proficiency. However, the debate on defi ning bilingualism is an ongoing process with researchers recognizing that, like a chameleon, it may take many different forms depending on the situation. This is exemplified by some 37 types of bilingualism recounted in just one table by Gass and Selinker (2008: 27–28), which includes terms like: ‘incipient bilingual: someone at the early stages of bilingualism where one language is not fully developed’; ‘dominant bilingual: someone with greater proficiency in one of his or her languages and uses it significantly more than the other language(s)’; ‘simultaneous bilingual: someone whose two languages are present from the onset of speech’; ‘unbalanced bilingual: someone who is not equally fluent in the two languages; ‘additive’ (vs. ‘subtractive’) bilingual: someone for whom both languages have socially and emotionally equal value, etc. The authors acknowledge that the table is incomplete and, while they themselves cite Valdés’ (2001) ‘mythical bilingual’ as just one more case, they admonish the reader to look into Wei (2000) for even more defi nitions. Among these defi nitions, heritage speakers are those bilinguals raised in a home where one language is spoken but they subsequently switch to another (Polinsky & Kagan, 2007). The present compilation contributes to heritage language research with the following papers: Alkhudidi et al.; Llama & López-Morelos; and Lleó (this volume: Chapters 8, 9 and 2, respectively). One can add to this the presence of a third (weaker?) language and dialectal variability in the input in one of the languages (e.g. Lleó, this volume: Chapter 2) for things to really start looking complicated for phonological acquisition. Bilingualism in terms of age and timing of first exposure

The most confounding factor within the field of bilingual development is age. The term refers to biological age, maturational age (brain lateralization and plasticity) and the individual speaker’s own level on the developmental trajectory. Deuchar and Quay (2000: 1) characterize bilingualism as the ‘acquisition of two languages in childhood’. Early bilingualism is bred within the family in an endogenous setting and ‘depends upon the family for encouragement if not for protection’ (Fishman, 1965: 71). Swain (1972) is cited fi rst proposing that a child’s two languages in bilingualism develop as if they were both a fi rst language. ‘Simultaneous’ and ‘successive’ language acquisition are terms devised by McLaughlin (1978) and are extensively employed (e.g. Hoff et al., 2011; the majority of chapters in this volume) with regard to child bilingualism

6

An Anthology of Bilingual Child Phonology

distinguishing between one set of languages that develop concurrently and another whereby one language succeeds the other. ‘Bilingual first language acquisition’ is a technical term devised by Meisel (1989), which also refers to the simultaneous acquisition of two languages with exposure to both of them before 3 years of age (see also Genesee, 1989; McLaughlin, 1978, 1995; Montrul, 2008) or one with exposure by at least the one-word stage (Bhatia & Ritchie, 2006). Padilla and Lindholm (1984) propose that exposure to one of the languages much later than birth ought to be regarded as ‘consecutive’ or ‘successive’ language acquisition. De Houwer (e.g. 1995, 2009) also espoused this point and introduced the acronym BFLA, standing for Meisel’s (1989) ‘bilingual first language acquisition’ to also refer to simultaneous bilingualism with exposure to the two languages by, at the latest, a week from birth. De Houwer (2009), additionally, proposed the terms ‘bilingual second language acquisition’ (BSLA), referring to exposure to one of the languages well after the first month but by the second birthday, and to ‘early second language acquisition’ (ESLA) which means exposure to a second language at a later stage ‘with some regularity over and above their L1 (e.g. through day care and preschool)’ (De Houwer, 2009: 2). The terms BSLA and ESLA match what is vaguely referred to in the literature as ‘early L2’. As seen, the breaching point between simultaneousness and successiveness is overall unclear and not distinctly resolved with empirical evidence to date (e.g. De Houwer, 1995; Hua & Dodd, 2006). ‘Dual’ language acquisition (e.g. Keshavarz & Ingram, 2002) is another synonym of simultaneous bilingualism that is being used broadly (e.g. Genesee, 2009; Hoff et al., 2011; Paradis et al., 2011, etc.), and that seems to be interchangeably used with ‘bilingualism’ by ASHA. ‘Sequential’ bilingualism is also used as a synonym for successiveness, meaning acquisition of the L2 either in childhood (aged 4–12 years; ‘early sequential bilingualism’) or in post-puberty and adulthood (‘late sequential bilingualism’) (McLaughlin, 1978). The term is very common these days (e.g. see Cilibrasi & Tsimpli; Weiss; Alkhudidi et  al.; Miller & Cardenas, this volume: Chapters 6, 7, 8, 11, respectively, using ‘sequential’, as compared to Llama & López-Morelos, this volume: Chapter 9, using ‘successive’). Bilingualism in terms of the status of the languages

Bilingualism is an asset with social-cultural and cognitive advantages that ought to be fostered (e.g. Genesee, 2009; McLaughlin, 1995). Language context and content and what determines language choice in social and situational contexts are equally important correlates in bilingualism. Language ‘context’ refers to the environment in which bilingual children experience each language, such as at home, school, other social contexts, etc., and language ‘content’ refers to the interactional context

Introduction 7

(who the interlocutor is), as well as the quality (grammatical accuracy and complexity of input) and quantity (how much exposure) of the linguistic input (e.g. Kohnert, 2008). Language context and content are underlying factors in measuring performance in all bilingual studies presented in this volume. The acronyms ‘L1/L2’ (e.g. Meisel, 2007) have been used alongside the terms ‘dominant/weak’ (e.g. Yip & Matthews, 2007) to distinguish between the two languages, because it is known that in bilingualism one language is generally more dominant than the other (e.g. Genesee, 2009). The term ‘dominant’ represents the stronger language although it may be an inaccurate term, as proficiency is often context-specific (e.g. Kohnert, 2008); a case in point is the study by Babatsouli (this volume: Chapter 5). Social interaction based on language is a prerequisite for language development (Trevarthen & Aiken, 2001). The ‘microsociological’ and ‘macrosociological’ aspects of bilingualism show that ‘societal concepts such as language vitality, ethnicity … have to use Hakuta’s words (1986: 192) “psychological reality as concepts in bilinguals”’ (Wei, 2007: 44) and are affected by domain variance. To Roeper (1999), bilingualism and ‘[i]ts cousins, dialects, interlanguage, foreign language, and speech register, all remain important social terms, but unclear theoretical terms’. Thus, dialect is defined as ‘a variation of language spoken by a group of speakers specific to a geographic region, socioeconomic factors, class, ethnicity, and/or educational background … People who are able to speak two languages effectively and code-switch [defi ned below] appropriately are considered bidialectal’ (Hoffer, 2007: 76). The importance of bidialectal and dialectal considerations in the study of bilingualism (e.g. FabianoSmith et al., 2014) is represented in this volume by Alkhudidi et al. (on bilinguals in English and a dialect of Arabic (Palestinian, Jordanian, or Saudi Arabian)), Petinou and Taxitari (on Cypriot Greek), and Miller and Cardenas (on the /s/ lenition of certain Spanish varieties) (this volume: Chapters 8, 10 and 11, respectively). The presence or absence of the community speaking what is a bilingual child’s L2 (or D2, standing for second dialect) accounts for the difference between endogenous and exogenous ‘bilinguality’ (Pienemann & Keßler, 2007). Where exogenous bilingualism takes place, it is likely that there is interlanguage ambiguity in the input (Paradis, 2000), in the sense that the bilingual person acquires his/her L2 via exposure to non-native (foreign, accented) input; it is not easy to determine how common this is, but if we are permitted to venture a conjecture, we would say that it is quite prevalent. De Houwer (2009: 158) says that children ‘do not hear accents, they hear people! And people do not always talk exactly the way that you would expect based on phonological descriptions for a particular language’. Hakuta (1986) states that the narrow view of bilingualism signifying native competence is less preferable, since only a few bilinguals have native-like control in both languages. With these arguments in mind, Grosjean’s (1989) definition of bilingualism as the everyday

8

An Anthology of Bilingual Child Phonology

use of two languages irrespective of native-like fluency seems to make sense. Babatsouli (this volume: Chapter 5) reports on such a case study on exogenous bilingualism with interlanguage input. Among the earliest identifi able sources discussing foreign accent is Nikolai Trubetzkoy (1939). It is notable that the field of child bilingual phonology has conventionally shied away from employing this less politically correct term (though not always so, e.g. Lleó; Llama & LópezMorelos; Nenonen, this volume: Chapters 2, 9, 4, respectively) which is, on the other hand, widely associated with interlanguage phonology (e.g. Major, 2001; Selinker, 1972; Simonet, 2014), or presented in clinical contexts to differentiate from impairment (e.g. ASHA, 2019; Hack et al., 2013). An everyday term, ‘accent’ may be synonymously used to replace more formal terms found in language research, such as ‘transfer’ (e.g. Major, 2008), ‘interaction’ (e.g. Paradis & Genesee, 1996) and ‘crosslinguistic interference/influence (CLI)’ (e.g. Flege & Port, 1981; Jarvis & Pavlenko, 2008), typically referring to ‘incorrect evaluations of the nonnative sounds, which in turn are caused by the differences between the phonological structure of the foreign language and the mother tongue of the speaker (Trubetzkoy [1939] 1969: 55) [the italics were quotations in the original]’ (Simonet, 2016: 1–2). Interaction in bilingualism is bidirectional between the languages, it may have either a positive and/or a negative effect on the languages involved (the negative effect being to the point of fi rst language attrition, e.g. Bardovi-Harlig & Stringer, 2010), it may even be absent or minimal at certain levels/contexts, or it may result from interlanguage ambiguity in the input, as mentioned in the previous paragraph. Research fi ndings have permitted elaboration on the meanings of these terms, and of other terms related to or ensuing from them, thus creating further semantic gradience (see reviews in, for example, Kehoe, 2018; Lleó, this volume;). Chapters 2 and 3 in this volume by Lleó and Kehoe, respectively, are two contributions here in this direction of research. It has been claimed by Odlin (2008) that existing defi nitions of transfer and interference should not be considered foolproof. Language mixing in the form of code-switching (CS) is the linguistic manifestation of the bilingual brain’s activation levels (e.g. Grosjean, 2008) that is evident in bilinguals, native-like second language speakers, and monolinguals of two speech registers (i.e. the standard language and the dialect, spoken and signed language). Code-switching refers to the intentional or unintentional alternation between two languages at phrase (‘intra-utterance code-mixing’) or sentence (‘inter-utterance code-mixing’) level (Genesee et al., 2008) during discourse among people with variable skills in the languages involved. Thus, code-switching is the manifestation of interference on the conversational level (Babatsouli, 2013: 47) as much as accent is. A viable link between code-switching and phonological/phonetic transfer known for a while to exist (e.g. Pardo,

Introduction 9

2006; Treffers-Daller, 2009). Lleó (this volume: Chapter 2) further elaborates on this theme. That language transfer and code-switching produce distinctive forms of learner language is also evident in several example blends of the names of the two languages in bilingual (in English-Other language) speakers, as in Chinglish (Chinese-English), Czenglish (Czech-English), Greeklish, etc. A much less explored field in bilingual research involves bimodal bilingualism, referring to hearing individuals that are bilingual in a signed and a spoken language (e.g. Emmorey & McCullough, 2009). Given that sign (gestural/manual) language is known to display by and large the same fundamental properties as spoken language (in that sign language exploits visual-spatial contrasts on all grammatical levels; e.g. Emmorey, 2015), bimodality provides a unique window into the neurocognitive changes that occur with the acquisition of two languages (registers). Bilingual children are very sensitive to the language behavior of the adults they are with (Genesee et al., 1996) and to language context (Lleó & Kehoe, 2002). They distinguish their languages before their first birthday and they match them to the interlocutors by their second birthday (e.g. Genesee et al., 1996; Maneva & Genesee, 2002). With regard to the ‘language choice’ of bilinguals, Fishman (1965) identifies three controlling factors: ‘reference group membership’ (with both objective and subjective sociopsychological criteria); ‘situation’ (that is, circumstances at the time of communication that include code-switching); and ‘topic’ as a regulator of language use in multilingual settings. Reference group (e.g. society) is more important to bilinguals than situation variance (e.g. family) or intimacy (e.g. role relations). Studies (e.g. Marian & Spivey, 2003) have shown that the two languages in bilinguals are constantly competing for attention. Grosjean (2001) had earlier identified this to be the ‘language mode continuum’. That is, ‘in their everyday lives, bilinguals fi nd themselves at various points along a situation continuum which induce different language modes’, with the ‘base’ language (L1) often impacting the ‘guest’ language (L2) at variable levels of the ‘base language effect’ (Macnamara & Kushnir, 1971) and vice versa. The move from base to guest mode, fluctuating back and forth along this continuum, is facilitated by a process of brain activation (Green, 1986). Green viewed the bilinguals’ control of the two languages in terms of varying levels of ‘activation’: ‘selected’ (controlling speech output); ‘active’ (an ongoing mental processing but no access to the speech production); and ‘dormant’ (stored in long-term memory but not surfacing in ongoing processing or production). Differences in environmental exposure to a language may account for the functional modulation in the bilingual’s brain (e.g. Perani et al., 2003). Numerous studies have shown that child bilingualism provides a brain boost and that bilinguals outperform monolinguals in various cognitive tasks (e.g. De Lange, 2012, and references therein).

10

An Anthology of Bilingual Child Phonology

Bilingualism in terms of degree of competence

Balance, fluency and native-like ability have equally concerned the field of bilingualism towards determining a definition. Edwards (2006: 7) states that bilinguals ‘differ in terms of degree’; her contention, at one end of the argument, is that if someone knows just a couple of words in a foreign language, this alone indicates ‘some command’ of that language and, as a result, ‘everyone is bilingual’. At the other end, the argument (as in SLA) is that the ‘true’ bilingual is equally fluent in both languages and that only balanced competence makes for a real bilingual (e.g. Bloomfield, 1933; Thiery, 1978). In this position, the fact that ‘not all bilingual exposure results in active bilingualism’ (De Houwer, 2002) would have no grounding at all. On the other hand, ‘passive’ bilingualism is an established phenomenon in the field (e.g. De Houwer, 2002, and references therein) and results when linguistic competence assumes receptive but not productive knowledge (Nation, 2001); that is, the child comprehends speech in both languages but consciously refuses to speak the weaker one. Bilingual households may neither compel bilingual development, as passive bilingualism attests (e.g. De Houwer, 1995; von Raffler-Engel, 1965), nor assure bilingualism. Bilingual children are known to eradicate the use of one language when linguistic and social settings shift (e.g. Major, 1977). Attrition in young children’s L1 (e.g. Kaltsa et al., 2015; Tsimpli, 2007) as well as incomplete acquisition in bilingualism (e.g. Montrul, 2008) are known facts. Grosjean (1989) postulates that, for successful bilingual acquisition to ensue, ‘the critical factor is need’. His holistic view of bilingualism advocates the idea that the bilingual speaker should not be considered as the sum of ‘two complete or incomplete monolinguals’, that is, as having ‘two separate and isolable language competencies’, but as someone ‘with a unique and specific linguistic configuration’ (Grosjean, 1985: 470). The argument is that bilinguals should not be described and evaluated in terms of fluency and balance in their languages and the monolingual speaker ought not to be the model of the ‘normal speaker-hearer’ (Grosjean, 1985: 470) against which bilingualism must be measured. This viewpoint, also hinted at earlier in Oksaar (1983), is widely accepted in the literature of bilingualism (e.g. De Houwer, 2007; Meisel, 2004; Norbert, 2011; Schlyter, 1993). Bilinguals are argued to be seldom balanced; there is an ‘ebb and flow’ in bilingualism (McLaughlin, 1995) as bilinguals show a trade-off between L1 and L2 proficiency (Grosjean, 1985). In ‘early bilingualism’, in particular, ultimate attainment is open ended (Gass & Selinker, 2008: 27) because bilinguals cannot always have the enormous amount of practice (De Keyser & Larson-Hall, 2005: 97) required due to the very circumstances of their bilingualism. Some bilinguals have proficient oral and writing skills in both languages, others may speak but are not educated in one of the languages, and others use one of the languages in certain environments such as only at home, only at work,

Introduction 11

or only with a group of friends. ‘The complementarity principle’ (Grosjean, 1989) demonstrates this, arguing that the languages in bilingualism are acquired for different purposes, with different people and in different situations. As a result, fluency in the languages is domain specific and will depend on the need for that language. The ‘weaker language hypothesis’ (Schlyter, 1993) addresses the issue of whether the weaker (non-dominant) language of simultaneous bilingual acquisition comes to resemble a second language. Affective factors (e.g. McLaughlin, 1995), cognition and personality, attention, motivation, memory and language ability also influence linguistic competence (e.g. Perani et al., 1998). Cook (1992) argues that the bilingual’s multicompetence is more than just one of degree, as it involves metalinguistic awareness not present in monolinguals. Valdés (2001) demonstrates the variability in bilingualism visually with a linear bilingual continuum. Here we provide the adapted version in Gass and Selinker (2008: 28): A Ab Ab Ab Ab Ba Ba Ba Ba Ba Ba B, where A represents the monolingual speaker of one language and B the monolingual speaker of the other language. Bilinguals of the two languages fall between the two in varying degrees of aptitude. Such a linear representation of the varying levels of aptitude in bilinguals, however, is relatively incomplete and one-dimensional for the simple reason that aptitude for each of these bilinguals in the line is not strictly a constant. Grosjean’s ‘language mode continuum’ and Green’s activation levels underline the inherently unsteady interplay of the languages in the bilingual person that are, by default, in flux not only constantly competing during daily language-mode shifts but also due to dominance variance in the lifespan. Bilingualism and impairment

Disparities between monolingual and bilingual acquisition initially led to the impression that bilingualism is a disadvantage for language acquisition and it was customarily identified as instances of ‘confusion’, speech/language impairment or delay (e.g. Genesee, 2009). This has recently been shown to be false and it is argued that, in fact, the opposite holds true, i.e. in terms of a bilingual advantage as reported earlier (e.g. De Lange, 2012). Despite vibrant work in this direction (e.g. see ASHA, 2019; Babatsouli, 2019; Babatsouli & Kehoe, 2018; Babatsouli et al., 2017; Battle, 2012; De Jong, 2009; Goldstein & Oller, 2012; Kraemer & Fabiano-Smith, 2017; McLeod & Goldstein, 2012; McLeod & Verdon, 2017; Paradis et al., 2011; Tsimpli et al., 2016), the interaction between bilingualism and speech-language pathology (like SSD, SLI, dyslexia, etc.) and speech delay, also known as protraction (e.g. Bernhardt & Stemberger, 2017), is still largely unexplored. Bilingual considerations in the clinical context include several diversity issues that pertain to dialectal (on national and international levels), multilingual and multicultural

12

An Anthology of Bilingual Child Phonology

diversity, as this affects knowledge, assessment, intervention and literacy in populations with atypical (that is showing impairment or disorder) language development and use (Babatsouli, in press). The present compendium of studies contributes in this arena with articles by Weiss (involving SLI), Petinou and Taxitari, and Bernhardt et al. (involving delay), and the review by Melloni and Vender (involving dyslexia) (this volume: Chapters 7, 10, 12, 13, respectively). A Few Words on the Contents

The aim of this section is not to provide a comprehensive list of studies on bilingual phonological acquisition, but it is worth pointing out that there is much research going on in an increasingly diverse number of language pairs (e.g. see Babatsouli & Ingram, 2018a; Cardoso et al., 2017; Kehoe, 2018; Munro et al., 2005; Tamburelli et al., 2015; Thomas & Mayr, 2010; Thomas & Mennen, 2014, etc.). Less work has been done on early trilingual acquisition (e.g. Oller, 2010; see Wrembel & Cabrelli Amaro, 2017, also the Introduction for discussion), although research in this field is also intensifying. The present anthology contributes to this surge with studies of underrepresented language and dialect combinations, group and individual case studies, and cross-sectional studies, principally novel data and analysis (Chapters 2–12), but also some revisited ones (Chapter 13), as well as sociolinguistic variables such as dialectal considerations, vital for the understanding of bilingualism (Miller & Cardenas, Chapter 11). The entirety of the volume reassesses longstanding themes in bilingual child phonology with a breath of fresh air, it tackles under-investigated problems and it advocates innovative solutions for deep-rooted struggles. Specifically, the book advances recent developments in typical bilingual child phonology (Lleó, Kehoe, Babatsouli, Nenonen, Cilibrasi & Tsimpli, Alkhudidi et al.: Chapters 2–6, 8), also including trilingualism (Llama & López-Morelos: Chapter 9) and bidialectalism (Alkhudidi et al., Petinou & Taxitari: Chapters 8, 10), without excluding instances of atypical child speech (SLI, delay, dyslexia) investigated alongside typically developing groups (Weiss, Petinou & Taxitari, Bernhardt et al., Melloni & Vender: Chapters 7, 10, 12, 13). The opening chapter touches on longstanding themes in phonological acquisition research, such as the various phonological outcomes of bilingualism, especially focusing on transfer and code-switching, and also making a proposal for how to run a code(or language) switching task (Chapter 2). The fi nal chapter reviews phonological processing, memory and awareness as issues affecting literacy, and it advances phonological processing and nonword repetition as a critical tool for the identification of dyslexia in bilingualism. The remaining chapters are original research contributions. The chapters by Cilibrasi and Tsimpli, and Weiss (Chapters 6, 7) examine phonological acquisition in bilingualism at the interface with morphology, while Petinou and Taxitari

Introduction 13

scrutinize the interconnectedness of various linguistic subsystems (Chapter 10). The book also comprises studies on typologically different languages, some underrepresented: Arabic, Czech, English, Finnish, German, Icelandic, French (European, Canadian), Greek (Standard, Cypriot), Hebrew, Italian, Portuguese, Russian, Slovenian, Spanish (Dominican, European, Mexican) and Swedish, a large number of which are involved in the crosslinguistic study of Chapter 12. This last article has been purposely included in the collection for its novel methodological advances, a useful device in the assessment of languages in either monolingual or bilingual/multilingual contexts. The bilingual pairs compared in this book collection are: Arabic-English, Czech-English, French-Romance language, French-Germanic language, German-Spanish, Greek-English, Hebrew-English, Italian-Other and Russian-Finnish; the various dialects involved are: Arabic (Palestinian, Jordanian, Saudi Arabian), English (European, Canadian), French (European, Canadian, Quebec), Greek (Standard, Cypriot) and Spanish (Dominican, Iberian, Mexican). Lastly in this section, some specific remarks on each chapter are presented in the order in which the chapters appear in the volume. Chapter 2: Investigating the Linguistic World of the Bilingual Child: Transfer and Code-switching Conxita Lleó

This chapter examines the various phonological outcomes of bilingualism, especially focusing on transfer and code-switching (CS). Although there are adequate analyses of transfer available, CS lacks even an acceptable working defi nition, because there is as yet no consensus as to whether a 2-year-old can be said to code-switch in the same way as an adult. Or does s/he do something similar to CS when s/he produces codemixes? After considering some predictions about the outcomes of bilingualism, the chapter directly faces CS from a linguistic and acquisition point of view, and concludes by proposing a type of oral (i.e. non-read) test to be run with the purpose of clarifying the very notion of CS using methodology that mimics spontaneous speech. Chapter 3: Seeking Crosslinguistic Interaction in French Bilingual Phonological Development Margaret Kehoe

This study seeks evidence of crosslinguistic interaction across different phonetic/phonological measures and age ranges of children. Specifically, it focuses on syllable structure, voice onset time (VOT) and word prosody in monolingual and bilingual French-speaking children aged between 2;6 and 6 years. Results show evidence of crosslinguistic interaction in the youngest

14

An Anthology of Bilingual Child Phonology

children in the area of syllable structure. Bilingual children whose L1 was characterized by highly complex syllable structure obtained better scores than monolinguals. There were no monolingual–bilingual differences in the use of short-lag voicing, but bilinguals aged 3–6 years produced fewer tokens with lead voicing. There was little influence of bilingualism on word prosody. Overall, crosslinguistic interaction was less present in the data than had been predicted. The chapter discusses the relevance of these results for the understanding of crosslinguistic interaction. Chapter 4: Case Studies of Phonological Development in Six Preschool-aged Russian-Finnish Bilingual Children Olga Nenonen

The purpose of this chapter was to assess the phonological performance of Russian-Finnish bilingual children in a longitudinal study investigating the evidence from six typically developing preschool-aged bilingual children. The data, collected using an articulation test in a 2.5year period, were compared to control group data of Russian monolinguals, Finnish monolinguals and bilinguals with specifi c language impairment (SLI). The focus is on: universals and individual variation in the bilinguals; bilinguals’ pronunciation errors; and acceleration or deceleration of bilingual phonetic development in comparison to monolinguals. Despite considerable individual variation and initial similarities with monolingual Russian and Finnish speakers, the bilinguals’ phonological development is slower, with error patterns specific to their bilingualism as well as errors resembling those of children with SLI. The study provides evidence of language interaction in the form of phonetic transfer, and of both a deceleration and an acceleration in the acquisition of the Russian and Finnish phonetic systems. Chapter 5: Enhanced Phonology in a Child’s Weaker Language in Bilingualism: A Portrait Elena Babatsouli

Does language dominance in morphology guarantee dominance in phonology in the same language? This question is addressed in the present study by examining a bilingual child’s speech data in English and Greek, two typologically different languages, over one month at the age of 2;7. The child’s speech was digitally recorded during daily interactions with the author and the data were subsequently entered in a CLAN database in orthographic and phonetic transcription. The data comprise 785 child utterances in English with 1516 word tokens, and 690 child utterances in Greek with 2374 word tokens. It is noted that the MLU in morphemes cannot be compared between the two languages because Greek is a morphologically richer language. The child’s MLU in words in Greek is 3.36

Introduction 15

while in English it is 2.01, indicating that Greek is the child’s morphologically dominant language. Nevertheless, the child’s phonological level (assessed by the accuracy of singletons, syllables and word length) in the two languages is the same despite their segmental and phonotactic differences. The results suggest that the morphologically dominant language boosts the child’s phonological performance in the weak language, which is phonologically as strong as the dominant language. Chapter 6: Sensitivity to Morphophonological Cues in Monolingual and Bilingual Children: Evidence from a Nonword Task Luca Cilibrasi and Ianthi Tsimpli

This study investigates whether bilingual children, like monolinguals studied earlier in the literature, are equally sensitive to the phonological strings that cue the presence of morphology in nonwords. A group of 45 bilingual children (Czech L1 and English L2, aged 8–11) and a group of 26 monolingual controls (aged 7–11) were presented with a same/different minimal pairs discrimination task with nonwords in which the items did or did not contain phonological cues to morphological structure. An analysis of the interaction between group (early, early sequential, late sequential bilinguals), type and condition shows that only simultaneous bilinguals approximated monolingual performance, while the other two groups did not. This fi nding is in line with previous claims showing the relevance of early exposure for the acquisition of English inflectional morphology and situates English morphological acquisition as a late phenomenon, depending primarily on input in order to be acquired by bilingual children. This crucially contrasts with the acquisition of inflectional morphology in richer languages that appears, instead, to be an early phenomenon in both L1 and child L2 acquisition. Chapter 7: Lexical-semantic Organization in Monolingual and Bilingual Hebrew Speaking Children: Evidence from a Word Association Task Atalia Hai Weiss

The aim of this study is to investigate the distribution of syntagmatic and paradigmatic as well as morphological responses of Hebrew speaking children in a word association task, commonly used as a means of inspecting lexical organization in children. The chapter examined the pattern of responses of 24 monolingual and 17 bilingual, 6–9 year-old, Hebrew speaking children. Initially, naming abilities were examined using a standardized naming task and, as expected, the average score of the bilingual children was significantly lower than that of their monolingual peers, and lower than expected monolingual norms. However, the pattern of associative responses was on average similar between the two groups: about 80%

16

An Anthology of Bilingual Child Phonology

of the responses were semantic, half syntagmatic and half paradigmatic. By contrast, the pattern of semantic responses was not found to be correlated with age; thus no clear evidence of the syntagmatic–paradigmatic shift was found. Importantly, only 3% of the responses were morphologically derived ones. Overall, the fi ndings suggest that a word association task can be a useful measure for inspection of the lexical-semantic organization of Hebrew speaking children, although the unique characteristics of Hebrew cannot be disregarded. Chapter 8: The Production of Marked Arabic Consonants by Arabic-English Children Living in Canada Anwar Alkhudidi, Yasmeen Hakooz, Madeline Walker, Ryan Stevenson and Yasaman Rafat

This study examines whether early exposure to English as L2 interferes with the acquisition of Arabic as the heritage language in ArabicEnglish speaking children living in Canada. Specifically, the production of the following L1-Arabic contrasts is investigated, not present in English, /t/-/t/, /d/-/d/, /s/-/s/, /h/-/ħ/ and /k-q/, in five Arabic-English bilingual children aged 8–13, who performed a word repetition task. The stimuli comprised 60 minimal or near-minimal pairs. Data were analyzed acoustically and the degree of F2 lowering in the previous vowels, a correlate of emphatics, pharyngeals and velars, was measured. While no significant effect of age or position was found, there was a significant effect of F2 lowering in bilingual children for all the marked sounds tested. Moreover, the degree of F2 lowering was constrained by consonant type. Importantly, the results suggest that bilingual Arabic-English speaking children may establish L1 categories that are different from those of monolingual children. Furthermore, the acquisition of a primary place of articulation may be easier than the acquisition of a secondary place of articulation. Chapter 9: On Heritage Accents: Insights from Voice Onset Time Production by Trilingual Heritage Speakers of Spanish Raquel Llama and Luz Patricia López-Morelos

This study extends heritage speaker (HS) research with a focus on trilingual children and (pre-)adolescents. Given that several studies suggest a correlation between voice onset time (VOT) and degree of foreign accent, the acquisition of VOT patterns is an interesting candidate for investigation in heritage phonetics and phonology (HPP). Research on Spanish HPP has often been carried out with bilinguals. The participants in this study spoke Spanish at home and were schooled in English (the community language) and French. The study explores the extent to which these early trilinguals produce native-like VOT patterns in their heritage language, as well as the interplay among their three sound systems. To this end, their

Introduction 17

speech was recorded while naming three sets of pictures (in English, Spanish and French) which led to the production of voiceless stops in stressed word-initial position. The results show monolingual-like production in Spanish. However, a fi ne-grained comparison to an age-matched Mexican control group also uncovered some isolated deviations from the monolingual norm, which could reveal traces of a ‘heritage accent’. Chapter 10: Linguistic Outcomes and the Role of Phonology in Typically Developing and Late Talking Toddlers Kakia Petinou and Loukia Taxitari

This investigation examines the existence of interconnectedness between developing linguistic subsystems (phonetics, lexicon and grammar) in bidialectal Cypriot-Greek toddlers. Spontaneous speech samples were collected from 31 typically developing (TD) toddlers across two age levels, at 28 and 36 months. Correlational analyses were performed synchronically (within ages) and predictively (across ages). Synchronic data revealed a significant positive relationship among all three language skills at 28 and 36 months. Predictive analyses showed that phonetic skills at 28 months predicted all skills at 36 months (phonetic, lexical, grammatical), while grammatical skills at 28 months only showed predictive value for grammatical skills at 36 months. In addition, a cluster analysis on the basis of performance on each individual skill revealed different linguistic profiles between groups: low performers showed multiple interactions between levels similar to the all-participants analysis, while high performers only showed minimal interactions between skills within and across ages, exhibiting ceiling effects. All in all, the current results reveal complex interdependencies among the different language skills examined both within and across age levels, with children of the same age level exhibiting variable linguistic profiles. Chapter 11: Stylistic Patterns in the Speech of Young Children and their Caregivers: A Study of Variable /s/ Lenition in Dominican Spanish Karen Miller and Rodrigo Cárdenas

This chapter presents a study on children’s acquisition of sociolinguistic variation, focusing specifically on when children acquire the variable patterns that are present in child-directed speech and how gender and speech style impact children’s use of variable forms. The empirical domain of the study is syllable final /s/ lenition in Dominican Spanish. Conversational data were collected from children and their caregivers while they played together in a playroom setting. Children also participated in a more formal repetition and retell task at school, which was administered by a native speaking research assistant. Tokens of syllable final /s/ were coded for pronunciation, and a variable-rule analysis examining the effect of various linguistic and

18

An Anthology of Bilingual Child Phonology

extralinguistic constraints was carried out. The results show that children’s variable production of /s/ lenition is similar to what is found in their caregiver’s speech, and that caregiver speech is consistent with what has been reported in previous studies on adult-to-adult speech. In addition, speech style and gender impact children’s variable /s/ lenition use, in that both girls and boys produce the stigmatized variant less often in school, and girls do this much more often than boys. Chapter 12: Identification of Protracted Phonological Development across Languages – The Whole Word Match and Basic Mismatch Measures Barbara May Bernhardt, Joseph Paul Stemberger, Daniel Bérubé, Valter Ciocca, Maria João Freitas, Diana Ignatova, Damjana Kogošek, Inger Lundeborg Hammarström, Thóra Másdóttir, Martina Ozbič, Denisse Perez and A. Margarida Ramalho

This chapter explores the potential utility of two measures that may be applied in identifying protracted phonological development (PPD) in monolingual and multilingual children: (1) a simple accuracy measure, Whole Word Match (WWM: yes-no congruence of adult and child productions of a word); and (2) for more borderline cases (not clearly typically developing, TD or with PPD), a composite mismatch measure (based on consonant deletion, vowel changes, consonant substitutions). Data are presented for eight languages: German, Icelandic, Swedish, Canadian French, European Portuguese, Granada Spanish, Bulgarian and Slovenian, comprising phonetically transcribed single-word elicitations of about 100 words per sample by child (full lists) and, for all but German, subsets of the full lists (screening probes). Statistical programs supported analysis which included a Bayesian analysis for Bulgarian as a preliminary statistical exploration of WWM. Screening and full word lists were compared within language for groups and individual children. Results showed the overall relevance of WWM as an identifier of PPD across languages (agreement with the original TD/PPD classification for 325/333 children) and similar levels of WWM by age across languages. Mismatch measures disambiguated most of the few borderline cases. The chapter concludes with implications for future research and clinical applications. Chapter 13: Phonological Processing and Nonword Repetition: A Critical Tool for the Identification of Dyslexia in Bilingualism Chiara Melloni and Maria Vender

This chapter deals with phonological processing and its interplay with literacy acquisition across children with typical and atypical development, and in sequential bilingual children. Because phonological processing is impaired in subjects with dyslexia, it is argued that phonological

Introduction 19

processing skills could be profitably assessed in sequential bilingual children misdiagnosed with dyslexia, while actually showing a still immature L2 competence. In particular, among the tasks assessing phonological memory skills, nonword repetition emerges as a critical tool for an early and more reliable assessment of dyslexia in bilinguals. The fi rst section of the chapter presents the main components of phonological processing and emphasizes its relevance for the acquisition of reading and spelling in alphabetic writing systems. The second section focuses on dyslexia, overviews research that underscores the limitations of phonological processing in this population and explains reading impairments as the result of these deficits. The third section explores the role of phonological awareness in several studies of early bilinguals and tries to reconcile apparently disparate results under a unitary hypothesis. The fi nal section focuses on nonword repetition, reporting the promising results of a recent study conducted on L2-Italian bilingual children with and without dyslexia.

Final Statement

This Introduction has presented the topic, the need and the organization of this book on child bilingualism. The intention of this collection of essays was to bring together child language studies with themes pertaining to bilingual phonological development. Because the impact of one language/dialect on another is of paramount importance in the study of bilingualism, the present volume has deemed crosslinguistic, crossdialectal and sociolinguistic variability research to be an essential backbone for investigating bilingualism. To this end, and with comparative linguistics in mind, the volume has included work on typical and atypical acquisition, different types of bilingualism (endogenous and exogenous, heritage language, dialectal and bidialectal variation) and trilingualism, and also encompasses two studies that target crosslinguistic and sociolinguistic variation comparisons even though the study participants are not bilinguals (Chapters 10 and 11). Evidence from a wide range of the language pairs investigated is brought to bear on several grammatical issues pertaining to phonology/phonetics and interfaces with morphology and semantics, and on pertinent enduring concerns like transfer, interaction, accent, code-switching/mixing, language dominance and performance, the role of context and input, as well as methodological resources for the assessment of child speech – normal, delayed or with impairment. Overall, the bilingual theme of this book brings together divergent papers, asks questions and seeks to formulate answers on protolanguage and its implications for interlanguage – a key data catering service for language acquisition research.

20

An Anthology of Bilingual Child Phonology

References Archibald, J. (ed.) (1995) Phonological Acquisition and Phonological Theory. Hillsdale, NJ: Lawrence Erlbaum. Aronin, L. and Hufeisen, B. (2009) On the Genesis and Development of L3 Research, Multilingualism and Multiple Language Acquisition. Amsterdam/Philadelphia, PA: John Benjamins. ASHA (American Speech-Language-Hearing Association) (2019) Speech Sound Disorders. See https://www.asha.org/public/speech/disorders/speech-sound-disorders/. Auer, P. and Wei, L. (eds) (2007) Handbook of Multilingualism and Multilingual Communication. Berlin/New York: Mouton de Gruyter. Babatsouli, E. (2013) Phonological development of a child’s L2 English in bilingualism. Unpublished PhD thesis, University of Crete. See https://phdtheses.ekt.gr/eadd/ handle/10442/35165. Babatsouli, E. (2016) Chaos in monolingual and bilingual speech. In C.H. Skiadas and C. Skiadas (eds) Handbook of Applications of Chaos Theory (pp. 789–890). Boca Raton, FL: CRC Press/Taylor & Francis. Babatsouli, E. (2019) A phonological assessment test for child Greek. Clinical Linguistics and Phonetics 33 (7), 601–627. doi:10.1080/02699206.2019.1569164. Babatsouli, E. (ed.) (2020) On Under-reported Monolingual Child Phonology. Bristol: Multilingual Matters. Babatsouli, E. (in press) Diversity considerations in speech and language disorders … a focus on training. In J.S. Damico, N. Müller and M.J. Ball (eds) Handbook of Language and Speech Disorders. Hoboken, NJ: Wiley-Blackwell. Babatsouli, E. and Ball, M.J. (2019) Editorial. Inaugural issue of the Journal of Monolingual and Bilingual Speech 1, 1–6. Babatsouli, E. and Ingram, D. (2018a) Prologue. In E. Babatsouli and D. Ingram (eds) Phonology in Protolanguage and Interlanguage (pp. 1–23). Sheffield: Equinox. Babatsouli, E. and Ingram, D. (eds) (2018b) Phonology in Protolanguage and Interlanguage. Sheffield: Equinox. Babatsouli, E. and Kehoe, M. (guest eds) (2018) Special issue in honour of M.J. Ball on ‘Phonological and phonetic studies of speech disorders across languages’. Clinical Linguistics and Phonetics 32 (12), 1067–1070. doi:10.1080/02699206.2018.1510981 Babatsouli, E., Ingram, D. and Müller, N. (eds) (2017) Crosslinguistic Encounters in Language Acquisition: Typical and Atypical Development. Bristol: Multilingual Matters. Baetens Beardsmore, H. (1982) Bilingualism: Basic Principles. Clevedon: Multilingual Matters. Ball, M.J. and Müller, N. (eds) (2016) Challenging Sonority: Cross-linguistic Evidence. Sheffield: Equinox. Ball, M.J., Perkins, R., Müller, N. and Howard, S. (2009) The Handbook of Clinical Linguistics. New York: Wiley/Blackwell. Bardovi-Harlig, K. and Stringer, D. (2010) Variables in second language attrition: Advancing the state of the art. Studies in Second Language Acquisition 32, 1–45. Battle, D. (2012) Communication Disorders in Multicultural and International Populations (4th edn). St Louis, MO: Elsevier. Benet, A., Cortés, S. and Lleó, C. (2012) Phonoprosodic corpus of spoken Catalan (PhonCAT). In T. Schmidt and K. Worner (eds) Multilingual Corpora and Multilingual Corpus Analysis (pp. 215–229). Hamburg Studies on Multilingualism No. 14. Amsterdam/Philadelphia, PA: John Benjamins. Bernhardt, B.M. and Stemberger, J.P. (1998) Handbook of Phonological Development: From the Perspective of Constraint-based Nonlinear Phonology. Bingley: Emerald Group.

Introduction 21

Bernhardt, B.M. and Stemberger, J.P. (2017) Investigating typical and protracted phonological development across languages. In E. Babatsouli, D. Ingram and N. Müller (eds) Crosslinguistic Encounters in Language Acquisition: Typical and Atypical Development (pp. 71–108). Bristol: Multilingual Matters. Bhatia, T.K. and Ritchie, W.C. (2006) The bilingual child: Some issues and perspectives. In W.C. Ritchie and T.K. Bhatia (eds) Handbook of Child Language Acquisition (pp. 569–643). San Diego, CA: Academic Press. Bialystok, E. (2001) Bilingualism in Development: Language, Literacy, and Cognition. Cambridge: Cambridge University Press. Bloomfield, L. (1933) Language. New York: Holt. Bratt Paulston, C. (1998) International Handbook of Bilingualism and Bilingual Education. Westport, CO: Greenwood Press. Brisk, M.E. and Harrington, M.M. (2006) Literacy and Bilingualism: A Handbook for All Teachers. Mahwah, NJ: Lawrence Erlbaum. Cabrelli Amaro, J., Flynn, S. and Rothman, J. (2012) Third Language Acquisition in Adulthood. Studies in Bilingualism No. 46. Amsterdam/Philadelphia, PA: John Benjamins. Cardoso, W., Kehoe-Winkler, M. and Yavas, M. (2017) Romance-Germanic Bilingual Phonology. Sheffield: Equinox. Cenoz, J. and Genesee, F. (2001) Trends in Bilingual Acquisition. Amsterdam/ Philadelphia, PA: John Benjamins. Cook, V. (1992) Evidence for multi-competence. Language Learning 42 (4), 557–591. Crystal, D. (1995) The Cambridge Encyclopedia of the English Language (2nd edn). Cambridge: Cambridge University Press. De Houwer, A. (1995) Bilingual language acquisition. In P. Fletcher and B. MacWhinney (eds) The Handbook of Child Language (pp. 219–250). Oxford: Blackwell. De Houwer, A. (2007) Parental language input patterns and children’s bilingual use. Applied Psycholinguistics 28, 411–424. De Houwer, A. (2009) An Introduction to Bilingual Development. Bristol: Multilingual Matters. De Houwer, A. (2002) Comparing monolingual and bilingual acquisition. Alkalmazott Nyelvtudomány 2 (1), 5–19. De Jong, J. (2009) Bilingualism and language impairment. In M.J. Ball, R. Perkins, N. Müller and S. Howard (eds) The Handbook of Clinical Linguistics. New York: Wiley/Blackwell. De Keyser, R. and Larson-Hall, J. (2005) What does the critical period really mean? In J.  Kroll and A. De Groot (eds) Handbook of Bilingualism: Psycholinguistic Perspectives (pp. 88–108). Oxford: Oxford University Press. De Lange, C. (2012) Bilingual brain boost: Two tongues, two minds. New Scientist 2863. Deuchar, M. and Quay, S. (2000) Bilingual Acquisition: Theoretical Implications of a Case Study. Oxford: Oxford University Press. Doughty, C.J. and Long, M.H. (2005) The Handbook of Second Language Acquisition. Malden, MA: Blackwell. Edwards, J. (2006) Foundations of bilingualism. In T. Bhatia and W. Ritchie (eds) The Handbook of Bilingualism (pp. 5–29). Oxford: Blackwell. Emmorey, K. (2015) The neurobiology of sign language. In A.W. Toga (ed.) Brain Mapping: An Encyclopedic Reference (Vol. 3, pp. 475–479). Oxford: Elsevier/ Academic Press. Emmorey, K. and McCullough, S. (2009) The bimodal bilingual brain: Effects of sign language experience. Brain and Language 109, 124–132. Fabiano-Smith, L., Shuriff, R., Barlow, J.A. and Goldstein, B. (2014) Dialect density in bilingual Puerto Rican Spanish-English speaking children. Linguistic Approaches to Bilingualism 4 (1), 34–60.

22

An Anthology of Bilingual Child Phonology

Fishman, J.A. (1965) Who speaks what language to whom and when? La Linguistique 2, 67–88. Flege, J. and Port, R. (1981) Cross-linguistic phonetic interference: Arabic to English. Language and Speech 24, 125–146. Gass, S.M. and Selinker, L. (2008) Second Language Acquisition: An Introductory Course. New York: Routledge. Genesee, F. (1989) Early bilingual development: One language or two? Journal of Child Language 6, 161–179. Genesee, F. (2009) Early childhood bilingualism: Perils and possibilities. Journal of Applied Research on Learning 2 (2), 1–21. Genesee, F., Boivin, I. and Nikoladis, E. (1996) Talking with strangers: A study of children’s communicative competence. Applied Psycholinguistics 17, 427–442. Genesee, F., Paradis J. and Crago, M. (2008) Dual Language Development & Disorders: A Handbook on Bilingualism & Second Language Learning. Baltimore: Paul Brookes. Goldstein, B. (1999) Cultural and Linguistic Diversity Resource Guide for Speechlanguage Pathologists. Boston, MA: Cengage Learning. Goldstein, B. and Oller, D. (2012) Bilingual Language Development and Disorders in Spanish-English Speakers. Baltimore, MA: Paul Brookes. Grech, H. and McLeod, S. (2012) Multilingual speech and language development and disorders. In D. Battle (ed.) Communication Disorders in Multicultural and International Populations (4th edn) (pp. 120–147). St Louis, MO: Elsevier. Green, D.W. (1986) Control, activation, and resource. Brain and Language 27, 210–223. Grosjean, F. (1985) The bilingual as a competent but specific speaker-hearer. Journal of Multilingual and Multicultural Development 4 (6), 467–477. Grosjean, F. (1989) Neurolinguists, beware! The bilingual is not two monolinguals in one person. Brain and Language 36, 3–15. Grosjean, F. (2001) The bilingual’s language modes. In J. Nicol (ed.) One Mind, Two Languages: Bilingual Language Processing (pp. 1–22). Oxford: Blackwell. Grosjean, F. (2008) Studying Bilinguals. Oxford: Oxford University Press. Grosjean, F. (2010) Bilingual: Life and Reality. Cambridge, MA: Harvard University Press. Grosjean, F. and Li, P. (2012) The Psycholinguistics of Bilingualism. New York: Wiley/ Blackwell. Gut, U. (2015) Bilingual Acquisition of Intonation: A Study of Children Speaking German and English. New York: Mouton De Gruyter. Hack, J., Marinova-Todd, S. and Bernhardt, B.M. (2013) Accent or impairment: Speech assessment of Chinese-English bilingual children. International Journal of SpeechLanguage Pathology 14 (6), 509–519. Hakuta, K. (1986) Mirror of Language: The Debate on Bilingualism. New York: Harper Collins. Hamers, J.F. and Blanc, M.H.A. (1983) Bilinguality and Bilingualism. Cambridge: Cambridge University Press. Haugen, E. (1953) The Norwegian Language in America. Philadelphia, PA: University of Pennsylvania Press. Hoff, E., Core, C., Place, S., Rumiche, R., Señor, M. and Parra, M. (2011) Dual language exposure and early bilingual development. Journal of Child Language 39, 1–27. Hoff er, K. (2007) Dialect variations of speech sound production. In M.E. GordonBrannan and C.E. Weiss (eds) Clinical Management of Articulatory and Phonologic Disorders (pp. 76–119). Baltimore, MD: Lippincott Williams & Wilkins. Hua, Z. and Dodd, B. (2006) A multilingual perspective on phonological development and disorders. In Z. Hua and B. Dodd (eds) Phonological Development and Disorders: A Multilingual Perspective (pp. 1–13). Clevedon: Multilingual Matters. Ingram, D. (1989) First Language Acquisition: Method, Description and Explanation. Cambridge: Cambridge University Press.

Introduction

23

International Expert Panel on Multilingual Children’s Speech (2012) Multilingual Children with Speech Sound Disorders: Position Paper. See https://cdn.csu.edu.au/__ data/assets/pdf_fi le/0010/392977/MultilingualSSDPositionPaper.pdf (accessed April 2019). Jarvis, S. and Pavlenko, A. (2008) Crosslinguistic Influence in Language and Cognition. London: Routledge. Johnson, W. and Reimers, P. (2010) Patterns in Child Phonology. Edinburgh: Edinburgh University Press. Kaltsa, M., Tsimpli, I.M. and Rothman, J. (2015) Exploring the source of differences and similarities in L1 attrition and heritage speaker competence: Evidence from pronominal resolution. Lingua. 164 (Part B), 266–288. Kehoe, M. (2018) Crosslinguistic interaction in early bilingual phonology: A critical review. In E. Babatsouli and D. Ingram (eds) Phonology in Protolanguage and Interlanguage (pp. 49–75). Sheffield: Equinox. Keshavarz, M.H. and Ingram, D. (2002) The early phonological development of a FarsiEnglish bilingual child. International Journal of Bilingualism 6 (3), 255–269. Kohnert, K. (2008) Language Disorders in Bilingual Children and Adults. San Diego, CA: Plural. Kraemer, R. and Fabiano-Smith, L. (2017) Language assessment of Latino Englishlearning children: A records abstraction study. Journal of Latinos and Education 16 (4), 349–358. Kupisch, T. and Rothman, J. (2018) Terminology matters! Why difference is not incompleteness and how early child bilinguals are heritage speakers. International Journal of Bilingualism 22 (5), 564–582. doi:10.1177/1367006916654355 Lauchlan, F. and Parafita Couto, M.C. (eds) Bilingualism and Minority Languages: Current Trends and Developments. Newcastle upon Tyne: Cambridge Scholar. Libben, M., Goral, M. and Libben, G. (2017) Bilingualism: A Framework for Understanding the Mental Lexicon. Amsterdam/Philadelphia, PA: John Benjamins. Lleó, C. (2018) The (un)predictable world of the bilingual child: Full and empty transfer and full acceleration. Special lecture presented at the International Child Phonology Conference 2018, Chania, Greece, 18–20 June. Lleó, C. and Kehoe, M. (2002) On the interaction of phonological systems in child bilingual acquisition. International Journal of Bilingualism 6, 233–237. Mackey, W.F. (1967) Bilingualism as a World Problem. Le Bilingualisme: Phenomene Mondial. Montreal: Harvest House. Macnamara, J. and Kushnir, S. (1971) Linguistic independence in bilinguals: The input switch. Journal of Verbal Learning and Verbal Behavior 10, 480–487. Major, R.C. (1977) Phonological differentiation of a bilingual child. Papers in Psycholinguistics and Sociolinguistics. Working Papers in Linguistics 22, 88–122. Major, R.C. (2001) Foreign Accent: The Ontogeny and Phylogeny of Second Language Phonology. Hillsdale, NJ: Lawrence Erlbaum. Major, R.C. (2008) Transfer in second language phonology. In J.G.H. Edwards and M.L. Zampini (eds) Phonology and Second Language Acquisition (pp. 63–94). Amsterdam/ Philadelphia, PA: John Benjamins. Maneva, B. and Genesee, F. (2002) Bilingual babbling: Evidence for language differentiation in dual language acquisition. In B. Skarabela, S. Fish and A. Do (eds) BUCLD 26: Proceedings of the 16th Annual Boston University Conference on Language Development (Vol. 1, pp. 383–392). Somerville, MA: Cascadilla Press. Marian, V. and Spivey, M. (2003) Competing activation in bilingual language processing: Within- and between-language competition. Bilingualism: Language and Cognition 6 (2), 97–115. Martin-Jones, M., Blackledge, A. and Creese, A. (2012) The Routledge Handbook of Multilingualism. London/New York: Routledge.

24

An Anthology of Bilingual Child Phonology

McLaughlin, B. (1978) Second Language Acquisition in Childhood. Hillsdale, NJ: Lawrence Erlbaum. McLaughlin, B. (1995) Fostering second language development in young children: Principles and practices. NCRCDSLL Educational Practice Reports. See http:// escholarship.org/uc/item/23s607sr. McLeod, S. (2012) Multilingual Children’s Speech. Bathurst, NSW: Charles Sturt University. McLeod, S. and Goldstein, B.A. (2012) Multilingual Aspects of Speech Sound Disorders in Children. Bristol: Multilingual Matters. McLeod, S. and Verdon, S. (2017) Tutorial: Speech assessment for multilingual children who do not speak the same language(s) as the speech-language pathologist. American Journal of Speech-Language Pathology 26, 691–708. Meisel, J.M. (1989) Early diff erentiation of languages in bilingual children. In K. Hyltenstam and L.K. Obler (eds) Bilingualism across the Lifespan: Aspects of Acquisition, Maturity and Loss (pp. 13–40). Cambridge: Cambridge University Press. Meisel, J.M. (2004) The bilingual child. In T. Bhatia and W.R. Ritchie (eds) The Handbook of Bilingualism (pp. 91–113). Oxford: Blackwell. Meisel, J.M. (2007) The weaker language in early child bilingualism: Acquiring a fi rst language as a second language? Applied Psycholinguistics 28, 495–514. Montrul, S.A. (2008) Incomplete Acquisition in Bilingualism: Re-examining the Age Factor. Amsterdam/Philadelphia, PA: John Benjamins. Munro, S., Ball, M.J., Müller, N., Duckworth, M. and Lyddy, F. (2005) Phonological acquisition in Welsh-English bilingual children. Journal of Multilingual Communication Disorders 3 (1), 24–49. doi:10.1080/14769670410001683467 Nation, I.S.P. (2001) Learning Vocabulary in another Language. Cambridge: Cambridge University Press. Norbert, F. (2011) Imbalances in bilingual development: A key to understanding the faculty of language. Language Sciences 33, 76–89. Odlin, T. (2008) Cross-linguistic influence. In C. Doughty and M.H. Long (eds) The Handbook of Second Language Acquisition (pp. 436–486). Cambridge, MA: Blackwell. Oksaar, E. (1983) Multilingualism and multiculturalism from the linguist’s point of view. In T. Husén and S. Opper (eds) Multicultural and Multilingual Education in Immigrant Countries (pp. 17–36). Wenner Gren Symposium Series No. 38. Oxford: Pergamon Press. Oller, K.D. (2010) All day recordings to investigate: A case study of a trilingual toddler. Communication Disorders Quarterly 31, 213–222. Ortega, L. (2009) Understanding Second Language Acquisition. London/NY: Routledge. Padilla, A. and Lindholm, K.J. (1984) Hispanic behavioural science research: Recommendations for future research. Hispanic Journal of Behavioural Sciences 6, 13–32. Paradis, J. (2000) Beyond ‘One system or two’: Degrees of separation between languages of French-English bilingual children. In S. Döpke (ed.) Cross-linguistic Structures in Simultaneous Bilingualism (pp. 175–200). Amsterdam/Philadelphia, PA: John Benjamins. Paradis, J. and Genesee, F. (1996) Syntactic acquisition in bilingual children: Autonomous or interdependent? Studies in Second Language Acquisition 18, 1 − 25. Paradis, J., Genesee, F. and Crago, M.B. (2011) Dual Language Development and Disorders: A Handbook on Bilingualism and Second Language Learning. Baltimore, MA: Paul Brookes. Pardo, J. (2006) On phonetic convergence during conversational interaction. Journal of the Acoustical Society of America 119, 2382–2393. Perani, D., Paulesu, E., Sebastian Galles, N. et al. (1998) The bilingual brain: Proficiency and age of acquisition of the second language. Brain 121, 1841–1852.

Introduction 25

Perani, D., Abutalebhi, J., Paulesu, E., et al. (2003) The role of age of acquisition and language usage in early high proficient bilinguals: A FMRI study during verbal fluency. Human Brain Mapping 19, 179–182. Pienemann, M. and Keßler, J. (2007) Measuring bilingualism. In P. Auer and L. Wei (eds) Handbook of Multilingualism and Multilingual Communication (pp. 247–267). The Hague: Mouton de Gruyter. Polinsky, M. and Kagan, O. (2007) Heritage languages: In the ‘wild’ and in the classroom. Language and Linguistics Compass 1 (5), 368–395. Qi, R. (2011) The Bilingual Acquisition of English and Mandarin: Chinese Children in Australia. Amherst, NY: Cambria Press. Roeper, T. (1999) Universal bilingualism. Bilingualism: Language and Cognition 2 (3), 169–186. Schlyter, S. (1993) The weaker language in bilingual Swedish-French children. In K. Hyltenstam and A. Viberg (eds) Progression and Regression in Language (pp. 289– 308). Cambridge: Cambridge University Press. Schmidt, T. and Wörner, K. (eds) (2012) Multilingual Corpora and Multilingual Corpus Analysis. Hamburg Studies on Multilingualism No. 14. Amsterdam/Philadelphia, PA: John Benjamins. Selinker, L. (1972) Interlanguage. International Review of Applied Linguistics 10, 209–231. Simonet, M. (2014) Phonetic consequences of dynamic cross-linguistic interference in proficient bilinguals. Journal of Phonetics 43, 26–37. Simonet, M. (2016) The Phonetics and Phonology of Bilingualism. Oxford Handbooks Online. Oxford: Oxford University Press. Smith, N.V. (1973) The Acquisition of Phonology: A Case Study. Cambridge Studies in Linguistics No. 25. Cambridge: Cambridge University Press. Swain, M. (1972) Bilingualism as a fi rst language. Unpublished PhD dissertation, University of California. Tamburelli, M., Sanoudaki, E., Jones, G. and Sowinska, M. (2015) Acceleration in the bilingual acquisition of phonological structure: Evidence from Polish-English bilingual children. Bilingualism: Language and Cognition 18 (4), 713–725. Thiery, C. (1978) True bilingualism and second language learning. In D. Gerver and H. Sinaiko (eds) Language, Interpretation and Communication (pp. 145–153). New York: Plenum Press. Thomas, E.M. and Mayr, R. (2010) Children’s acquisition of Welsh in a bilingual setting: A psycholinguistic perspective. In D. Morris (ed.) Welsh in the 21st Century (pp. 99–117). Cardiff : Cardiff University Press. Thomas, E.M. and Mennen, I. (eds) (2014) Advances in the Study of Bilingualism. Bristol: Multilingual Matters. Treff ers-Daller, J. (2009) Code-switching and transfer: An exploration of similarities and differences. In B.E. Bullock and A.J. Toribio (eds) The Cambridge Handbook  of  Linguistic Codeswitching (pp. 58 –74). Cambridge: Cambridge University Press. Trevarthen, C. and Aiken, K.J. (2001) Infant intersubjectivity: Research, theory, and clinical applications. Journal of Child Psychology and Psychiatry 42, 3–48. Trubetzkoy, N. ([1939] 1969) Grundzüge der Phonologie. In Principles of Phonology (Travaux du cercle linguistique de Prague 7, trans. C. Baltaxe). Berkeley, CA: University of California Press. Tsimpli, I. (2007) First language attrition from a minimalist perspective: Interface vulnerability and processing effects. In B. Köpke, M. Schmid, M. Keijzer and S. Dostert (eds) Language Attrition: Theoretical Perspectives (pp. 83–98). Amsterdam/Philadelphia, PA: John Benjamins. Tsimpli, I.M., Peristeri, E. and Andreou, M. (2016) Narrative production in monolingual and bilingual children with specific language impairment. In N. Gagarina, I.M.

26

An Anthology of Bilingual Child Phonology

Tsimpli and J. Walters (guest eds) Special issue on ‘Narrative abilities in bilingual children’. Applied Psycholinguistics 37 (S1), 195–216. Tucker, G.R. (1998) A global perspective on multilingualism and multilingual education. In J. Cenoz and F. Genesee (eds) Beyond Bilingualism: Multilingualism and Multilingual Education (pp. 3–15). Clevedon: Multilingual Matters. Valdés, G. (2001) Heritage language students: Profi les and possibilities. In J.K. Peyton, D.A. Ranard and S. McGinnis (eds) Heritage Languages in America: Preserving a National Resource (pp. 37–80). McHenry, IL: Center for Applied Linguistics and Delta Systems. Verdon, S., Blake, H.L., Hopf, S.C., Phạm, B. and McLeod, S. (2016) Cultural and linguistic diversity in speech-language pathology. Editorial. International Journal of Speech-Language Pathology 18 (2), 109–110. doi:10.3109/17549507.2015.1122838 Vihman, M.M. (1996) Phonological Development: The Origins of Language in the Child. Cambridge. MA: Blackwell. Vihman, M.M. (2014) Phonological Development: The First Two Years. Cambridge, MA: Wiley Blackwell. von Raffler-Engel, W. (1965) Del bilinguismo infantile. Archivio Glottologico Italiano 50, 175–180. Wei, L. (2000) The Bilingualism Reader. London/New York: Routledge. Wei, L. (ed.) (2007) The Bilingualism Reader (2nd edition). Routledge: London & New York. Wrembel, M. and Cabrelli Amaro, J. (2017) Advances in the Investigation of L3 Phonological Acquisition. New York: Routledge. Yavaș, M. (ed.) (2015) Unusual Productions in Phonology: Universals and Languagespecifi c Considerations. London/New York: Psychology Press. Yip, V. and Matthews, S. (2007) The Bilingual Child: Early Development and Language Contact. Cambridge: Cambridge University Press.

2 Investigating the Linguistic World of the Bilingual Child: Transfer and Code-switching Conxita Lleó

Introduction

In the past few decades, much research on language acquisition has continued to focus on the acquisition of one language at a time, but concomitantly there has been an increase in research on bilingualism and multilingualism, topics that had previously been rather ignored. Languages in contact and especially cross-language interaction (CLI) have been the most frequently discussed issues in the latter topic areas, with the aim of discovering relevant factors for predicting the outcomes of bilingual acquisition. Factors include input and output, heritage and dominant language, frequency, markedness, simplicity, etc. Several theories have been proposed to account for the simultaneous acquisition of two languages, and their two phonologies. Out of the various factors and theories available, in this chapter I describe bilingual acquisition from the perspective of optimality theory (OT; Tesar & Smolensky, [1993] 2000). By now it is well known that OT does not describe phonological patterns in terms of rules but rather by means of constraints, which determine what structures are permitted. For instance, the NoCoda constraint indicates that codas are not permitted. However, each time a coda is produced, this constraint is violated, and often violated constraints must be demoted from their outranking position. Initially, child grammar has markedness constraints outranking faithfulness constraints (Gnanadesikan, 1995). One theory of phonological development is that the child’s grammar develops through constraint demotion (Tesar & Smolensky, [1993] 2000) and, in spite of the outranking position attributed to markedness constraints by some researchers, the question has been posed as to whether constraints are available from the very beginning of language acquisition (Fikkert & Levelt, 2008; Lleó, 2015, 2018). A further issue relates to the question of 27

28

An Anthology of Bilingual Child Phonology

how constrained child grammar should be, taking continuity between the parental and the child grammars into consideration. During the fi rst years of life, the creative mind of the bilingual child may arrive at surprising solutions from the point of view of adult grammar, unexpectedly including transfer and language switching, which suggests the question of whether these notions (transfer and code-switching) can be posited as equivalent to the adults’ notions, regardless of age. In this chapter I will deal with the simultaneous acquisition of the phonology of two languages from the point of view of OT. The aim of the chapter is twofold. On the one hand, I will try to fi nd out the role of constraints in the development of the child’s grammar. On the other hand, the focus will be on outcomes of bilingualism, especially on the effects of transfer (T) and code-switching (CS) in the child’s grammar. Preamble: Beginnings of research on family multilingualism

When, in 1966, I expressed my wish to study bilingual acquisition to my thesis advisor (at the Linguistics Department of the University of Washington in Seattle), his reaction was that, since we hardly knew anything about the acquisition of a single language, it would be adventurous to attempt the study of the simultaneous acquisition of two languages. Because I myself had been exposed to Catalan and Spanish from birth (although to different degrees, as exposure to Spanish had not been as frequent at fi rst), I tried to tackle the topic autodidactically. But it was with the birth of our daughters, Laura and Ariadna, that I found the opportunity to observe language acquisition empirically. And since at home we practiced some bilingualism (Catalan was spoken by myself and Spanish by my husband (hereafter Miguel) and me), some trilingualism (Catalan, Spanish and German by all of us) and some bidialectalism (two varieties of Spanish were spoken in the family: Iberian or Peninsular by myself and Argentinean by Miguel), I began to do research on bilingual acquisition based on the limited (psycho-)linguistic theories of that time and the linguistic situation of our family’s languages in contact. Our daughters were exposed to all those languages and varieties; however, they heard and used Catalan at home and German at kindergarten, which meant that since I observed and recorded them at home, I only had access to their Catalan, to their Spanish and to some German that they brought home from kindergarten. The children’s situation was comparable to Grosjean’s (1997) monolingual versus bilingual modes. He proposes that language use by the bilingual speaker involves a continuum from monolingualism to bilingualism, which implies that when one of the languages is activated, the other language of the bilingual speaker is also active, although not to the same degree. The monolingual mode was prevalent in kindergarten, but not at home, where Catalan, (two varieties of) Spanish and German were used. One could say that I had a ‘language lab’ at home.

Investigating the Linguistic World of the Bilingual Child 29

However, my real research on bilingual and multilingual acquisition began in 1986, when I received the support of the German Research Foundation (DFG; Deutsche Forschungsgemeinschaft) for the PEDSES project (see below). A long way towards approaching bilingualism

After Laura’s birth, at home I kept speaking Spanish, which was the language Miguel and I used to communicate with each other. At that point we did not practice bilingualism yet, but bidialectalism, i.e. Laura was exposed to the Argentinean and Peninsular varieties of Spanish. However, in a visit to my family in Barcelona, a few months after Laura’s birth (Christmas, 1980), I realized that my fi rst language, Catalan – and not Spanish – was the most suitable language to communicate with my daughter, and I made the switch to Catalan as the language of communication with her. Soon afterwards, I began to observe and (audio-)record Laura’s Catalan productions. During the first year she was growing up multilingual, exposed to Catalan (spoken by myself and by my mother who came from Barcelona to visit us often), some Spanish (from Miguel, i.e. Laura’s father) and some German (from German babysitters and friends), but at home her few productions were mainly Catalan. Her first word spoken to me was [tε] for Catalan target [te], meaning something like ‘here, take it’, as she was handing a (raw) potato to me (for me to peel). After that her linguistic environment changed, as in her second year of life she began to attend kindergarten. There, she was exclusively exposed to German from Monday to Friday, from 9am to around 4pm. In the evenings she was with me and the language of the environment was Catalan. At the weekends and some evenings, Miguel’s Argentinean Spanish also came into play. When our daughter Ariadna was born, the language constellation at home was the one here described, with an important difference, namely that at the time of Ariadna’s birth Laura was 3;9 (years;months), and had already acquired a lot of language knowledge (Bullock et al., 2006; Toribio et al., 2005). When Ariadna was 1 year old (March 1985), the family moved from Göttingen to Hamburg, and in October 1984 both girls began to attend a public kindergarten in Hamburg, from Monday to Friday, from 9am to around 5pm. At home, Laura spoke Catalan (especially with me and my mother), some Spanish with her father, and generally German with her sister. German had begun to be her strongest language because of the kindergarten, where she spent so many hours on the weekdays. Ariadna spent the first months at home, and when she was about seven months old she began to attend the same kindergarten as Laura. German became their main language of communication with people outside the family, and with each other. At the end of 1986 I began to do research on German-Spanish bilingualism with the PEDSES project (Phonological Acquisition of German

30

An Anthology of Bilingual Child Phonology

and Spanish as fi rst language(s)), which allowed me to study the development of three German-Spanish bilingual children in Hamburg (with alias names Irene, Robert and Stefan). However, I soon realized that it was difficult to understand how the simultaneous development of the two languages was taking place, if we did not have sufficient information on the individual development of German on the one hand and Spanish on the other. We clearly needed to study the monolingual acquisition of the two phonologies, which I was able to do with the German-Spanish project PAIDUS (Parameter Setting in German and Spanish), with five monolingual German children in Hamburg and three monolingual Spanish children in Madrid. Both projects, PEDSES and PAIDUS, received the support of the DFG (Lleó, 2012) and the DAAD (Acciones Integradas). The reason for studying monolingual acquisition in these projects was to provide a basis of comparison to simultaneous bilingual acquisition. However, applying a monolingual approach to the study of bilingualism is controversial, because it may ignore the fact that the two language contexts (monolingual and bilingual) are very different (DeKeyser, 2012; Grosjean, 1997; Lardière, 2003). Some recent studies argue against studying each language of the bilingual speaker separately, because languages in contact may exert much influence on one another, with the consequence that the input may become questionable; one or both languages (generally the heritage language; HL) may be strongly modified, and this modified language cannot be taken as a valid input for the language to be learned. It is certainly not the case that the modified language (HL) should be studied as if it were acquired in isolation. However, much of the language that serves as input to the following generation is spoken by a very limited number of speakers, as in the case of guest workers in Germany who may represent a small minority with limited linguistic consequences or no consequences at all. Some researchers defend the position that monolingual language acquisition is hardly relevant to the simultaneous acquisition of two languages, either an L2 or the HL. Nonetheless, when doing research on the bilingual acquisition of, for example, German and Spanish, the researcher needs to have some information on the acquisition of these languages in isolation in order to attribute a certain property to the right source. For instance, the bilingual child at about 2;0 may substitute a stop consonant for a approximant, and say [el 'ɡa.to] instead of [el 'ɣa,to] ‘the cat’. The crucial question is whether this substitution comes from: (1) the lack of such spirants in German; or (2) the universal preference for stops over approximants and approximants. If we do not know how the monolingual Spanish child develops in a monolingual context, we will not know the answer to our question. However, if we are familiar with this development, we will know that the preference for stops over fricatives is not typical for the development of the Spanish child, who as soon as at 2;0, and sooner, already prefers approximants over stops, given the Spanish

Investigating the Linguistic World of the Bilingual Child 31

process of spirantization (see Barlow, 2003; Lleó, 2015, 2018). At the same time, the alleged preference for stops over approximants will be called into question, as such preference appears only in relation to those languages that show many processes of assimilation and lenition. My Research on Bilingualism: Several Things We (Don’t) Know

The bilingual projects PEDSES and the monolingual PAIDUS were developed at the University of Hamburg from the late 1980s through the 1990s. Furthermore, between 1999 and 2011 a Research Center on Multilingualism was created at the University of Hamburg, where I also led the bilingual project PhonBLA (Phonological Bilingual Language Acquisition). As a result of the research activity of those years, the Hamburg Center for Language Corpora received the support of the DFG and of the University of Hamburg. Moreover, 15 volumes of the Hamburg Studies on Multilingualism (HSM), all of them dealing with issues of bilingualism and multilingualism, were edited by members of the Research Center and published by John Benjamins (Amsterdam/Philadelphia, PA), also with the support of the DFG and the University of Hamburg (see Schmidt & Wörner, 2012). Data and hypotheses

Despite continuous research on phonological acquisition during more than half a century, some basic issues are not yet settled. In the area of bilingualism in particular, the following are some open questions which will be addressed in this chapter on the basis of extant literature and of related research projects: (a) Are there quantitative similarities and differences between monolinguals, heritage speakers (HS) and L2 speakers? (b) Is transfer a necessary phonological result of languages in contact? (c) Is child code-mixing comparable to adult code-switching? Heritage language (HL) and dominant language (DL)

Bilingualism can take several forms, e.g. family bilingualism, which refers to cases of immigration of, for example, Spanish speakers in a country where another language is spoken, as in this case, German in Germany. The former (i.e. Spanish in our example) is considered the HL, whereas the latter (German) is the societal or dominant language (DL). There are many defi nitions of HL. For the purposes of this chapter, I will adopt Rothman’s (2007: 156) defi nition: ‘A language qualifies as a heritage language if it is a language spoken at home or otherwise readily available to young children, and crucially this language is not a dominant language of

32

An Anthology of Bilingual Child Phonology

the larger (national) society.’ It is also important to note that HL speakers should have some degree of language mastery, as it is not sufficient to have social or cultural interest in the language in order to belong to the HL community (Kupisch, 2013). An important aspect of a HL is its grammar (HG), which may not be equally developed in all its areas. In fact, it is not clear which aspects of grammar remain undeveloped or rather underdeveloped into adulthood, e.g. Spanish Heritage speakers have good control of object clitics, but difficulties with other objects of grammar (Montrul, 2008). In this sense, research on bilingual acquisition leads to the question of whether HG may remain incomplete, although some researchers reject this notion of incompleteness when referring to grammar (Kupisch & Rothman, 2018; Rothman, 2007). Other researchers prefer to ask which specific aspects of the HG are more likely to be different from those of monolingual speakers and why. Factors responsible for the outcome of language contact

Scholars doing research on language acquisition generally distinguish factors internal to the learner and those external to the learner (Eckman, 2012; Skehan, 2012) when talking about outcomes of language contact in bilingual acquisition. Among the former, the following factors are often noted: age at onset of exposure (AoE), knowledge of another language, cognitive maturity and language learning aptitude. Among the factors external to the learner, the following are often mentioned as predictors: socio-economic status (SES), maternal education, L2 proficiency, number of siblings, length of exposure to the language, input quantity, input quality and language use or output (DeKeyser, 2012). Another way of tackling the analysis of predictive factors for bilingual outcomes is by focusing on the language itself and not on the learner. According to this more recent view, the following purely linguistic factors are considered internal: dependent on the specificities of a particular language, markedness (Jakobson, [1941] 1968), uniformity (Kenstowicz, 1998) – reducible to complexity, because it maintains forms unmodified, and this leads to the reduction of allophony/allomorphy – and frequency (including additiveness, i.e. the presence of a given unit or phenomenon in both languages). Lleó and Cortés (2013) found evidence in favor of the following hierarchy of internal factors: frequency > markedness > uniformity

where ‘frequency’ is more relevant than ‘markedness’, which is also more relevant than ‘uniformity’. Factors external to the language under this new perspective based on language (versus the learner) are sociolinguistic factors. They are dependent on the external conditions of language use,

Investigating the Linguistic World of the Bilingual Child 33

dominance (language of the large social environment), language spoken with mates and friends, family’s language (language spoken between parents to each other), language spoken with grandparents, etc. Clearly, this is only a partial result which should be submitted to a critical analysis. This result should especially be tested in order to see whether it has the same validity when submitted for proof. Interaction patterns in German-Spanish simultaneous bilinguals

Going back to input, the two languages of a bilingual child are in contact with one another, which will be briefly referred to in this section, later focusing on T and CS. One of the first studies of the relevant factors predicting bilingual outcomes was by Paradis and Genesee (1996). They proposed three potential manifestations of CLI (or ‘inter-dependence’): delay, acceleration and transfer. They observed three French-English bilinguals at 2;0, 2;6 and 3;0, growing up in Montreal, Canada, a city of French-English bilinguals, where French is generally dominant although not in all districts. In the syntax (fi niteness, negation and pronominal subjects), they found no delay, no acceleration and no transfer. Nevertheless, not fi nding CLI in syntax does not preclude interaction in phonology or phonetics. In fact, research during these last decades has manifested many cases of phonetic and phonological CLI, such as voiced stops instead of spirants in Spanish (Lleó, 2018; Lleó & Rakow, 2003; Rakow & Lleó, 2003), because the allophonic pattern of spirantization was not yet acquired. Acceleration is manifested in syllabic coda production by bilinguals in Spanish who outperform monolinguals. Delay appears in the pre-tonic (unfooted) syllables in Spanish (see Figures 2.2a and 2.2b). Moreover, Lleó (2018) also showed cases of delay and acceleration, as well as two more types of interaction, one affecting the order of acquisition and the other one showing no effect at all. The former case involves differences in the order of acquisition, such as Spanish monolinguals producing trisyllabic words before monosyllables, whereas bilinguals produced monosyllables before producing trisyllables. The case with no effect can be exemplified by the acquisition of the Spanish vowels which, being the five cardinal vowels /i/, /u/, /e/, /o/, /a/, do not pose any difficulties to the monolingual or bilingual learners. Finally, fusion or blending, by which two categories of one of the languages are reduced to a single category, can also show interactions between the languages. Nils at age 2;0–2;3 produced voiced stops with lead voicing in both German and Spanish, although only target Spanish has lead voicing; at age 2;3–2;6 he produced voiceless stops with long lag in both Spanish and German, although only target German has long lag (see Kehoe et al., 2004).

34

An Anthology of Bilingual Child Phonology

Segmental and prosodic categories

The syllable can be structured around the nucleus (a vowel), the onset (not obligatory in Spanish), associated to a consonant and, in case the syllable is closed, a coda, associated to a consonant, as in the syllable which is only comprised of nucleus and coda, without onset. Depending on the position of consonants and vowels within the word, and on the type of segment being associated with the prosodic positions, segments may appear in a modified form, as for instance voiced stops, which occur with a certain continuity feature, depending on the type of segment preceding the voiced consonant, e.g. [uŋ ɡa.to] ‘a cat’ (with a nasal – nasals are [–cont] – preceding the stop), versus [u.na.ɣa.ta] ‘a female cat’ (with a vowel – vowels are [+cont] – preceding the spirant). Also in the case of nasals, the place of articulation (PA) varies, depending on the PA of the following obstruent as in, for example, [uŋ ɡa.to] ‘a cat’ (with a dorsal nasal preceding a dorsal stop), versus [um.pe.ro] ‘a dog’ (with a labial nasal preceding a labial stop). At the beginnings of research on phonological acquisition, the focus was on segments (Chiat, 1989; Clark & Clark, 1995; Fabiano-Smith & Goldstein, 2010) but, as phonological theory developed and prosody was established as a central component of language, it was the association of the various prosodic positions with the corresponding segments that attracted the interest of researchers. Among the prosodic categories we examine syllables and their codas; among segmental categories we focus on collocations on those that are variable, i.e. segments with allophonic variants. Based on the literature and on research carried out at the University of Hamburg, the following section outlines some key fi ndings of our bilingual research. The type of prosodic category analyzed in phonology, beside the syllable and the foot, refers to the prosodic positions for consonants and vowels, i.e. where the latter are associated. All data were collected in semi-spontaneous situations, playing and engaging in a conversation with the child. Research questions were selected on the basis of (similarities and) differences between German and Spanish. Finally, I will show some new data on code-switching that needs further analysis.

Different grammars imply different hierarchies

In OT, grammar is derived from a certain hierarchy of constraints. Different hierarchies result in different grammars. Assuming that, initially, negative markedness constraints are highest ranked (Gnanadesikan, 1995), which accounts for the initial distance between the phonetic shapes as produced by children and by adults, such forms must change at the point where the child is supposed to behave phonetically similarly to an adult. Among other things, markedness constraints must be demoted, in case the

Investigating the Linguistic World of the Bilingual Child 35

target language contains marked forms. In the following sections I will discuss some of the bilingual outcomes of Paradis and Genesee’s (1996) study, introduced earlier, namely closed syllables, unfooted syllables, spirantization, assimilation of PA of nasals, transfer and code-switching. Closed Syllables in German and Spanish

In German, more than 60% of syllables have codas (are closed syllables), whereas in Peninsular Standard Spanish less than 30% of syllables are closed (Delattre, 1965). Closed syllables are considered to be marked as they violate the NoCoda constraint, which is a markedness constraint. In German, closed syllables seem to respond to two factors, markedness and frequency, because despite being marked, they are very frequent. Thus, because markedness constraints are by defi nition in a dominant position in the constraint hierarchy, at fi rst the frequent violations of markedness make the learner aware of codas, metaphorically speaking, and this leads to the demotion of the constraint that shows so many violations. Before we studied the development of codas in the German-Spanish bilingual population (Figure 2.1a), we had asked the question as to whether bilinguals would behave like Spanish monolinguals (with a slow development of codas) or like German monolinguals (with a fast coda development). (1) NoCoda: Syllables must not have a coda. (Prince & Smolensky, 1993: 85) Since there are many words with codas in the world languages, such a constraint is often violated. Results, in terms of percentages of coda production by the monolingual and bilingual children, were as follows. 100 90 80 70 60

M-Ger

50 Bi-Spa

40

M-Spa

30 20 10 0 1;2-1;4

1;5-1;6

1;7-1;8 1;9-1;10 1;11-2;0 2;1-2;2

2;3-2;4

Figure 2.1a Percentages of coda production by German speaking monolinguals (M-Ger; highest curve), Spanish speaking monolinguals (M-Spa; lowest curve) and German-Spanish bilinguals in Spanish (Bi-Spa; middle curve)

36

An Anthology of Bilingual Child Phonology

Figure 2.1b Percentages of coda production by four 2- to 3-year-old SpanishGerman bilingual children living in Madrid

Monolingual German children produced more than 80% of codas after 1;11, but monolingual Spanish did not yet produce 50% of codas at 2;4, i.e. at the end of the study. There were a variety of answers to our query about the speed of coda development in bilinguals. German-Spanish bilinguals (where the order of the two languages from left to right indicates that German is the DL and Spanish the HL) produced more than 50% of codas in Spanish after 1;9. Furthermore, coda production by bilinguals in German (not included in Figure 2.1b) did not differ from that of monolinguals. The German-Spanish bilingual children have a high-ranked constraint banning codas which, given the many codas of German, incurs many violations, leading the bilingual child to soon accede to the German low ranking of NoCoda. That is, in agreement with Tesar and Smolensky ([1993] 2000)’s Recursive Constraint Demotion Algorithm, the NoCoda constraint is soon demoted. With the purpose of comparing German-dominant bilinguals to Spanish-dominant ones, we also analyzed the percentages of coda production by a group of Spanish-German bilinguals being raised in Madrid (Spain) (Figure 2.1b). Spanish coda production by bilinguals living in Madrid also reached higher values than those of Spanish monolinguals. We found that coda production in Spanish by bilinguals from Madrid tended to converge towards the German values, whereas coda production in German was not delayed in spite of the limited coda production of Spanish monolinguals. Unfooted Syllables in Spanish and German

Both German and Spanish contain many trisyllabic words which occur in adult as well as in child language. However, trisyllabic words have received different prosodic analyses depending on the language they belong to: in Spanish, with a preference for trochaic prosodic feet, such as

Investigating the Linguistic World of the Bilingual Child 37

zapato [θa'pa.to] ‘shoe’, the unstressed initial syllable of the three syllables is considered to be an unfooted syllable, preceding the following foot, whereas in German the initial syllable of such words can be either stressed or unstressed. If it is stressed, as in zumachen ['tsu.ma.xen] ‘close’, it constitutes a foot that precedes a second foot; if unstressed, the syllable is considered unfooted, as in Spanish (see Figures 2.2a and 2.2b). See Pater (1997) and Ota (1998) for comparable analyses.

Figure 2.2a Percentages of unfooted syllable truncation by three Spanish speaking (low curve) and three German speaking (high curve) monolinguals Source: Adapted from Lleó (2002).

Figure 2.2b Percentages of unfooted syllables in Spanish by the Spanish speaking (low curve) and three German-Spanish bilinguals (higher curves) Source: Adapted from Lleó (2002).

38

An Anthology of Bilingual Child Phonology

The Spanish lexicon also contains many disyllabic and a large number of trisyllabic words, generally stressed on the penultimate syllable, i.e. they are paroxitonic: SW and WSW. In the case of disyllables, the pattern SW (or trochaic) is the most frequent stress pattern in Spanish, and also in the case of three syllables, the majority of them with an unstressed initial syllable (Lleó, 2002; Lleó & Demuth, 1999; Ota, 1998; Pater, 1997). Monolingual Spanish children began to produce unfooted syllables very soon, whereas German children truncated unfooted syllables until 2;0 and later. Bilinguals had about as much truncation of unfooted syllables in Spanish as German monolinguals until about 2;0, as shown in Figures 2.2a and 2.2b (Lleó, 2002). Two markedness constraints, Align(Right) and Align(Left) can account for the different behavior of truncation in both languages: (2) Align(Ft,R,PW,R): The right edge of all feet must be aligned with the right edge of a PW. (3) Align(PW, L, Ft, L): All PWs must have their left edge aligned with a foot. Constraint (2) requires that each foot be aligned with a prosodic word to the right, which limits the number of feet to the number of words, as is generally the case in German, where unfooted syllables are not preferred. And according to Constraint (3), prosodic words must begin with a foot, excluding unfooted syllables, as is generally preferred in German (see Lleó, 2002, for further discussion). Accordingly, the two constraints AlignRight and AlignLeft lead to different grammars, depending on their relative rankings. Spanish allows for unfooted syllables earlier than German, thus AlignLeft is demoted first, whereas German allows for words comprised of two feet, which means that AlignRight is demoted first. Prosodic Categories and Allophonic Variation: Assimilation of Voiced Obstruents in Spanish

The chapter turns next to the Spanish pattern of spirantization, which requires further description of prosodic categories and allophonic variation. These variants are allophones; being based on variation, I tentatively suggest that children might have difficulties learning them, or they might experience delay, transfer or fusion. Standard phonology defi nes spirantization as a process by which the voiced stops /b d ɡ/ are produced as the approximants [β ð ɣ], respectively, if preceded by a [+cont] sound, whereas they keep their [−cont] status in initial position, or if preceded by a nasal consonant, i.e. a [−cont] sound (Barlow, 2003; Harris, 1984; Lleó & Rakow, 2006). Monolinguals and bilinguals being raised in Spain produce high percentages of target-like spirants. German-Spanish bilinguals growing up in Germany produce few spirants at 3;0, and although their percentages increase at 5;0 and at 7;0, for some of these bilinguals the

Investigating the Linguistic World of the Bilingual Child 39

percentages of spirants are still very low. Thus, there is much individual variation. It is again the case that two constraints are involved – Uniform Exponence and Agree[cont] – which result into different grammars depending on their relative rankings. Above I discussed the interactions in German and Spanish bilingual acquisition affecting coda use and the unfooted syllables in Spanish. I outline these further here from the point of view of OT, starting with the example of Spirantization. (4) Uniform Exponence (UE): A lexical item (morpheme, word) is invariable for property (P). (5) Agree[cont]: After a vowel or a [+cont] sound, voiced obstruents are [+cont]. In German, since there is no process of spirantization similar to the Spanish one, Agree[cont] is outranked by UE, which excludes form modifications (Kenstowicz, 1998). However, Spanish grammar has the opposite order, i.e. Agree[cont] outranks UE in Spanish: AGREE[cont] >> UE. (6) German hierarchy (without spirantization): UE >> Agree[cont]. (7) Spanish hierarchy (with spirantization): Agree[cont] >> UE. More on the Acquisition of Allophones: Assimilation of Nasals in Spanish

In word-medial onset position, both Spanish and German have three nasal phonemes each: in Spanish /m/, /n/, and /ɲ/, in German /m/, /n/, and /ŋ/, shown by the Spanish minimal pairs: ['ka.ma] ‘bed’, ['ka.na] ‘grey/white hair’, ['ka.ɲa] ‘cane’, and by the German ones ['ʀa.men] ‘to ram’, ['ʀa.nen] ‘past of rinnen’, ['ʀa.ŋen] ‘to be ranged’. In spite of these similarities, there is a very important difference between the two languages. In Spanish, nasals in coda position do not have a fi xed PA, but the PA results from assimilation: the nasal consonant adopts the PA of the following obstruent, and this assimilation process takes place word internally as well as between words (Harris, 1982). In German, a similar process may take place, but only word-internally; between words the nasal consonant often appears as a non-assimilated coronal. Spanish examples are shown in (8) and German examples in (9). (8) e.g. in Spanish [um pe.ro] [uɲ tʃo.ro] [uŋ ɡa.to] (9) e.g. in German [ajn bajn] [ajn ʃlawx] [ajn kʊs] Monolingual children being raised in Spain show high percentages of assimilation. However, among the German-Spanish bilinguals at about 3

40

An Anthology of Bilingual Child Phonology

100 80

[m] [ŋ]

60 40 20 0

JB SB TR TIK TOK PW EW CW LW LG VA JP

JC ND AL PA

Figure 2.3 Percentages of target-like assimilation of nasals to the LABIAL place of articulation (PoA) (black columns) and the DORSAL PoA (light gray columns) produced by German-Spanish bilinguals in Hamburg (Germany)

years of age there are few cases of assimilation and much individual variation. For bilinguals at 7;0 growing up in Germany, percentages of assimilation show much variation too (see Figure 2.3). This figure shows the percentages of target-like assimilation of the PA of nasals to the PA of the following obstruent, as produced by the German-Spanish bilinguals living in Hamburg (Germany). In a parallel study to this one, with data recorded in Madrid (Spain), we could show that within the population growing up in Madrid, both spirantization and PA-assimilation reached a high degree of assimilation. The example of above has shown some differences between Spanish and German, the latter showing a reduced extent to which the assimilation of the nasal PA is applicable. This difference has as a result that the Spanish nasals occurring in the syllabic coda can be analyzed as variants or allophones, since the various PAs depend on the PA of the following onset. Children being raised in Spain, both monolinguals and bilinguals, tend to produce high percentages of assimilation for nasals. However, those living in Germany show much less assimilation at about age 7 years (Figure 2.3). On the one hand, the examples in (8) show that the different forms adopted by the nasals in Spanish constitute allophonic variants, whereas (9) shows that the coronal PA does not assimilate in German, and it is maintained. Transfer and Types of Transfer

We turn now to issues of transfer. Since the time of its formulation in second language acquisition (SLA) research, several distinctions have been drawn, such as positive and negative types of transfer. Under positive transfer one assumes identical categories in the L1 and L2 or L3, if present. These should not pose any difficulties to the learner. On the contrary,

Investigating the Linguistic World of the Bilingual Child 41

the presence of the same category in both languages (L1 and L2) should facilitate acquisition. Negative transfer assumes different categories in the L1 and the other language(s) being acquired, with potential learning problems ensuing. The (bilingual) child who produces mostly voiced obstruents as such, without the alternation between obstruents and approximants, can be characterized as transferring voiced obstruents to the position that should be reserved for approximants. This phenomenon corresponds to transfer, too. However, saying ['lo.bo] ‘wolf’ for ['lo.βo] does not create a new segment, as the voiced stop produced by the child in ['lo.bo] was already contained in her lexicon, e.g. in ['be.so] ‘kiss’. One thus considers the expansion of the contexts in which /b/ occurs as ‘empty’ transfer. One type of transfer has been addressed, namely forward transfer, from L1 to L2. More recently, reverse transfer, from L2 to L1, has also been reported. In a way, transfer can be compared to attrition. However, attrition implies that something has been lost, whereas here either some category is added or an already existing category expands its context (e.g. voiced stops occur between vowels as in ['lo.bo]), i.e. after a [+cont] segment. Grosjean (2011) makes an attempt to differentiate the terminology. He suggests that ‘transfer’ be used ‘for static phenomena, which reflect permanent traces of one language (La) on the other (Lb) in the bilingual’, whereas ‘interference’ should be used ‘for dynamic phenomena,’ i.e. linked to ‘processing and have to be accounted for by encoding mechanisms’. Grosjean’s attempt to clarify the terminology was welcome in the field of phonological acquisition. Suprasegmental transfer

Despite the increase of suprasegmental transfer discussion in recent decades, studies involving prosody are still limited. This situation can be better understood by considering a brief survey of major international journals on L2 acquisition, embracing 39 years of editorial activity (1969– 2008). The survey was carried out by Gut (2009) with the following results: out of 133 empirical studies on L2 phonology, only 17 dealt with prosody. The first studies tackled categorical domains around segments and prosodic categories like syllable structure; later on, the focus was on gradient domains like VOT and intonation (with gradient phonetic differences). Code-switching and Types of Code-switching

The next major topic I will discuss is code-switching. CS consists of using different languages (generally two) in one single sentence or in a clause. Muysken (2000) distinguishes three CS patterns: alternation (between the structures of the two languages), insertion (of material from one language into the structure of the other language) and congruent

42

An Anthology of Bilingual Child Phonology

lexicalization (of material from different lexicons into a shared grammatical structure). The majority of researchers use the expression ‘codeswitching’ for adults and some reserve the expression ‘code-mixing’ for children’s mixing of language (Lanza, 1992). Muysken (2000) prefers the term ‘code-mixing’ to refer to ‘all cases where lexical items and grammatical features from two languages appear in one sentence’. Researchers have proposed diff erent types of code or language switches, such as intersentential CS (between different sentences) or intrasentential (within a sentence). Myers-Scotton (2002) proposed two categories: matrix language (ML) and embedded language (EL). The former is the dominant language, which sets the stage or background, into which the lexical categories of the latter non-dominant language are inserted. Nonetheless, Auer (2002) provides empirical evidence that issues are not so straightforward, because ‘insertions do not necessarily take on the morphology of the matrix language, but may be treated according to the rules of the embedded language’. There is still an open debate as to whether CS abides by some specific rules, or whether no new rules are necessary but only the two grammars of the languages involved need to be fulfi lled. This and other questions regarding CS still await further discussion. The fi rst theoretical position involves principles like the equivalence constraint or the free morpheme condition (Poplack, 1980). Both children and adults use CS in a general, often unpredictable way, sometimes because in their community it is normal to do code-switching, or because a word or a structure is often used in CS and it is probably more activated than the equivalent word in the language serving as the ML. When referring to CS, investigators have claimed that it is often caused by a certain lack of competence, e.g. because a word is missing in the ML and the speaker borrows it from the other language. On the other hand, it has also been argued that CS is the result of a ‘specific skill of the bilingual’s pragmatic competence’, i.e. the ability to select the language according to the interlocutor (Meisel, 1993). Table 2.1 refers to 11 publications, involving seven languages (see last column in Table 2.1): English in contact with Spanish, French, Greek and Cantonese; Dutch in contact with Spanish and with Greek; and German in contact with Spanish. Most pairs of languages involve a Germanic language (English, Dutch or German) with a Romance language (Spanish or French). In one single case, a tonal language (Cantonese) is in contact with a non-tonal language (English or Norwegian). Most studies witness some kind of CLI: four cases of transfer and four of code-switching. In two cases there is no interaction. As far as the phenomena dealt with are concerned (see third column of Table 2.1), beside one segmental case of allophony (spirantization), the majority of cases (four) were related to intonation (yes/ no questions, rising contours, focus, narrow focus and tones). Further, the corpus shows that many studies focused on individual segments. Out of 39 studies concerned with individual consonants, 21

Investigating the Linguistic World of the Bilingual Child 43

Table 2.1 Research on transfer and code-switching Researchers (chronological)

Phenomena

Languages

Results

Grosjean and Miller (1994)

VOT

French/English

No interaction

Mennen (2004)

LH* rise

Dutch/Greek

Bi-directional T

Lleó et al. (2004)

L > H* rise

German/Spanish

Uni-directional T

Holm and Dodd (2006)

Tones

Cantonese/English

No interaction

Bullock and Toribio (2005)

VOT

English/Spanish

Bi-directional CS

Alvord (2010)

Yes/no questions

Spanish/English

Uni-directional T

Antoniou et al. (2011)

VOT

Greek/English

Uni-directional CS

Olson (2012)

Pitch/V-duration

Spanish/English

Narrow focus

Olson (2013)

VOT

Spanish/English

Bi-directional CS

MacSwan and Colina (2014)

Spirant/voiced

Spanish/English

Uni-directional CS

van Maastricht et al. (2016)

Focus marking

Dutch/English

Bi-directional T

investigated the VOT of stops. Perceptual abilities were examined in 29 studies (n = 29). Aspects of L2 intonation feature in nine studies, whereas both speech rhythm and phonological processes such as flapping and coarticulation are investigated in four studies each. Three studies are concerned with the production of tone in L2 acquisition. Acquisition of L3

The acquisition of an L3 is also very often investigated nowadays, which poses the question of whether L1 or L2 will be more influential regarding the transference of properties into L3. In Chapter 2 of Leung (2009: 71–88), Chin fi nds more transfer from L2 English to L3 Spanish than transfer of L1 Chinese. Jin examines the results of L1 Mandarin, L2 English and L3 Norwegian grammaticality judgements and correction tasks and fi nds that ‘despite the similarities of the L2 and the L3, in which null objects are not allowed, the L1 grammar still has an effect on task performance in the L3. Thus, it can be concluded that ‘the influence of the L1 cannot be overlooked, even when the L2 and L3 share more similar features,’ such as when ‘L1 is a tone language like Mandarin,’ and neither L2 nor L3 are tonal. Summarizing, it is difficult to try and bring all the data under one statement, as in one case L2 is more influential than L1, but in another case the L1 Mandarin is more influential than L2 English on L3 Norwegian, in spite of the similarities between L2 and L3. Table 2.2 The acquisition of an L3 L2 English + (Chin)

L3 Spanish

L1 Mandarin + (Jin)

L2 English

L3 Norwegian

L2 Rom +

L2 English/Rom +

L3 English/Rom

L1 Chinese (Chin)

44

An Anthology of Bilingual Child Phonology

Experimental Research

CS has inspired a lot of experimental research. The fi rst tests adopted structures that had become well-established in phonetic experimentation, e.g. on words preceded by a carrier sentence as in the following examples: Escriba la palabra /pa/. (Elman et al., 1977) Il faut qu’on categorise /de/. (Bürki-Cohen et al., 1989) Il faut qu’on copie /Paul/ constamment. (Grosjean & Miller, 1994)

First stimuli to test VOT in CS utterances

Grosjean and Miller (1994) reached the conclusion that the CS is phonetically immediate, cost-free and complete. This conclusion was challenged by Toribio et al. (2005). The switch between the ML (also called basic or precursor language) and the EL (embedded language or guest stimuli) is a total switch, both a switch of the morphemes and a phonetic switch (Sankoff, 2001). However, Bullock and Toribio (1994) and Toribio et al. (2005), applying a more sensitive method to the phonetic analysis of the switched VOT stimuli, found reasons to support the view that the phonetic switch was not complete, and that there was some CLI.

More on experimental research into CS

Following up on CS and VOT research, Toribio et al. (2005) designed two experiments. Experiment 1 was run on Spanish L2 speakers of English with age of exposure (AoE) around 12 years. Speakers produced separate phonetic categories for Spanish /p t k/ (25 ms) versus English (55 ms). CS promotes convergence in VOT, but with an asymmetric effect: it affects the L2, not the L1. Experiment 2 was run with L1 English late-learners of Spanish. If L1 is more resistant than L2, their L2 Spanish language VOTs should rise towards their dominant language, English. The hypotheses were confi rmed. Bullock and Toribio (2005) continued to investigate language dominance and performance outcomes in bilingual pronunciation. L1 speakers of Spanish had to switch from Spanish to English, and L1 speakers of English had to switch from English into Spanish. Results showed that L1 Spanish speakers converged with monolingual Spanish especially before switch and also at switch. L1 speakers of English’ convergence towards Spanish appeared at pre-switch and at switch, but with reduced values. The authors conclude that ‘the dominant L1 phonetic system is not deterministic of bilingual performance’ (Bullock & Toribio, 2005: 291).

Investigating the Linguistic World of the Bilingual Child

45

Three phases of phonetic CS: Pre-switch, switch and post-switch

The most recent studies on CS have focused on phonology and phonetics. This is rather new, as in the past the phonetics of CS were ignored, or rather taken for granted, because some researchers were convinced of fi nding total switch from one language to the other. However, studies like Toribio et al. (2005) had some years ago already shown that there was more to it than perfect convergence. Nowadays most recent work on CS has confi rmed such a hypothesis, although we still do not know how sounds converge in general to certain aspects of language. One of the few things we do know is that there are three points at which the speaker doing CS may be influenced by the ML, which can exert its CLI into the language involving CS: pre-switch, switch and post-switch. In other words, despite the fact that CS involves both a switch at the lexical level and a switch at the phonological level, there is a dearth of studies addressing the phonetics of code-switching. (a) Spanish to English Todos mis amigos | pre-switch

talked Spanish as | switch

kids | post-switch

(b) English to Spanish The typhoon damaged | pre-switch

techos | switch

paredes | post-switch

y

Segmental code-switching again: Non-continuous features (MacSwan & Colina, 2014)

Tests applied to bilinguals – HL as well as L2 – confirmed that it was advisable to measure the phonetics and phonology of CS together with other measurements, like VOT (short lag, long lag and pre-voicing), applied to vowel length or to delayed peak, etc. All three phenomena (VOT, vowel length and delayed peak) go back to the use of milliseconds (ms) as the main unit of calculation, together with percentages (%) of pronunciation of certain categories like codas, unfooted syllables, spirants and further phenomena. In fact, besides VOT, only a few such tests have been implemented. However, ms can be used in other measuring tasks, certainly not only in relation to VOT. Let us imagine that one runs a test along the lines of MacSwan and Colina (2014), with the following hypotheses in mind (repeated here for convenience): H1. Although the extant literature has evidence to propose that phonetic CS implies three phases related to any phenomenon involving categorical perception – pre-switch, switch and post-switch, where the ML is

46

An Anthology of Bilingual Child Phonology

Spanish and the EL is German or vice versa – I predict that the preswitch and switch will show values converging towards the values of the EL. H2. The post-switch could go in one of two directions, either back to the ML or ahead towards the values of the EL.

Back to segmental code-switching? (MacSwan & Colina, 2014)

In ‘Some consequences of language design: Code-switching and the PF interface’, MacSwan and Colina (2014) propose to go back to segmental CS, but in a new way, based on the Minimalist Program for Syntax and Optimality Theory for Phonology/Phonetics. They ran two experiments: (a) five simultaneous Spanish-English bilinguals from Central Arizona had to pronounce 27 sentences, three times each. In Experiment 1, sentences were of the type Hablamos de mi ___ yesterday. In the blank space, English words of the type book, disk, ghost, i.e. monosyllables beginning with a voiced stop consonant, should be produced, in order to measure the VOT in the pre-switch, switch and post-switch. (b) MacSwan and Colina (2014) analyze CS with the aim of discovering the points at which the switches are to be expected. (11) Ha(β)lamos de mi ↑ book yesterday [b] or [β]? disk ghost ↑= code-switching point from Spanish to English In Experiment 2, phonetic voicing rules of Spanish of the type /s/ → [z], i.e. the sibilant /s/ in the coda is voiced when followed by a voiced consonant like /ð/ as in →['dezðe] ‘since’, or /m/ as in →['miz. mo] ‘same’, →[laz.ðos] ‘the two’, →['trez. 'ma. nos] ‘three hands’; but not if followed by a vowel, as in →[tɾe. 'so.sos] ‘three bears’. The feature [+voice] can serve as a trigger for the Spanish voicing process independently of its belonging to an English lexical item. The question was whether a word-initial English segment would serve as a trigger for a Spanish process that has as its target a word-fi nal consonant. Both hypotheses could be confi rmed. (12) [dezðe], [laz.ðos]>, [tɾez 'ma. nos]> [s] or [z]? [+voice] = trigger for Spanish process that has as target a word-fi nal consonant. (13) Ha(β)lamos(z) de(ðe) mis(z)

Investigating the Linguistic World of the Bilingual Child 47

MacSwan and Colina (2014) made a very strong claim, namely that CS involves phonetic switch. Before such a claim is endorsed, more research is needed. Similarities and Differences between Transfer and Code-switching

Here I integrate the discussion of transfer and code-switching. Transfer is something not yet acquired in one of the languages of the bilingual which is replaced by some equivalent from the other language. It is not clear from the literature whether functions or lexemes are more frequently replaced. If a lexical unit is replaced by a lexical unit from the other language, this is generally considered ‘borrowing’; researchers have also argued that it is probably due to individual preferences (Poplack & Sankoff, 1984). Codeswitching involves producing a change or switch into the other language. The motivation for one or the other seems to be different. Psycholinguists often use the term ‘language switching’ (LS). As Treffers-Daller (2009) writes, what often happens is that different terminology is used to characterize data and phenomena that are the same, but each discipline tends to use its own terminology. Methods: Measuring Criteria

Given the limitations of studies as noted above, the field needs to establish new methods for experimentation. The criteria used to predict bilingual outcomes are coarse rather than fi ne (e.g. input is characterized as ‘rich’, ‘incomplete’, etc.). Unsworth (2005) proposed quantifying the input (e.g. in percentages). For the time being, one should try to reach two purposes: on the one hand, I suggest quantifying child outputs (in percentages, ms, Hz, etc.), focusing on measurements that provide distinctive information. On the other hand, one should not renounce discovering the expected predictability of bilingual outcomes. Universal criteria are also relevant: children apply processes (or constraints) that are typical of child language, such as truncation of unfooted syllable, deletion of coda consonant, reduplication of the stressed syllable of the word replacing the initial syllable, etc. (see Lleó, 1990). On the one hand, phonological criteria are categorical, because phonology is categorical, and thus provides limited information on what precise values are being transferred. On the other hand, phonetic criteria are gradient and can provide precise information on what values are being transferred. Measurements Involving CS (Lleo & Ulloa, 2019)

Code/language switching is ‘phonological switching’ (MacSwan & Colina, 2014) and phonetic switching (Lleó, 2018). If we want to decide

48

An Anthology of Bilingual Child Phonology

whether 2-year-olds produce CS, we need more precise tools to measure and compare. But we must also know what the cause of the switch is. For instance, it is important to know whether the cause is sociolinguistic. In this case, it would be appropriate to try collaborative work together with sociolinguists. If the speaker rather wants to brag, s/he will make a total twist. However, if s/he uses a word of the other language because s/he lacks the word in the L1, s/he might not switch completely. Here there are some suggestions for measurements that can be successfully applied to CS: • • • • • • • •

VOT: short lag, long lag, pre-voicing or voicing lead; Vowel length/duration: long versus short in ms; Spirants: % of spirant production; Codas: % of coda production; Delayed peak in ms from end of stressed syllable; Pitch ranges in Hz; Rhythm in PVIs (pairwise variability indices); etc.

VOT has been one of the most productive areas in which to study phonological acquisition of L1 as well as L2, which has included monolingual children (Allen, 1985; Macken & Barton, 1980), monolinguals and bilinguals (Kehoe et al., 2004; Splendido, 2014). These are possible suggestions for expanding our views of data analysis of CS, involving bilingual children. A fi rst analysis of some data produced by Laura and Ariadna have led to the following generalizations: First stage: one word utterances, either German or Catalan, random at first and progressively in the right context. Second stage: functional-lexical mixings. The German Det (including the negative ) should win, because with three features (gender, number and case) it is the more transparent Det system. Children’s productions may contain some mixed utterances at the syntactic level. However, grammar is often German (e.g. word order is German), and phonetics contains some German, too. Third stage: multiword utterances. Words are in general Catalan, but depending on how complete the acquisition of the heritage language is, grammar (i.e. word order) may still be German. ‘Code/language switching is phonological switching’ (MacSwan & Colina, 2014). This cannot be the case, given CLI. At least the switch is phonological but phonetic as well. Note that there have been some earlier attempts to summarize bilingual children’s development, such as Volterra and Taeschner (1978), who also distinguish three stages of development. In the first stage there is only one lexicon with words from both languages; in the second stage there are two lexica but one single system of syntactic rules; in the third stage, both the phonology and syntax are already differentiated. Despite the similarities of

Investigating the Linguistic World of the Bilingual Child 49

these three stages with the stages posited for Laura and Ariadna, there are also important differences between Volterra and Taeschner (1978) and the present chapter. In fact, the similarities end here, as the characterization proposed here is compatible with two lexica and two syntactic components. These authors report on a faster acquisition of the two lexica compared to the slow acquisition of Catalan and Spanish, as the HLs of Laura and Ariadna, probably due to the fact that Laura and Ariadna were acquiring three languages at the same time, not just two. Moreover, the second and third stages here are distinguished by extreme CLI, whereas in Volterra and Taeschner (1978) the authors claim that both lexicon and syntax are already acquired in both languages, German and Italian. Proposal to Run a Code (or Language) Switching Task

The ideal conditions in which to study CS would be to observe some population practicing CS often, that is, studying it in a naturalistic way. Unfortunately, things are not that easy because there are few such populations, or even if there are, they are not readily accessible. What is generally done is to stimulate the production of CS, which is then observed and analyzed. So far so good. The problem, though, is that when they are observed, people often avoid behaviors that they consider incorrect or clumsy, so that the researcher must introduce some kind of stimulus to accelerate the process of getting more naturalistic data. What has generally been practiced in the analyses of CS has been to give the participant some kind of text (e.g. a dialogue) to be read. This method, though, poses an important problem that I would want to avoid. Briefly put, the input that one obtains by means of reading is very different (more formal) from the natural use of language, and the text that reaches the analyst will probably lack representability. Instructions for E(xperimenter) to run the test: P(articipants) and data-gathering procedure

In order to contribute to our understanding of CS, a test based on MacSwan and Colina (2014) will be run. First, one will begin with interviews of ca. 20-year-old adults (German-Spanish bilingual students of a German University). The intention is to record the reactions of the participants (Ps), whose task will be to describe some scenes on paper or on the screen of a tablet, for example a girl getting a haircut in the hair salon. Pictures should support P’s descriptions. P is expected to produce about 30 sentences in German and 30 in Spanish, and these should contain stops with the three PAs. The stimuli will not have to be the same in both languages, provided that they require certain sounds, such as the stops with their various PAs. Since reading must be avoided, one will use flags to indicate whether the ML in each case is German or Spanish. At crucial spots (pre-switch, switch, post-switch) there will be switches that allow us

50

An Anthology of Bilingual Child Phonology

to measure the phonetic values of certain consonants, such as VOT values for the Spanish and for the German stops. Once the test has been done with students, it should be run with ca. 6-year-olds. P should produce the sentence in the language of the flag (in German if the flag is German or in Spanish if the flag is Spanish). Although this instruction could be given orally, doing it with flags avoids possible phonetic influences from the experimenter (E) onto the P. E shows the picture of an object or an action, and P names it with a sentence in the same language as before. The stimuli will not have to be the same in both languages, provided that they require certain sounds, such as the stops with their various PAs. That is, P should produce around 30 sentences in Spanish and sentences in German, out of which 10 must be labials, 10 coronals and 10 dorsals. The sentences with CS begin in German and continue in Spanish, or vice versa, beginning in Spanish and continuing in German. E alternates flags and P repeats the sentence with and without CS. Flags should help to clarify the actions, and the exact point of language switches. They can be substituted by some other object, especially when running the test on children.

Experimental data: CS in German and Spanish

Since the reading of a text is methodologically problematic, one will use images that present people in action, that is, performing some action that the participant should describe and also code-switch depending on the position of a little flag that, being German or Spanish, indicates what language is expected. See the following examples. SPANISH-SPANISH: Sp –cs (14) Pedro cree que Pablo tiene coche car’ Sp –cs

–‘Pedro thinks that Paul owns a

GERMAN-GERMAN: Ge –cs (15) Petra glaubt, dass Paul zum Geburtstag kommt ‘P. thinks that Pa. comes to the birthday party’ Ge –cs SPANISH-GERMAN: Sp +cs (16) Maria dice dass sie ihre Haare schneiden lassen wird ‘Maria says that she will have a hair-cut’ Sp +cs GERMAN-SPANISH: Ge +cs (17) Carmen glaubt, dass Peter vendrá a la fiesta de cumpleaños / Ge +cs ‘Carmen believes that C. vendrá a la fiesta de cumpleaños SPANISH-GERMAN: Sp +cs (18) María dice dass sie ihre Haare schneiden lassen wird ‘María dice que se hará cortar el pelo’ / Maria sagt dass sie ihre Haare schneiden lassen wird.’ Ge +cs

Investigating the Linguistic World of the Bilingual Child 51

SPANISH-GERMAN: (19) Carmen cree dass Pablo zum Geburtstag kommt ‘C. thinks that P. owns a car.’ Sp +cs GERMAN-SPANISH: Ge +cs (20) Clara glaubt que Pablo tiene coche car’ Ge +cs

‘Ca. believes that Pa. owns a

Questions and Answers

Earlier on, I posed some research questions related to the bilingual outcomes of languages in contact depending on the type of bilingualism, i.e. the modality of language acquisition that has taken place. The questions are repeated here for convenience. (a) Are there quantitative similarities and differences between monolingual, heritage speakers (HS) and L2 speakers? (b) Is transfer a necessary phonological result of languages in contact? (c) Is child code-mixing comparable to adult code-switching? Answers to the research questions can be derived from the information in (10), which refers to the percentages of target-like productions by monolinguals and some types of bilinguals (from Lleó & Ulloa, 2019: 277). ‘Monolinguals’ refer to speakers of one single language, which is generally called L1, without knowing whether there will be an L2, i.e. before we know whether other languages will be learned by that child; ‘cL2’ refers to an L2 acquired by a child, the cutting-off point being around the age of 5 years; ‘AoE’ is age of exposure, which here is defined as ca. 5-year-old children, i.e. monolingual children who begin to learn an L2 at about age 5 years; ‘HL’ is the heritage language (see the Introduction to this chapter), which can be acquired simultaneously to or earlier than the dominant language (2L1). (10) L1 (monolinguals) > cL2 (bilinguals) > HL (heritage or simultaneous bilinguals) Conclusion

(Qa) The scale in (10) refers to the percentages of target-like productions by monolinguals and some types of bilinguals to defi ne the possible outcomes of languages in contact. It shows that from left to right there is a decrease of scores in relation to the various groups of speakers: in our studies of German-Spanish bilinguals and monolinguals, monolinguals have obtained the best scores both in relation to spirants and to PA assimilation (98%). The cL2 group follows, with 85% of spirants and 72% of PA assimilation. Finally, HL had 59% of spirants and 90% of PA assimilation. Speakers who had the lowest scores were the HL speakers with regard to spirantization. They not only had lower percentages than L1 speakers, but also lower than those of cL2.

52

An Anthology of Bilingual Child Phonology

Once researchers became interested in phenomena that resulted from bilingualism or languages in contact, two of these outcomes prevailed as the most important and the most frequent: transfer and code-switching. Although for a long time transfer and CS were approached as being very different, with the sole shared property of involving language contact, lately their similarities are seen as greater than their differences (TreffersDaller, 2009). One of the reasons for seeing them as distinct was that transfer is often used in SLA to account for production errors, whereas CS or LS is observed in normal conversations. In agreement with the literature and contemporary beliefs, I had expected that HL speakers would probably have almost as good a proficiency as monolingual speakers, or anyway better scores than cL2. (Qb) Is transfer a necessary phonological result of languages in contact? According to the literature and to several results obtained in our research, transfer seems indeed to be unavoidable, in those cases when something is being acquired. As already discussed, research over the last decades has found much cross-language influence between the two languages of the bilingual speaker. However, we do not yet have a clear global picture of the precise outcomes of these situations. Much more research will be necessary to clarify interaction between languages. (Qc) Code-switching: Is child code-mixing comparable to adult codeswitching? Some researchers do not distinguish between switching and mixing. However, before we understand CS and its context, unifying the various properties does not appear to be an optimal solution. If we follow those researchers who consider CS to have sociolinguistic causes, and code-mixing to be due above all to incomplete lexical information, we will be able to explain the relatively limited percentages of code-mixing produced by young children and the fact that as children grow older CS is generally not reduced but increased. When in 1996 Paradis and Genesee tried to defi ne the possible outcomes of languages in contact based on the empirical acquisition data of that time, they did an excellent job, trying to account for essentially different outcomes, considering the negative light under which bilingualism had appeared from the beginning of the 20th century. They came up with delay, acceleration and transfer. Delay had often been associated with bilingualism, but acceleration was rather new in this context and it came as a surprise, which in fact had to wait until OT was capable of providing a plausible explanation, showing that markedness and frequency go hand in hand. The latter outcome (T) had to be mentioned as well, because SLA had often found it in the data during the second half of the 20th century, especially transfer from L1 to L2. Later, further factors, namely different order of acquisition and fusion or blending, were found. Transfer and code (or language) switching have a clear similarity, namely both notions involve CLI. Both notions refer to the influence that one language may exert on the other when they are in a contact situation.

Investigating the Linguistic World of the Bilingual Child 53

The precise value of measuring phenomena related to contact was not considered a few decades ago, because phonetics was left out of the analyses. However, phonetics is now included, and studies dealing with prosodic CLI are increasing constantly (cf. Table 2.1). In this context, running a test on CS deepens the new perspective towards a more specific relationship between transfer and CS. The hypotheses that I formulated had a parallel format with regard to spirantization and PA assimilation of nasals; that is, I hypothesized that sequential bilinguals would fail to apply these two processes, which would yield some foreign accent in their variety of Spanish. This was indeed the case, because bilingual children omitted assimilation processes quite often, especially spirantization which was omitted more frequently than PA assimilation of nasals. One more surprising result involved the mean percentages of the HL group, which were not only lower than those of monolinguals but also lower than those of the cL2 group. I had expected that HL speakers, being exposed to the HL very early, would probably have better scores than cL2, who are exposed to L2 later on. Much more research will be necessary to clarify the CLI picture. Later, further factors, namely different order and fusion or blending were found, too. Once CS appeared as one more phenomenon to expand the outcomes of bilingualism or languages in contact, out of those outcomes two prevailed as the most important and the most frequent ones: transfer and CS. The perceived similarity between transfer and CS is relatively new, but in my view very appealing, since both notions involve CLI. Considering that in 1966 we hardly had any knowledge about the simultaneous acquisition of two languages from a (psycho-)linguistic point of view, much progress has been made in the research into this area. We have not only learned a lot about bilingualism but also about multilingualism in general. The most appealing aspect of this development is theoretical: at one point the role of phonology in acquisition was not yet understood and phonologists tried to defi ne it as generative (see the Introduction to this chapter). However, not even syntax is seen as generative nowadays, because on the one hand the tools of analysis are not rules but constraints, and the resulting entities are not derivations but representations. On the other hand, phonology and prosody are the parts of grammar that have witnessed more development. And still, the area covered by a chapter like this one is minimal, and in need of more research. Acknowledgements

I am very thankful to the Deutsche Forschungsgemeinschaft, to the University of Hamburg, to the SFB (Sonderforschungsbereich) 538 and to Dr Margaret Kehoe research assistant for the project PhonBLA during the first three years, as well as to the children and their parents who participated in the projects.

54

An Anthology of Bilingual Child Phonology

References Allen, G.D. (1985) How the young French child avoids the pre-voicing problem for wordinitial voiced stops. Journal of Child Language 12, 37–46. Alvord, S.M. (2010) Variation in Miami Cuban Spanish Interrogative Intonation. Hispania 93 (2), 235–255. Antoniou, M., Best, C.T., Tyler, M.D. and Kroos, C. (2011) Inter-language interference in VOT production by L2 dominant bilinguals: Asymmetries in phonetic codeswitching. Journal of Phonetics 39, 558–570. Auer, P. (2002) ‘Embedded language’ and ‘matrix language’ in insertional language mixing: Some problematic cases. Rivista di Linguistica 17 (1), 35–54. Barlow, J.A. (2003) The stop-spirant alternation in Spanish: Converging evidence for a fortition account. Southwest Journal of Linguistics 22, 51–86. Bullock, B.E. and Toribio, A.J. (eds) (2005) Perseverative phonetic effects in bilingual code-switching. Amsterdam Studies in the Theory and History of Linguistic Science Series 4, 291–306. Bullock, B.E., Toribio,A.J., González, V. and Dalola, A. (2006) Language dominance and performance outcomes in bilingual pronunciation. In M.G. O’Brien, C. Shea and J. Archibald (eds) Proceedings of the 8th Generative Approaches to Second Language Acquisition Conference (GASLA 2006): The Banff Conference (pp. 9–16). Somerville, MA: Cascadilla Proceedings Project. Bürki-Cohen, J., Grosjean, F. and Miller, J.L. (1989) Base language effects on word identification in bilingual speech: Evidence from categorical perception. Language and Speech 32, 355–371. Chiat, S. (1989) The relation between prosodic structure, syllabification and segmental realization: Evidence from a child with fricative stopping. Clinical Linguistics and Phonetics 3, 223–242. Clark, J. and Clark, C.Y. (1995) An Introduction to Phonetics and Phonology. Oxford: Blackwell. DeKeyser, R. (2012) Age effects in second language learning. In S.M. Gass and A. Mackey (eds) The Routledge Handbook of Second Language Acquisition (pp. 442–460). London/New York: Taylor & Francis. Delattre, P. (1965) Comparing the Phonetic Features of English, French, German and Spanish: An Interim Report. Heidelberg: Groos. Eckman, F.R. (2012) Second language phonology. In S.M. Gass and A. Mackey (eds) The Routledge Handbook of Second Language Acquisition (pp. 91–105). London/New York: Taylor & Francis. Elman, J.L., Diehl, R.L. and Buchwald, S.E. (1977) Perceptual switching in bilinguals. Journal of the Acoustical Society of America 62, 4. Fabiano-Smith, L.C. and Goldstein, B.A. (2010) Phonological acquisition in bilingual Spanish-English speaking children. Journal of Speech, Language, and Hearing Research 53 (1), 160–178. Fikkert, P. and Levelt, C. (2008) How does Place fall into place? The lexicon and emergent constraints in children’s developing phonological grammar. In P. Avery, B.E. Dresher and K. Rice (eds) Contrast in Phonology: Perception and Acquisition (pp. 219–256). Berlin: Mouton. Flege, J.E. (2007) Language contact in bilingualism: Phonetic system interactions. In J. Cole and J.I. Hualde (eds) Laboratory Phonology 9 (pp. 353–382). Berlin/New York: Mouton de Gruyter. Gnanadesikan, A.E. (1995) Markedness and faithfulness constraints in child phonology. MS ROA–67, University of Massachusetts at Amherst. Grosjean, F. (1997) The bilingual individual. Interpreting 2 (1/2), 163–187. Grosjean, F. (2011) An attempt to isolate, and then differentiate, transfer and interference. International Journal of Bilingualism 16 (1), 11–21.

Investigating the Linguistic World of the Bilingual Child 55

Grosjean, F. and Miller, J.L. (1994) Going in and out of languages: An example of bilingual flexibility. Psychological Science 5 (4), 201–206. Gut, U. (2009) Non-native Speech: A Corpus-based Analysis of Phonological and Phonetic Properties of L2 English and German. Berlin: De Gruyter. Harris, J.W. (1982) Theories of phonological representation and nasal consonants in Spanish. XIIth Linguistic Symposium on Romance Languages (pp. 153−168). State College, PA: Penn State University. Harris, J.W. (1984) La espirantización en castellano y la representación fonológica autosegmental. Estudis Gramaticals 1. Working Papers in Linguistics (pp. 149−167). Bellaterra: Universitat Autònoma de Barcelona. Holm, A. and Dodd, B. (2006) Phonological development and disorder of bilingual children acquiring Cantonese and English. In Z. Hua and B. Dodd (eds) Phonological Develop ment and Disorders in Children (pp. 286−325). Clevedon: Multilingual Matters. Jakobson, R. ([1941] 1968) Kindersprache, Aphasie und allgemeine Lautgesetze [Child Language, Aphasia and Phonological Universals]. Uppsala: Almqvist & Wiksell. Kehoe, M., Lleó, C. and Rakow, M. (2004) Voice onset time in bilingual German-Spanish children. Bilingualism: Language and Cognition 7, 71−88. Kenstowicz, M. (1998) Uniform exponence: Exemplifi cation and extension. Rutgers Optimality Archive 218. Kupisch, T. (2013) A new term for a better distinction? A view from the higher end of the proficiency scale. Theoretical Linguistics 39, 203–214. Kupisch, T. and Rothman, J. (2018) Terminology matters! Why difference is not incompleteness and how early child bilinguals are heritage speakers. International Journal of Bilingualism 22 (5), 564–582. Lanza, E. (1992) Can bilingual two-year-olds code-switch? Journal of Child Language 19, 633–658. Lardière, D. (2003) Revisiting the comparative fallacy: A reply to Lackshmanan and Selinker 2001. Second Language Research 19, 129–143. Leung, Y.I. (ed.) (2009) Third Language Acquisition and Universal Grammar. Clevedon: Multilingual Matters. Lleó, C. (1990) Homonymy and reduplication: On the extended availability of two strategies in phonological acquisition. Journal of Child Language 17, 267–278. Lleó, C. (2002) The role of markedness in the acquisition of complex prosodic structures by German-Spanish bilinguals. International Journal of Bilingualism 6 (3), 291–313. Lleó, C. (2012) Monolingual and bilingual phonoprosodic corpora of child German and child Spanish. In T. Schmidt and K. Wörner (eds) Multilingual Corpora and Multilingual Corpus Analysis (pp. 107–122). Hamburger Studies on Multilingualism No. 14. Amsterdam/Philadelphia, PA: John Benjamins. Lleó, C. (2015) On the permeability of German-Spanish bilinguals’ phonological grammars. In E. Babatsouli and D. Ingram (eds) Proceedings of the International Symposium on Monolingual and Bilingual Speech 2015 (pp. 76–108). See http:// ismbs.eu/publications. Lleó, C. (2018) German-Spanish bilinguals’ phonological grammars: Permeable or resilient? In E. Babatsouli and D. Ingram (eds) Phonology in Protolanguage and Interlanguage (pp. 76–108). Studies in Phonetics and Phonology. Sheffield: Equinox. Lleó, C. and Cortés, S. (2013) Modelling the outcome of language contact in the speech of Spanish-German and Spanish-Catalan bilingual children. In J. Kabatek and L. Loureido (eds) Special Issue on ‘Language Competition and Linguistic Diff usion: Interdisciplinary Models and Case Studies’. International Journal of the Sociology of Language 221, 101–125. Lleó, C. and Demuth, K. (1999) Prosodic constraints on the emergence of grammatical morphemes: Crosslinguistic evidence from Germanic and Romance languages. In A.

56

An Anthology of Bilingual Child Phonology

Greenhill, H. Littlefield and C. Tano (eds) Proceedings of the 23rd Annual Boston University Conference on Language Development (Vol. 2, pp. 407–418). Somerville, MA: Cascadilla Press. Lleó, C. and Rakow, M. (2003) Markedness effects in voiced stop spirantization in bilingual German-Spanish children. In J. Cohen, K.T. McAlister, K. Rolstad and J. MacSwan (eds) Proceedings of the 4th International Symposium on Bilingualism (ISB4) (pp. 1353–1371). CD-Rom. Somerville, MA: Cascadilla Press. Lleó, C. and Rakow, M. (2006) Nasalassimilation und Prosodische Hierarchie im monolingualen und bilingualen Erwerb des Spanischen und des Deutschen. In C. El Mogharbel and K. Himstedt (eds) Phonetik und Nordistik. Festschrift für Magnús Pétursson zum 65. Geburtstag (pp. 95–117). Frankfurt am Main: Theo Hector. Lleó, C. and Ulloa, M. (2019) An analytical approach to teaching Spanish pronunciation to native speakers of German: First language and age of exposure as crucial factors. In R. Rao (ed.) Key Issues in the Teaching of Spanish Pronunciation: From Description to Pedagogy (pp. 268–287). London: Routledge. Lleó, C., Kuchenbrandt, I., Kehoe, M. and Trujillo, C. (2003) Syllable fi nal consonants in Spanish and German monolingual and bilingual acquisition. In N. Müller (ed.) (In) vulnerable Domains in Multilingualism (pp. 191–220). Amsterdam/Philadelphia, PA: John Benjamins. Lleó, C., Rakow, M. and Kehoe, M. (2004) Acquisition of language-specific pitch accent by Spanish and German monolingual and bilingual children. In T. Face (ed.) Laboratory Approaches to Spanish Phonology (pp. 3–27). Berlin/New York: Mouton. Lleó, C., Ulloa, M. and García Sánchez, I. (2013) Promoting the weak language of German-Spanish bilingual children living in Germany. In K. Bührig and B. Meyer (eds) Transferring Linguistic Know-how into Institutional Practice (pp. 99–120). Hamburg Studies on Multilingualism (HSM 15). Amsterdam/Philadelphia, PA: John Benjamins. Macken, M.A. and Barton, D. (1980) The acquisition of the voicing contrast in Spanish: A phonetic and phonological study of word-initial stop consonants. Journal of Child Language 7, 41–74. MacSwan, J. and Colina, S. (2014) Some consequences of language design: Codeswitching and the PF interface. In J. MacSwan (ed.) Grammatical Theory and Bilingual Codeswitching (pp. 185–210). Cambridge, MA: MIT Press. Meisel, J.M. (1993) Grammatical constraints on code-switching in young bilinguals. Paper presented at the Sixth International Congress for the Study of Child Language, Trieste. Mennen, I. (2004) Bi-directional interference in the intonation of Dutch speakers of Greek. Journal of Phonetics 32, 543 –563. Montrul, S.A. (2008) Incomplete Acquisition in Bilingualism: Re-examining the Age Factor. Amsterdam/Philadelphia, PA: John Benjamins. Muysken, P. (2000) Bilingual Speech: A Typology of Code Mixing. Cambridge: Cambridge University Press. Myers-Scotton, C. (2002) Contact Linguistics. Oxford: Oxford University Press. Olson, D.J. (2012) The phonetics of insertional code-switching: Suprasegmental analysis and a case for hyper-articulation. Linguistic Approaches to Bilingualism 2 (4), 439–457. Olson, D.J. (2013) Bilingual language switching and selection at the phonetic level: Asymmetrical transfer in VOT production. Journal of Phonetics 41, 407–420. Ota, M. (1998) Phonological theory and the acquisition of prosodic structure: Evidence from child Japanese. PhD thesis, Georgetown University. Paradis, J. and Genesee, F. (1996) Syntactic acquisition in bilingual children: Autonomous or interdependent? Studies in Second Language Acquisition 18, 1–25.

Investigating the Linguistic World of the Bilingual Child 57

Pater, J. (1997) Minimal violation and phonological development. Language Acquisition 6, 201–253. Poplack, S. (1980) Sometimes I’ll start a sentence in Spanish y termino en español: Toward a typology of code-switching. Linguistics 18 (7/8), 581–618. Poplack, S. and Sankoff, D. (1984) Borrowing: The synchrony of integration. Linguistics 22, 99–135. Prince, A. and Smolensky, P. (1993) Optimality Theory: Constraint Interaction in Generative Grammar. ROA Version 8/2002. Rakow, M. and Lleó, C. (2003) On the (un)markedness of spirantization: Evidence from fi rst language acquisition. In M.J. Solé, D. Recasens and J. Romero (eds) Proceedings of the 15th International Congress of Phonetic Sciences (pp. 1025–1028). Barcelona: Universitat Autònoma. Rothman, J. (2007) Heritage speaker competence differences, language change, and input type: Inflected infi nitives in Heritage Brazilian Portuguese. International Journal of Bilingualism 11, 359–369. Sankoff, G. (2001) Linguistic outcomes of language contact. In P. Trudgill, J. Chambers and N. Schilling-Estes (eds) Handbook of Sociolinguistics (pp. 638–668). Oxford: Basil Blackwell. Schmidt, T. and Wörner, K. (eds) (2012) Multilingual Corpora and Multilingual Corpus Analysis. Hamburg Studies on Multilingualism No. 14. Amsterdam/Philadelphia, PA: John Benjamins. Skehan, P. (2012) Language aptitude. In S.M. Gass and A. Mackey (eds) The Routledge Handbook of Second Language Acquisition (pp. 381–395). London/New York: Taylor & Francis. Splendido, F. (2014) Le développement d’aspects phonético-phonologiques du français chez des enfants bilingues simultanés et successifs. Le VOT et la liaison dans une étude de cas multiples. Etudes Romanes de Lund 92, 314. Tesar, B. and Smolensky, P. ([1993] 2000) Learnability in optimality theory. Linguistic Inquiry 29, 229–268. Toribio, A.J., Bullock, B.E., Botero, C.G. and Davis, K.A. (2005) Perseverative phonetic effects in bilingual code switching. In R. Gess and E. Rubin (eds) Theoretical and Experimental Approaches to Romance Linguistics (pp. 206–291). Amsterdam/ Philadelphia, PA: John Benjamins. Treffers-Daller, J. (2009) Code-switching and transfer: An exploration of similarities and differences. In B.E. Bullock and A.J. Toribio (eds) The Cambridge Handbook of Linguistic Codeswitching (pp. 58 –74). Cambridge Handbooks in Language and Linguistics. Cambridge: Cambridge University Press. Unsworth, Sh. (2005) Comparing the role of input in bilingual acquisition across domains. In T. Grüter and J. Paradis (eds) Input and Experience in Bilingual Development. Amsterdam: John Benjamins. van Maastricht, L., Krahmer, E. and Swerts, M. (2016) Prominence patterns in a second language: Intonational transfer from Dutch to Spanish and vice versa. Language Learning 66 (1), 124–158. Volterra, V. and Taeschner, T. (1978) The acquisition and development of language by bilingual children. Journal of Child Language 5 (2), 311–326.

3 Seeking Crosslinguistic Interaction in French Bilingual Phonological Development Margaret Kehoe

Introduction

In recent years, researchers have compared the phonetic/phonological production skills of monolingual and bilingual children in order to determine whether the two groups differ from one another. When differences have been found, researchers have been interested in identifying what factors account for these differences. One factor that might lead bilinguals to differ from monolinguals is the grammatical influence of one of the bilingual’s languages upon the other, a phenomenon referred to as crosslinguistic interaction (Paradis & Genesee, 1996). Interaction between two languages may manifest itself in changes in the rate of development. For example, bilinguals may acquire phonological structures at a faster or slower rate than monolinguals. Interaction may also include the presence of non-native sounds in one language because of transfer from the other. Many studies provide evidence of crosslinguistic interaction (Almeida et al., 2012; Fabiano-Smith & Goldstein, 2010; Keffala et al., 2018; Lleó et al., 2003; Tamburelli et al., 2015), but the variability of fi ndings makes it difficult to form a cohesive picture of when and under what circumstances it takes place (Kehoe, 2015). Furthermore, some types of crosslinguistic interaction such as transfer are very infrequent (Fabiano-Smith & Goldstein, 2010; Goldstein et al., 2005), making researchers query how robust the phenomenon is. The aim of this study is to seek crosslinguistic interaction across different age groups and phonetic/phonological measures. Its presence across different ages and measures should provide further information about when and under what conditions crosslinguistic interaction takes place. Our population consists of French speaking monolinguals and bilinguals. The bilinguals have been exposed to French before the age of 3 years and 58

Seeking Crosslinguistic Interaction in French Bilingual Phonological Development 59

have differing L1s (i.e. the home language that is other than French). We test the French of the bilingual children only. We code the bilinguals’ L1s in terms of the complexity of the phonetic/phonological structures and from this we formulate precise predictions about the nature and direction of crosslinguistic interaction. We also include information about the bilinguals’ language experience and lexical knowledge to ensure that any monolingual–bilingual differences that are present are due to crosslinguistic interaction and not to other factors. The phonetic/phonological measures that we examine are syllable structure (word-fi nal codas and onset clusters), voice onset time (VOT) and word prosody. In the remaining sections of the Introduction, we summarize fi ndings on crosslinguistic interaction in young bilinguals in these three domains and end with a statement of the research predictions. Crosslinguistic interaction in syllable structure development

When discussing crosslinguistic interaction in syllable structure, it is important to distinguish between frequency and complexity. Frequency refers to the low or high presence of a structure as determined by syllabletype counts, whereas complexity refers to typological markedness (Gierut, 2001). A phonetic/phonological property that contains more elements and more structure is more complex than a phonetic/phonological property that contains fewer elements and less structure. Our notion of complexity is particularly influenced by treatment studies on children with phonological disorders which indicate that treatment of a complex phonological target leads to greater and more widespread phonological learning than treatment of a non-complex target (Gierut, 2001). Several studies show that the high frequency and complexity of structures in the L1 lead to accelerated production of these same structures in the L2 (i.e. the majority language). For example, Lleó et al. (2003) found that bilingual Spanish-German children produced more codas in Spanish than Spanish monolingual children due to the influence of German. German has a higher proportion of closed syllables than Spanish (67% versus 27%; Meinhold & Stock, 1980). Similarly, Keffala et  al. (2018) found that bilingual Spanish-English children produced more codas in Spanish than Spanish monolingual children due to the influence of English. English also has a higher proportion of closed syllables than Spanish (60% versus 32%; Delattre & Olsen, 1969; Guffey, 2002). In the production of onset clusters, several authors have found accelerated production of clusters in bilinguals in comparison to monolinguals due to the increased segmental or structural complexity of the L1 (Keffala et al., 2018; Tamburelli et al., 2015). For example, Tamburelli et al. (2015) found accelerated production of /s/ + obstruent clusters in the English of Polish-English bilinguals in comparison to monolinguals. Both Polish and English have clusters with small sonority differences (e.g. /sp/), but only

60

An Anthology of Bilingual Child Phonology

Polish has clusters containing sonority plateaus (e.g. /pt/) in word-initial position. The more marked sonority patterns in Polish relative to English had a facilitative effect on cluster development in English. One problem when making claims about crosslinguistic interaction is that it is not always possible to separate out the independent effects of frequency and complexity (Keffala et al., 2018). Kehoe and Havy (2019) found similar effects on word-fi nal coda production irrespective of whether frequency or complexity was coded. For the sake of simplicity, we focus only on the parameter of complexity in our analyses of crosslinguistic interaction in syllable structure. Crosslinguistic interaction in VOT

VOT refers to the time interval between the release of the articulators and the onset of vocal fold vibration. The timing relationship between these two events is expressed on a continuum such that voicing that occurs before the release of the articulators is given a negative value and voicing that occurs after the release of the articulators is given a positive value. Lisker and Abramson (1964) established three categories of voicing based on VOT: voicing lead, short lag and long lag. Many patterns of crosslinguistic interaction in VOT development have been identified in bilingual children (Fabiano-Smith & Bunta, 2012; Johnson & Wilson, 2002; Kehoe et al., 2004; Khattab, 2000; Stoehr, 2018; Stoehr et al., 2018). Some studies report evidence of delay. Fabiano-Smith and Bunta (2012) documented VOT values in the short lag region for English /p/ by their bilingual Spanish-English 3-year-olds, whereas monolingual English speaking children produce target voiceless stops in the short lag region only at the early stages of development. By 3 years, sometimes even by 2 years, monolingual English children produce them in the long lag region (Macken & Barton, 1980a). Johnson and Wilson (2002) found that their English-Japanese bilingual children produced voiceless stops with long lag values in English but they also did so in Japanese, whereas they should have produced them in the short lag region. This result suggests transfer of long lag voicing from English into Japanese. Still other studies have not documented any patterns of crosslinguistic interaction. Deuchar and Clark’s (1996) study of a Spanish-English child indicates similar findings to that of monolinguals. Their subject, Manuela, produced voiced stops as short lag and voiceless stops as long lag in English, and made a contrast in the short lag region for Spanish, which is also typical of Spanish monolingual children (Macken & Barton, 1980b). All of the above studies have examined VOT in children whereby the majority language was characterized by a long lag–short lag distinction. We are aware of only one study that has examined acquisition of VOT in bilingual children whereby the majority language contained a lead–short lag distinction. Stoehr (2018) studied the acquisition of VOT in bilingual

Seeking Crosslinguistic Interaction in French Bilingual Phonological Development

61

German-Dutch children growing up in the Netherlands. Dutch contains a lead–short lag whereas German contains a long lag–short lag distinction. Stoehr (2018) found few differences in the acquisition of VOT in the majority language, Dutch. The monolingual and bilingual children produced voiceless stops with similar VOTs (i.e. in the short lag region), although the bilingual children still produced significantly fewer stops with lead voicing than the monolingual children (50% versus 30%). Stoehr (2018) found greater differences in the acquisition of the VOT in the minority language, German. The bilinguals produced voiceless plosives with shorter VOTs than monolingual German children. They also produced a higher percentage of stops with lead voicing in German than the monolinguals. Stoehr’s (2018) fi ndings are of interest since we focus on acquisition of VOT in the majority language, French, which is characterized by a lead–short lag distinction. Crosslinguistic interaction in word prosody

The seminal work of Delattre (1966) indicates that languages differ in the prosodic characteristics of stressed and unstressed syllables. In Germanic languages such as English and German, acoustic differences between stressed and unstressed syllables are greater than in Romance languages such as Spanish, Italian and French. French differs from other Romance languages, however, in not having lexical but phrasal stress. Primary stress falls on the fi nal syllable of the last lexical item in a phonological phrase (Dell, 1984). It is characterized for the most part by increased vowel duration (Delattre, 1966); the fi nal syllable is approximately two times longer than the non-fi nal syllable (Vaissière, 1991). In addition to primary stress, an optional pitch accent may occur on the initial syllable of lexical words (Goad & Buckley, 2006; Jun & Fougeron, 2002). This pitch accent is conditioned by different factors such as speech register, the number of syllables in the word and the position of the word in the sentence (Jun & Fougeron, 2002). In this study we examine the word prosody of two-syllable words in French by comparing the duration and fundamental frequency (F0) characteristics of the initial and fi nal syllables. Several authors have investigated the rhythmic patterns of monolingual and bilingual children (Bunta & Ingram, 2007; Kehoe et al., 2011; Mok, 2013; Schmidt & Post, 2015). They have found that bilinguals may manifest delays in the acquisition of rhythm in comparison to monolinguals. Fewer authors have studied the acoustic correlates of stress but some studies exist, namely those by Rose and Champdoizeau (2007) and Dodane and Bijeljac-Babic (2017), who tested bilingual English-French children. Rose and Champdoizeau (2007) found that their bilingual participant, aged 2;0–3;0 years, already realized the language-specific acoustic characteristics of stress in her two languages. The stressed (initial) syllable versus the unstressed (fi nal) syllable

62

An Anthology of Bilingual Child Phonology

in English was characterized by increased F0 and intensity with only minimal duration differences. The stressed (fi nal) syllable versus unstressed (initial) syllable of French was characterized by increased duration with only marginal F0 and intensity differences. Thus, Rose and Champdoizeau’s (2007) bilingual child realized stress in her two languages in a similar way to monolinguals. In contrast, Dodane and Bijeljac-Babic (2017) found evidence of crosslinguistic interaction in their study of French-English bilinguals aged 3;3–6;0. The bilinguals produced overly long fi nal syllables in French, longer even than those of the French monolinguals, possibly to mark the contrast between their English and French productions. They produced high pitch accents on the initial syllable of their disyllabic productions in French, more so than they did in their English productions, which the authors interpret as transfer of the English trochaic F0 pattern to French. With this background on the three phonetic/phonological measures, we outline our research predictions. Research Predictions Syllable structure

The basis of our predictions of crosslinguistic interaction in syllable structure is that a structure that has a higher complexity in the L1 compared to the L2 should facilitate acquisition in the L2, whereas a structure that has lower complexity in the L1 should inhibit acquisition. Facilitation effects result in acceleration which we define as significantly higher correct performance for a target structure in the bilinguals’ L2 in comparison to monolinguals. Inhibition effects lead to delay which is significantly lower correct performance for a target structure in the bilinguals’ L2 in comparison to monolinguals. The L1s of the bilingual children in this study are diverse and include Romance languages (Spanish, Italian, Portuguese), Germanic languages (English, German) and other languages (e.g. Polish, Turkish, Armenian, Arabic, Cantonese). We did not use L1 as an inclusionary criterion; hence this set of languages reflects the L1s of bilinguals whose parents agreed to take part in the study. We have categorized languages into two or three complexity groupings by pooling information from multiple sources (e.g. Dryer & Haspelmath, 2013; World Atlas of Language Structures Online; phonological grammars). Word-fi nal codas in French may consist of different manners (stops, nasals, fricatives and liquids) and places of articulation (PoA; labial, coronal and dorsal). They may also be complex. We predict delay in coda production if the bilingual’s L1 belongs to a low-complexity language such as Spanish and Portuguese, since the codas in these languages are less complex than in French. We predict no change in coda production if the bilingual’s L1 belongs to a high-complexity group such as English and German, since these languages have similar degrees of complexity to French.

Seeking Crosslinguistic Interaction in French Bilingual Phonological Development

63

Onset clusters in French contain mainly obstruent liquid (OL) clusters, although they may also contain /s/ + stop clusters, which appear in loan words (e.g. ski, sport, stop) and are infrequent in children’s speech (Andreassen, 2013). We predict that children speaking languages with no onset clusters such as Arabic or Turkish will be at a disadvantage for producing clusters in French and thus we anticipate delay. We predict no change in French cluster development for children speaking Spanish and Portuguese since these languages contain OL clusters only, and thus have similar levels of cluster complexity to French. In contrast, children speaking languages that contain an extensive range of word-initial /s/ + C clusters or that contain clusters with sonority plateaus and falls (e.g. /pt, mʃ/) should show accelerated production of clusters, since these languages have higher complexity onset clusters than French. A summary of research predictions and examples of the languages included in the high-, mid- and low-complexity categories are shown in Table 3.1. Table 3.1 A summary of predictions on crosslinguistic interaction in syllable structure Linguistic

Complexity grouping

Example of languages

Prediction of crosslinguistic interaction

Word-final codas

Low coda

Italian, Portuguese, Spanish, Mandarin, Japanese

Delay

High coda

Romanian, Russian, Polish, Albanian, Czech, English, German

No change

Low onset

Arabic, Turkish

Delay

Mid onset

Catalan, Portuguese, Spanish

No change

High onset

Italian, English, German, Dutch, Polish, Romanian, Russian

Acceleration

Onset clusters

Voice onset time

In making predictions about crosslinguistic interaction in VOT, we divided bilingual children into those speaking lead–short lag and those speaking long lag–short lag languages. Since French is also characterized by a lead–short lag contrast, we predicted no difference between the monolingual and the bilingual children speaking a language with the same VOT contrast. We predicted differences, however, between the monolinguals and the bilinguals speaking languages with a long lag–short lag contrast. In the case of target voiceless stops, bilinguals may produce them with longer VOTs than monolinguals due to transfer of long lag VOT values from their L1. In the case of target voiced stops, bilinguals may produce lead voicing less often than monolinguals due to the influence of their other language which is not characterized by lead voicing. A summary of the research predictions for VOT are given in Table 3.2.

64

An Anthology of Bilingual Child Phonology

Table 3.2 A summary of predictions on crosslinguistic interaction in VOT Bilingual

Example of languages

Target voiceless stops

Target voiced group stops

Lead–short lag

Spanish, Italian, Portuguese

No change

No change

Long lag–short lag

English, German

Longer VOTs

Less lead voicing

Word prosody

In making predictions on crosslinguistic interaction in word prosody, we divided bilingual children into those speaking Romance languages such as Spanish and Italian and those speaking Germanic languages such as English or German. In both groups of bilingual children, we predicted that the duration ratio of fi nal to initial syllables should be reduced in comparison to monolingual children. That is, we considered the possibility of acoustic compromise in which the ratio fell in between the monolingual French values and those of the L1. Acoustic compromise has been reported in VOT (Flege, 1991; Flege & Port, 1981) and rhythm measures (Kehoe et al., 2011) in bilingual children and adults. Dodane and Bijeljac-Babic (2017) indicate final to non-final ratios of 1.73 for French monolinguals aged 3;3–6;0. The final to non-final ratios for the bilinguals in their L1s should be 1.0 or less than 1.0 since their languages are characterized by trochaic stress in which the initial stressed syllable is longer than the final unstressed syllable or at least the same length due to phrasefinal lengthening. Delattre’s (1966) work shows that the ratio of stress to unstress is greater in a Germanic than a Romance language. Consequently, we considered the possibility of graded effects in which the final to non-final ratio would be smaller in the bilinguals speaking Germanic as compared to Romance languages. Thus, we predicted that bilinguals speaking Germanic languages would have reduced final to non-final duration ratios compared to the bilinguals speaking Romance languages, who would in turn have reduced final to non-final duration ratios compared to the monolinguals. Pitch accent on the initial syllable is an optional feature of French prosody whereas it is an integral part of the stress system in trochaic languages such as English, German and Spanish. If bilingual children are influenced by the prosodic patterns of their L1, we predicted that the initial pitch accent should be present more frequently in the bilinguals than in the monolinguals. We did not predict any differences between the bilingual children speaking Germanic versus Romance languages since they all spoke trochaic languages. A summary of the research predictions for word prosody is given in Figure 3.1. In sum, this study seeks evidence for crosslinguistic interaction in syllable-structure, VOT and word prosody by comparing French speaking bilinguals and monolinguals at three different age ranges: 2;6, 3–4 and 5–6 years.

Seeking Crosslinguistic Interaction in French Bilingual Phonological Development

65

a) duration French monolingual

Contact with trochaic language

French bilingual

(phonetic compromise)

Predicon: Duraon rao of final to inial syllable will be reduced in bilinguals compared to monolinguals. There will be greater reducon in bilinguals speaking Germanic as compared to Romance languages.

b) pitch accent French monolingual

Contact with trochaic language

*H (oponal)

French bilingual

*H

*H (oponal)

*H

Predicon: High pitch accent on the inial syllable of disyllabic words will be more frequent in bilinguals than monolinguals.

Figure 3.1 A graphic representation of research predictions for word prosody

Method Database

The data come from two separate studies: Kehoe and Havy (2019) and Kehoe and Giradier (2020). In the fi rst study, 40 children (17 monolinguals, 23 bilinguals) aged 2;6 were tested at the university laboratory. In the second study, 37 children (15 monolinguals, 22 bilinguals) aged 3–4 years (2;11–4;11) and 64 children (22 monolinguals, 42 bilinguals) aged 5–6 years (5;0–6;10) were tested at their crèches or public schools. The children were tested in quiet rooms in the crèches and schools to ensure low ambient noise levels which would allow optimal conditions for phonetic transcription and acoustic analysis. The bilinguals had all been exposed to French before the age of 3 years and could thus be classified as simultaneous bilinguals. In both sets of studies, children took part in an object and picture naming task in which stimuli

66

An Anthology of Bilingual Child Phonology

were selected to test phonological features such as word-final codas, onset clusters and initial stop consonants. The bilingual’s first languages were coded in terms of the complexity of these phonological features. In addition, the parents completed a questionnaire on their child’s language experience. In the Kehoe and Havy (2019) study, parents indicated the percentage of time children were exposed to each language, whereas in the Kehoe and Giradier (2020) study, the parents rated the child’s language use on a 5-point scale. Bilingual children were later categorized as dominant or non-dominant in French. As for vocabulary, parents of children in the Kehoe and Havy (2019) study completed L’Inventaire Français du Développement Communicatif (IFDC) (Kern & Gayraud, 2010) (the French adaptation of the MacArthur-Bates Communicative Development Inventory; MCDI) (Fenson et al., 1993) and the MCDI of the child’s other language if the children were bilingual. In the Kehoe and Girardier (2020) study, children were administered a French vocabulary test. The study of syllable structure was based on all children in the Kehoe and Havy (2019) and Kehoe and Giradier (2020) studies, whereas the study of VOT and word prosody was based on subsets of children from these studies. In the case of VOT, bilingual children who spoke languages with long lag–short lag distinctions were in the minority. We selected all of these children and then selected monolingual and bilingual children who spoke languages with lead–short lag distinctions so that all of the groups had the same average age. In the case of word prosody we followed a similar procedure. Bilingual children who spoke Germanic languages were in the minority. We selected all of these children and then selected monolinguals and bilingual children who spoke Romance languages so that all of the groups had the same average age. The data on VOT are based on 42 children from the Kehoe and Giradier (2020) study. In the 3–4 years group, nine children are monolinguals, seven are bilinguals speaking a long lag–short lag and seven are bilinguals speaking a lead–short lag language. In the 5–6 years group, seven children are monolinguals, six are bilinguals speaking a long lag– short lag and six are bilinguals speaking a lead–short lag language. We refer to the bilinguals speaking a long lag–short lag distinction as Bi-Eng since many of the children in this group spoke English. We refer to the bilinguals speaking a lead–short lag distinction as Bi-Span since many of the children in this group spoke Spanish. The data on word prosody are based on 16 children from the Kehoe and Havy (2019) study and 36 children from the Kehoe and Giradier (2020) study. In the 2;6 group, eight children are monolinguals and eight are bilinguals speaking a Romance language. There were insufficient numbers of bilinguals speaking Germanic languages to form a second group. In the 3–4 and 5–6 years groups, six children are monolinguals, six are bilinguals speaking a Germanic language and six are bilinguals speaking a Romance language.

Seeking Crosslinguistic Interaction in French Bilingual Phonological Development

67

Data analyses

Using Phon, a software program designed for the analysis of phonological data (Rose & MacWhinney, 2014), each child’s wave file was segmented, and stimulus words were identified and transcribed. Four French speaking graduate students who had experience in phonetic transcription, including training in the speech laboratory, performed the analyses. All words containing word-fi nal codas and onset clusters were extracted for the analysis of syllable structure. Words starting with a stop consonant were extracted for the analysis of VOT, and disyllabic words were extracted for the analysis of word prosody. Fifteen participants (across both studies) were re-transcribed by a second transcriber using the Blind Transcription function of the Phon program. Point-to-point agreement in terms of consonant transcription (excluding voicing errors) was good (88–96%). Acoustic analyses were conducted in Praat (Boersma & Weenink, 2007). VOT (in ms) was the interval between the release burst and the first glottal pulse as shown on the waveform and corresponding spectrogram. Following Davis (1995), two values were measured for target voiced stops: (1) lead VOT (duration of voicing prior to burst); and (2) lag VOT (period between release burst and fi rst glottal pulse after the release). Two hundred tokens were reanalyzed for VOT by a second examiner. The correlation coefficient between the VOT values obtained by the two examiners was 0.98. To measure the acoustic correlates of disyllabic words, we focused on the vocalic nucleus of each syllable. The onset and offset of each vowel was the fi rst and last detectable periodic cycle in the time waveform. Inspection of the time waveform, spectrographic display and auditory judgement were used to aid boundary identification. Once the vocalic nucleus of each syllable was defined, we extracted the following measures: duration of initial and fi nal syllables; and maximum pitch of initial and final syllables. We then calculated: (1) duration ratio = duration of fi nal/ duration of initial syllable; and (2) F0 difference = maximum pitch of initial – maximum pitch of fi nal syllable. The presence of pitch accent was defi ned as 30 Hz difference between initial and fi nal syllables, which was equivalent to a 10% change in F0, the average maximum F0 of the first syllable being 300 Hz. Statistical analyses

The statistical analyses were performed using R statistical software (R Development Core Team, 2015) and the lme4 package (Bates et al., 2015) for mixed-effects models. To evaluate the contribution of each predictor in the model, we performed pairwise model comparisons between a saturated and a more restricted model. The saturated model included all main effects. The more restricted model omitted the predictor under

68

An Anthology of Bilingual Child Phonology

consideration. Comparisons were made using likelihood ratio tests (LRT) which yielded a chi-squared statistic. Multilevel variables were further analyzed using Wald z-statistics. Random factors included intercepts for participant and item. Results Analyses of syllable structure Word-final codas

First we examined whether the complexity of codas in the bilingual’s L1 influences coda production when children are aged 2;6. We ran mixedeffects logistic regression entering the predictor variable, coda complexity, along with variables, % French exposure and total vocabulary. Coda complexity significantly improved model fit to data (χ2(2) = 6.05, p = 0.048). Wald test results indicated that children who spoke high-complexity languages had higher scores than monolinguals (β = 1.22, z = 2.58, p = 0.01); children who spoke low-complexity languages had marginally higher scores (β = 1.16, z = 1.9, p = 0.06). In addition, % French exposure and total vocabulary were significant predictors in the model. Graphic representations of coda accuracy according to coda complexity for children aged 2;6 are shown in Figure 3.2. We then examined whether the complexity of codas in the bilingual’s L1 influences coda production when children are aged 3–4 and 5–6 years. We ran mixed-effects logistic regression entering the predictor variable,

Figure 3.2 Box plot representation of % word-final coda accuracy in French based on complexity groupings of codas for children aged 2;6 Notes: mon = monolinguals; high = bilinguals who speak languages with high-complexity codas; low = bilinguals who speak languages with low-complexity codas.

Seeking Crosslinguistic Interaction in French Bilingual Phonological Development

69

coda complexity, along with age (in months), dominance and French vocabulary score. Coda complexity did not improve model fit to data at either age group. At age 3–4, age (in months) was the only significant variable (β = 0.04, χ2(1) = 4.59, p = 0.03), whereas at age 5–6 years, vocabulary was significant (β = 0.03, χ 2(1) = 16.75, p < 0.001). Graphic representations of coda accuracy according to coda complexity for children aged 3–4 and 5–6 years are shown in Figures 3.3 and 3.4. In sum, only at age 2;6 were results consistent with predictions that high complexity in the bilingual’s L1 would increase phonological accuracy in the L2. In actual fact, bilinguals with high-complexity codas did better than monolinguals, although we predicted no change; bilinguals with low-complexity codas had similar results to monolinguals, although we predicted delay. Age influenced coda production at age 3–4 years whereas vocabulary played a role at age 5–6 years. Onset clusters

Next we examined whether the complexity of onset clusters in the bilingual’s L1 influences cluster production when children are aged 2;6. We ran mixed-effects logistic regression entering the predictor variable, cluster complexity, along with variables % French exposure and total vocabulary. Results indicated that the high cluster group had scores that were significantly different from monolinguals (β = 3.24, z = 2.53, p = 0.01); the mid cluster group had scores that were marginally different

Figure 3.3 Box plot representation of % word-final coda accuracy in French based on complexity groupings of codas for children aged 3–4 years Notes: mon = monolinguals; high = bilinguals who speak languages with high-complexity codas; low = bilinguals who speak languages with low-complexity codas.

70 An Anthology of Bilingual Child Phonology

Figure 3.4 Box plot representation of % word-final coda accuracy in French based on complexity grouping of codas for children aged 5–6 years Notes: mon = monolinguals; high = bilinguals who speak languages with high-complexity codas; low = bilinguals who speak languages with low-complexity codas.

from monolinguals (β = 2.47, z = 1.88, p = 0.06); and the low cluster group were not significantly different from monolinguals (β = 1.99, z = 1.58; p = 0.11). In addition, total vocabulary marginally improved model fit to data. Graphic representations of cluster accuracy according to complexity are given in Figure 3.5.

Figure 3.5 Box plot representation of % onset cluster accuracy in French based on complexity groupings of clusters for children aged 2;6 Notes: Mon = monolinguals; high = bilinguals who speak languages with high-complexity clusters; mid = bilinguals who speak languages with mid-complexity clusters; low = bilinguals who speak languages with low-complexity clusters.

Seeking Crosslinguistic Interaction in French Bilingual Phonological Development

71

We then examined whether the complexity of onset clusters in the bilingual’s L1 influences cluster production when children are aged 3–4 and 5–6 years. We ran mixed-effects logistic regression entering the predictor variable cluster complexity, along with age (in months), dominance and French vocabulary score. Cluster complexity was not significant in the model at either age group. At both age groups, vocabulary was the sole predictor of cluster accuracy (age 3–4: β = 0.05, χ2(1) = 12.26, p < 0.001; age 5–6: β = 0.06, χ2(1) = 15.38, p < 0.001). Figures 3.6 and 3.7 display the fi ndings of cluster accuracy according to complexity at ages 3–4 and 5–6 years. As can be seen, graded effects in complexity appear to be present, particularly at age 3–4 years. In actual fact, initial analyses indicated that complexity of onset clusters contributed to model fit at age 3–4 years. Monolinguals obtained superior results to children who had mid-complexity clusters (z = −2.08, p < 0.04) but not to those who had high-complexity clusters (z = −1.35, p = 0.17). Children who had low-complexity clusters were not included because of the small number of children in this group. When vocabulary was added to the model, these results did not turn out to be significant. An analysis with a larger sample of children might have led to more robust fi ndings, however. In sum, at age 2;6, results were consistent with our predictions that complexity in the bilingual’s L1 influences cluster accuracy in the bilingual’s L2. The only exception was that we predicted delay for children who spoke languages with low-complexity clusters whereas instead we

Figure 3.6 Box plot representation of % onset cluster accuracy in French based on complexity groupings of clusters for children aged 3–4 years Notes: mon = monolinguals; high = bilinguals who speak languages with high-complexity clusters; mid = bilinguals who speak languages with mid-complexity clusters; there were no children who spoke low-complexity clusters in the group.

72

An Anthology of Bilingual Child Phonology

Figure 3.7 Box plot representation of % onset cluster accuracy in French based on complexity groupings of clusters for children aged 5–6 years Notes: mon = monolinguals; high = bilinguals who speak languages with high-complexity clusters; mid = bilinguals who speak languages with mid-complexity clusters; low = bilinguals who speak languages with low-complexity clusters.

observed that the low-complexity group did not differ from monolinguals. At age 3–4 years, graded effects of cluster complexity were present (see Figure 3.6), but they did not contribute to the fi nal model. In the older age groups, children with larger vocabularies had superior accuracy scores to children with smaller vocabularies. Analyses of VOT

In the next analyses, we examined the effect of predictor variables on VOT in the 3–4 and 5–6 years group. First we examined the influence of speaking a language with a lead–short lag versus long lag–short lag distinction on VOT. We predicted that only bilinguals who spoke languages containing long lag–short lag distinctions should differ from monolinguals. In this analysis, we included all positive VOT values for target voiced and voiceless stops. In a separate analysis, we examined the effect of predictor variables on the presence of lead voicing for target voiced stops only. In addition to the predictor variable, bilingual status (Mon, Bi-Span, Bi-Eng), we also included control variables: age (in months), voice (target voiced versus voiceless), PoA (labial, coronal or dorsal) and vowel type (/a, i, o/). Predictor variables, dominance and French vocabulary did not prove significant in any of the analyses. The results of a linear mixed model for children aged 3–4 years indicated that bilingual status did not improve model fit to data (χ2(2) = 0.09, p = 0.96). As shown in Figure 3.8, monolinguals and both groups of

Seeking Crosslinguistic Interaction in French Bilingual Phonological Development

73

Figure 3.8 Positive VOT values for target voiced and voiceless stops for monolinguals (Mon) and bilinguals speaking lead–short lag (Bi-Span) and long lag–short lag (BiEng) languages in the 3–4 year group

bilinguals produced target voiceless stops with median values of approximately 23 ms and target voiced stops with values of approximately 11 ms. They all demonstrated a great deal of variability and realized many stops with VOTs in the long lag region. Although bilingual status was not significant, control variables, voice, PoA and vowel type were significant. The results of a linear mixed model for children aged 5–6 years indicated that bilingual status did not improve model fit to data (χ2(2) = 3.09, p = 0.21). Figure 3.9 displays the VOT values for target voiced and voiceless stops for monolinguals and bilinguals speaking languages containing lead–short lag and long lag–short lag distinctions. Similar to what we observed for children aged 3–4 years, monolingual and bilingual children alike produced VOT values for target voiced and voiceless stops of a comparable magnitude. As with the 3–4 years age group, control variables were significant. Voice, PoA, vowel type and age (in months) significantly contributed to model fit. We used binomial logistic regression to examine whether bilinguals differed from monolinguals in the realization of lead voicing. When we examined the 3–4 and 5–6 year-old groups separately, there were no significant effects of predictor variables on the presence of lead voicing. This may have arisen due to lack of statistical power since lead voicing was infrequent in the current database, being present in only 20% of the productions of target voiced stops. Thus, we reanalyzed the data including the two age groups together. This analysis indicated that two factors significantly improved model fit to data, age (in months) (χ2(2) = 6.15, p = 0.013) and PoA (χ2(2) = 12.10, p = 0.002). Older children produced more

74

An Anthology of Bilingual Child Phonology

Figure 3.9 Positive VOT values for target voiced and voiceless stops for monolinguals (Mon) and bilinguals speaking lead–short lag (Bi-Span) and long lag–short lag (BiEng) languages in the 5–6 year group

lead voicing than younger children and the presence of lead voicing differed according to PoA. In addition, Wald test results indicated that children who belonged to the long lag–short lag group had less lead voicing than monolinguals (β = −2.14, z = −2.07, p = 0.038); children who belonged to the lead–short lag group were not significantly different from monolinguals (β = −0.44, z = −0.48, p = 0.62). The percentage of lead voicing in the monolingual and bilingual groups is shown in Figure 3.10. In sum, results indicated no significant differences between monolinguals and bilinguals in their realization of short lag values for voiced and voiceless stops. There were significant differences between monolinguals 35 % lead voicing

30 25 20 15 10 5 0 Mon Bi-Span Bi-Eng Monolingual and bilingual groups

Figure 3.10 Percentage of target voiced stops with lead voicing in the monolinguals (Mon) and bilinguals speaking lead–short lag (Bi-Span) and long lag–short lag (BiEng) languages in the 3–4 and 5–6 year group combined

Seeking Crosslinguistic Interaction in French Bilingual Phonological Development

75

and bilinguals in their realization of lead voicing when age groups were examined together. Bilinguals speaking long lag–short lag languages produced less lead voicing than monolinguals. Analyses of word prosody

Finally, we examined the effect of predictor variables on the realization of word prosody in two-syllable words. Our dependent variables were the duration ratio of fi nal to initial syllable and the presence/absence of pitch accent on the initial syllable. Duration ratio

The results of a linear mixed model examining the effects of the predictor variables, bilingual status, French exposure and total vocabulary, on duration ratios indicated no significant effects for the 2;6 age group. The mean ratio of fi nal to initial syllable was 1.55 (s.d. = 1.03) for the monolinguals and 1.63 (s.d. = 0.89) for the bilinguals. Similarly, the results of linear mixed models examining the effects of the predictor variables, bilingual status, dominance and French vocabulary as well as the control variable, age (in months), indicated no significant effects on vowel duration for the 3–4 and 5–6 year-old groups. At age 3–4 years, the mean ratio of final to initial syllables was 1.69 (s.d. = 0.81) for the monolinguals, 1.92 (s.d. = 0.79) for the bilinguals speaking Romance languages and 1.83 (s.d. = 0.94) for the bilinguals speaking Germanic languages; at age 5–6 years, the mean ratio of fi nal to initial syllables was 2.05 (s.d. = 0.81) for the monolinguals, 2.18 (s.d. = 1.32) for the bilinguals speaking Romance languages and 2.14 (s.d. = 1.30) for the bilinguals speaking Germanic languages. Presence of pitch accent

We used binomial logistic regression to examine whether bilinguals differed from monolinguals in the utilization of pitch accent on the initial syllable. Analysis of the data for the 2;6 year-olds indicated that bilingual status significantly improved model fit to data (χ2(1) = 5.42, p = 0.02). The fi ndings were contrary to our predictions, however. Monolinguals realized a pitch accent on the initial syllable more frequently (67.83%) than bilinguals (52.48%). There were no other significant variables in the model. At age 3–4 years, bilinguals realized a pitch accent on the initial syllable more frequently (48.10%) than monolinguals (36.62%); however, statistical models revealed no significant effect of bilingual status on the realization of pitch accent. At age 5–6 years, monolinguals and bilinguals realized a pitch accent on the initial syllable to similar degrees (mon: 34.21%; bi: 32.30%). Once again, there was no significant effect of bilingual status on the realization of pitch accent. No other variable proved significant in the models.

76

An Anthology of Bilingual Child Phonology

In sum, statistical models that tested the influence of bilingual status on duration ratios of fi nal to initial syllables as well as on the realization of pitch accent on the initial syllable yielded few significant fi ndings. The only significant effect was at 2;6 whereby monolinguals were found to produce a pitch accent more frequently on the initial syllable than bilinguals. Since this pattern runs opposite to our prediction, we query whether it is evidence of crosslinguistic interaction. We return to this in the Discussion.

Discussion

The aim of the study was to seek crosslinguistic interaction across different phonetic/phonological measures and age groups. The three domains that we investigated, namely syllable structure, VOT and word prosody, are ones in which crosslinguistic interaction in young bilinguals has often been observed (Almeida et  al., 2012; Dodane & Bijeljac-Babic, 2017; Fabiano-Smith & Bunta, 2012; Johnson & Wilson, 2002; Keffala et al., 2018; Kehoe et al., 2004; Lleó et al., 2003; Stoehr, 2018; Tamburelli et al., 2015). Furthermore, the three age groups we examined, namely 2;6, 3–4 and 5–6 years, span a period of childhood in which crosslinguistic evidence has often been documented. Despite the inclusion of several measures and age ranges, our study yielded only a few clear examples of crosslinguistic interaction. There were complexity effects on coda and onset cluster development in the youngest but not in the oldest groups of children. Bilingual children aged 3–6 years produced fewer target voiced stops with lead voicing, but they did not display differences in their use of short lag voicing for target voiced and voiceless stops. There was little influence of bilingualism on word prosody. The fi ndings on crosslinguistic interaction are summarized in Table 3.3. In the following paragraphs we discuss them in more detail and consider how they contribute to our understanding of crosslinguistic interaction. Table 3.3 A summary of the findings that provide evidence of crosslinguistic interaction Age groups

Syllable structure

VOT

Word prosody

2;6

Complexity effects on codas and onset clusters

Not tested

No influence of bilingualism on duration ratios. Reduced presence of pitch accent in bilinguals

3–4 years

No complexity effects

5–6 years

No complexity effects

No influence of bilingualism on short lag voicing Reduced amounts of lead voicing in Bi-Eng group

No influence of bilingualism No influence of bilingualism

Seeking Crosslinguistic Interaction in French Bilingual Phonological Development

77

Crosslinguistic interaction in syllable structure

Our study indicated that bilingual children aged 2;6 who spoke languages with high-complexity codas or onset clusters obtained superior coda and cluster accuracy scores to monolinguals. Perceiving and producing complex structures in one language appears to facilitate production of these structures in the other language. These fi ndings are consistent with Keffala et al. (2018) and Tamburelli et al. (2015) who found similar effects of complexity when comparing monolingual and bilingual children in syllable structure development. Nevertheless, significant results were not obtained for the two older groups of children. In their case, vocabulary was a significant predictor of outcomes. Children who scored higher on a French vocabulary test obtained superior coda and cluster accuracy scores to children who scored lower. What could explain the different fi ndings for the younger versus older groups of children? First, it must be pointed out that these are cross-sectional and not longitudinal data. There may be important differences between the groups of bilingual children tested which could have contributed to the different fi ndings. For example, the children aged 2;6 were tested in the university laboratory, whereas the children aged 3–4 and 5–6 years were tested at their crèches or public schools. Children whose parents respond to an announcement for testing at a university site may represent a different population from those tested in a crèche or school. Keeping aside these methodological differences, however, there may have been other factors that contributed to the presence of complexity effects in the younger but not in the older group. One relevant point is that accuracy scores were higher in the older children, often in excess of 90%. It could be surmised that grammatical effects from one language to the other are stronger when the phonological system is developing. This may arise because the phonologies of both languages share a common speech-motor base as well as have many segments and structures in common. At later stages of development, children need to acquire language-specific phonological structures for which a large language-specific vocabulary may be more important. This may explain why we observed crosslinguistic interaction in the younger age group and lexical influences in the older age group. Not all complexity effects were consistent with predictions, however. We had predicted delay for the low coda group and no change for the high coda group, whereas we documented no change (or some marginal improvement) and acceleration, respectively, in bilinguals aged 2;6. We predicted delay for the low onset group whereas we documented no change. Thus, there appeared to be a general tendency for these young bilinguals to perform better than monolinguals, a fi nding that may relate to the increased perceptual and productive advantages that arise from being exposed to a larger inventory of sounds and syllable types across both languages (see Kehoe & Havy, 2019). In addition, there were

78

An Anthology of Bilingual Child Phonology

complexity effects in the 3–4 year-old group which were not consistent with predictions. Bilinguals who spoke languages with high-complexity onset clusters had similar scores to monolinguals, whereas we had hypothesized that they would have better scores than monolinguals. It is, nevertheless, possible that this is an example of acceleration, but bilinguals also need to counteract the inhibitive effects of reduced language exposure and vocabulary; thus, those who have an advantage due to speaking languages with high-complexity structures obtain comparable rather than superior results in comparison to monolinguals. Fabiano-Smith and Goldstein (2010) referred to this type of pattern as a variation of acceleration. Crosslinguistic interaction in VOT

Many studies have shown that exposure to two languages influences children’s development of VOT (Fabiano-Smith & Bunta, 2012; Johnson & Wilson, 2002; Kehoe et al., 2004; Khattab, 2000; Stoehr, 2018; Stoehr et  al., 2018). The bulk of the research has focused on bilingual VOT acquisition when the majority language has a long lag–short lag distinction. In contrast, this study examined the influence of bilingualism on VOT acquisition when the majority language had a lead–short lag distinction. The only previous study that has investigated such a situation is Stoehr et  al. (2018), which examined acquisition of VOT in bilingual Dutch-German children growing up in the Netherlands. They found few effects of bilingualism on Dutch, the majority language, with the exception of reduced presence of lead voicing. Our study on the acquisition of VOT in French bilingual children yielded similar fi ndings. The only difference we found was that bilingual children who spoke a language with a long lag–short lag contrast produced fewer tokens with lead voicing than monolingual children. The fi ndings were subtle, however, being only evident when the two older age groups were combined. We had also hypothesized that bilingual children who spoke languages with long lag–short lag distinctions might produce target voiceless stops with longer VOTs than monolinguals due to the influence of their L1. Similar effects have been reported in bilingual children acquiring Japanese (Johnson & Wilson, 2002), whereby they produce target voiceless stops in the long lag region due to the influence of the majority language, English. In our study, these effects were not evident. Monolingual and bilingual children alike produced most of their stops in the short lag region. Their voicing patterns were not stable, however, since they produced many stops in the long lag region, but this tendency was apparent across all groups of children (see Figures 3.8 and 3.9). We assume that short lag stops are unmarked. Developmentally, they are the fi rst stops to be acquired prior to long lag and lead forms (Macken & Barton, 1980a). Lead stops are articulatorily difficult and considered marked (Lowenstein & Nittrouer, 2008). Thus, it is not surprising that

Seeking Crosslinguistic Interaction in French Bilingual Phonological Development

79

short lag stops were resistant to language influence whereas lead stops were not. Stoehr (2018) found that bilingual children’s VOT patterns correlated with those of their parents. Thus, it cannot be excluded that what is being interpreted here as crosslinguistic interaction reflects instead the influence of the input on VOT: the Bi-Eng children may have been exposed to non-native French in which lead voicing was not always employed. In fact, the percentage of tokens produced with lead voicing was very low in the current study (less than 30% in monolingual children aged 3–6 years). In contrast, MacLeod (2016) reports percentages of lead voicing as high as 69% in Canadian French monolingual children aged 2,6–4;6. Okalidou et al. (2010) also found that children learning standard Greek acquired adult-like values of lead voicing as early as 2;0–2;6. We are currently testing Swiss French adults in order to determine whether there is something special about the children’s input which may explain the low percentages of tokens with lead voicing. Crosslinguistic interaction in word prosody

Our analyses of bilingual influence on word prosody yielded few significant fi ndings. We had hypothesized that bilingual children would produce smaller fi nal to non-fi nal duration ratios than monolingual children due to the influence of their other language. Instead, we observed that bilingual children produced two-syllable words with similar duration ratios to monolingual children. At 2;6, the ratio of fi nal to non-fi nal was approximately 1.6 and this increased to adult-like values of 2.0 by age 5–6 years. Dodane and Bijejac-Babic (2017) observed that bilingual children produced ratios even greater than monolingual children, a pattern that they interpreted as an attempt by the bilingual children to mark the contrast between their two languages. We also documented a tendency by the bilinguals to produce mean ratios greater than monolinguals (3–4: Mon – 1.69, Bi-Span – 1.92, Bi-Eng – 1.83; 5–6: Mon – 2.05, Bi–Span – 2.18, Bi–Eng – 2.14). However, these tendencies did not prove to be statistically significant. Thus, in the end our results based on duration ratios were similar to Rose and Champdoizeau (2007), who found no influence of bilingualism on children’s word prosody patterns. Turning to the realization of pitch accent on the initial syllable, we had hypothesized that bilinguals would produce a pitch accent more often on the initial syllable than monolinguals, consistent with the way stress is marked in their L1. Instead, we observed that monolingual French children produced a pitch accent on the initial syllable more often than bilingual children at age 2;6 and to comparable degrees at later ages. How do we interpret the increased use of pitch accent in monolinguals as compared to bilinguals in the younger age group? In French, a high pitch accent on the initial syllable is not a correlate of stress but is considered part of the intonation system (Jun & Fougeron,

80

An Anthology of Bilingual Child Phonology

2002). It is conditioned by different factors including discourse type. It may have an expressive or didactic function. We have observed it being used frequently by young children in a word-naming task (e.g. C’est un cádeau! ‘That is a present’). It is possible that the bilingual children have not yet acquired this language-specific aspect of French intonation, which is cued by discourse or pragmatic context. Thus, the reduced presence of initial accent in the bilingual children aged 2;6, in comparison to the monolinguals, may still be an example of crosslinguistic interaction but one of delay in acquisition of the initial pitch accent rather than what we had originally predicted, which was transfer of the non-optional pitch accent from the bilingual’s L1 onto French. At later age ranges, no differences between monolinguals and bilinguals were observed, possibly because bilingual children have had greater experience using initial pitch accent in different pragmatic contexts. Other investigators have reported crosslinguistic interaction in word prosody (Dodane & Bijeljac-Babic, 2017) or in rhythm (Bunta & Ingram, 2007; Kehoe et al., 2011; Schmidt & Post, 2015). Thus, it is surprising that there was little evidence of it in the current study. One possibility is that the prosodic characteristics of two-syllable words in child French are consistent with universal tendencies, making them less susceptible to bilingual influence. Vihman et al. (1998) present a model of the way in which biological (natural physiological) and ambient language factors interact in the acquisition of word prosody. They propose that the natural tendency for duration is that the final syllable is longer than the initial. Final syllable lengthening is found in many languages and is acquired at an early age (Vihman, 1996). The natural tendency for pitch is a falling contour, in which the initial syllable has higher pitch than the fi nal (Kent & Murray, 1982). Both of these patterns were evident in the youngest children’s productions. Ambient language influences seemed to take over at later age ranges since the presence of pitch accent became less frequent on the initial and more frequent on the final syllable over time – although the falling contour predominated at all ages. Thus, the absence of crosslinguistic interaction in word prosody may be consistent with what we observed above in VOT: unmarked phenomena such as short lag stops or word prosody in French may be resistant to grammatical influence. Conclusion: Understanding Crosslinguistic Interaction

This study sought evidence for crosslinguistic interaction across different phonetic/phonological measures and age ranges. We found evidence for crosslinguistic interaction on a small number of occasions but not across all the ones that we investigated. One reason as to why crosslinguistic interaction was infrequent was that the children were simultaneous bilinguals who had all been exposed to French before the age of 3 years. We tested their majority language. More evidence for crosslinguistic

Seeking Crosslinguistic Interaction in French Bilingual Phonological Development

81

interaction might have come from studying sequential bilinguals or from investigating the minority language. That is, we may have recorded more examples of crosslinguistic interaction in the ‘other’ direction: French influence onto the L1. Overall, our results suggest that crosslinguistic interaction in the phonetic/phonological domain when tested in the majority language is not an inevitable outcome of bilingual acquisition. It may be short-lived as it was documented only at the youngest age range for syllable structure development. It may be more frequent for marked phenomena as suggested by its manifestation for lead but not for short lag voicing and by its relative absence in the area of word prosody. It may emerge in certain phonetic/ phonological domains more than others. In the current study, the clearest evidence of crosslinguistic interaction was in the phonological domain (syllable structure), which suggests that grammatical influence is stronger for categorical structures (e.g. presence or absence of coda consonant) than for continuous (acoustic) measures. At the beginning of this study, we mentioned that one of the difficulties inherent in bilingual phonological research is the variability of fi ndings. Unfortunately, our fi ndings did not resolve this problem since our research predictions were largely unsupported, adding more variability to the current knowledge base. Nevertheless, studies such as ours, which examine crosslinguistic interaction across different measures and ages, contribute important information to what happens when two languages meet in the mind of a young bilingual, which should eventually lead to more elaborate models of crosslinguistic interaction. References Almeida, L., Rose, Y. and Freitas, J. (2012) Prosodic influence in bilingual phonological development: Evidence from a Portuguese-French fi rst language learner. In A. Biller, E. Chung and A. Kimball (eds) Proceedings of the 36th Annual Boston University Conference on Language Development (pp. 42–52). Somerville, MA: Cascadilla Press. Andreassen, H. (2013) Schwa distribution and acquisition in light of Swiss French data. Unpublished Doctoral dissertation, University of Tromsø. Bates, D., Maechler, M., Bolker, B. and Walker, S. (2015) Fitting linear mixed-effects models using lme4. Journal of Statistical Software 67, 1–48. Boersma, P. and Weenink, D. (2007) PRAAT: Doing Phonetics by Computer [Computer program]. Bunta, F. and Ingram, D. (2007) The acquisition of speech rhythm by bilingual Spanishand English-speaking 4- and 5-year-old children. Journal of Speech, Language, and Hearing Research 50, 999–1014. Davis, K. (1995) Phonetic and phonological contrasts in the acquisition of voicing: Voice onset time production in Hindi and English. Journal of Child Language 22, 275–305. Delattre, P. (1966) A comparison of syllable length conditioning among languages. International Review of Applied Linguistics 4, 183–198. Delattre, P. and Olsen, C. (1969) Syllabic features and phonic impression in English, German, French and Spanish. Lingua 22, 160–175.

82

An Anthology of Bilingual Child Phonology

Dell, F. (1984) L’accentuation dans les phrases en français. In F. Dell, D. Hirst and J.-R. Vergnaud (eds) Forme Sonore du Langage (pp. 65–122). Paris: Hermann. Deuchar, M. and Clark, A. (1996) Early bilingual acquisition of the voicing contrast in English and Spanish. Journal of Phonetics 24, 351–365. Dodane, C. and Bijeljac-Babic, R. (2017) Cross-language influences in the productions of bilingual children: Separation or interaction. In M. Yavas, M. Kehoe and W. Cardoso (eds) Romance-Germanic Bilingual Phonology (pp. 38–55). Sheffield: Equinox. Dryer, M. and Haspelmath, M. (eds) (2013) The World Atlas of Language Structures Online. Leipzig: Max Planck Institute for Evolutionary Anthropology. See http:// wals.info (accessed 4 May 2017). Fabiano-Smith, L. and Bunta, F. (2012) Voice onset time of voiceless bilabial and velar stops in 3-year-old bilingual children and their age-matched monolingual peers. Clinical Linguistics and Phonetics 26, 148–163. Fabiano-Smith, L. and Goldstein, B. (2010) Phonological acquisition in bilingual SpanishEnglish speaking children. Journal of Speech, Language, and Hearing Research 53, 1–19. Fenson, L., Dale, P., Reznick, S., Thal, D., Bates, E., Hartung, J., Pethick, S. and Reilly, J. (1993) MacArthur Communicative Development Inventories: User’s Guide and Technical Manual. San Diego, CA: Singular Publishing. Flege, J. (1991) Age of learning affects the authenticity of voice onset time (VOT) in stop consonants produced in a second language. Journal of the Acoustical Society of America 89, 395–411. Flege, J. and Port, R. (1981) Cross-linguistic phonetic interference: Arabic to English. Language and Speech 24, 125–146. Gierut, J. (2001) Complexity in phonological treatment: Clinical factors. Language, Speech, and Hearing Services in Schools 32, 229–241. Goad, H. and Buckley, M. (2006) Prosodic structure in child French: Evidence for the foot. Catalan Journal of Linguistics 5, 109–142. Goldstein, B., Fabiano, L. and Washington, P. (2005) Phonological skills in predominantly English-speaking, predominantly Spanish-speaking, and Spanish-English bilingual children. Language, Speech, and Hearing Services in Schools 36, 201–218. Guffey, K. (2002) Spanish Syllable Structure. Lanham, MD: University Press of America. Johnson, C. and Wilson, I. (2002) Phonetic evidence for early language differentiation: Research issues and some preliminary data. International Journal of Bilingualism 6, 271–289. Jun, S.A. and Fougeron, C. (2002) The realizations of the accentual phrase in French intonation, Probus 14, 147–172. Keffala, B., Barlow, J. and Rose, S. (2018) Interaction in Spanish-English bilingual’s acquisition of syllable structure. International Journal of Bilingualism 22, 16–37. Kehoe, M. (2015) Cross-linguistic interaction: A retrospective and prospective view. In E. Babatsouli and D. Ingram (eds) Proceedings of the International Symposium on Monolingual and Bilingual Speech 2015 (pp. 141–167). See http://ismbs.eu/ publications. Kehoe, M. and Girardier, C. (2020) What factors influence phonological production in French-speaking children, aged three to six years? Journal of Child Language Early Online. DOI: https://doi.org/10.1017/S0305000919000874 Kehoe, M. and Havy, M. (2019) Bilingual phonological acquisition: The influence of language-internal, language-external, and lexical factors. Journal of Child Language 46, 292–333. Kehoe, M., Lleó, C. and Rakow, M. (2004) Voice onset time in bilingual German-Spanish children. Bilingualism: Language and Cognition 7, 71–88. Kehoe, M., Lleó, C. and Rakow, M. (2011) Speech rhythm in the pronunciation of German and Spanish monolingual and German-Spanish bilingual 3-year-olds. Linguistische Berichte 227, 323–351.

Seeking Crosslinguistic Interaction in French Bilingual Phonological Development

83

Kent, R. and Murray, A. (1982) Acoustic features of infant vocalic utterances at 3, 6, and 9 months. Journal of the Acoustical Society of America 72, 353–365. Kern, S. and Gayraud, F. (2010) L’inventaire Français du Développement Communicatif. Grenoble: Editions La Cigale. Khattab, G. (2000) VOT in English and Arabic bilingual and monolingual children. In D. Nelson and P. Foulkes (eds) Leeds Working Papers in Linguistics and Phonetics 8, 95–122. Lisker, L. and Abramson, A.S. (1964) A cross-language study of voicing in initial stops: Acoustical measurements. Word 20, 384–422. Lleó, C., Kuchenbrandt, I., Kehoe, M. and Trujillo, C. (2003) Syllable fi nal consonants in Spanish and German monolingual and bilingual acquisition. In N. Müller (ed.) (Non)Vulnerable Domains in Bilingualism (pp. 191–220). Amsterdam: John Benjamins. Lowenstein, J.H. and Nittrouer, S. (2008) Patterns of acquisition of native voice onset time in English-learning children. Journal of the Acoustical Society of America 124 (2), 1180–1191. Macken, M.A. and Barton, D. (1980a) The acquisition of the voicing contrast in English: Study of voice onset time in word-initial stop consonants. Journal of Child Language 7 (1), 41–74. Macken, M.A. and Barton, D. (1980b) The acquisition of the voicing contrast in Spanish: A phonetic and phonological study of word-initial stop consonants. Journal of Child Language 7 (3), 433–458. MacLeod, A. (2016) Phonetic and phonological perspectives on the acquisition of voice onset time by French-speaking children. Clinical Linguistics and Phonetics 30 (8), 584–598. Meinhold, G. and Stock, E. (1980) Phonologie der deutschen Gegenwartssprache. Leipzig: Bibliographisches Institut. Mok, P. (2013) Speech rhythm of monolingual and bilingual children at age 2;6: Cantonese and English. Bilingualism: Language and Cognition 16, 693–703. Okalidou, A., Petinou, K., Theodorou, E. and Karasimou, E. (2010) Voice onset time development in Cypriot-Greek and Greek-speaking children age 2–4 years: A crosssectional study. Clinical Linguistics and Phonetics 24 (7), 503–519. Paradis, J. and Genesee, F. (1996) Syntactic acquisition in bilingual children: Autonomous or interdependent? Studies in Second Language Acquisition 18, 1–25. R Development Core Team (2015) R: A Language and Environment for Statistical Computing. Vienna: R Foundation for Statistical Computing. Rose, Y. and Champdoizeau, C. (2007) There is no innate trochaic bias: Acoustic evidence in favour of the neutral start hypothesis. In A. Gavarró and M.J. Freitas (eds) Proceedings of the Generative Approaches to Language Acquisition Conference 2007. Newcastle upon Tyne: Cambridge Scholars. Rose, Y. and MacWhinney, B. (2014) The PhonBank initiative. In J. Durand, U. Gut and G. Kristoffersen (eds) The Oxford Handbook of Corpus Phonology (pp. 380–401). Oxford: Oxford University Press. Schmidt, E. and Post, B. (2015) The development of prosodic features and their contribution to rhythm production in simultaneous bilinguals. Language and Speech 58, 24–47. Stoehr, A. (2018) Speech production, perception and input of simultaneous bilingual preschoolers: Evidence from Voice Onset Time. Unpublished Doctoral dissertation, Radboud University and International Max Planck Research School for Language Sciences. Stoehr, A., Benders, T., van Hell, J. and Fikkert, P. (2018) Heritage language exposure impacts voice onset time of Dutch–German simultaneous bilingual preschoolers. Bilingualism: Language and Cognition 21 (3), 598–617.

84

An Anthology of Bilingual Child Phonology

Tamburelli, M., Sanoudaki, E., Jones, G. and Sowinska, M. (2015) Acceleration in the bilingual acquisition of phonological structure: Evidence from Polish-English children. Bilingualism: Language and Cognition 18, 713–725. Vaissière, J. (1991) Rhythm, accentuation and fi nal lengthening in French. In J. Sundberg, L. Nord and R. Carlson (eds) Music, Language, Speech and Brain (pp. 108–120). London: Macmillan. Vihman, M. (1996) Phonological Development: The Origins of Language in the Child. Oxford: Blackwell. Vihman, M., DePaolis, R. and Davis, B. (1998) Is there a ‘trochaic bias’ in early word learning? Evidence from infant production in English and French. Child Development 69, 935–949.

4 Case Studies of Phonological Development in Six Preschool-aged Russian-Finnish Bilingual Children Olga Nenonen

Introduction

Traditionally, language development in bilingual children has been thought to resemble that of monolinguals, although it often appears to be slower (Bialystok & Feng, 2011; Bialystok et al., 2010; Hoff, 2015; Silvén et al., 2014). The same can be said with regard to phonology: young bilingual children acquire phonology in two languages in the same way as their monolingual peers. Still, monolingual children are often ahead of bilinguals in the acquisition of some phonological features (e.g. Fabiano-Smith & Barlow, 2010; Gildersleeve-Neumann et al., 2008). Furthermore, bilingual children tend to demonstrate language patterns similar to those of children with specific language impairment (SLI) (Armon-Lotem et al., 2015). This accounts for bilingual children repeatedly being considered as having problems with language development (Armon-Lotem et  al., 2015; Launonen, 2007; Paradis, 2010). As a result, it is necessary to conduct further research on the phonological development of bilingual children in order to obtain additional information on the specific mechanisms involved in typical bilingual language acquisition. The present chapter explores a number of points in relation to this issue. In particular, we focus on the following: (a) individual variation and common features in the phonetic production of bilingual children; (b) bilinguals’ errors in production; and (c) the acceleration or deceleration of bilingual phonetic development in comparison to monolingual phonetic development. We also address the question of whether young bilingual children have one phonological system for both languages or they have two separate phonological language systems. This chapter is structured as 85

86

An Anthology of Bilingual Child Phonology

follows. The background section below presents a very brief review of previous work that has been done in the area of phonological assessment. We then describe the current study and discuss the results with regard to the assessment of bilingual phonological development. Finally, we outline the limitations of the study and propose directions for further research. The phonological systems of Russian and Finnish and a hypothesis for bilingual acquisition

A review of theory helps to create specific predictions that can be tested by empirical research. In the present study, we compare the phonological systems of Russian and Finnish in order to generate a linguistic prognosis of possible problems in bilingual phonological acquisition (i.e. bilingual errors) involving these languages. Information will be presented in the following order: vocalic and consonantal inventory (Tables 4.1 and 4.2, respectively), duration of vowels and consonants, and word-level prosody. The Russian inventory of six vocalic phonemes is rather modest compared to that of Finnish with eight monophthongs (which can have short and long varieties) and 18 diphthongs. However, Russian vocalic phonemes can have a large number of allophones depending on their position in the word and their distance from the stressed syllable. While unstressed vowels are the subject of qualitative and quantitative reduction in Russian, in Finnish quantitative reduction only occurs in unstressed vowels. De Silva (1999: 46, 167) and Ljubimova (2010: 346) point out that the most problematic Russian vowels for Finns are [i] and [ɨ]. Correspondingly, [æ], [y] and [ø], the long vowels (all short vowels can also occur doubled) and the diphthongs found in Finnish are challenging for Russians. These typologically less common (language-specific) vowels ([i], [ɨ] in Russian, and [æ], [y], [ø], long vowels and diphthongs in Finnish) are also infrequent in both languages (Bondarko, 2009: 35; Iivonen, 2009: Table 4.1 The vocalic inventory of Russian and Finnish Russian

Finnish

/a/

/a/

/o/

/o/

/u/

/u/

/e/

/e/

/i/

/i/

/ɨ/

/æ/ (/ä/)

reduced [ɐ] / [ʌ]

/y/

reduced [ə]

/ø/ (/ö/)

Notes: Language-specific phonemes are in bold.

Case Studies of Phonological Development

87

Table 4.2 The consonant inventory of Russian and Finnish Russian

Finnish

Stops

p, b, t, d, k, g

p, b, t, d, k, g

Fricatives

f, v, s, z, ʂ, ʐ, ɕː, h

pʲ, bʲ, tʲ, dʲ, kʲ, gʲ f, v, s, ʃ, h

fʲ, vʲ, sʲ, zʲ, hʲ Affricates

t͡ s, t͡ ɕ

Nasals

m, n

m, n, ŋ

mʲ, nʲ Laterals

l, lʲ

l

Rhotics

r, rʲ

r

Glides

j

j

Notes: Language-specific phonemes are in bold; all phonemes in the table are independent phonemes, not allophones.

62; Suomi et al., 2008: 21–23). Taking these into account, all the aforementioned vowels ([i], [ɨ], [æ], [y], [ø]) could prove difficult for bilingual children to acquire. For example, the Finnish [æ] could be perceived as [e] or [ja], and [ø] could be perceived as [e] or [jo] (Nenonen, 2016: 29). Conversely, primary vowels are usually acquired early (Cejtlin, 2000; Gvozdev, 1961; Stoel-Gammon & Herrington, 1990) and therefore might not be problematic for bilingual children. The Russian consonantal inventory is rather large according to different classifications (Avanesov, 1972: 34; Bondarko, 1998: 7, 32; Bulanin, 2009: 1–53, 81–84; Maslov, 2007: 59). It has between 34 and 37 consonants. The large number of consonants can be explained by the existence of soft (palatalized) sounds. The most important typological characteristics of consonants in Russian are the binary oppositions of hard–soft (unpalatalized–palatalized) and voiceless–voiced. These features, together with Russian language specific consonants – including the sibilants /z, t͡ s, ʂ, ʐ, t͡ ɕ, ɕː/ – are problematic for Finnish speakers. According to Karlsson (1983), the consonant paradigm in Finnish is different, being defi ned as polysystemic with respect to its consonantal variants. In short, there are 17 consonant phonemes in Finnish, of which 11 consonants are common in all variants (dialects) of the language. The phoneme /ŋ/ occurs in all variants of the language but has a narrow distribution. The defective phoneme /d/ (Karlsson, 1983: 57–58; Swadesh, 1995: 15) also has a narrow distribution, but it does not occur in all the variants of the language. Finally, there are the marginal consonants, /f, b, ʂ, g/, which appeared in Finnish only recently and are found in loan words (Kallioinen, 1969: 7; Karlsson, 1983: 57–59; Suomi et al., 2008: 23–38). One of the most important typological features of Finnish is the opposition of consonants with

88

An Anthology of Bilingual Child Phonology

regard to duration; this feature concerns phonetically short-long sounds (e.g. [k]–[k:], [m]–[m:]) and phonemically single-double phonemes (e.g. /k/–/kk/, /m/–/mm/). Geminates that are considered to be double consonants (sequences of two similar consonants) (Kraehenmann, 2001) and /ŋ/ are usually difficult for Russian speakers (Toivola, 2011). Vowel and consonant duration is an important feature of the Finnish phonological system. According to the standard interpretation, quantitatively long segments are described as sequences of two identical phonemes (i.e. as double vowels and consonants) in contrast to short or single phonemes, and diphthongs are viewed as sequences of two dissimilar vowels. When examined qualitatively, phonetically short and long (phonemically single and double) vowels appear very similar to native speakers (Karlsson, 1969: 354; Suomi et al., 2008: 19, 39). In the Russian language, vowel and consonant duration is not phonemically significant. This explains why one may hear the Finnish accent in Russian words when Russian stressed vowels are perceived as diphthongs or long vowels, or if Russian reduced vowels are reproduced as non-reduced stressed vowels. Correspondingly, the Russian accent can be heard when Finnish long vowels are produced as short vowels – i.e. as reduced both qualitatively and quantitatively. When discussing word-level prosody, we must point out that the primary stress in Finnish is fi xed on the first syllable of a word, and the secondary stress usually falls on the third or fourth syllable. In Russian, the syllabic stress of a word is free and unfi xed; it may also be still or mobile in the inflectional paradigm of the word. Due to these prosodic features, Russian speakers might move the word stress to the second or third syllable in Finnish words, and Finnish speakers might stress the fi rst syllable in Russian words (Nenonen, 2016: 29). Based on this information, a linguistic prognosis in this study would be that bilinguals may experience difficulties in the acquisition of language-specific features in each language, and in defective and marginal consonants in Finnish. Additionally, some phonemically similar consonants that have different articulations in Russian and Finnish (e.g. /t, s, n, h/) are also potentially problematic and could be pronounced with an accent. As the Russian consonantal inventory is much larger than the Finnish consonant inventory, it may require more time for bilinguals to acquire. Child phonological development

Early child phonological development is generally described as a complex process that does not follow a linear route (e.g. Bernhardt & Stemberger, 1998; Mohanan, 1992). On the path to adult-like production, the child passes through different phases of both fast and arrested development, showing substitutions as well as distortions. The presence of errors is not necessarily problematic in child speech productions; on the

Case Studies of Phonological Development

89

contrary, such errors are considered a normal feature in child language which is a result of the child’s undeveloped resources in speech production. In fact some of these errors are so systematic that they are regarded as specific to developmental child speech rather than deviations from the target form. Further discussion on this stance may be found in several sources (e.g. Bialystock, 2001; Cejtlin, 2000, 2008; Eliseeva, 2008; Menn & Stoel-Gammon, 2005). In this study we use the terms ‘error’, ‘pronunciation error’, ‘mistake’ or ‘mispronunciation’ synonymously as shorthand to denote a deviation from the typical adult realization of the phonetic form of the word or word segment. We also point out pronunciation errors on the paradigmatic level that comprise both phonemic and phonetic errors. By phonemic error here we mean substitutions of the target phoneme by other sounds in the phonemic inventory of the same language. Phonetic errors involve cases of sound distortions or developmental errors, e.g. pronouncing interdental [θ̟ ] instead of Russian [s]. Correspondingly, pronunciation errors on the syntagmatic level denote word structure transformation, such as omissions, assimilations, additions, etc. (Nenonen, 2016: 70). Monolingual phonological development in Russian and Finnish

Numerous researchers have described early monolingual phonological development in Russian and Finnish (e.g. Eliseeva, 2008; Gvozdev, 1961; Kunnari, 2000; Kunnari & Savinainen-Makkonen, 2012; Piotrovskaja, 2011; Savinainen-Makkonen, 2001). Phonemic errors (sound substitutions) are generally defi ned as early developmental errors which tend to disappear at the start of the period of complete phonological acquisition (around 4;0 years). However, some exceptions hold for both the Finnish and Russian languages. Some Finnish children still substitute the defective phoneme /d/ by the age of 6;0 or 7;0 years (Kunnari & SavinainenMakkonen, 2012). In Russian monolinguals, all sibilants are marked and substituted by other consonants for a long time; they are fi nally acquired between 4;0 and 5;0 (Belʹtjukov & Salakhova, 1975; Eliseeva, 2008; Gvozdev, 1961). In addition, in both languages liquid phonemes, /l r/, are subject to steady substitutions. Phonetic errors (sound distortions) mostly concern late-acquired sounds in both languages: /s r l j k g h/ and the sibilants in Russian. In contrast to phonemic errors, phonetic errors appear at a later stage of development within the age range 5;0–8;0 (Kunnari & Savinainen-Makkonen, 2012: 97; Piotrovskaja, 2011: 99–101) (see Tables 4.3 and 4.4). A number of researchers (e.g. Ingram, 1976; Kehoe, 2010; Menn & Stoel-Gammon, 2005; Stoel-Gammon, 1985, 1998; Stoel-Gammon & Herrington, 1990; Vihman, 1988, 1996, 2010; Vihman & Keren-Portnoy, 2013) have described children’s strategy of simplification during different stages of phonological development. Simplification results in

90

An Anthology of Bilingual Child Phonology

Table 4.3 Age of the acquisition of similar consonants by monolingual Finnish and Russian children Finnish

Russian

[v]

3;0–3;11

1;10–2;0

[h]

3;0–3;11

1;10–2;0

[s]

3;0–3;11

2;8–3;3

[l]

4;0–4;11

2;8–3;3

[r]

5;0–6;11

3;1–5;0

[d]

4;0–4;11

2;0–2;7

Table 4.4 Normative age of acquisition of consonantal sounds in monolingual Russian children Age of acquisition

Consonants

1;0–1;9

p, t, k, m, b

1;10–2;0

j, f, v, tʲ, dʲ, nʲ, kʲ, g, gʲ, h

2;0–2;7

sʲ, zʲ, pʲ, bʲ, mʲ, d, n

2;8–3;3

fʲ, vʲ, s, z, hʲ, l, lʲ

3;4–3;8

ɕː , t͡ s, t͡ ɕ

3;1–5;0

ʂ, ʐ, r, rʲ

pronunciation errors on the syntagmatic level, such as sound and syllable omissions, cluster simplifications and errors belonging to phonological processes. Both Russian and Finnish monolingual children omit consonants, and it is reported that consonants in Russian may be omitted in any syllable but more commonly at the end of the word (Švačkin, 1995: 107), while in Finnish consonant omission is more common at the beginning of the word (Savinainen-Makkonen, 2001: 43). Cluster simplification is also observed in children acquiring both languages (Cejtlin, 2000: 78–79; Eliseeva, 2008: 45–48; Gvozdev, 1961: 98; Jortikka, 1993: 81–83; Kunnari & Savinainen-Makkonen, 2012: 122–126; Savinainen-Makkonen, 2001: 39–43). Monolingual Finnish speaking children are sometimes unable to produce different vowels within a single word; instead, they assimilate all targets to a single vowel in all syllables. Some children also tend to reduce diphthongs to monophthongs (Jortikka, 1993: 83; Kunnari & SavinainenMakkonen, 2012: 121). Syntagmatic pronunciation errors are much more common in the production of consonants. They are produced with the effect of such phonological processes as consonant assimilation, cluster simplification, consonant and syllable omission, metathesis (transposition of both sounds and syllables), compensatory elongation or prolongation (addition), contamination, etc. In Finnish, the aforementioned errors mostly affect /v l n j h/, especially at the beginning of a word and in

Case Studies of Phonological Development

91

Table 4.5 Normative age of acquisition of consonantal sounds in monolingual Finnish children Age of acquisition

Consonants

2;0–2;5

p, t, k, n

2;6–2;11

m

3;0–3;11

ŋ, s, h, v, j

4;0–4;11

d, l

5;0–6;11

r

clusters. In Russian, nearly all the consonants can be subject to the aforementioned phonological processes. A review of studies cited earlier reveals that Russian and Finnish speaking monolingual children undergo the same stages of phonological development. There are similarities, particularly in respect of the order and age of acquisition as well as the typology of the children’s errors. Thus, the fi rst vowels and consonants (minimum vocalic and consonant systems) appear in children’s language approximately at the same time (/a, i, o, u, e/, /p, t, k, m, n/). The next stage is characterized by the acquisition of language-specific features in each language. Russian and Finnish speaking monolingual children attain vowel accuracy by the age of 3;0 after mastering late-acquired /ɨ/ in Russian and /y, ø, æ/ in Finnish. The process of consonant acquisition differs in Russian and Finnish monolingual children; many consonants that do not belong to the minimum consonantal system and are similar in both languages are mastered by Finns later than by Russians, as shown in Table 4.3. The fi nal consonants to be established are the lateral /l/, trill /r/, fricative /s/, the language-specific sibilants in Russian and the defective phoneme /d/ in Finnish. Geminates are usually mastered by Finnish children during the early stages of development, around 2;0 (Kunnari, 2000: 28–29). The age of complete acquisition of consonants (see Tables 4.4 and 4.5) varies between 3;0 and 8;0 in Russian monolinguals (Belʹtjukov & Salakhova, 1975; Gvozdev, 1961) and between 4;0 and 7;0 in Finnish monolinguals (Kunnari & Savinainen-Makkonen, 2012). Typically, children achieve complete consonant accuracy before primary school age (7;0) in both Finland and Russia. Phonological development in bilingual children

In the current study, bilingual phonological development is understood as the acquisition of the phonology in two languages, i.e. the acquisition of two phonological systems. Phonetic development comprises the acquisition of phonetic inventories, the adult-like production of sounds and sound combinations, the growth of articulatory abilities and the identification and practice of word stress.

92

An Anthology of Bilingual Child Phonology

Bilingual language acquisition is generally associated with language contact and interaction (e.g. Döpke, 2000; Lanza, 2000; Paradis & Genesee, 1996; Weinreich, 1972). Language contact often results in crosslinguistic influence which means that separate phonological systems interact. This interaction may have a different impact on phonological development. For example, Paradis and Genesee (1996) and Fabiano-Smith and Barlow (2010) provide evidence of the deceleration (delay) and acceleration (facilitation) of bilingual phonological development. This means that the transfer is no longer considered only as ‘negative’; it can have positive manifestations as well. ‘Positive transfer’ can affect identical features, such as identical phonemes. The impacts of ‘negative transfer’ are more evident; they can be seen in the avoidance of features that do not exist in either the L1 or the L2 – like language-specific phonemes – or in the form of a foreign accent (a nonnative pattern of pronunciation). For a comprehensive review of phonological interaction, see Kehoe (2018). To our knowledge, there are few studies on Russian accent (that is, second language ‘transfer’, Major, 2008) in Finnish (Toivola, 2011) or Finnish accent in Russian (Ljubimova, 2010), and what studies there are focus on adult second language productions. Some observations on the accent of Russian-Finnish bilingual children before primary school age are given in a study by Protasova and Rodina (2005). One of the most important questions in bilingual language development deals with that of language systems: Do children start with one system and later move to a gradual differentiation of separate language systems (the one-system model), or do they start with two language systems that are differentiated from the start and develop separately (the twosystem model)? There is experimental evidence for the latter viewpoint. For instance, Polka and Sundara (2003) argue that small children are already able to differentiate the sounds of different phonological systems in the preverbal period. In addition, Fabiano-Smith and Barlow (2010) have shown that English-Spanish bilingual small children (3;0–4;0) had two different phonological systems organized in the same way as the phonological system in monolinguals – the small amount of transfer did not affect the fi ndings. Vihman (2002) has shown that small children (at age 0;6) differentiated their two languages in perception, while they used the same phonetic templates in production. In addition, their fi rst words, which emerged subsequently, were so unclear that one could hardly identify the source language. Based on these findings, Vihman proposes a nonsystem hypothesis, asserting that it is too early to talk about one or two phonological systems in the initial stages of phonological development. Phonological development in monolingual children with specific language impairment

SLI is defi ned as a primary deficit in linguistic skills and language development which is unrelated to hearing loss, intelligence or neurological

Case Studies of Phonological Development 93

problems (Leonard, 1998). In children with SLI, the patterns of language disorder exhibited have a systematic nature across different language levels, although different linguistic skills are not equally impaired. Phonological processing and auditory memory are reported to be impaired in children with SLI and intact in typically developing bilingual children. For example, previous research has shown that monolingual and bilingual children with SLI perform poorly on nonword repetition tasks, while typically developing monolingual and bilingual children have no difficulties in repeating nonwords (Armon-Lotem et al., 2015). Finnish and Russian children with SLI make errors related to developmental phonology that are found in typically developing children, as well as non-typical errors. Detailed information may be found in works on impaired phonological development (Filičeva & Čeveleva, 1987: 73–75; Ingram, 1976: 29–44, 120; Jortikka, 1993: 79–80, 96–97; Korpinen & Nasretdin, 2009: 60–67; Kunnari & Savinainen-Makkonen, 2012: 442–443; Ljakso, 2008: 60–76). Among the general markers of SLI, the following features can be found: (1) the child is far behind peers in language development; (2) the child has imprecise articulation; (3) the child makes errors in the minimal vocalic and consonantal systems, such overwhelming omissions; (4) the child uses very few patterns of word structure and tries to simplify the structure of words; (5) the child makes unsystematic pronunciation errors; and (6) the child makes a large number of substitutions and demonstrates many phonological processes, so that words and phrases become distorted beyond recognition. The Current Study

The current study addresses phonological development in preschoolaged Russian-Finnish bilingual children. In this chapter we describe six independent case studies that were conducted as longitudinal research. We observed six typically developing bilingual (BL) children over a period of 2.5 years (for age ranges, see section on participants in Methodology below). The study describes the development of sound inventories in the group of young bilingual children compared to three control groups: (a) monolingual Russian peers (MLR), (b) monolingual Finnish peers (MLF) and (c) bilingual children with SLI. All six typically developing bilingual children were born in Finland and grew up learning Russian and Finnish at preschool age. By the term ‘monolingual’ (ML), we refer to (a) Russian native speakers who grew up in Moscow learning only Russian and (b) Finnish native speakers who grew up in Helsinki learning only Finnish. The third control group represents bilingual children who were diagnosed with SLI in Helsinki and attended speech therapy lessons. We chose the bilingual SLI control group in order to compare phonetic production in typically developing bilinguals and bilinguals with SLI and thus to attempt to disentangle features in bilingualism from those of SLI. All three control

94

An Anthology of Bilingual Child Phonology

groups participated in a cross-sectional study from 2010 to 2014. The fi ndings of this earlier study are given in Nenonen (2016). The main objective here was to describe phonological development in Russian-Finnish bilingual children with a view to outlining the trajectories of typical bilingual phonological development in children speaking this pair of languages. In assessing the phonology of young children, the study was guided by the following aims: • • • • •

to compare the time of the acquisition of vowels and consonants in BL children with that of control groups of ML children and those with SLI; to analyze phonological and phonetic errors and processes in the BL children; to describe transfer in the BL children; to present fi ndings on bilingual phonological development, with a focus on those phonetic characteristics that are easier or more difficult to acquire for BL children; to discuss the question of whether BL children construct two different phonological systems or create a synthesis of two systems where the sounds of the two languages coexist.

Stoel-Gammon and Stone (1985: 25) state that a child with a large vocabulary and a capacity for word combinations will have ‘an expanding phonological system, with a full range of sound classes and syllable and word shapes’. Correspondingly, a child with delayed language acquisition is expected to have a more limited phonological system. Considering previous research, as cited in the Introduction, the following hypothesis is made: Phonological development of the bilingual children in the study is expected to be slower (or more limited) than that of monolingual controls; the bilinguals’ productions will include: (1) typical developmental errors, (2) typical language-specific errors, (3) transfer/ interaction and (4) errors found in children with SLI. Methodology

The present chapter is based on mixed-methods research conducted in the framework of child language development studies and contrastive and contact linguistics. The study utilizes both qualitative and quantitative data analyses. Participants

The sample of participants of this longitudinal study consists of six typically developing Russian-Finnish bilingual children from a RussianFinnish bilingual kindergarten. Four children (B, C, E, F) are simultaneous bilinguals from Russian-Finnish bilingual families (Russian mothers, Finnish fathers). Participant A is a sequential bilingual from a Russian

Case Studies of Phonological Development

95

Table 4.6 Subjects of the longitudinal study A

B

C

D

E

F

Sex

f

f

m

m

f

m

Date of birth

09.10.08 15.09.07 30.08.07 17.08.07 30.07.07 26.04.07

Age at the beginning of the study 3;1

4;2

4;3

4;3

4;4

4;7

Age at the end of the study

5;7

6;7

6;8

6;8

6;9

7;0

1st, 2nd or 3rd child in the family

3rd

2nd

2nd

2nd

1st

1st

Age at emergence of speech

2;0

1;6

1;6

2;0

1;0

2;6

Age when the child started to attend daycare

3;0

4;0

3;0

3;0

3;0

3;5

speaking family. Participant D is a trilingual who acquired two L1s (Russian and Farsi) with parental exposure and Finnish as an L3 later (at age 3;0) in daycare. The parents of all participants filled out a background questionnaire, providing information on the child, family, parents’ education, languages, and parents’ evaluation of the child’s language skills as well as dominant language. All children come from middle-class families with educated parents. More detailed information on the participants is presented in Table 4.6. The sample of the cross-sectional study in Nenonen (2016) used as control groups for the present study consists of: (i) 40 typically developing Russian monolinguals (MLR), (ii) 20 typically developing Finnish monolinguals (MLF) and (iii) 20 Russian-Finnish bilinguals with SLI. Procedure

The study explores data collected using the articulation test designed by Remes and Ojanen (1997). The test, which involves picture-naming tasks, was used to assess pronunciation in both Finnish and Russian. The test included the original words in Finnish and an adaptation made for Russian (see Appendix). The 63 target words have various phonetic structures and contain all the vowels and consonants of both languages in different positions and combinations. Each test session was documented by fi lling in a form in which the errors were registered. Some spontaneous utterances of the participants were also analyzed. The test sessions were video-recorded. The longitudinal study lasted for 2.5 years, and every child participated in five test sessions (for details see Tables 4.7–4.12). Participants were tested individually in each language separately. They were given the choice of which language to start with. According to our observations, the children preferred to start with their dominant language. The time needed to execute the test varied greatly among the children and correlated with their language proficiency: informants used less time to perform the test in the stronger language; older children were

96

An Anthology of Bilingual Child Phonology

faster; and the later sessions of the test were much shorter (around 11 minutes) than the first sessions (the maximum length of the fi rst session was 45 minutes). In the cases of balanced bilingualism, the time for test execution in Russian and Finnish was nearly the same. However, some of the fi nal testing sessions required more time because the children interrupted testing with spontaneous speech. Overall, the participants tried to do their best and therefore pronounced the words accurately, sometimes stressing nearly every syllable, which made it impossible to assess the prosodic features of the words or evaluate the realization of word stress. Nevertheless, the collected data enabled us to analyze their phonemic inventories in Russian and Finnish, which was the main task of this study. Analysis

The participant’s production of target words was transcribed phonetically. Next, phonetic and phonological errors were analyzed. In addition, errors in the informant’s spontaneous utterances were also registered during the sessions. The pronunciation was assessed by auditory analyses, and the results were analyzed according to the principles of relational analysis, which compares the child’s productions with the corresponding adult target productions (Stoel-Gammon, 1985). The pronunciation errors were approached with an orientation towards a ‘nativeness’ principle (i.e. with the target of native-like pronunciation). The test data were transposed into tables and graphics. During the analyses, the results were analyzed in terms of the linguistic prognosis made (difficult and language-specific targets) and in comparison with the data on the phonological development of monolingual controls and the bilingual children with SLI. Results

The results of the study on the six bilingual children are discussed below. First, we present the case studies of four simultaneous bilingual children (F, C, E, B) from Russian-Finnish families, then we describe the case study of a trilingual (D) in Russian, Farsi and Finnish, and fi nally we introduce the case study of the sequential bilingual (A) from a Russian speaking family. Overall, the data show considerable individual differences, so the children’s productions are introduced separately. We start with the child’s background information and then move on to a brief overview of the pronunciation errors. Case 1: F

Participant F is the first child in a Russian-Finnish family. At the age of 2;6 he started to speak both languages. At the beginning of the test

Case Studies of Phonological Development

97

period, the boy had good proficiency in both languages and he preferred to speak Russian with his siblings. During the longitudinal study, F showed unstable but progressive development in Russian pronunciation; there were no vowel errors at the beginning of the study (4;7). With regard to Russian consonants, he had certain difficulties in the acquisition of the hard-soft opposition, the late acquired [r], [rʲ], [l], [lʲ] and the sibilants. The latter were especially difficult; in the last testing session (7;0), [s] and [ts ⁀] errors still remained (see Table 4.7a). As for the nature of the error, the trills were omitted by F at the beginning of the study and substituted by [l], [lʲ] and [j] later on; the sibilants were interchangeable. Phonological processes (assimilation, addition, omission and metathesis) were observed mostly in Russian words. In Finnish, there was defi nite progress in phonological development, although some regression of certain vowels and consonants was observed at the end of the study (7;0). Vowel errors (/æ, ææ, øi/) persisted until 7;0, and consonant errors remained in the lateacquired and language-specific sounds (ŋ and geminates) (see Table 4.7b). The boy’s [s] sounds in Finnish were transferred from Russian. Regression in the pronunciation of Finnish sounds coincided with attendance in the preparatory grade, when most attention was given to the Russian language, so this might explain why F’s Finnish pronunciation temporarily worsened. Other reasons for such regression could be the influence of new Russian speaking schoolmates with weaker Finnish or even the influence or imitation of a younger sibling.

Table 4.7a Pronunciation errors in Russian made by Participant F during the longitudinal study 4;7

4;10

5;4

6;0

7;0

r

64%

79%

79%

64%

0%

rʲ

100%

67%

67%

100%

0%

l

33%

0%

0%

0%

0%

lʲ

33%

17%

0%

0%

0%

s

56%

50%

56%

81%

44%

sʲ

50%

50%

50%

0%

0%

t͡ ɕ

20%

40%

20%

0%

0%

t͡ s

33%

100%

67%

33%

67%

ʐ

50%

50%

0%

0%

0%

z

100%

100%

67%

33%

0%

ʂ

33%

17%

17%

17%

0%

f

50%

0%

0%

0%

0%

Notes: 100% means that the participant mispronounced the sound every time in all the words; 0% means that the participant did not make a single error in the sound.

98

An Anthology of Bilingual Child Phonology

Table 4.7b Pronunciation errors in Finnish made by Participant F during the longitudinal study 4;7

4;10

r

100%

100%

5;4 50%

6;0 50%

7;0 0%

s

7%

21%

14%

21%

43%

l

16%

5%

5%

5%

0%

ŋ

100%

100%

100%

100%

100%

j

17%

17%

17%

17%

0%

t

0%

0%

0%

5%

0%

pp

0%

0%

0%

0%

50%

mm

0%

0%

0%

0%

50%

kk

0%

0%

0%

0%

33%

ie

100%

100%

0%

0%

0%

a

3%

0%

0%

3%

0%

æ

0%

0%

0%

0%

8%

ææ

0%

0%

0%

0%

50%

øi

0%

0%

0%

0%

100%

Case 2: C

Participant C is the second child in a Russian-Finnish family and an early bilingual speaker; he started to speak both languages at the age of 1;6. During the entire testing period, his Russian was stronger than his Finnish, so he preferred to speak Russian with his sibling. C made visible progress in Russian. Minor vowel errors appeared until the age of 5;8, and certain consonant errors became less frequent. Nevertheless, he produced a large amount of consonant mismatches expected of his bilingual status, with regard to the hard–soft opposition, the late-acquired [r], [rʲ], [l], [lʲ] and the sibilants (see Table 4.8a). In Finnish there was uneven and slow development with a regression in difficult sounds. A number of errors expected to result from his bilingual status were registered, such as consonant mismatches in late-acquired and language-specific sounds. The informant also made atypical mistakes, which are typically found in the production of children with SLI, specifi cally consistent vowel errors in monophthongs and diphthongs, and some unexpected consonant errors (e.g. [d], [n], [h]; see Table 4.8b). Vowel errors (substitutions) (4;3–4;9) were noticed not only in language-specific vocalic phonemes but also in the primary vowels [i], [a] and [u], which could be the result of poor skills in Finnish; the boy was unsure when pronouncing the endings of some words (e.g. tuoli /tuoli/ [tuola] ‘chair’, lapsi /lapsi/ [lapse] ‘child’, kampa /kampa/ [kampi] ‘hairbrush’, viulu /viulu/ [viula] ‘violin’). C’s productions showed a large number of phonological processes (sound omissions, assimilations and metatheses), especially in Finnish words. Several cases of transfer were noted; the transfer was mostly

Case Studies of Phonological Development

99

from dominant Russian to weaker Finnish (a Russian accent was observed in Finnish words), but also vice versa, which was an unexpected result. Table 4.8a Pronunciation errors in Russian made by Participant C during the longitudinal study 4;3

4;6

4;9

5;8

6;8

r

64%

57%

21%

29%

29%

rʲ

100%

75%

25%

25%

0%

l

17%

17%

8%

17%

0%

lʲ

17%

17%

17%

17%

0%

s

81%

75%

81%

94%

81%

sʲ

50%

50%

50%

50%

100%

t͡ ɕ

60%

40%

0%

0%

0%

t͡ s

100%

100%

67%

67%

100%

ʐ

67%

67%

67%

100%

67%

z

100%

100%

100%

100%

100%

ʂ

60%

80%

100%

60%

60%

tʲ

33%

33%

0%

33%

0%

b

0%

0%

0%

0%

40%

г

0%

0%

0%

0%

40%

ə

0%

0%

0%

33%

0%

Table 4.8b Pronunciation errors in Finnish made by Participant C during the longitudinal study 4;3

4;9

4;9

5;8

6;8

r

83%

33%

33%

17%

17%

s

57%

64%

64%

93%

93%

l

11%

11%

11%

58%

63%

ŋ

100%

100%

100%

100%

100%

j

17%

17%

17%

17%

0%

d

0%

50%

0%

0%

0%

v

0%

0%

0%

25%

0%

n

0%

0%

0%

6%

0%

h

0%

0%

0%

0%

14%

k

7%

0%

0%

7%

0%

kk

0%

33%

33%

0%

33%

tt

33%

0%

33%

33%

0%

ll

0%

0%

0%

33%

0%

a

3%

10%

0%

0%

0%

i

0%

13%

0%

0%

0%

(continued)

100

An Anthology of Bilingual Child Phonology

Table 4.8b (continued) Pronunciation errors in Finnish made by Participant C during the longitudinal study 4;3

4;9

4;9

5;8

6;8

u

0%

0%

0%

0%

9%

y

29%

0%

0%

0%

0%

æ

0%

8%

0%

0%

8%

uo

100%

100%

0%

100%

0%

au

50%

0%

0%

0%

0%

ie

0%

100%

0%

0%

0%

Case 3: E

Participant E is the fi rst child in a Russian-Finnish family. She started to speak both languages early, at the age of 1;6. According to her parents, her Russian was dominant before the age of 3;0 when she started to attend daycare. After that, her Finnish got stronger and it became unquestionably dominant at the age of 6;8. E spoke Finnish with her sister and also constantly tried to speak it with her Russian speaking mother. Her language proficiency in Finnish was strong and the phonological system in Finnish was complete; in Finnish, E showed nearly correct pronunciation. In particular, there were no vowel errors, while in the consonants only /s/ and /r/ mismatches were observed (see Table 4.9b). E’s acquisition of her dominant language phonology followed the same pattern as that of monolingual children. For example, while acquiring the most difficult sounds, E went through the following stages: omissions, substitutions, later in some cases distortions, and gradually fi nal acquisition. In Russian, which was the weaker language, E developed her own individual trajectory. Some sound groups were formed in a rather chaotic way, e.g. all fricatives and aff ricates were interchangeable and there did not seem to be any systematic substitutions. Testing revealed a large amount of pronunciation errors, slow progress and non-linear development in Russian. Errors in vowels persisted from the age of 4;4 to 5;8. Consonantal errors were observed especially in some soft and voiced consonants, the sibilants, and the late-acquired [r], [s] and [l] (see Table 4.9a). The following types of phonological processes were noted in Russian words: omission of syllables, addition, assimilation and metathesis. Transfer from Finnish to Russian was especially evident in the production of hard–soft pairs; E usually failed to pronounce the Russian hard or soft consonant and therefore substituted it with the corresponding Finnish sound. In addition, the Russian vowels [i] and [ɨ] were substituted with the Finnish [I]. Another typical transfer was a voiced consonant substitution by a voiceless sound, as voiced consonants are illegal elements in the Finnish phonological system. The words in Russian were often pronounced with a Finnish accent, which was sustained during the entire longitudinal study. Although the normal direction of transfer was from Finnish to

Case Studies of Phonological Development

101

Russian, evidence of the reverse direction in transfer was also found. The fi ndings suggest that at the age of 5;8, E had not yet acquired the main phonological contrasts in Russian, although the situation improved by the age of 6;8. In addition, E had difficulties with the acquisition of reduced vowels. All of the above indicated that E was lagging behind her monolingual Russian peers. Table 4.9a Pronunciation errors in Russian made by Participant E during the longitudinal study 4;4

4;10

5;1

5;8

6;8

r

64%

71%

71%

29%

rʲ

25%

75%

75%

100%

36% 0%

l

8%

0%

0%

0%

0%

lʲ

33%

17%

0%

0%

0%

s

31%

25%

38%

38%

19%

sʲ

0%

0%

0%

0%

50%

t͡ ɕ

71%

100%

29%

86%

29%

t͡ s

33%

67%

33%

33%

0%

ʐ

67%

100%

67%

33%

33%

z

33%

67%

33%

67%

67%

ʂ

50%

100%

17%

17%

0%

tʲ

0%

0%

0%

33%

0%

d

100%

100%

100%

0%

0%

b

17%

0%

0%

0%

0%

g

100%

25%

50%

0%

0%

n

10%

0%

10%

0%

0%

nʲ

0%

0%

100%

100%

0%

mʲ

100%

100%

100%

100%

0%

vʲ

33%

33%

33%

33%

0%

bʲ

50%

50%

50%

50%

0%

h

50%

0%

0%

0%

0%

f

50%

50%

0%

0%

0%

i

13%

13%

0%

25%

0%

ɨ

0%

50%

0%

0%

0%

ə

100%

80%

50%

0%

0%

Table 4.9b Pronunciation errors in Finnish made by Participant E during the longitudinal study 4;4

4;10

5;1

5;8

6;8

r

100%

100%

83%

100%

0%

s

21%

57%

21%

21%

36%

102

An Anthology of Bilingual Child Phonology

Case 4: B

Participant B is the second child in a Russian-Finnish family. She also started to speak both languages at the age of 1;6. The parents reported that her Finnish was dominant, and she usually preferred to speak her stronger language when possible, e.g. with her sister and relatives. In Finnish, B had only a few consonant errors during the fi rst stages of testing. She showed very fast progress in phonological development, so that at the fi nal stage (6;7) only /s/ errors remained (as indicated in Table 4.10b). B’s phonological development was in line with the monolingual phonological development of Finnish children. The acquisition of Russian phonology was rather fast. The girl never produced vowel errors. Among the consonants, the following sounds proved difficult: voiced consonants, sibilants and the late-acquired [r]. On the whole, B was consistent in using substitutions, and the overall number of substituted sounds was small. All the problematic sounds seemed to be acquired at the end of the study, except for the affricate [ts ⁀], in which B showed no progress during the 2.5 years of the research (see Table 4.10a). It is also worth noting that the girl systematically transferred word stress to the fi rst syllable in Russian words, thus following the Finnish prosodic pattern. The results of the study show that phonological development was faster in Finnish than in Russian.

Table 4.10a Pronunciation errors in Russian made by Participant B during the longitudinal study 4;2

4;6

4;11

5;7

6;7

r

93%

86%

7%

7%

0%

rʲ

75%

75%

75%

0%

0%

s

58%

58%

51%

47%

5%

sʲ

100%

100%

100%

50%

50%

t͡ ɕ

62%

62%

100%

38%

0%

t͡ s

100%

100%

100%

100%

100%

ʐ

100%

67%

100%

67%

33%

z

100%

40%

80%

40%

0%

ʂ

80%

80%

100%

40%

0%

ɕː

100%

100%

100%

0%

0%

d

0%

0%

100%

100%

100%

g

100%

100%

0%

0%

0%

Case Studies of Phonological Development

103

Table 4.10b Pronunciation errors in Finnish made by Participant B during the longitudinal study 4;2

4;6

4;11

5;7

6;7

r

83%

100%

17%

0%

0%

s

75%

81%

88%

100%

44%

l

11%

11%

0%

0%

0%

d

0%

50%

0%

0%

0%

v

13%

25%

0%

0%

0%

p

0%

8%

0%

0%

0%

h

14%

0%

0%

0%

0%

k

7%

0%

0%

0%

7%

a eu

0%

0%

0%

3%

0%

100%

0%

0%

0%

0%

Case 5: D

Participant D is trilingual. He is the second child in a Russian-Persian family. The emergence of speech for both Russian and Farsi was at 2;0. Finnish, being the L3 language, appeared at the age of 3;0 when the boy started to attend daycare. The boy mostly spoke Russian at home and Finnish in kindergarten. The results of the fi rst testing showed that the boy’s language proficiency (including phonetic and phonological skills) in Table 4.11a Pronunciation errors in Russian made by Participant D during the longitudinal study 3;6

4;8

5;0

5;8

6;9

r

7%

14%

0%

7%

7%

s

5%

5%

21%

16%

37%

sʲ

0%

0%

50%

50%

0%

t͡ ɕ

13%

13%

50%

13%

25%

t͡ s

0%

0%

0%

0%

33%

ʐ

33%

0%

33%

0%

0%

z

60%

0%

0%

40%

0%

ʂ

0%

10%

0%

0%

0%

j

33%

33%

33%

33%

0%

b

17%

0%

0%

0%

0%

g

100%

100%

0%

0%

0%

0%

0%

100%

100%

0%

nʲ vʲ

0%

0%

33%

0%

0%

i

0%

13%

0%

0%

0%

ɨ

0%

0%

0%

50%

0%

104

An Anthology of Bilingual Child Phonology

Table 4.11b Pronunciation errors in Finnish made by Participant D during the longitudinal study 3;6

4;8

5;0

5;8

6;9

r

0%

0%

17%

0%

17%

s

0%

0%

93%

0%

7%

l

11%

0%

0%

0%

0%

j

0%

17%

0%

17%

0%

t

0%

0%

5%

5%

0%

ll

0%

0%

33%

0%

0%

both Russian and Finnish was already very high – in fact clearly better compared to the other participants. D demonstrated a small number of expected pronunciation errors typical of his bilingual status, which did not have an impact on his good progress in Russian. Among D’s errors were vowel errors in [i] and [ɨ] (4;8–5;8), some problems with voiced sibilants – ⁀], the substitution of voiced consonants by the regression in [s], [ts ⁀j] and [ts voiceless equivalents, the substitution of soft consonants by their hard pairs, and the omission of [j] (see Table 4.11a). Phonological development in Finnish was definitely faster than in Russian, and no vowel errors were observed. D made minor consonant errors in late-acquired sounds, [j] was omitted a few times, [t] was substituted by [d] and double consonants were shortened. There is evidence that some difficult sounds were acquired in the same way in both Russian and Finnish. At the age of 5;0, D started to distort Russian /s/ and Finnish /s/ in the same way – interdental sigmatism appeared (see Table 4.11b). The findings suggest that this trilingual child was ahead of the other participants in the longitudinal study in phonological development in both Russian and Finnish. Regrettably, we were not able to assess the boy’s phonological skills in Farsi. Case 6: A

Participant A is the third child in a Russian speaking family. She started to speak Russian at the age of 2;0. At the age of 3;0 she started to attend daycare, where she became acquainted with her L2, the Finnish language. In the very first test session (at 3;1) she showed very low proficiency in both languages, although Russian was naturally her dominant language. The girl’s productions – with very unclear and indistinct pronunciation, substituted and distorted sounds, together with a large number of phonological processes – sometimes resembled the protowords produced by an infant and the utterances of a child with SLI. She demonstrated very little progress in Russian and Finnish between the ages of 3;1 and 4;6. Her pronunciation remained unclear, although the rhythmic structure of words was well reproduced. Despite the long period with no noticeable improvement, A made certain progress in the acquisition of some Russian sounds. The pronunciation errors in Russian comprised

Case Studies of Phonological Development

105

Table 4.12a Pronunciation errors in Russian made by Participant A during the longitudinal study 3;1

3;5

3;10

4;6

5;7

r

64%

79%

79%

79%

0%

rʲ

25%

50%

25%

25%

0%

l

67%

33%

50%

42%

0%

lʲ

83%

17%

67%

33%

0%

s

50%

50%

69%

44%

6%

sʲ

50%

0%

0%

0%

0%

t͡ ɕ

69%

85%

100%

69%

0%

t͡ s

67%

33%

67%

100%

0%

ʐ

75%

100%

75%

75%

0%

z

100%

40%

60%

60%

0%

ʂ

100%

71%

100%

100%

0%

ɕː

100%

100%

100%

100%

0%

tʲ

0%

0%

0%

0%

33%

d

100%

100%

50%

0%

0%

b

50%

67%

17%

17%

0%

bʲ

100%

0%

0%

0%

0%

p

0%

0%

0%

20%

0%

g

50%

25%

25%

25%

0%

nʲ

100%

100%

100%

100%

0%

m

17%

0%

0%

0%

0%

vʲ

33%

33%

0%

0%

0%

f

50%

0%

0%

0%

0%

j

67%

0%

0%

0%

0%

i

0%

13%

0%

0%

0%

ɨ

100%

100%

50%

50%

0%

vowel errors in [i] and [ɨ], and a large amount of various consonant errors that affected hard-soft and voiced consonants, sibilants and the lateacquired [r], [s] and [l] (see Table 4.12a). Additionally, A demonstrated errors typical of children with SLI, such as overuse of a combination of phonological processes. In Finnish there were numerous vowel errors (at 3;1) and consonant errors (between 3;1 and 4;6), as well as errors typical of children with SLI in the form of combining various phonological processes. Transfer from Russian to Finnish occurred in A’s production for a long time, and the girl had a strong Russian accent while speaking Finnish. However, following the very slow progression in her development that lasted nearly two years, the girl did achieve very high scores in Russian and Finnish by age 5;7 which was the last testing session. At that time, A spoke both languages fluently with her siblings, friends and relatives. Shortly after exposure to her L2 started, A’s phonological development in Finnish accelerated. At the age of 5;7 only a few errors in Finnish were

106

An Anthology of Bilingual Child Phonology

Table 4.12b Pronunciation errors in Finnish made by Participant A during the longitudinal study 3;1

3;5

3;10

4;6

5;7

r

83%

83%

83%

67%

0%

s

71%

79%

71%

79%

14%

l

63%

58%

58%

42%

0%

ŋ

100%

100%

100%

0%

0%

j

17%

0%

0%

0%

0%

d

50%

0%

0%

0%

0%

h

43%

0%

0%

0%

0%

p

0%

0%

0%

8%

0%

nn

100%

100%

100%

100%

0%

kk

33%

33%

33%

67%

0%

ll

67%

67%

67%

67%

0%

a

3%

0%

0%

0%

0%

i

0%

0%

6%

0%

0%

e

33%

0%

0%

0%

0%

æ

30%

20%

10%

0%

0%

y

14%

14%

0%

0%

0%

yø

100%

0%

0%

0%

0%

au

50%

50%

0%

0%

0%

ie

100%

0%

0%

0%

0%

ou

100%

100%

100%

0%

0%

eu

100%

0%

0%

0%

0%

iu

100%

100%

0%

0%

0%

æy

100%

0%

0%

0%

0%

øy

100%

0%

0%

0%

0%

observed (see Table 4.12b). Overall, her progress in Finnish, which was initially the weaker language, was faster than in Russian. She showed the delayed onset and slow development of expressive language of a late talker, but she was subsequently able to follow an otherwise age-appropriate developmental path (e.g. Roos & Weismer, 2008), denoting her sequential bilingualism. The findings show that this girl produced the largest number of pronunciation errors, including developmental errors (i.e. typical of monolingual children), and errors typical of bilingual children and of children with SLI. At the early stage of her development, many sounds were shared in the two languages (i.e. they sounded identical) and her two systems showed clear evidence of interaction. However, at the end of the study, A showed fast progress in both lexical and phonological development. The transfer from Russian to Finnish disappeared, and the girl was very close to balanced bilingualism.

Case Studies of Phonological Development

107

Discussion

The pronunciation assessment of six bilingual children in the study revealed not only individual development trajectories but also some characteristics typical of bilingualism in the acquisition of sound inventories. In Russian, the most problematic sounds were [r] and [s]. In addition, nearly all children in the study had difficulties with [rʲ], [l] and [lʲ], and major problems concerning fricatives and affricates, and the sound oppositions of hard–soft and voiced–voiceless. Three out of six children made systematic errors while pronouncing the vowels [i], [ɨ] and the unstressed [ə]. In Finnish, the most difficult sounds appeared to be: [r], [s] and [l]; the language-specific consonant [ŋ]; double consonants; and nearly all vowels, even the ‘universal’ (primary) vowels that are usually acquired early by monolingual children. Still, the most problematic Finnish vowels were the language-specific [y], [ø] and [æ] and the diphthongs. Five out of six children (all except the trilingual D) made a large number of pronunciation mistakes in Russian. Three children – the trilingual D and the bilingual B and E – were ahead of the other children in the acquisition of Finnish sound inventories. For Participants B and E, Finnish was the dominant language during the whole period of the longitudinal study, while for Participant D Finnish developed fast and became dominant soon after he started attending daycare. The phonological development of the other three children was faster and easier in Finnish than in Russian, even when Finnish was the L2 (i.e. the weaker language). Participant A demonstrated how readily a child is able to cope with a significant delay in both the L1 and L2: after a difficult start, the girl became a rather balanced sequential bilingual (by 5;7). The case of Participant A questions the view of simultaneous delay in two languages being a marker of language impairment. The longitudinal study also helped to depict the different and sometimes changing profi les of the bilingual participants over the course of their development. A monolingual girl (A) became a sequential bilingual, a bilingual boy (D) with two languages (Russian and Persian) became trilingual having acquired an L3 (Finnish), one boy (C) retained Russian as his dominant language, and one girl (E) retained Finnish as her dominant language. One girl (B) had Russian as her dominant language in the early stages of development, but later on Finnish became her stronger language. The data show that the phonological development of these bilingual children is not an even, steady process. Instead, development has a wavelike trajectory with ups and downs: earlier acquired phonetic features may get lost, new substitutions of speech sounds may appear and the phonetic systems of two languages may interact. Rather progressive and rectilinear development was observed in the acquisition of the trill [r] in both languages. Three participants (D, B and A) acquired this sound in Russian faster (3;1–5;0) than the Russian monolinguals (5;0–7;0;

108

An Anthology of Bilingual Child Phonology

Belʹtjukov & Salakhova, 1975; Eliseeva, 2008, 2014; Gvozdev, 1961). In Finnish, all six children acquired the trill faster (3;1–6;6) than the Finnish monolinguals in general (around 7;0; Iivonen, 1994, 1998, 2009; Korpinen & Nasretdin, 2009; Kunnari, 2000; Saaristo-Helin et  al., 2011; Savinainen-Makkonen, 2001; Toivainen, 1990). The trilingual child acquired the trills in both languages even earlier than the bilingual participants in the study. This example could be evidence of acceleration in this particular sound’s acquisition in bilinguals. It is important to point out that the rhotics in Russian and Finnish are very similar, although the Finnish trill is longer and comprises more vibration movements than the Russian trill. While the Russian sound is produced with one to two contacts, the Finnish trill vibrates for three and more contacts (de Silva et al., 1999; Skalozub, 1963). There were different pathways to the acquisition of the fricatives [s] and [sʲ] in Russian and [s] in Finnish. Many participants, after periods of adult-like realization of these sounds, started to substitute them with other fricatives or affricates, and sometimes the substitutions were borrowed from the other language – i.e. from Russian to Finnish or vice versa. All six participants acquired [sʲ] much later (5;0–7;0) than their monolingual peers (2;0–2;7). The Russian unpalatalized [s] was also mastered by bilinguals later (around 6;0) than by monolinguals (2;8–3;3; Belʹtjukov & Salakhova, 1975; Gvozdev, 1961). While monolingual children usually acquire the palatalized [sʲ] before the unpalatalized [s], bilinguals acquire them in the reverse order, probably because palatalized sounds are generally difficult for them. Similarly, in Finnish [s] was acquired more slowly by bilinguals (not yet acquired by the age of 6;0–7;0) than by monolinguals (the sound is usually acquired by 3;0–3;11, although interdental sigmatism may still remain; Kunnari, 2000; Savinainen-Makkonen, 2001). The described case could be an example of decelerated sound acquisition in the bilinguals of the study. The present study revealed the presence of transfer in the productions of five out of the six participants. The main direction of the transfer was from the dominant to the weaker language, e.g. from Finnish to Russian in Participants B, E and A, and from Russian to Finnish in Participant C. However, Participant C also showed some evidence of reverse transfer at the last stage of the longitudinal study, which is an indication of the boy’s growing proficiency in Finnish. The transfer was especially noticeable in the productions of Participant E, who had a strong Finnish accent in Russian. According to our observations, transfer tends to disappear in time, which was registered in the case of Participant A. Overall, the study has shown that interaction is rather common among the different bilingual children studied here. Our fi ndings also support previous research that has shown the interaction of two phonological systems in bilingual children (e.g. Fabiano-Smith & Barlow, 2010; Kehoe, 2018; Polka & Sundara, 2003). The question of how closely the two

Case Studies of Phonological Development

109

systems are interconnected could be answered in different ways depending on the case. For instance, in the case of Participant F there was only minor interaction between the two phonological systems, while for Participant E the Finnish phonological system deeply infiltrated the Russian phonological system. We suppose that a clear marker of interaction between the two systems is isomorphism – that is, using shared sounds in both phonological systems. At the same time, there is plenty of evidence in favour of the two-system model in the participants of the present study. In particular, occurrences of transfer were limited, and the children’s production in a certain language contained mostly phonetic material of this language. Finally, not a single participant – not even A, who had very poor language skills in Russian and Finnish at the beginning of the study – appeared to have a single phonological system. Taken together, these fi ndings show that at the ages between 3;0 and 7:0 Russian-Finnish bilinguals have two different – albeit interacting – phonological systems, supporting previous fi ndings in the literature. Conclusion

Despite the considerable individual variation in phonetic production, the fi ndings of the present case studies suggest that bilinguals acquire Russian and Finnish phonetic inventories, by and large, later than their monolingual peers. The difference is evident both in the speed of acquisition and in the number and nature of the errors. With regard to the nature of mispronunciation, four types of errors were distinguished in the bilingual participants of this longitudinal study: (1) developmental errors that are commonly made by bilinguals and monolinguals; (2) language-specifi c errors made by both monolinguals and bilinguals (however, the latter group made considerably more mistakes, especially at an older age); (3) crosslinguistic transfer mistakes (caused by the differences in the Russian and Finnish phonological systems) made by bilinguals, resembling the errors of second language learners; and (4) unpredictable errors found in bilingual typically developing children and children with SLI. The analysis from a longitudinal perspective reveals that phonological development is faster and easier for bilinguals in Finnish than in Russian. However, the relatively simpler Russian vocalic inventory is acquired faster than the Finnish vocalic inventory, whereas the complex system of Russian consonants takes longer to develop than the Finnish consonantal system. Furthermore, language-specific features appear to be the most problematic in acquisition. The fi ndings indicate that, although bilinguals show initial similarities with Russian and Finnish speaking monolingual peers, their phonological development is by and large slower, and they make specific errors as a result of their bilingual status, as well as errors that resemble those of children with SLI.

110

An Anthology of Bilingual Child Phonology

This study provides further evidence of language interaction in bilingual phonological development, e.g. in the form of cross-language transfer, delay and acceleration. As a result, some bilingual children may have either a Russian or a Finnish accent. However, this accent tends to disappear gradually. As predicted, bilingual language acquisition demonstrates predominantly a deceleration in the acquisition of the Russian and Finnish phonological systems. Nevertheless, bilinguals also showed a few cases of accelerated sound acquisition in Finnish and Russian. Limitations and Further Research

Several limitations of this study need to be acknowledged. The fi rst limitation relates to the small sample size: six participants took part in the longitudinal experiment. Of the six, four were simultaneous bilinguals, one was a sequential bilingual and one a trilingual. It would be useful to carry out broader research focused on a larger group of participants. Secondly, the study was based on a single articulation test (picture naming) and a small number of spontaneous comments by the participants. Future research could focus on various tasks and the spontaneous speech production of the participants. Finally, further studies should aim at dense and more detailed observations of individual learning trajectories (for instance, every two weeks), since this type of research highlights important aspects of the nature of the developmental process. This would, in turn, help explore questions such as when and why ‘phase shifts’ occur in the speech of developing children, like those portrayed in the cases of Participants A and D. Appendix: Articulation Test in Finnish and Russian Word in Finnish

Targeted IPA

Word in Russian

Targeted IPA

English translation

1

jäätelö

[jæætelø]

мороженое

[mʌroʐɨnəjə]

icecream

2

pöllö

[pøllø]

сова

[sʌva]

owl

3

tyyny

[tyyny]

подушка

[pʌduʂkə]

pillow

4

tuoli

[tuoli]

стул

[stul]

chair

5

sieni

[sieni]

гриб

[grʲip]

mushroom

6

syö

[syø]

ест

[jest]

(he) eats

7

auto

[auto]

машина

[mʌʂɨnə]

car

8

joulukuusi

[joulukuusi]

ёлка

[jolkə]

Christmas tree

9

neula

[neula]

иголка

[igolkə]

Needle

10

viulu

[viulu]

скрипка

[skrʲipkə]

violin

11

laiva

[laiva]

корабль

[kʌrablʲ]

ship

Case Studies of Phonological Development

Word in Finnish

Targeted IPA

Word in Russian

Targeted IPA

English translation

12

poika

[poika]

мальчик

[malʲt͡ ɕik]

boy

13

ui

[ui]

плавает

[plavəjit]

(he) swims

14

leipä

[leipæ]

хлеб

[hlʲep]

bread

15

äiti

[æiti]

мама

[mamə]

mother

16

päärynöitä

[pæærynøitæ]

груши

[gruʂɨ]

pears

17

lyijykynä

[lyijykynæ]

карандаш

[kərʌndaʂ]

pencil

18

täynnä

[tæynnæ]

полный

[polnɨj]

full

19

pöytä

[pøytæ]

стол

[stol]

table ball

20

pallo

[pallo]

мяч

[mʲat͡ ɕ]

21

piippu

[piippu]

трубка

[trupkə]

tobacco pipe

22

talo

[talo]

дом

[dom]

house

23

tutti

[tutti]

соска

[soskə]

pacifier

24

sukat

[sukat]

носки

[nʌskʲi]

socks

25

sydän

[sydæn]

сердце

[sʲert s͡ə]

heart

26

kala

[kala]

рыба

[rɨbə]

fish

27

kukka

[kukka]

цветок

͡ [tsvʲitok]

flower

28

vauva

[vauva]

малыш

[mʌlɨʂ]

baby

29

ovi

[ovi]

дверь

[dvʲerʲ]

door

30

suu

[suu]

рот

[rot]

mouth

31

kissa

[kissa]

кошка

[koʂkə]

cat

32

lammas

[lammas]

овца

[ʌft͡ sa]

sheap

33

juna

[juna]

поезд

[pojist]

train

34

leija

[leija]

змей

[zmʲej]

kite

35

hattu

[hattu]

шляпа

[ʂlʲapə]

hat

36

puhelin

[puhelin]

телефон

[tʲilʲfon]

telephone

37

radio

[radio]

радио

[radʲio]

radio

38

orava

[orava]

белка

[bʲelkə]

squirrel

39

lippu

[lippu]

флаг

[flak]

flag watch

111

40

kello

[kello]

часы

[t͡ ɕisɨ]

41

muna

[muna]

яйцо

[jijt͡so]

egg

42

mummo

[mummo]

бабушка

[babuʂkə]

grandmother

43

nenä

[nenæ]

нос

[nos]

nose

44

hevonen

[hevonen]

лошадь

[loʂətʲ]

horse

45

avain

[avain]

ключ

[klʲut͡ ɕ]

key

46

kengät

[keŋæt]

ботинки

[bʌtʲinkʲi]

boots

47

lapsi

[lapsi]

ребёнок

[rʲibʲonək]

child

48

itkee

[itkee]

плачет

[plat͡ ɕit]

(he) cries (to cry) (continued)

112

An Anthology of Bilingual Child Phonology

Word in Finnish

Targeted IPA

Word in Russian

Targeted IPA

English translation

49

veitset

[veitset]

ножи

[nʌʐɨ]

knives

50

sakset

[sakset]

ножницы

[noʐnʲit͡ sɨ]

scissors

51

tasku

[tasku]

карман

[kʌrman]

pocket

52

istuu

[istuu]

сидит

[sʲidʲit]

(he) seats (to seat)

53

lehmä

[lehmæ]

корова

[kʌrovə]

cow

54

tähti

[tæhti]

звезда

[zvʲizda]

star

55

perhonen

[perhonen]

бабочка

[babət͡ ɕkə]

butterfly

56

porkkana

[porkkana]

морковка

[mʌrkofkə]

carrot

57

korva

[korva]

ухо

[uhə]

ear

58

hylje

[hylje]

тюлень

[tʲulʲenʲ]

seal

59

pulkka

[pulkka]

санки

[sankʲi]

sled

60

silmä

[silmæ]

глаз

[glas]

eye

61

kampa

[kampa]

расчёска

[rʌɕːoskə]

comb

62

kynttilä

[kynttilæ]

свечка

[svʲet͡ ɕkə]

candle

63

kaktus

[kaktus]

кактус

[kaktus]

cactus

Acknowledgements

The author is indebted to Arto Mustajoki and Ekaterina Protasova for their invaluable advice, guidance and institutional help. Thanks are also given to Aleksandr Ovchinnikov, Natalia Rodina, the Kalinka and Mishka bilingual kindergartens, the Engel Finnish kindergarten and the Finnish-Russian School in Helsinki for their help in collecting the data. Recognition also goes to COST Action IS0804 and the annual scientific seminar The Problems of Ontolinguistics at Herzen State Pedagogical University, Russia. References Armon-Lotem, S., de Jong, J. and Meir, N. (eds) (2015) Assessing Multilingual Children: Disentangling Bilingualism from Language Impairment. Bristol: Multilingual Matters. Avanesov, R.I. (1972) Russkoe literaturnoe proiznošenie [Russian Literary Pronunciation]. Moscow: Prosveščenie. [In Russian] Belʹtjukov, V.I. and Salakhova, A.D. (1975) Ob usvoenii detʹmi zvukovoj (fonemnoj) sistemy jazyka [About the acquisition of sound (phonemic) system of language in children]. Voprosy psihologii [Psychological Questions] 5, 71–76. [In Russian] Bernhardt, B.M. and Stemberger, J. (1998) Handbook of Phonological Development: From a Nonlinear Constraints-based Perspective. San Diego, CA: Academic Press. Bialystok, E. (2001) Bilingualism in Development: Language, Literacy, and Cognition. Cambridge: Cambridge University Press. Bialystok, E. and Feng, X. (2011) Language proficiency and its implications for monolingual and bilingual children. In A.Y. Durgunoğlu and C. Goldenberg (eds) Language and Literacy Development in Bilingual Settings (pp. 121–138). New York: Guilford Press.

Case Studies of Phonological Development

113

Bialystok, E., Luk, G., Peets, K.F. and Yang, S. (2010) Receptive vocabulary differences in monolingual and bilingual children. Bilingualism: Language and Cognition 13, 525–531. Bondarko, L.V. (1998) Fonetika sovremennogo russkogo jazyka [Phonetics of the Modern Russian Language]. St Petersburg: Saint Petersburg University. [In Russian] Bondarko, L. (2009) Short description of Russian sound system. In V. de Silva and R. Ullakonoja (eds) Phonetics of Russian and Finnish: General Description of Phonetic Systems – Experimental Studies on Spontaneous and Read-aloud Speech (pp. 23–35). Frankfurt am Main: Peter Lang. Bulanin, L.L. (2009) Fonetika sovremennogo russkogo jazyka [Phonetics of the Modern Russian Language]. Moscow: Librokom. [In Russian] Cejtlin, S.N. (2000) Jazyk i rebenok [Language and child]. Lingvistika detskoj reči [Linguistics of Child Speech]. Moscow: Vlados. [In Russian] Cejtlin, S.N. (2008) Rečevye ošibki i ih predupreždenie [Speech Errors and Prevention of Errors]. St Petersburg: MiM. [In Russian] de Silva, V. (1999) Quantity and Quality as Universal and Specifi c Features of Sound Systems: Experimental Phonetic Research on Interaction of Russian and Finnish Sound Systems. Jyväskylä: University of Jyväskylä. Döpke, S. (2000) On the status of cross-linguistic structures in research on young children growing up with two languages simultaneously. In S. Döpke (ed.) Cross-linguistic Structures in Simultaneous Bilingualism (pp. 1–10). Amsterdam: John Benjamins. Eliseeva, M.B. (2008) Fonetičeskoe i leksičeskoe razvitie rebenka rannego vozrasta [The Early Phonetic and Lexical Development of the Child]. St Petersburg: Izdatelʹstvo RGPU im. A.I. Gercena. [In Russian] Eliseeva, M.B. (2014) Stanovlenie individualʹnoj jazykovoj sistemy rebenka: Rannie etapy [The Development of Individual Language System of a Child: Early Stages]. Moscow: Jazyki slavjanskih kulʹtur. [In Russian] Fabiano-Smith, L. and Barlow, J.A. (2010) Interaction in bilingual phonological acquisition: Evidence from phonetic inventories. International Journal of Bilingual Education and Bilingualism 13 (1), 81–97. Filičeva, T.B. and Čeveleva, N.A. (1987) Logopedičeskaja rabota v specialʹnom detskom sadu [Logopedic Work in a Special Kindergarten]. Moscow: Prosveščenie. [In Russian] Gildersleeve-Neumann, C., Kester, E., Davis, B. and Peña, E. (2008) English speech sound development in preschool-aged children from bilingual Spanish-English environments. Language, Speech, and Hearing Services in Schools 39, 314–328. Gvozdev, A.N. (1961) Voprosy izučenija detskoj reči [Questions about Studying Child Speech]. Moscow: Izd-vo APN SSSR. [In Russian] Hoff, E. (2015) Language development in bilingual children. In E.L. Bavin and L. Naigles (eds) The Cambridge Handbook of Child Language (pp. 483–503). Cambridge: Cambridge University Press. Iivonen, A. (1994) Paradigmaattisia ja syntagmaattisia näkökohtia lapsen foneettis-fonologisessa kehityksessä [Paradigmatic and syntagmatic aspects in the phonetic and phonological development of children]. In A. Iivonen, A. Lieko and P. Korpilahti (eds) Lapsen normaali ja poikkeava kielen kehitys [The Normal and Abnormal Development of Child Language] (pp. 34–77). Helsinki: Finish Literature Society. [In Finnish] Iivonen, A. (1998) Suomen ja saksan vokaalien laadun ja keston vertailu. Vieraan kielen fonetiikan opetuksen näkökohtia [Comparison of Finnish and German vowels’ quality and duration. Aspects of teaching foreign language phonetics]. In A. Iivonen and T. Nevalainen (eds) Helsingin yliopiston fonetiikan laitoksen julkaisuja [The Publications of Phonetics Department of the University of Helsinki] (pp. 47–70). Helsinki: Hakapaino Oy. [In Finnish] Iivonen, A. (2009) Major features of standard Finnish phonetics. In V. de Silva and R. Ullakonoja (eds) Phonetics of Russian and Finnish: General Description of Phonetic

114

An Anthology of Bilingual Child Phonology

Systems – Experimental Studies on Spontaneous and Read-aloud Speech (pp. 45–65). Frankfurt am Main: Peter Lang. Ingram, D. (1976) Phonological Disability in Children. London: Edward Arnold. Jortikka, T. (1993) Dysfonologia lapsen kielessä [Dysphonology in child language]. In A. Iivonen, A. Lieko and P. Korpilahti (eds) Lapsen normaali ja poikkeava kielen kehitys [The Normal and Abnormal Development of Child Language] (pp. 78–98). Helsinki: SKS. [In Finnish] Kallioinen, V. (1969) Suomen fonetiikan rakenneharjoituksia [Structural Exercises in Finnish Phonetics]. Helsinki: Helsingin Liikekirjapaino. [In Finnish] Karlsson, F. (1969) Suomen yleiskielen segmentaalifoneemien paradigma [The paradigm of segmental phonology in standard Finnish]. Virittäjä 73, 351–362. [In Finnish] Karlsson, F. (1983) Suomen kielen äänne- ja muotorakenne [The Sound and Form Structure of the Finnish Language]. Helsinki: WSOY. [In Finnish] Kehoe, M. (2010) Relationships between lexical and phonological development: A look at bilingual children – a commentary on Stoel-Gammon’s ‘Relationships between lexical and phonological development in young children’. Journal of Child Language 38 (1), 75–81. Kehoe, M. (2018) Crosslinguistic interaction in early bilingual phonology: A critical review. In E. Babatsouli and D. Ingram (eds) Phonology in Protolanguage and Interlanguage (pp. 49–75). Sheffield: Equinox. Korpinen, L. and Nasretdin, F. (2009) Artikulaatio ja äännevirheet. In K. Launonen and A.-M. Korpijaakko-Huuhka (eds) Kommunikoinnin häiriöt. Syitä, ilmenemismuotoja ja kuntoutuksen perusteita [Articulation and Sound Errors. Communication Disorders. Reasons, Forms of Manifestation and Arguments for Rehabilitation] (pp. 59–76). Helsinki: Yliopistopaino. [In Finnish] Kraehenmann, A. (2001) Swiss German stops: Geminates all over the word. Phonology 18, 109–145. Kunnari, S. (2000) Characteristics of early lexical and phonological development in children acquiring Finnish. Acta Universitatis Ouluensis. PhD thesis, University of Oulu. Kunnari, S. and Savinainen-Makkonen, T. (eds) (2012) Pienten sanat: Lasten äänteellinen kehitys [The Words of Children: Sound Development in Children]. Jyväskylä: PS-kustannus. [In Finnish] Lanza, E. (2000) Language contact – a dilemma for the bilingual child or for the linguist? In S. Döpke (ed.) Cross-linguistic Structures in Simultaneous Bilingualism (pp. 227– 245). Amsterdam: John Benjamins. Launonen, K. (2007) Monikielisyyden haasteet kielihäiriöiden tunnistamisessa ja kuntoutuksessa [The challenges of multilingualism in identification and rehabilitation of language impairment]. Virittäjä 2, 240–243. [In Finnish] Leonard, L. (1998) Children with Specifi c Language Impairment. Cambridge, MA: MIT Press. Ljakso, L.L. (2008) Stanovlenie reči rebenka v ontogeneze: Longitjudnoe issledovanie [Child speech development in ontogenesis: A longitudinal study]. In T.N. Ušakova (ed.) Reč rebenka – problemy i rešenija [Child Speech – Problems and Solutions] (pp. 54–89). Moscow: Institut Psihologii RAN. [In Russian] Ljubimova, N. (2010) Projavlenie fi nsko-russkoj fonetičeskoj interferencii pri vosprijatii i vosproizvedenii slov v uslovijah stanovlenija subordinativnogo dvujazyčija [The occurrence of Finnish-Russian phonetic transfer during the comprehension and production of words while becoming sequential bilingual]. In A. Mustajoki (ed.) Izučenie i prepodavanie russkogo jazyka v Finljandii [Learning and Teaching Russian Language in Finland] (pp. 341–348). St Petersburg: Zlatoust. [In Russian] Major, R.C. (2008) Transfer in second language phonology. In J.G.H. Edwards and M.L. Zampini (eds) Phonology and Second Language Acquisition (pp. 63–94). Amsterdam/ Philadelphia, PA: John Benjamins.

Case Studies of Phonological Development

115

Maslov, Ju.S. (2007) Vvedenie v jazykoznanie [Introduction to Linguistics]. St Petersburg: Faculty of Philology, Saint Petersburg University. [In Russian] Menn, L. and Stoel-Gammon, C. (2005) Phonological development: Learning sounds and sound patterns. In J. Berko Gleason (ed.) The Development of Language (6th edn) (pp. 62–111). Boston, MA: Pearson, Allyn & Bacon. Mohanan, K.P. (1992) Emergence of complexity in phonological development. In C.A. Ferguson, L. Menn and C. Stoel-Gammon (eds) Phonological Development: Models, Research, Implications (pp. 635–662). Timonium, MD: York Press. Nenonen, О. (2016) Fonetičeskoe razvitie russko-fi nskih bilingvov doškolʹnogo vozrasta [The phonetic development of Russian–Finnish preschool-aged bilingual children]. PhD thesis, Helsinki: Slavica Helsingiensia 48. [In Russian] Paradis, J. (2010) The interface between bilingual development and specific language impairment. Applied Psycholinguistics 31 (2), 227–252. Paradis, J. and Genesee, F. (1996) Syntactic acquisition in bilingual children: Autonomous or interdependent? Studies in Second Language Acquisition 18, 1–25. Piotrovskaja, L.A. (2011) Psiholingvističeskie mehanizmy fonetičeskih substitucij v ontogeneze [Psycholinguistic mechanisms of phonetic substitutions in ontogenesis]. In T.A. Krugljakova (ed.) Ontolingvistika – nauka XXI veka [Ontolinguistics – the Science of the XXI Century] (pp. 97–102). St Petersburg: Zlatoust. [In Russian] Polka, L. and Sundara, M. (2003) Word segmentation in monolingual and bilingual infant learners of English and French. In M.J. Solé, D. Recasens and J. Romero (eds) Proceedings of the 15th International Congress of Phonetic Sciences (pp. 1021–1024). Barcelona: ICPhS Archive. Protasova, E.Ju. and Rodina, N.M. (2005) Mnogojazyčie v detskom vozraste [Multilingualism in Childhood]. St Petersburg: Zlatoust. [In Russian] Remes, K. and Ojanen, A. (1997) Artikulaatiotesti. Äänteenmukainensanakuvatesti [Articulation Test. Word Naming Task]. Helsinki: Early Learning. [In Finnish] Roos, E.M. and Weismer, S.E. (2008) Language outcomes of late talking toddlers at preschool and beyond. See https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2862655/ (accessed 27 February 2019). Saaristo-Helin, K., Kunnari, S. and Savinainen-Makkonen, T. (2011) Phonological development in children learning Finnish: A review. First Language 31, 342–363. Savinainen-Makkonen, T. (2001) Suomalainen lapsi fonologiaa omaksumassa [Finnish children acquiring phonology]. Helsingin yliopiston Fonetiikan laitoksen julkaisuja [Publication of the Department of Phonetics] 42. PhD thesis, University of Helsinki. [In Finnish] Silvén, M., Voeten, M., Kouvo, A. and Lundén, M. (2014) Speech perception and vocabulary growth: A longitudinal study of Finnish-Russian bilinguals and Finnish monolinguals from infancy to three years. International Journal of Behavioral Development 38 (4), 323–332. Skalozub, L. (1963) Palatogrammy i Rentgenogrammy Soglasnyx Fonem Russkogo Literaturnogo Jazyka [Palatograms and Roentgenograms of the Russian Language Consonant Phonemes]. Kiev: Izdatelstvo Kievskogo Universiteta. [In Russian] Stoel-Gammon, C. (1985) Phonetic inventories, 15–24 months: A longitudinal study. Journal of Speech and Hearing Research 28 (4), 505–512. Stoel-Gammon, C. (1998) Sounds and words in early language acquisition: The relationship between lexical and phonological development. In R. Paul (ed.) Exploring the Speech-language Connection (pp. 25–52). Baltimore, MD: Brookes. Stoel-Gammon, C. and Herrington, P. (1990) Vowel systems of normally developing and phonologically disordered children. Clinical Linguistics and Phonetics 4, 145–160. Stoel-Gammon, C. and Stone, J.R. (1985) Assessing phonology in young children. Clinics in Communication Disorders 1 (2), 25–39.

116

An Anthology of Bilingual Child Phonology

Suomi, K., Toivanen, J. and Ylitalo, R. (2008) Finnish Sound Structure: Phonetics, Phonology, Phonotactics and Prosody. Studia Humaniora Ouluensia 9. Oulu: University of Oulu. See http://jultika.oulu.fi /fi les/isbn9789514289842.pdf (accessed 7 June 2018). Švačkin, N.H. (1995) Razvitie rečevyh form u mladšego doškol’nika [The development of speech forms in a young preschool-aged child]. In A.N. Leontjev and A.V. Zaporožec (eds) Voprosy psihologii rebenka doškol’nogo vozrasta [Psychological Questions Regarding Preschool Children] (pp. 99–110). Moscow: Meždunarodnyj obrazovatel’nyj i psihologičeskij colledg. [In Russian] Swadesh, M. (1995) The phonemic principle. In E.P. Hamp, M. Joos, F.W. Householder and R. Austerlitz (eds) Readings in Linguistics I & II (abridged edn) (pp. 14–19). Chicago, IL: University of Chicago Press. Toivainen, J. (1990) Acquisition of Finnish as a First Language: General and Particular Themes. Publications of the Department of Finnish and General Linguistics of the University of Turku No. 35. Turku: Turun Yliopisto. Toivola, M. (2011) Vieraan aksentin arviointi ja mittaaminen suomessa [The Assessment and Measuring of Foreign Accent in Finnish]. Helsinki: Unigrafia. [In Finnish] Vihman, M.M. (1988) Early phonological development. In J. Bernthal and N. Bankson (eds) Articulation and Phonological Disorders (pp. 60–110). Englewood Cliff s, NJ: Prentice Hall. Vihman, M.M. (1996) Phonological Development: The Origins of Language in the Child. Oxford: Blackwell. Vihman, M.M. (2002) Getting started without a system: From phonetics to phonology in bilingual development. International Journal of Bilingualism 6 (3), 239–254. Vihman, M.M. (2010) Phonological templates in early words: A cross-linguistic study. In C. Fougeron, B. Kühnert, M. D’Imperio and N. Vallée (eds) Variation, Phonetic Detail and Phonological Representation (pp. 261–284). Laboratory Phonology 10. Berlin: Mouton de Gruyter. Vihman, M.M. and Keren-Portnoy, T. (eds) (2013) The Emergence of Phonology: Whole Word Approaches, Cross-linguistic Evidence. Cambridge: Cambridge University Press. Weinreich, U. (1972) Odnojazyčie i mnogojazyčie [Monolingualism and multilingualism]. Novoe v lingvistike VI: Jazykovie kontakty [New Questions in Linguistics VI: Language Contacts] (pp. 25–60). Moscow: Progress. [In Russian]

5 Enhanced Phonology in a Child’s Weaker Language in Bilingualism: A Portrait Elena Babatsouli

Introduction

Current research in child language acquisition encourages a crosssectional methodological design, i.e. one involving several child participants, hoping to guarantee a more representative sample of actual populations and, thus, aspiring to come closer to accurately deciphering universal mechanisms. As valid and rigorous as this approach may be, it often results in research where the documentation and elaboration of language data per se is implied. On the other hand, prodigious works like those by Werner Leopold (1949) and Neil Smith (1973) are examples of meticulous language research that zooms in on individual children’s developmental data in great detail and along sizeable lengths of time but these, being earlier contributions, have failed to provide quantitative information and to scrutinize the influence of bilingual exposure/use on the development of the language(s) studied. Since then, an ever-increasing number of bilingual studies crosslinguistically have been furnishing our understanding of child phonologies that develop bilingually, examining the extent of the match between a bilingual’s languages and those of respective monolinguals, and/or the extent of interaction between a bilingual’s languages, since absolute language separation is not considered a routine scenario even among fluent adult bilinguals (Weinreich, 1953). Nevertheless, ‘[t]he criteria used to predict bilingual outcomes are coarse more than fi ne’ and suggestions are being made to quantify both input and output (Lleó, this volume). Despite the plethora of bilingual studies (see reviews in Babatsouli & Ball, this volume; Babatsouli & Ingram, 2018; also involving L3, see Wrembel & Cabrelli Amaro, 2018), however, (1) little has been done in terms of quantifying a child’s input and output (e.g. Babatsouli & Nicoladis, 2019), especially with regard to phonology, and (2) little is known about child bilingual phonological performance with exclusive interlanguage (IL) (Selinker, 1972) exposure, meaning second language input, in an exogenous environment. 117

118

An Anthology of Bilingual Child Phonology

The current study presents such a case, drawing the portrait of a bilingual child’s developing phonologies as a snapshot in time, and aiming to investigate whether the phonology of the language with compromised exposure is also compromised in acquisition level when compared to that of the language with ambient environment exposure. In particular, the study evaluates a girl’s speech performance at age 2;7 in two typologically different languages, English (stressed-timed) and Greek (syllable-timed). Assessing speech and language skills across a bilingual’s languages is vital for the understanding of bilingualism as the status quo in miscellaneous contexts of exposure and use on a global scale (Babatsouli, 2019c), as well as for the insights this provides for reliably differentiating between language pair differences crosslinguistically, and in the context of language disorder (e.g. Babatsouli, forthcoming; Babatsouli & Ball, this volume; Babatsouli et al., 2017; McLeod & Goldstein, 2012). Bilingual children crosslinguistically run the double risk of either being misdiagnosed with speech delay/impairment or, conversely, of failing to be accurately identified when they do have a speech disorder (see also Vender & Melloni, this volume). Studies of child bilingualism also carry the potential to inform and be informed by developing second language phonologies in adults (e.g. Babatsouli, 2017; Babatsouli & Ingram, 2018). The sub-sections that follow review relevant literature, introduce the child participant of this study and her milieu, and subsequently present results and discuss the fi ndings of the study. Bilingual speech development

Any evaluation of bilingual speech acquisition involves a comparison to monolingual acquisition since the two are concomitant; bilingual children follow comparable developmental paths with similar patterns/processes and mistakes to those of monolingual children (e.g. Babatsouli, 2017; Babatsouli & Sotiropoulos, 2018; Bunta et al., 2009; Ingram, 1981a; Lleó & Kehoe, 2002), while analogous patterns are also evidenced in bilinguals with speech sound disorders (e.g. Bunta & Douglas, 2013; Ingram, 2012). Individual child variation renders minute differences between monolingual and bilingual acquisition harder to determine (e.g. Paradis, 2000). In a comparison of bilingual skill between the languages, however, several issues emerge: (a) language dominance/status (which language is stronger or first (L1), weaker or second (L2)) (e.g. Meisel, 2007); (b) a lack of absolute degree of separation of the systems on an underlying level (e.g. Bunta et al., 2006; Paradis, 2000); (c) there are subtle differences in the individual languages of bilingual children compared to monolinguals (e.g. Ball et al., 2001; Law & So, 2010; Lleó et  al., 2003); and (d) there is interaction between the languages facilitating or hindering acquisition (e.g. Babatsouli, 2019d; Babatsouli & Nicoladis, 2019; Fabiano-Smith & Barlow, 2010; Kehoe, 2018, this volume; Kopečková et al., 2016; Lleó, 2018a, 2018b).

Enhanced Phonology in a Child’s Weaker Language in Bilingualism

119

A determination of bilingual performance also depends on other mitigating factors such as age, degree and sociolinguistic context of exposure, degree of language use or activation, and affective factors (e.g. Gollan et al., 2008; Grosjean & Li, 2013; Jasinska & Petitto, 2013; Wei, 2010). Despite the abundance of research on child bilingual acquisition, however, there is a dearth of research on early bilingual development with nonnative, exogenous exposure (one language not readily available in its native form in the ambient environment) to one of the languages, or of studies in the linguistic development of bilingual Greek-English (e.g. Antoniou et al., 2010; Babatsouli, 2017; Babatsouli & Nicoladis, 2019). Assessing bilingual skill Vocabulary

When we speak about the lexicon, we refer to ‘the inventory of words in a language’ alongside those structural rules that determine the semantics, morphology and phonology of those words; words in a bilingual’s languages are directly interconnected via shared concepts, via separate lexical nodes or both (Babatsouli, 2019b). Work on vocabulary evaluation in bilinguals includes batteries that test both conceptual and actual vocabulary. Conceptual vocabulary tests include fast mapping and dynamic assessment of vocabulary in development (e.g. Eviatar et  al., 2018; Maragkaki & Hessels, 2017). On the other hand, static vocabulary tests, either standardized or informal, rely on information based on individual snapshots in time (e.g. Hasson et al., 2013). The child’s lexical knowledge and breadth of vocabulary influences the rate and pace of phonological acquisition (e.g. Ervin & Miller, 1963; Kehoe et al., 2018; Stoel-Gammon, 2011). Lexical qualities that predict speech development include word frequency and phoneme sequence frequency (phonotactic probability) in the targeted language(s) (e.g. Babatsouli, 2019a; Bernhardt & Stemberger, 1998). Grammatical level

Utterance length has been a measure of child speech progress for almost 100 years, known as average length of sentence (ALS; Nice, 1925), and renamed mean length of utterance in words (MLUw; Bernardini & Schlyter, 2004), following Brown’s mean length of utterance in morphemes (MLU; Brown, 1973). MLU has often been a preferred gauge for measuring bilingual performance and identifying dominance patterns (e.g. Treffers-Daller, 2011). Since one of the languages examined in this study is Greek, which is morphologically richer than the other language, English, MLU will not be an appropriate measure to compare language performance, but MLUw will. Bernardini and Schlyter (2004) have found the two measures are comparable and suggest that the developing language is weak in a child’s bilingualism whose MLUw is at least one word less than that of the other language.

120

An Anthology of Bilingual Child Phonology

Phonology

Quantitative approaches to evaluating the acquisition of phonology have been statistical in nature revealing the extent of acquisition of grammatical structures, measuring correctness of, for example, segments (e.g. Shriberg et al., 1997), segment sequences (Babatsouli & Sotiropoulos, 2018) or whole words (Ingram & Ingram, 2001). Production accuracy is determined as degree of correct (match, adult-like) use of a structure in obligatory environments with several criteria in use: 70% (e.g. Ingram et al., 1980), 75% (e.g. Diedrich & Bangert, 1980) or 90% (Brown, 1973) correct production. In cross-sectional studies, correctness is viewed in terms of group performance (e.g. Smit et al., 1990). The only quantitative approach to monolingual Greek phonology ascertaining norms at the 75% criterion of group performance is PAL (1995). Schmitt et al. (1983) proposed the whole-word accuracy (WWA) formula during a single production of a given word; this is a measure that favourably complements existing measures of speech performance. With regard to singleton consonant performance, Shriberg et al.’s (1997) proportion of consonants correct (PCC) formula, computing the cumulative consonant correctness for each word type and then averaging them arithmetically, is widely accepted. Babatsouli and Sotiropoulos (2018) have proposed a measure for cluster proximity (MCP) which distinguishes between different two-member (CC) cluster productions, giving credit to various non-adult-like productions; so, one adult-like cluster member scores 25%, two adult-like members with vowel epenthesis score 62.5% and CC production with one member substituted scores 87.5%. Beyond these, assessment of phonological similarity among children has been viewed in terms of phonological mean length of utterance (PMLU) (Ingram & Ingram, 2001) which accounts for singleton performance in an utterance; unlike MLU (Brown, 1973) which counts all morphemes equally, PMLU doubles the count of correct-in-context consonant segments, because errors occur more often on targeted consonants (e.g. Ingram, 1981b). The relation between produced and target words in child speech is accounted for by phonological whole-word proximity (PWP) (Ingram & Ingram, 2001), which divides produced PMLU by the targeted PMLU. In a multidimensional phonological approach, Ingram and Dubasik (2011) propose nine measures that address whole-word, word position and prosodic aspects. A number of studies have used whole-word measures to quantify the level of phonological development in normal (e.g. Bunta et al., 2009) and phonologically impaired (Burrows & Goldstein, 2010) bilingual children. Comparing English and Greek

As a fusional language, Greek (el, ISO 639.1, see https://www.loc.gov/ standards/iso639-2/php/code_list.php) is grammatically more complex than English (en, ISO 639.1). Being a syllable-timed language, it also has

Enhanced Phonology in a Child’s Weaker Language in Bilingualism

121

Table 5.1 Consonantal inventories in Greek and English Common (15)

Greek only (16)

English only (17)

Plosives

p, b, t, d, k, g

[mp, mb, ŋg, c, ɟ]

[ph, th, kh]

Nasals

m, n

[ŋ, ɱ, ɲ]

ŋ, [m̩ , n̩ ]

Laterals

l

[ʎ]

[ɫ, l̩ ]

ɾ

ɹ, [ɹ̩ ]

[ç, ʝ], x, ɣ

ʃ, ʒ, h

ts, dz

tʃ, ʤ

Rhotics Fricatives

f, v, θ, ð, s, z

Affricates Glides

j, w

Note: Bold symbols denote phonemic and phonetic status and brackets denote allophones.

a simpler syllabic structure (C 0–3VC 0–1) and mostly longer words (bisyllabic and multisyllabic) than English (a stress-timed language) whose syllabic structure is more complex (C 0–3VC 0–4) (Mennen & Okalidou, 2007); the few monosyllabic words in Greek are function words. For a recent comprehensive review of Greek phonology, see Babatsouli (2019a). There is a larger proportion of consonants to vowels in stress-timed languages (Nespor et al., 2003), which also holds for Greek and English: English monosyllabic words are more complex, involving a larger variety of consonants and consonant clusters than Greek, reflecting the respective phonotactic differences of the two languages with regard to permissible word-fi nal consonant singletons; so, with the exception of loans, only /n, s/ are allowed at the end of Greek words (e.g. Babatsouli & Nicoladis, 2019). Table 5.1 shows consonantal inventories in the languages. With regard to their consonantal inventories, the two languages differ phonemically in that: Greek has two velar fricatives, /x ɣ/, while English has two alveopalatals, /ʃ ʒ/, and a glottal consonant, /h/; aff ricates in Greek are alveolar, /ts, dz/, and in English alveopalatals, /tʃ dʒ/; glides /j w/ and the velar nasal /ŋ/ are phonemic in English but allophonic and dialectal in Greek; and the English rhotic is an approximant, /ɹ/, while in Greek it is a flap, /ɾ/. Among their shared phonemes, /p b f v t d s z θ ð m n l k g/, allophonic processes such as aspiration (en), velarization (en, el), palatalization (el) and pre-nasalization (el) differentiate the two languages phonetically, giving ten allophones in Greek, [c ɟ ç ʝ ɱ (ɱp ɱb ɱf ɱv) ŋ (ŋk ŋg ŋx ŋγ) ɲ ʎ r ɹ], and eight in English, [ph th kh m ̩ n̩ ɫ l̩ ɹ̩ ]. Method Child and milieu

Born and raised in Greece by native Greek parents, the child participant in this study was exposed to the ambient language (Greek) from birth. Following the parents’ decision to raise her bilingually, consistent

122

An Anthology of Bilingual Child Phonology

exposure to English began on a daily basis at age 1;0, coming exclusively through the mother’s English input. The mother is a fluent L2 speaker of English – a result of a combination of learning English in a foreign instructional setting when young (Greece), subsequent higher education studies in the UK – and a total of more than 10 years’ residence experience in the UK and the US. Her English is native-like; intermittent production of both English targets, and transfers (Flege & Davidian, 1984; Gass & Selinker, 1983; Major, 2008; Weinreich, 1953) from Greek are marked by a parenthesis in the following representation of the mother’s consonantal phonetic inventory in English: p (ph), b, d, g, t (th), k, (kh) m, (m̩ ), n, (n̩ ), ŋ, l, ɫ, l̩ , ɹ, (ɹ̩ ), θ, ð, f, v, s, z, ʃ, ʒ, tʃ, ʤ, j, w, h Targeted English → Transfer from Greek k, kh → (c), g → (ɟ), ɹ → (ɾ), j → (ʝ), h → (x), (ç)

English productions

The child’s father (also a second language speaker of English) only spoke Greek to the child. During the child’s first two years, the child spent the overwhelming majority of time with her mother. Exposure to English, therefore, was bigger than to Greek in the year prior to age 2 years. At 2;0 the child started attending daycare in monolingual Greek for 7 hours a day 5 days a week, thus interacting in English with her mother during the remaining intervals. Data set

The onset of the child’s recognizable speech at 1;7 (Greek) and 1;9 (English) meets expectations in both monolingual and bilingual development (e.g. Brulard & Carr, 2003; Ingram, 1989). She entered the twoword stage (Ingram, 1989) in English at 2;6. Voluble and stable enough production in English, facilitating recorded data collection, ensued at 2;7 (the month of the study). The child’s speech was digitally recorded at home with a hand-held Olympus WS11-311M three to four times a week during unstructured routine interaction between mother and child. The recordings – a total of 182 minutes – mostly targeted capturing the child’s English and were of short duration, reflecting the child’s overall reluctance to speak English. Full utterances were IPA phonetically transcribed and entered by the author in a CLAN (MacWhinney, 2000) database of 41 CHAT fi les, with 785 utterances in English and 688 in Greek, as they naturally occurred. The database fi les include utterances in each language and mixed utterances, as code-switching and joint activation rather than a strictly monolingual mode (Grosjean & Li, 2013) was typical in the child’s speech at 2;7 and during data collection. Acoustic analyses (using Praat; Boersma &

Enhanced Phonology in a Child’s Weaker Language in Bilingualism

123

Weenink, 2018) of representative consonant realizations during frequent intervals verify the reliability of phonetic transcription, by and large. Results and Discussion

The child’s bilingual data are presented and discussed next in three sections: lexicon and grammar, whole-word accuracy and consonant performance, in terms of acquisition level, substitutions and phonotactic differences. Lexicon and grammar

There are 540 Greek word types (2374 tokens) and 317 English types (1516 tokens), of which 20% and 22% in English and Greek, respectively, are function words. A comparison of types-to-tokens ratio (TTR) (English: 0.21, Greek: 0.23) shows that each type is produced on average five times during the month, an indication of balanced use of the languages despite the child’s overall reluctance in English. A comparison to monolingual English vocabulary development norms of between 300 words (by age 2;0) and 1000 (by 3;0) (e.g. Ingram, 1989) suggests that her English lexicon is weaker than her Greek, but richer than the mean of 216.58 words between 2;6 and 2;11 of the 276 Greek children (aged 1;6– 2;11) studied by Papaeliou and Rescorla (2011) using a Greek adaptation of the Language Development Survey. Given that comparisons of bilingual children’s vocabulary should include words in both languages (e.g. Werker et al., 2009), the child’s sum of 857 types in the data at 2;7 is also notable. Furthermore, an examination of semantically different words between the languages (English types: 96%, Greek types: 71%) reveals that her Greek vocabulary appears richer because of the richer grammatical variance typical of Greek. This study focuses on consonants, so vowel-only words are ignored hereon. Next, the child’s results on the proportion of consonants-tovowels (PV), English-PV: 43% (2262/5206), Greek-PV: 50% (4551/9172), demonstrate the self-similarity of her speech to each targeted language. Similarly, monosyllabic word tokens (English: 1214, Greek: 897) account for 80% and 38% of her English and Greek, respectively, again mirroring language-specific patterns. This information on vocabulary provides evidence of typical linguistic development, indicating that the child is in the third, multi-word stage of phonological acquisition (Ingram, 1989). To compare the child’s English grammatical level to respective monolingual norms, MLU is calculated for all 785 utterances, resulting in 2.3, close to Brown’s (1973) Stage III: 2.5–3.0 (between 2;7 and 2;10). Because Greek is morphologically richer, grammatical comparison between the two languages cannot be made based on MLU. Instead, ALS (Nice, 1925) is used as a measure for comparison. ALS measures in both English (2.02)

124

An Anthology of Bilingual Child Phonology

and Greek (3.68) show more words per utterance in Greek: single-word utterances are frequent in English while complicated ones including clauses appear only in Greek. Her Greek ALS is higher than the averages (between 2.5 and 3.0 at 2;7) of five Greek speaking children in Marinis (2003). The phonological status of each language is investigated next. Whole-word accuracy and phonological measures

The child’s whole-word accuracy (WWA) (e.g. Ingram & Ingram, 2001; Schmitt et al., 1983) is investigated next (see Table 5.2). When all words are considered, WWA is the same in both languages; balance across languages is also evidenced in singleton-consonant words (i.e. without clusters, 52%). Monosyllabic words are produced considerably better than multisyllabic words in each language, matching universal expectations in phonological acquisition (e.g. Ingram & Dubasik, 2011). Greek monosyllabic words, e.g. μου [mu] ‘me’, να [nɐ] ‘to’, της [tis] ‘her’, perform better because there are fewer types (mostly function words) with fewer singletons and no clusters word-finally, compared to her English, e.g. bit, last, leave. Only word-final /s, n, v/ are targeted in her Greek, while the following are targeted in her English: /p, ps, b, v, t, ts, nt, d, m, n, nz, nt, nd, ndz, s, ʃ, st, z, ʧ, ʤ, ɫ, ɫp, ɫd, ɫz, ɫk, ɹt, ɹz, ɹst, ɹʧ, ɹɫ, ɹk, k, ks, g, gz/. Overall, the child’s English multisyllabic words (with fewer word types, English: 111 versus Greek: 483), are produced slightly better than in Greek because they consist of three times fewer three-or-more syllables (34/302 = 11% in tokens, 20/111 = 18% in types) than in Greek (446/1477 = 30% in tokens, 246/483 = 51% in types). Two-syllable words in both languages are produced better than three-or-more syllable words. Two-syllable words in English are 37% (99/268) correct, while three-ormore syllable words are 21% (7/34) correct. In Greek, two-syllable words are 39% (407/1033) correct, while three-or-more syllable words are 20% (87/446) correct. Comparing words that contain clusters in the languages, we see that the child’s English performance is overall better than her Greek. It is noted, however, that the ‘better’ language (English) had fewer word types for both singleton (209) and cluster (108) words than her Greek (356 and Table 5.2 Whole word accuracy in English and Greek English WWA

Greek WWA

All word tokens

665/1516 = 44%

1057/2374 = 45%

Multisyllabic words

106/302 = 35%

498/1477 = 34%

Monosyllabic words

559/1214 = 46%

559/897 = 62%

Singleton-consonant words

635/1232 = 52%

1023/1967 = 52%

Consonant-cluster words

30/284 = 11%

34/407 = 8%

Enhanced Phonology in a Child’s Weaker Language in Bilingualism

125

184, respectively). Also, Greek multisyllabic words that involve clusters are produced slightly better than those in English, because there are clusters in English targeted in codas, [-ts], [-st], [-nt], [-nd], [-nz], [-ɹpl̩ ], [-ɹz], [-ʃt] (not so in Greek, Babatsouli, 2019a), that are produced mostly incorrectly. By and large, Greek is more correct in monosyllabic singleton and cluster words because of less complex and fewer word types, while English is ahead in multisyllabic words because of fewer three-or-more syllable words. PMLU (Ingram, 2002) is calculated next based on the 688 English utterances and the 785 Greek utterances. The child’s produced PMLU is 4.3 and 4.82 for a targeted PMLU of 5.38 and 5.81 in English and Greek, respectively. Consequently, we see that there is no lag in her English performance, as her results match Ingram’s PMLU Stage III: 4.5–5.5 for normative English. These differences confi rm separation of the child’s phonologies to the degree permitted by the specificities of each language but, regarding language dominance, the cumulative WWA discussed shows that both languages are equally strong phonologically; this is additionally corroborated by her PWP in each language: 80% (English), 83% (Greek). The child’s phonetic repertoire Acquisition level: Segments

This section evaluates the child’s consonant acquisition level quantitatively and qualitatively and discusses universal order of acquisition of segments (Jakobson, [1941] 1968) and successive feature contrasts (e.g. Bernhardt & Stemberger, 1998; Dinnsen, 1992; Ervin & Miller, 1963; Jakobson, [1941] 1968). The child’s substitution patterns are also presented and compared to those of respective monolingual norms in each language. This is done to establish universal and individual variation patterns in this child’s acquisition of consonants and to pinpoint differences between the two languages on both a qualitative and a quantitative basis. It is noted that analysis of consonantal sounds in this chapter includes singleton and cluster members cumulatively. The child’s consonantal inventory (portrait) in each language is given in Tables 5.3 (English) and 5.4 (Greek). Each portrait shows her consonantal use in exact proportions with regard to correctness, deletions and the substitutions she used. Column A: /C/ lists targeted consonants, while Row 1, [C], shows the child’s respective realizations (adult-like in context, substitutions and deletions). Therefore, proportions in each row add up to 1. The consonants of both languages are included in each portrait, so when there is no target, the corresponding cellin the diagonal is empty. Also, cells corresponding to deletions are left empty. However, proportions in the portraits are shown in two decimal places resulting in cells showing zeros when the proportion is smaller than 0.005. The numbers in bold diagonally in the portraits are the proportions of adult-like in-context realizations.

0.01

0.02

0.05

0.03 0.17

h

0.37

0.01 0.07

0.03

0.03

ʎ

0.02 0.12 0.01

0.01 0.26

0.01 0.47

0.78 0.01

0.01

0.53

0.57

0.06 0.09 0.18 0.18

0.01 0.04

0.21 0.65 0.01

0.86 0.02

0.01 0.01

0.38

0.21 0.07 0.07

0.01

0.29 0.66

l

0.08 0.25

n

w

0.03 0.01

0.01

0.10

0.00

0.01

0.01

dz

ŋ

ɫ

g

k

j

dʒ

tʃ

ʃ

ɹ

0.01

0.01

n

l

z

ts

0.01 0.03

z

0.87 0.02 0.01

0.22 0.69

0.33

0.06

s

0.03

0.01

d

0.14 0.05 0.01

0.17

0.01

t

s

0.03

0.00

ð

d

0.01

0.08

0.36

θ

0.83 0.02 0.02

0.01

0.03

v

0.77 0.02

f

t

ð

θ

v

0.07

0.95

0.02

m

f

0.95 0.03

0.03 0.93 0.01

m

p

b

b

[C]//C/ P

Table 5.3 The child’s English consonant portrait

0.01

ɹ

0.01 0.05

ɾ

ʒ

0.46

0.03 0.13

0.50

0.02

0.00

0.42

0.04

ʃ

0.01

0.56

tʃ

j

0.00 0.02

0.01 0.01

ɲ c

0.03

0.03 0.04 0.52 0.01

0.55

dʒ

0.05

ɟ

0.00

ç

0.03

k

0.00

g

0.01

x

0.33

0.02

ɫ

0.23

ŋ

0.15

0.01

w

0.07

0.00

0.01

h

0.02

0.04 0.08

0.23

0.24 0.26

0.03

0.03

0.03

0.14

0.55

0.47

0.19

0.04

0.03

0.09

0.12

0.36

0.57

0.09

0.00 0.05

0.01

0.01

vcl Ø

126 An Anthology of Bilingual Child Phonology

0.03

ʎ

0.01

0.05

0.03

0.01

x

γ

0.90 0.02

0.08 0.33

0.03

g

k

ç

1.00

0.82 0.01

ɟ

0.21

0.01

0.01 0.00

j

0.11

0.00

1.00

c

ɲ

ɾ

0.05

0.02

0.00 0.02

l

d

z

ts

0.23

0.27

0.00

0.02

0.03

0.78

0.02

dz

0.03

0.01

0.02

0.01

0.68 0.02

0.81 0.00 0.01

0.01

0.01 0.01

0.27 0.01 0.01

0.20

s

0.00 0.85 0.00 0.02

n

0.00

0.09

0.02

0.00

0.01

t

0.05 0.07 0.01

ð

0.05

0.02

0.04

0.01

θ

ts

z

s

0.00

0.00

t

0.00

0.02

ð

d

0.01 0.01

θ

0.01

0.02 0.03 0.92

v

0.05

0.69

0.00 0.00 0.98

m

v

f

0.05 0.94

f

0.96 0.01 0.00 0.01

m

b

b

p

[C]//C/ p

Table 5.4 The child’s Greek consonant portrait l

ʎ

0.01

0.52 0.03

0.16

0.01 0.46

0.23 0.15

0.02 0.83

0.84 0.02

0.01 0.40 0.01

0.02

0.00 0.01

n

ɹ

0.00

0.01 0.01

ɾ

ʒ

0.38

0.41

0.00

0.02

0.01

0.14

0.08

0.09

0.01 0.01

0.45

0.01

ʃ

0.20

tʃ

0.50

0.02

dʒ

0.01

j

0.90

0.01

0.05

0.29

0.54

0.00

0.04

0.01 0.02

ɲ

ɟ

0.00

0.04 0.01

c

0.16

0.00

0.01

ç

0.04

0.01

0.01

k

0.08

g

0.03

x

0.06

0.00

γ w

0.00 0.00

0.01

0.01

ɫ

0.01

0.14

0.01

0.16

0.10

0.49

0.14

0.11

0.02

0.02

0.14

0.10

0.31

0.10

0.09

0.00

0.01

0.00

Ø

0.25

0.03 0.23

h

Enhanced Phonology in a Child’s Weaker Language in Bilingualism 127

128

An Anthology of Bilingual Child Phonology

Table 5.5 Correct productions of segments English

Greek

Completely acquired: 90%

p, b, m

p, b, v, m, d, ɲ

very frequent: 75%–89%

f, s, t, n

t, s, z, n, l

Frequent: 50–74%

d, z, ʃ, ʧ, ʤ, j

f, ʦ

Not acquired: 1–49%

v, ð, l, ɹ, ɫ, ŋ, k, w, h

θ, ð, ʎ, ɾ, ʝ, c, ç, k, g, x, γ

Not used in context: 0%

θ, g

ɟ

No targets

ʒ

ʣ

In Table 5.5, the consonants for the two languages are grouped according to the level of acquisition. The level is taken as the in-context adultlike productions in proportion to the consonant targets, that is, as the weighted average (W). Similar to the gradient classification for the acquisition of segments at any certain point in time that was proposed by Ingram (1981), ‘used’, ‘frequent’, ‘infrequent’ and ‘not-used’, the following terms are used here: ‘completely acquired’ referring to those segments produced adult-like in context at 90% and over; ‘very frequent’ for adult-like production between 75% and 89%; ‘frequent’ for those between 50% and 74%; ‘not acquired’ for those segments between 1% and 49%; and ‘notused’ for 0% production rate. Previous approaches in phonology have set the 70% (e.g. Ingram et al., 1980) and 75% correct use of a structure in obligatory environments (e.g. Diedrich & Bangert, 1980; PAL, 1995) as criteria of adequate performance. The criterion of 90% correct use of consonants has also been utilized in developmental phonology (e.g. PAL, 1995; Smit et al., 1990). To complement this information, the number of targets/tokens for each consonant and the number of word types containing the consonant are also shown in Table 5.6, where each word type occupies a row, while the number of targets, adult-like tokens and substitution tokens occupy separate cells. In this way, both a weighted (W) average (i.e. adult-like tokens divided by their corresponding total tokens) and an arithmetic (A) average (the sum of the weighted means divided by the total number of types) may be obtained, with the arithmetic average being a more sensitive measure (Ingram, 1981b; Shriberg et al., 1997). It is observed that when the consonant’s sample size is relatively small ( >labial > >coronal, where dorsals are distinctly marked. Substitutions

The child’s substitutions largely abide by norms and universal phonological processes, e.g. fronting, stopping, vocalization, assimilation (e.g. Babatsouli, 2019a; Bernhardt & Stemberger, 1998; Ingram, 1989; McLeod & Bleile, 2003; PAL, 1995). Examples of exceptional productions indicating individual variation are /ð/ → [l] in both languages (e.g. there, ðio ‘two’) and /f/ → [s] (e.g. breakfast [bεstats], εftɐ [εstɐ] ‘seven’), contrasting monolinguals’ /ð/ → [d], /f/ → [p], respectively, although these substitutions are present to a lesser degree in monolingual studies as well (e.g. Ingram et al., 1980). The choice of the child’s substitutions provides evidence of system separation; some phonological interference is also supported at both the surface and the underlying levels, as discussed in detail in Babatsouli (2019d). Phonotactic differences on performance

Evidence of language differentiation in terms of phonotactics in the child’s performance further reflects bilingual status along the acquisition path. Because the data here are based on running speech rather than on isolated words, reference will be made to consonant position within the syllable rather than within the word.

Enhanced Phonology in a Child’s Weaker Language in Bilingualism

135

WF coda singletons in English that do not exist in Greek show the following level of acquisition: p (75%), m (79%), v (50%), t (88%), d (43%), z (59%) (e.g. cup, come, give, bit, bed, boys). The child also produces English WF ʧ (50%) and ʤ (57%) (e.g. catch, page), even though these do not exist in Greek. The child’s performance is similar to that of English monolingual children at age 3;0 (Smit et al., 1990), showing that phonotactic differences between the languages do not hinder bilingual performance even in the specific circumstances of this study. Word-fi nal n, s perform better in English (88%, 91%), because these sounds in Greek (54%, 89%) are exclusively found in function words (stin, stis ‘to the’) that are prosodically weaker (e.g. Bernhardt & Stemberger, 1998) and carry more grammatical load than English content words. Conclusion

This study has shown that only the initial unary and binary oppositions of the Minimal Consonantal System (Ervin & Miller, 1963; Jakobson, [1941] 1968) have been completely acquired in both languages and that, despite the slight advance in Greek, the child’s phonological system in both languages is at the beginning of the third intermediate stage of phonological acquisition. Furthermore, the child’s patterns of consonantal acquisition in both languages emphasize the child’s adherence to the universal Front < Back (Jakobson, [1941] 1968) contrast and establish it quantitatively. With the exception of her velars and palatals idiosyncratically lagging behind, the child’s level at 2;7 is, by and large, in accordance with monolingual norms, and slightly better than norms for some of the sounds in English. Finally, universal phonological processes such as harmonies, coda devoicing and deletion evidenced in her speech are age appropriate and occurring as in monolingual norms. These provide evidence that her developmental course and milestones at 2;7 are similar to those in respective monolingual development, irrespective of the disparity in the quality and quantity of input she has received in her two languages. Moreover, the study has utilized a quantitative approach at many levels in order to compare speech performance crosslinguistically in terms of vocabulary, whole-word accuracy, utterance length, phonological mean length of utterance, phonological word proximity and cumulative consonantal performance across singleton and cluster contexts, as well as cumulative consonant performance based on word complexity. The approach has enabled comparison of the bilingual child’s speech in L2 English and L1 Greek, given that the languages differ grammatically and phonologically, the former being stress-timed and the latter syllabletimed. It is found that the developing phonologies in the child’s bilingualism are equally strong despite the compromised input in English. Exogenous, non-native exposure in one language (English) has not hindered the acquisition of core developmental milestones and patterns in

136

An Anthology of Bilingual Child Phonology

either of the languages. On the contrary, Greek has embraced and facilitated the development of English, leading to earlier consonantal acquisition than the norm, even in phonotactically different contexts. The languages are shown to share a common underlying system in acquisition level and substitution patterns that, while child-specific in some respects, largely adheres to universal expectations. Separation of systems is supported by production of language-specific phonological patterns. This child’s portrait of developing phonologies enhances knowledge on bilingual speech acquisition and may guide future evaluations of crosslinguistic performance in child bilingual and adult second language speech. References Antoniou, M., Best, C.T., Tyler, M.D. and Kroos, C. (2010) Language context elicits nativelike stop voicing in early bilinguals’ productions in both L1 and L2. Journal of Phonetics 38, 640–653. Babatsouli, E. (2017) Bilingual development of theta in a child. Poznan Studies in Contemporary Linguistics 53 (2), 157–195. doi:10.1515/psicl-2017-0007 Babatsouli, E. (2019a) A phonological assessment test for child Greek. Clinical Linguistics and Phonetics 33 (7), 601–627. doi:10.1080/02699206.2019.1569164 Babatsouli, E. (2019b) Lexicon. In J.S. Damico and M.J. Ball (eds) The SAGE Encyclopedia of Human Communication Sciences and Disorders. Thousand Oaks, CA: SAGE. Babatsouli, E. (2019c) Multilingualism. In J.S. Damico and M.J. Ball (eds) The SAGE Encyclopedia of Human Communication Sciences and Disorders. Thousand Oaks, CA: SAGE. Babatsouli, E. (2019d) Melodies of child Greek-English phonological interference. Estudos da Língua(gem) 17 (2), 209–236. Babatsouli, E. (forthcoming) Diversity considerations in speech and language disorders: a focus on training. In J.S. Damico, N. Müller and M.J. Ball (eds) The Handbook of Language and Speech Disorders (2nd edn). Malden, MA: Wiley-Blackwell. Babatsouli, E. and Ingram, D. (2018) Prologue. In E. Babatsouli and D. Ingram (eds) Phonology in Protolanguage and Interlanguage (pp. 1–23). Sheffield: Equinox. Babatsouli, E. and Nicoladis, E. (2019) The acquisition of English possessives by a bilingual child: Do input and usage frequency matter? Journal of Child Language 46 (1), 170–183. doi:10.1017/S0305000918000429 Babatsouli, E. and Sotiropoulos, D. (2018) A measure for cluster proximity (MCP) in child speech. Clinical Linguistics and Phonetics 32 (12), 1071–1089. doi:10.1080/0269920 6.2018.1510982 Babatsouli, E., Ingram, D. and Müller, N. (eds) (2017) Crosslinguistic Encounters in Language Acquisition: Typical and Atypical Development. Bristol: Multilingual Matters. Ball, M.J., Müller, N. and Munro, S. (2001) The acquisition of the Rhotic Consonants by Welsh-English bilingual children. International Journal of Bilingualism 5 (1), 71–86. https://doi.org/10.1177/13670069010050010401 Bernardini, P. and Schlyter, S. (2004) Growing syntactic structure and code-mixing in the weaker language: The Ivy Hypothesis. Bilingualism: Language and Cognition 7 (1), 49–69. Bernhardt, B. and Stemberger, J. (1998) Handbook of Phonological Development: From a Nonlinear Constraints-based Perspective. San Diego, CA: Academic Press. Boersma, P. and Weenink, D. (2018) Praat: Doing Phonetics by Computer [Computer program]. Version 6.0.43. See http://www.praat.org/ (accessed 8 September 2018).

Enhanced Phonology in a Child’s Weaker Language in Bilingualism

137

Brown, R. (1973) A First Language: The Early Stages. Cambridge, MA: Harvard University Press. Brulard, I. and Carr, P. (2003) French-English bilingual acquisition of phonology: One production system or two? International Journal of Bilingualism 7, 177–202. Bunta, F. and Douglas, M. (2013) The effects of dual language support on the English skills of bilingual children with cochlear implants and hearing aids as compared to monolingual peers. Language, Speech, and Hearing Services in Schools 44, 281–290. Bunta, F., Davidovich, I. and Ingram, D. (2006) The relationship between the phonological complexity of a bilingual child’s words and those of the target languages. International Journal of Bilingualism 10 (1), 71–88. Bunta, F., Fabiano-Smith, L., Goldstein, B.A. and Ingram, D. (2009) Phonological wholeword measures in three-year-old bilingual children and their age-matched monolingual peers. Clinical Linguistics and Phonetics 23, 156–175. Burrows, L. and Goldstein, B.A. (2010) Whole word measures in bilingual children with speech sound disorders. Clinical Linguistics and Phonetics 24, 357–368. Diedrich, W. and Bangert, J. (1980) Articulation Learning. Houston, TX: College Hill Press. Dinnsen, D.A. (1992) Variation in developing and fully developed phonologies. In C.A. Ferguson, L. Menn and C. Stoel-Gammon (eds) Phonological Development: Models, Research, Implications (pp. 191–210). Timonium, MD: York Press. Ervin, S.M. and Miller, W.R. (1963) Language development. In H.W. Stevenson (ed.) Child Psychology, NSSEY (pp. 108–143). Chicago: Chicago University Press. Eviatar, Z., Taha, H., Cohen, V. and Schwartz, M. (2018) Word learning by young sequential bilinguals: Fast mapping in Arabic and Hebrew. Applied Psycholinguistics 39 (3), 649–674. doi:10.1017/S0142716417000613 Fabiano-Smith, L. and Barlow, J. (2010) Interaction in bilingual phonological acquisition: Evidence from phonetic inventories. International Journal of Bilingual Education and Bilingualism 13 (1), 81–97. Ferguson, C.A. (1978) Fricatives in child language acquisition. In V. Honsa and M.J. Hardman-de Bautista (eds) Papers on Linguistics and Child Language (pp. 93–115). The Hague: Mouton. Flege, J.E. and Davidian, R. (1984) Transfer and developmental processes in adult foreign language speech production. Applied Linguistics 5, 323–347. Gass, S.M. and Selinker, L. (eds) (1983) Language Transfer in Language Learning. Cambridge, MA: Newbury House. Gollan, T.H., Montoya, R.I., Cera, C. and Sandoval, T.C. (2008) More use almost always means a smaller frequency effect: Aging, bilingualism, and the weaker links hypothesis. Journal of Memory and Language 58 (3), 787–814. Grosjean, F. and Li, P. (2013) The Psycholinguistics of Bilingualism. Malden, MA/ Oxford: Wiley-Blackwell. Hasson, N., Camilleri, B., Jones, C., Smith, J. and Dodd, B. (2013) Discriminating disorder from diff erence using dynamic assessment with bilingual children. Child Language Teaching and Therapy 29, 57. doi:10.1177/0265659012459526 Ingram, D. (1981a) The emerging phonological system of an Italian-English bilingual child. Journal of Italian Linguistics 2, 95–113. Ingram, D. (1981b) Procedures for the Phonological Analysis of Children’s Language. Baltimore, MD: University Park Press. Ingram, D. (1989) First Language Acquisition: Method, Description and Explanation. Cambridge: Cambridge University Press. Ingram, D. (2002) The measurement of whole-word production. Journal of Child Language 29, 713–733. Ingram, D. (2012) Prologue: Cross-linguistic and multilingual aspects of speech sound disorders in children. In S. McLeod and B. Goldstein (eds) Multilingual Aspects of Speech Sound Disorders in Children (pp. 3–12). Bristol: Multilingual Matters.

138

An Anthology of Bilingual Child Phonology

Ingram, D. and Dubasik, V.L. (2011) Multidimensional assessment of phonological similarity within and between children. Clinical Linguistics and Phonetics 25, 962–967. Ingram, D. and Ingram, K. (2001) A whole-word approach to phonological analysis and intervention. Language, Speech and Hearing Services in Schools 32, 271–283. Ingram, D., Christensen, L., Veach, S. and Webster, B. (1980) The acquisition of wordinitial fricatives and aff ricates in English by children between 2 and 6 years. In G.H. Yeni-Komshian, J.F. Kavanaugh and C.A. Ferguson (eds) Child Phonology, Vol. I: Production (pp. 169–192). New York: Academic Press. Jakobson, R. ([1941] 1968) Child Language, Phonological Universals and Aphasia (trans. A. Keiler). The Hague: Mouton. Jasinska, K.K. and Petitto, L.A. (2013) How age of bilingual exposure can change the neural systems for language in the developing brain: A functional nearinfrared spectroscopy investigation of syntactic processing in monolingual and bilingual children. Developmental Cognitive Neuroscience 6, 87–101. Kehoe, M. (2018) Crosslinguistic interaction in early bilingual phonology: A critical review. In E. Babatsouli and D. Ingram (eds) Phonology in Protolanguage and Interlanguage (pp. 49–75). Sheffield: Equinox. Kehoe, M., Patrucco-Nanchen, T., Friend, M. and Zesiger, P. (2018) The relation between phonological and lexical development in French-speaking children. Clinical Linguistics and Phonetics 32 (12), 1103–1125. doi:10.1080/02699206.2018.1510984 Kopečková, R., Marecka, M., Wrembel, M. and Gut, U. (2016) Interactions between three phonological subsystems of young multilinguals: The influence of language status. International Journal of Multilingualism 13 (4), 426–443. Law, N.C.W. and So, L.K.H. (2010) The relationship of phonological development and language dominance in bilingual Cantonese-Putonghua children. International Journal of Bilingualism 10 (4), 405–427. Leopold, W.F. (1949) Speech Development of a Bilingual Child: A Linguist’s Record. Evanston, IL: Northwestern University Press. Lleó, C. (2018a) The (un)predictable linguistic world of the bilingual child: Delay, acceleration, transfer and code-switching. Special lecture given at the International Child Phonology Conference 2018, 18–20 June, Chania, Greece. Lleó, C. (2018b) German-Spanish bilinguals’ phonological grammars: Permeable or resilient? In E. Babatsouli and D. Ingram (eds) Phonology in Protolanguage and Interlanguage (pp. 76–108). Sheffield: Equinox. Lleó, C. and Kehoe, M. (2002) On the interaction of phonological systems in child bilingual acquisition. International Journal of Bilingualism 6, 233–237. Lleó, C., Kuchenbrandt, I., Kehoe, M. and Trujillo, C. (2003) Syllable fi nal consonants in Spanish and German monolingual and bilingual acquisition. In N. Müller (ed.) (In) vulnerable Domains in Multilingualism (pp. 191–220). Amsterdam: John Benjamins. MacWhinney, B. (2000) The CHILDES Project: Tools for Analyzing Talk. Mahwah, NJ: Lawrence Erlbaum. Major, R.C. (2008) Transfer in second language phonology. In J.G.H. Edwards and M.L. Zampini (eds) Phonology and Second Language Acquisition (pp. 63–94). Amsterdam/ Philadelphia, PA: John Benjamins. Maragkaki, I. and Hessels, M.G.P. (2017) A pilot study of dynamic assessment of vocabulary in German for bilingual preschoolers in Switzerland. Journal of Studies in Education 7 (1), 32–49. Marinis, T. (2003) The Acquisition of the DP in Modern Greek. Amsterdam/Philadelphia, PA: John Benjamins. McLeod, S. (2007) The International Guide to Speech Acquisition. Clifton Park, NY: Thomson Delmar Learning. McLeod, S. and Bleile, K. (2003) Neurological and developmental foundations of speech acquisition. Invited seminar presentation at the American Speech and Hearing Association Convention, November, Chicago, IL.

Enhanced Phonology in a Child’s Weaker Language in Bilingualism

139

McLeod, S. and Goldstein, B. (eds) (2012) Multilingual Aspects of Speech Sound Disorders in Children. Bristol: Multilingual Matters. Meisel, J.M. (2007) The weaker language in early child bilingualism: Acquiring a fi rst language as a second language? Applied Psycholinguistics 28, 495–514. Mennen, I. and Okalidou, A. (2007) Acquisition of Greek phonology: An overview. In S. McLeod (ed.) The International Guide to Speech Acquisition (pp. 398–407). Clifton Park, NY: Thomson Delmar Learning. Nespor, M., Peña, M. and Mehler, J. (2003) On the different roles of vowels and consonants in speech processing and language acquisition. Lingue e Linguaggio 2, 203–229. Nice, M. (1925) Length of sentence as a criterion of child progress in speech. Journal of Educational Psychology 16, 370–379. Olmsted, D. (1971) Out of the Mouth of Babes. The Hague: Mouton. PAL (Panhellenic Association of Logopedics) (1995) Assessment of Phonetic and Phonological Development. Athens: PAL. [In Greek] Papaeliou, C.F. and Rescorla, L.A. (2011) Vocabulary development in Greek children: Across-linguistic comparison using the Language Development Survey. Journal of Child Language 38 (4), 861–887. doi:10.1017/S030500091000053X Paradis, J. (2000) Beyond ‘One system or two’: Degrees of separation between languages of French-English bilingual children. In S. Döpke (ed.) Cross-linguistic Structures in Simultaneous Bilingualism (pp. 175–200). Amsterdam/Philadelphia, PA: John Benjamins. Prather, E.M., Hedrick, D.L. and Kern, C.A. (1975) Articulation development in children aged two to four years. Journal of Speech and Hearing Disorders 40, 179–191. Prince, A. and Smolensky, P. (2004) Optimality Theory: Constraint Interaction in Generative Grammar. Oxford: Basil Blackwell. Schmitt, L.S., Howard, B.H. and Schmitt, J.F. (1983) Conversational speech sampling in assessment of articulation proficiency. Language, Speech, and Hearing Services in Schools 14, 210–214. Selinker, L. (1972) Interlanguage. International Review of Applied Linguistics 10, 209–231. Shriberg, L.D., Austin, D., Lewis, B.A., McSweeny, J.L. and Wilson, D.L. (1997) The percentage of consonants correct (PCC) metric: Extensions and reliability data. Journal of Speech, Language, and Hearing Research 40, 708–722. Smit, A.B. (1993) Phonologic error distributions in the Iowa-Nebraska articulation norms project: Consonant singletons. Journal of Speech and Hearing Research 36, 533–547. Smit, A.B., Hand, L., Freilinger, J.J., Bernthal, J.E. and Bird, A. (1990) The Iowa articulation norms project and its Nebraska replication. Journal of Speech and Hearing Disorders 55, 779–798. Smith, N.V. (1973) The Acquisition of Phonology: A Case Study. Cambridge Studies in Linguistics 25. Cambridge: Cambridge University Press. Stoel-Gammon, C. (2011) Relationships between lexical and phonological development in young children. Journal of Child Language 38, 1–34. Treffers-Daller, J. (2011) Operationalizing and measuring language dominance. International Journal of Bilingualism 15 (2), 147–163. Wei, L. (ed.) (2010) Bilingualism and Multilingualism: Critical Concepts in Linguistics, Vols 1–4. London: Routledge. Weinreich, U. (1953) Languages in Contact: Findings and Problems. The Hague: Mouton. Werker, J.F., Byers-Heinlein, K. and Fennell, C.T. (2009) Bilingual beginnings to learning words. Philosophical Transactions of the Royal Society of London B 364, 3649–3663. Wrembel, M. and Cabrelli Amaro, J. (2018) Advances in the Investigation of L3 Phonological Acquisition. Routledge Special Issues as Books (SPIB). London: Routledge.

6 Sensitivity to Morphophonological Cues in Monolingual and Bilingual Children: Evidence from a Nonword Task Luca Cilibrasi and Ianthi Tsimpli

Introduction Age of onset effects on the acquisition of inflectional morphophonology

Effects of age of onset in the second language have received an increasing amount of attention in recent years in bilingualism research (Montrul, 2008). While it is becoming clear that in several linguistic domains bilingual children do not perform as monolinguals, it has also become clear that considerable variation is related to the age of onset in the second language, and the subsequent exposure to that language (Flege, 2009). Kovelman et al. (2008), for example, have shown that children with an age of onset of 3 years or below are able to attain native-like performance in phonological and in reading tasks in primary school, while children with a later age of onset may lag behind way beyond the first stages of schooling. The present study aims to investigate the role of age of onset in the acquisition of inflectional morphophonology. More specifically, this study examines the sensitivity to morphophonological cues in a group of CzechEnglish primary school bilingual children (mean age 10;09) with varying ages of onset of exposure to English. The study focuses thus on morphophonological perception, a domain that has received less attention than production. Sensitivity to bound morphemes is examined using a nonword task, in order to isolate morphophonological effects from 140

Sensitivity to Morphophonological Cues in Monolingual and Bilingual Children

141

semantic and lexical effects. Since it is well attested that bilingual children have smaller lexica than monolingual children do (Bialystok et al., 2010), the use of nonwords is an attempt to control for item-familiarity confounds and focus on the actual processing of morphophonology. In other words, this study aims at offering some answers to the following questions: Are bilingual children sensitive to morphophonological cues? Is their performance comparable to that of monolingual children? Does age of onset of exposure in the second language modulate the result? The Introduction consists of three sections: in the first section we present crosslinguistic evidence that sensitivity to morphophonological cues can be assessed using nonword tasks. In the second section we outline the theoretical approach that we used to design our study (i.e. Tsimpli, 2014). In the third section we explain how the study of inflectional morphology in bilinguals relates to the theoretical approach described in the second section. Sensitivity to inflectional morphophonology in the absence of meaning

Several studies across different languages have shown that adult speakers are sensitive to the presence of bound morphology in the absence of meaning as well. For instance, a classic study by Caramazza et  al. (1988) on Italian showed that nonwords that are inflected following regular inflectional rules take longer to be recognized than nonwords that cannot be construed as carrying inflectional morphology. In their study, the authors present a series of tasks in which participants were asked to decide whether a given item was a real word or not. The nonwords used were created using two variables: they could consist of real or invented stems, and they could end with bound morphemes or with endings that could not be bound morphemes. The results showed that nonwords ending with existing bound morphemes were slower to be recognized, and the effect was larger when the stem was also an existing stem (although applied to the wrong suffi x). In a subsequent work on German, Clahsen et al. (1997) showed that participants can apply decomposition and inflection on nonwords and that they build expectations on inflections of nonwords based on the regular paradigms. This experiment consisted of two phases: a familiarization phase and a testing phase. In the familiarization phase, participants were presented with a list of regularly inflected nonwords. Then, in the testing phase, participants were asked to perform a lexical decision task, and the items in this task were the same nonwords as those used in the familiarization phase. Crucially, the nonwords in the testing phase could be inflected regularly (but with a different tense from that in the familiarization phase), or irregularly. Participants were quicker when the nonwords were inflected regularly, showing that during the

142

An Anthology of Bilingual Child Phonology

familiarization phase they had built expectations on the kind of inflections that could appear in these nonwords. A different study performed by Post et al. (2008) in English presents similar findings, but it additionally explores the role of phonology in cueing the detection of bound morphology. In this experiment, participants were presented with a same/different minimal pairs discrimination task. The items in the pair, when different, were separated by the distinction of one single phoneme, absent in one item and present in the other. The task contained several different types of pairs, including pairs with real words. Of interest to our discussion, one item in each minimal pair of nonwords contained an extra phoneme that could either be or not be a regular inflection of the nonword. For example, one possible distinction was between gubbed/ gub, where the first nonword is potentially the inflected form of the second, and another possible distinction was between steet/stee, where the first nonword is not a possible inflected form of the second. Crucially, the ending /t/ can be a bound morpheme in certain phonological contexts, i.e. following a devoiced consonant, but it cannot be a bound morpheme when following a vowel. The study showed that participants are slower when the phonological context cues the presence of morphology (Condition 1, i.e. gubbed/gub). The task used in this study is a development of the same line of research initially presented in Cilibrasi (2016). The task maintained the phonological attention of Post et al. (2008), but it differed from it in two ways. First, in Cilibrasi’s experiment participants were never presented with bare stems so that participants would not be aware of the manipulation performed and would not be able to understand that the word ending was an ending to a stem. Secondly, a third condition was added in which nonwords ended with voiced consonants that were not, however, morphological. This was done in order to differentiate the morphological effects of the consonants /d/ and /z/ from purely voicing effects. Also in Cilibrasi’s (2016) experiment, carried out with native speakers of English, the prediction was confirmed in that nonwords with endings that were potentially bound morphemes were recognized more slowly than nonwords that did not contain potential bound morphology. Early and late phenomena

In her epistemological paper, Tsimpli (2014) makes a distinction between early and late phenomena in language acquisition. In her words: ‘early phenomena are core, parametric and narrowly syntactic, in contrast to late and very late phenomena, which involve syntax-external or even language-external resources too’ (Tsimpli, 2014: 283). This distinction roughly reflects age of acquisition in monolingual children and refers to the fact that certain aspects of language are acquired quickly and proficiently, while others require a large amount of time and remain difficult for a longer period of time.

Sensitivity to Morphophonological Cues in Monolingual and Bilingual Children

143

In the same article, Tsimpli reviews fi ndings that show that amount of input is a good predictor of performance for late phenomena, while for early phenomena minimal input may be sufficient to master the relevant linguistic property. A good example of an early phenomenon is headcomplement directionality. In monolingual speakers, directionality is acquired very early: Poeppel and Wexler (1993), for instance, showed that monolingual German 25-month-old children already produce utterances that respect the head-complement distribution. This fi nding may be explained in terms of parameter setting (Chomsky, 1993) in that core parametric properties may be set early in development, while non-parametric properties are relatively delayed (Wexler, 1998). How do bilingual children deal with early skills of this kind? Studies in bilingual acquisition show that this core type of linguistic properties is set quite early and proficiently by bilinguals as well. Möhring and Meisel (2003) investigated the directionality parameter in GermanFrench bilinguals. German and French are two languages that crucially differ in the canonical order of verb and object, with German being OV and French VO. The study of Möhring and Meisel (2003) showed quite clearly that children were able to produce correct word orders in both languages very early, approximately from the age of 2 years, an age comparable to that of monolinguals attaining the same skill (Poeppel & Wexler, 1993). As Tsimpli (2014) notes, the acquisition of head-complement ordering must be possible with a small amount of input; on the assumption that bilingual children receive less input in each language compared to that of monolingual children, the early mastery of headdirectionality in monolinguals and bilinguals alike seems to support the suggestion that (very) early phenomena do not depend on large amounts of input. In contrast, the amount of input is crucial in the acquisition of several other late properties. For instance, the acquisition of grammatical gender requires between 4 and 6 years in monolinguals depending on the language (Blom et al., 2008; Tsimpli, 2003) and can thus be defi ned as a late phenomenon. How do bilinguals deal with a late phenomenon such as grammatical gender? To answer this question, Unsworth et  al. (2014) investigated the acquisition of grammatical gender in English-Greek and English-Dutch bilinguals using a series of elicitation tasks. First of all, the study confi rmed the relatively late nature of the phenomenon in the two languages, which was additionally modulated by the crosslinguistic differences between genders in Greek which is earlier acquired by monolingual children than gender in Dutch. In addition, the study showed that amount of input has a larger power than age of onset in predicting the acquisition of this skill. Specifically, for both languages, an early age of onset was not sufficient to predict accuracy in the production of grammatical gender, while input was a good predictor in both cases, independently from age of onset.

144

An Anthology of Bilingual Child Phonology

Inflectional morphophonology: Early or late?

Where does inflectional morphology stand? Generally, inflectional morphology appears as an early phenomenon: a collection of studies on the acquisition of mini-paradigms (Bittner et al., 2011) shows that children, across languages, master a number of inflectional morphemes in the early stages of acquisition, approximately from the age of 2 years. The result is reported, with some variation, in the following languages: Spanish, Turkish, German, Greek, Russian, Dutch, Croatian, Austrian, Finnish, Italian and Lithuanian. As the authors suggest, English children behave slightly differently, and only the inflectional morpheme -ing is mastered early. The other two morphemes, -s and -ed, instead, require more time. Johnson et al. (2005), for example, showed that 4- and 5-yearold English children perform at chance in picture-matching tasks with sentences of this type: (1) The cats sleep on the bed (2) The cat sleeps on the bed Speakers of Italian, Greek or Russian do not present with these difficulties. This difference in performance between English and the other languages is likely to be related to inflectional richness (Guasti, 2017): the English language includes a total of three verb-bound morphemes, a number significantly smaller than that of the other languages tested. While it may seem surprising that children fi nd it more difficult to learn a poorly inflected system than a richly inflected one, there may be reasonable explanations for the pattern: fi rst of all, in morphologically rich languages inflectional morphemes are usually realized phonologically and prosodically more prominently than in English. In Italian, for example, inflections require a full-vowel and in certain cases entire syllables to express (i.e. Mangi-o ‘eat.1s’, Mangi-ano ‘eat.3p’). In English, bound morphemes appear as either null-morphemes or individual consonants, and only in certain phonological contexts is a vowel added (i.e. ‘I eat-0’, ‘she eat-s’, ‘she fi nish-es’). A second reason is productivity: while in rich languages there is an inflectional morpheme for each person, and inflections vary according to tense, in English the appropriate choice is a nullmorpheme for most persons in the present tense, and the same morpheme (-ed) for all persons inflected in the past. English children are thus likely to use the null as the default option, while children acquiring Italian, Greek or Russian would not consider this an option, as it is rare or nonexistent in the morphological paradigms. In short, while it is generally the case that inflectional morphology appears as an early phenomenon, its age of acquisition is influenced by the morphological richness of the language learned: in poorly inflected languages, paradigms are acquired later than in richly inflected languages. For this reason, inflectional morphology appears as a late phenomenon in

Sensitivity to Morphophonological Cues in Monolingual and Bilingual Children

145

children acquiring English (in comparison to most of the other European languages). It should be stressed that inflectional morphology is not a monolithic concept, and the reasoning we just outlined does not necessarily apply to all morphemes. In English, for example, the bound morpheme -ing is acquired earlier than the others, possibly thanks to its phonological prominence. While in this chapter we will keep referring to English inflectional morphology in general terms, our claims may not apply equally to all morphemes. English inflectional morphophonology in bilingual children

Studies on the L2 acquisition of English inflectional morphophonology show that bilingual children experience an overall delay when compared to monolinguals. However, input appears to play an important role and bilingual children that receive large amounts of input in the tested language are closer to native performance than those with less exposure. Pearson (2002), for example, tested Spanish-English bilinguals with a story retelling task, and reported two fi ndings: fi rst, bilingual children had lower performance than their monolingual peers overall, with a smaller number of inflections produced; and secondly, when input was taken into account, morphological performance was modulated accordingly, in that the difference between monolinguals and bilinguals with high exposure was considerably smaller than the difference between monolinguals and bilinguals with little exposure. In a different study on Spanish-English bilingual children, GutiérrezClellen et al. (2006) showed that the differences in the acquisition of morphology between monolinguals and bilinguals are present only if one asks a bilingual to perform in their non-dominant language. When comparing monolingual to bilingual performance in the bilingual’s dominant language, group differences tend to disappear. A similar result is reported in Paradis et al.’s (2007) study with French-English bilinguals. Morphological performance is poorer in bilinguals only when they are tested in their non-dominant language. Interestingly, this fi nding was reported with 4-year-old preschoolers, suggesting that the pattern is observable very early during language development and that it is, to some extent, independent from schooling and education. Considering the fi ndings reported so far in monolingual and bilingual acquisition, there are good reasons to suspect that English inflectional morphology belongs to the class of late phenomena: current research suggests that English inflectional morphology is acquired rather late by bilingual children and that its acquisition depends heavily on input. Our study aims at testing this claim with a nonword task, and it aims at extending the fi nding to a new pair of languages, Czech and English. Paradis (2010) noticed that task type can modulate performance as well. The harder the task is, the more the importance of input quantity. For this reason, our

146

An Anthology of Bilingual Child Phonology

experiment focusing on nonwords and substantially reducing the effect of semantics and lexical knowledge may be an interesting ground for investigating sensitivity to morphophonology at a more implicit level. The research questions we address are: How sensitive are bilingual children to inflectional morphophonology in the absence of verb meaning? How does input modulate recognition of inflected nonwords? Materials

The task was developed by Cilibrasi (2016) and it included a list of 120 monosyllabic nonwords. To create the nonwords we selected the fi rst four starting consonants to be /v/, /n/, /θ/, /dʒ/, on the grounds that these consonants have a relatively low frequency in word initial position. As such, the nonwords created were less likely to resemble existing words. The vowels used in the nonwords were five: /ɪ/, /aɪ/, /æ/, /ɔ/, /ʌ/. The nonwords were created using rules that allowed the combination of onsets, nuclei and codas as follows. Each onset was combined with each nucleus. This enabled the creation of 20 base forms. After the creation of the 20 base forms, we proceeded with the combination of base forms with codas. Each nonword contained a /l/ following the nucleus. The choice of using /l/ was motivated by the morphophonological rules that apply to it: when /l/ is followed by a voiced /d/ and /z/, the nonword can be interpreted as inflected; when /l/ is followed by /t/ or /s/ the nonword cannot be construed as inflected but the sequence is still phonotactically allowed. In addition, /l/ can be followed by /b/ and /m/ as well, which allowed for the creation of the voicing control condition: having a third condition in which /l/ is followed by voiced consonants that are not morphological was necessary in order to separate voicing and morphological effects. If we found that items such as /vʌld/ are slower than their unvoiced counterpart /vʌlt/, we could not tease apart whether the difference was due to voicing or to morphology. The fi rst item may be processed more slowly because the fi nal consonant is voiced or because the fi nal consonant is carrying morphological information (or a combination of these two aspects). If /vʌld/ is additionally slower than the voicing control condition /vʌlb/, then a voicing explanation alone is not sufficient and morphology must be put into the picture. This task is the fi rst to contain an additional condition of this kind. The six different codas were then added to each base form, thus generating 120 nonwords. The six codas used were: /ld/, /lz/, /lt/, /ls/, /lb/, /lm/. The full list of stimuli can be found in the Appendix. Phonotactic probabilities were calculated for each nonword using Vitevitch and Luce (2004). These include positional segment frequency and biphone segment frequency. The values were never zero, confi rming that the nonwords used were phonotactically allowed. Rhyme probabilities were additionally calculated using Moreland (2011), and in this case too values never reached zero.

Sensitivity to Morphophonological Cues in Monolingual and Bilingual Children

147

The nonwords were created according to the phonotactic probabilities of English, without considering the phonotactic constraints of Czech. The syllabic structure used for these nonwords is also allowed in Czech, where words can be made of the sequence CVCC. However, Czech does not allow voiced consonants in word fi nal position, and this makes all of the nonwords in the morphological and in the control conditions not allowed in Czech. Method

The current study was approved by both the University of Cambridge and Charles University Ethics Committees. Our research assistant (Alzbeta Brabcova) contacted several schools in the Prague metropolitan area and we were eventually able to recruit from three of those. Several schools were contacted in the Cambridge area to recruit controls, and two of the schools agreed to participate. The three schools tested in Prague were in either middle- or upper-middle-class areas of the capital. Two follow the British educational system; one follows the American one. One of the schools in Cambridge was in a middle-class area and the other in an upper-middle-class area of the town. The vast majority of  the children in this study have parents with a bachelor’s degree or above. Procedure

Testing took place in a quiet room in the school. After providing the consent form previously signed by their parents, the children were instructed on the nature of each task. Each child was tested on the nonword task, a task on syntactic processing (not discussed in this chapter) and the background tasks (more details on the background tasks will be presented in the next section). In addition, the children were asked the questions from a questionnaire that was used to understand their linguistic background. Parents were also given a questionnaire to complete at home. The questionnaires were adapted from Tsimpli et al. (2014), who used it in a different study on German-Greek bilingual children. White and black stickers were put on two buttons of the keyboard. The instructions for the experimental task were the following: ‘Scientists have been recording some words produced by aliens in space, and we need your help to analyze them. We need your help because children are better than adults at recognizing words. You will hear these words presented in pairs. In each pair there will either be two different words, or the same word repeated twice. When you think that the two words you hear are different, press white as fast as you can; when you think that you are actually listening to the same word repeated twice, press black as fast as you can.’ Instructions were in English for all children.

148

An Anthology of Bilingual Child Phonology

Table 6.1 Summary of conditions Condition

Morphological pairs

Non-morphs

Control

Example

vʌld – vʌlz

vʌlt – vʌls

vʌlb – vʌlm

Point of articulation

plosive/fricative

plosive/fricative

plosive/nasal

Voicing

voicing coherent

voicing incoherent

voicing coherent

Morphological status

+ bound morpheme

− bound morpheme

− bound morpheme

The children were then presented with a practice trial. If the practice trial was successful, the children would then proceed with the actual testing phase. The task contained 120 randomized trials: 60 with same minimal pairs and 60 with different minimal pairs. Each set of 60 pairs included three conditions: the morphological condition (minimal pairs with a possible morphological ending); the non-morphological condition (minimal pairs with an ending that could not be morphological in that phonological context); and a control condition (minimal pairs with a voiced ending that could not be morphological). The design is summarized in Table 6.1. Participants Control group

A total of 26 children were recruited in primary schools in the Cambridge area. All children were monolingual and none of them was previously reported as having any language difficulty. The children were additionally assessed with two background tests: the Children’s Test of Nonword Repetition (CNRep) and the Coloured Progressive Matrices (CPM). The CNRep (Gathercole et al., 1994) is a widely used nonword repetition task which assesses verbal working memory by measuring children’s ability on repeating invented words of different lengths. The CPM (Raven, 1998) is a widely used test of fluid intelligence which assesses children’s ability on completing small puzzles, created with logical patterns and geometrical figures. All children performed as expected for their age in the background tests. Descriptive statistics for the control group are presented in Table 6.2. Bilinguals

A total of 47 bilingual children participated in the study, but two were excluded because the questionnaire revealed knowledge of a third Table 6.2 Descriptive statistics control group RT

CNREP

CPM (A + B)

AGE

Mean

1.51

33

20

9;03

SD

0.26

4

3

1;02

Sensitivity to Morphophonological Cues in Monolingual and Bilingual Children

149

language. The children were divided into three groups based on the answers in the parents’ questionnaire: simultaneous bilinguals, early sequential bilinguals and late sequential bilinguals. The division was based on age of onset of the second language (English): when age of onset was 0, children were assigned to the simultaneous group; when age of onset was 3 (nursery), children were assigned to the group of early sequential bilinguals; and when age of onset was 5 or later (primary school or pre-primary), children were assigned to the group of late sequential bilinguals. All children had Czech as L1, and no language difficulty was reported for any of the children. Descriptive statistics for bilinguals are presented in Table 6.3. Predictions

Only monolingual children and bilingual children with high exposure in the tested language (English) will show sensitivity to the phonological cues in morphological structure. As previously reported by Post et  al. (2008) and Cilibrasi (2016) for English monolingual adults, we expect English monolingual children to be slower in recognizing nonwords that carry morphological information in comparison to nonwords that do not carry morphological information and nonwords in the control condition. We expect bilinguals to show a weaker sensitivity to these cues, and we expect age of onset in English to be a predictor of their success. Specifically, while we do not expect children with a late age of onset to be slower in the morphological condition in comparison to the other two conditions, we expect this to be the case for children with an early age of onset in English. Results

Data were analyzed using linear mixed models (Baayen et al., 2008; Bates, 2005). Initially, a full model including condition*type*group and random effects of item and participant was used. The dependent variable was reaction times, and only reaction times from correct responses were used for the analysis (see Cilibrasi et al., 2019, for a discussion of why reaction times are a more reliable measure than accuracy in interpreting this task). The independent variables were group (control/simultaneous/ early-sequential/late-sequential), type (same/different) and condition (morpho/non-morpho/control). It should be stressed that a morphological Table 6.3 Descriptive statistics bilingual group RT

CNREP

CPM (B)

AGE

10

10;09

Mean

1.59

32

SD

0.2

7

1.6

0;10

150

An Anthology of Bilingual Child Phonology

Table 6.4 Main effects and interactions in the full model Mean square

F-value

Group

0.0273

0.1988 ns

Cond

0.0859

0.6264 ns

Type

3.6525

26.6418**

Group:Cond

0.4646

3.3889*

Group:Type

0.5118

3.7330*

Cond:Type

0.3345

2.4395*

Group:Cond:Type

0.8611

6.2807*

Notes: ns = not significant; * = p < 0.05, ** = p < 0.001.

effect is considered to occur only if the morphological condition differs from both the non-morphological and the control conditions. Random slopes were also tested but the best random model was revealed to be M1: (1|part + 1|item). The full model was thus: Rt-correct → condition*type*group + (1|part) + (1|item) The full model revealed the presence of a significant three-way interaction (see Table 6.4). We then decided to analyze each group separately. Control group

We report the results for the full model with the two-way interaction: cond*type + (1|part) + (1|item). The analysis shows that for the control group there is a main effect of type, a main effect of condition and no significant interaction between type and condition (see Table 6.5). Table 6.5 Main effect and interaction in the control group

Cond

Mean Sq

F-value

4.29

26.57**

Type

1.93

11.98*

Cond:Type

0.08

0.55 ns

Notes: ns = not significant; * = p < 0.05, ** = p < 0.001.

We report the summary of the fi xed effects directly in the text, because this exemplifies the direction of the main effects, with no need for post hocs (see Field, 2013). Condition 1 (morphological) is set as the default condition by r, so Table 6.6 shows how both Condition 2 (nonmorphological) and Condition 3 (control) are significantly different from it. In both cases, Condition 1 is slower, as shown by the sign of the estimate and of t. Similarly, same pairs were overall faster than different pairs, as reported in Cilibrasi (2016) for adult monolinguals. The effects are represented in Figure 6.1.

Sensitivity to Morphophonological Cues in Monolingual and Bilingual Children

151

Table 6.6 Fixed effects in the control group Estimate

SE

t

Cond2

− 0.16

0.02

5.53*

Cond3

− 0.06

0.02

− 2.37*

TypeSame

− 0.08

0.02

− 2.8*

Cond2:TypeSame

0.02

0.04

0.61 ns

Cond3:TypeSame

0.04

0.04

1.05 ns

Notes: ns = not significant; * = p < 0.05, ** = p < 0.001.

Figure 6.1 Effects of condition and type in the monolingual group. The morphological condition differs from the phonological control condition and the non-morphological condition. Same minimal pairs are overall faster than different minimal pairs

Bilingual children

In order to keep the groups comparable to each other, we analyzed each sub-group of bilinguals with the same model used for the control group: cond*type + (1|part) + (1|item). The analysis of the simultaneous group shows a main effect of type and a significant interaction between type and condition (Table 6.7).

Table 6.7 Main effects and interactions in the simultaneous bilinguals

Cond

Mean Sq

F-value

0.29

2.36 ns

Type

0.73

5.87*

Con:Type

0.37

2.98*

Notes: ns = not significant; * = p < 0.05, ** = p < 0.001.

152

An Anthology of Bilingual Child Phonology

Post hoc analyses were carried out to understand the nature of the interaction. Since four t-tests were run, we used an adjusted alpha value of 0.05/4, following Bonferroni correction. Post hocs revealed a significant difference between the morphological and the non-morphological conditions in both the same and different minimal pairs: t(15) = 6.52, p < 0.001 for different pairs, and t(15) = 3.81, p = 0.001 for same pairs. The morphological condition differed from the phonological control condition in the same minimal pairs, t(15) = 3.63, p = 0.002, but it did not reach significance in the different minimal pairs, t(15) = −1.48, p = 0.15. The effects are represented in Figure 6.2.

Figure 6.2 Effects of condition and type in the group of simultaneous bilinguals. In the same minimal pairs, the morphological condition differs from the non-morphological and from the phonological control condition; in different pairs, the morphological condition differs from the non-morphological condition, but not from the phonological control condition

In the group of early sequential bilinguals, the main model reveals a main effect of type and an interaction between condition and type (Table 6.8). Post hocs revealed a significant difference between the morphological and the non-morphological conditions in both same and different minimal pairs: t(14) = 5.01, p < 0.001 for different pairs, and t(14) = 5.40, p < 0.001 for same pairs. The morphological condition, instead, did not differ from the phonological control condition in the same pairs, t(14) = 2.55, p = 0.02, and it did differ in the different pairs, but in the opposite direction to that in simultaneous bilinguals, t(14) = −4.74, p < 0.001. The effects are represented in Figure 6.3. The contrast between morphological and control does not reach significance in the same pairs and it does reach significance in the different pairs: the control condition is slower than the morphological condition.

Sensitivity to Morphophonological Cues in Monolingual and Bilingual Children

153

Table 6.8 Main effects and interactions in the early sequential bilinguals

Cond

Mean Sq

F-value

0.06

0.59 ns

Type

2.36

25.42**

Con:Type

1.02

8.97*

Notes: ns = not significant; * = p < 0.05, ** = p < 0.001.

Figure 6.3 Effects of condition and type in the group of early sequential bilinguals. The morphological condition differs from the non-morphological condition in same and different pairs

The same model was used to analyze the group of late sequential bilinguals. This group showed only a main effect of type, and no main effect of condition or interaction between type and condition (see Table 6.9). Fixed effects were used to understand the direction of the main effect: the sign of the estimate shows that same minimal pairs were overall faster than different minimal pairs (as previously reported for monolingual adults in Cilibrasi, 2016). The effects are represented graphically in Figure 6.4. Discussion

This study investigated sensitivity to bound morphemes in monolingual and bilingual children using inflected nonwords. The results show that monolingual children are the only group that is consistently sensitive to these cues. Simultaneous bilinguals show some sensitivity, but only in a subset of the task, while sequential bilinguals do not show sensitivity to the presence of bound morphemes. These results on monolingual children

154

An Anthology of Bilingual Child Phonology

Table 6.9 Main effect and interactions in the late sequential bilinguals

Cond

Mean Sq

F-value

0.5

0.38 ns

Type

3.70

25.42**

Cond:Type

0.02

0.15 ns

Notes: ns = not significant; * = p < 0.05, ** = p < 0.001.

are coherent with previous studies on adults, such as Post et al. (2008) and Cilibrasi (2016). The performance of bilinguals, on the other hand, is shown to be predicted by the age of onset in English: only children with an early onset of exposure to English manifest some sensitivity to morphophonological cues, while a later age of onset corresponds to a lack of sensitivity to these cues. The results we obtained with nonwords are consistent with results obtained with real words. Paradis and colleagues investigated the acquisition of inflectional morphology in French-English bilingual children with different levels of exposure to English, and showed that only children with high exposure performed as monolinguals. Our results confi rm this fi nding, although the study of Paradis et al. (2007) differs from ours in two respects. A fi rst difference is that the L1 of the children tested by Paradis et  al. was different from the L1 of the children in this study. In Paradis et al. (2007), the first language was French, while in this study it is Czech. The other important difference is in the use of real verbs. Our results demonstrate that monolingual and, to some extent, simultaneous

Figure 6.4 Effect of type in the group of late sequential bilinguals. The same minimal pairs are overall faster than different minimal pairs. The effect of condition, instead, does not reach significance

Sensitivity to Morphophonological Cues in Monolingual and Bilingual Children

155

bilingual children do not actually require real verbs (and meaning) in order to show sensitivity to inflectional morphology: they just need something that resembles a verb. As with Paradis et al. (2007), the only group that is replicating the behaviour of monolingual adults is the group of monolingual children. Simultaneous bilinguals are moving in the direction of monolingual performance, although significantly differing from it, and show the monolingual pattern in a subset of the trials (that is, in the same minimal pairs). For the two sequential bilingual groups, however, performance differed from monolinguals. This fi nding can be associated with broad proposals on bilingual acquisition, such as the one advocated in Tsimpli (2014). According to her proposal, the distinction between early and late phenomena in monolingual acquisition raises certain expectations regarding the role of input in bilingual development. Specifically, while early phenomena require minimal input and seem to be less sensitive to L2 age of onset, late phenomena are dependent on quantity of input. Tsimpli’s argument is based on the evidence that certain structures are acquired effortlessly by bilinguals while some are not and require more exposure. Interestingly, despite coming from a rather different theoretical environment, the suggestion in Tsimpli (2014) shows important similarities to the proposition that Paradis et al. (2007) make to explain their fi ndings regarding English inflectional morphology. As Paradis et al. suggest: This research has revealed that the extent to which bilingual-monolingual differences are apparent depends on factors such as variation in the amount of input between the two languages bilinguals receive, and relative complexity of the target structure examined. For example, bilingual children who received relatively more input in language a than language b at home and at school were more likely to approach monolingual levels of performance in language a (Erdos et al., 2005; Gathercole, 2002, 2006). In addition, acquisition of more transparent structures was less sensitive to the reduced input received by bilinguals, since for these structures bilinguals were more likely to approach monolingual levels of performance than for more opaque structures (Gathercole, 2002, 2006; Gathercole & Hoff, 2007). Thus, the research indicates that bilinguals may lag behind monolinguals selectively, but not globally, in acquisition milestones. (Paradis et al., 2007: 1)

These claims are substantially in line with those in Tsimpli (2014) and may be summarized by saying that an explanation strictly driven by input (such as in Tomasello, 2009) is to be used for certain (opaque/late) structures, while for other structures input is a poorer predictor and, thus, linguistically based explanations must be found. Certainly, it remains to be discussed how the terms ‘opaque and shallow’ relate to the terms ‘early and late’ in Tsimpli (2014). The two pairs of terms refer to two qualitatively different aspects of the structures in question: in the fi rst case they refer to the regularity of the structure, while in the second they refer to

156

An Anthology of Bilingual Child Phonology

the timing of acquisition. In general, shallow and regular structures are also those that are acquired earlier by children, and this makes the parallelism between Paradis et  al. (2007) and Tsimpli (2014) particularly reasonable. For example, this type of explanation seems acceptable if we  look again at structures such as head-complement directionality and  grammatical gender. The fi rst is shallow (regular in the input, inversions are not attested) and acquired very early, while the second is opaque (largely unpredictable and arbitrary in most languages) and acquired late. Does sensitivity to inflectional morphology belong to the opaque/late input-driven group of structures? Paradis et al. (2007) suggest that, in English, it does, particularly when it comes to irregular forms. Our data suggest that regular forms too, at least in English, require a large amount of input if tested with complex tasks. Only children that have had a lot of exposure to L2 are able to successfully detect the presence of inflectional morphology in nonwords. In fact, while the exposure simultaneous bilinguals experience is less than that of monolinguals (since the former grew up being exposed to English only part of the time), the amount of exposure for simultaneous bilinguals is considerably more than that for sequential bilinguals, in this sample. This may explain why simultaneous bilinguals partly behaved as monolingual children, while sequential bilinguals behaved very differently. Only the former group had received a sufficient amount of exposure to English to allow for partly native-like detection of bound morphemes. The pattern of performance of sequential bilinguals does not appear to be random, at least in the case of early sequential bilinguals (since the effect of condition was not significant in the late sequential bilinguals, it is difficult to make claims about that group). While the non-morphological items were recognized very quickly, the morphological and the phonological-control items required a significantly larger amount of time. We can try to explain and understand the foundation of the observed pattern in the early sequential bilinguals. As Paradis (2001) suggests, bilingual children do not have encapsulated phonological systems in the two languages; rather, the two systems are separated but in partial interaction with each other. One possible explanation for the performance of early sequential bilinguals in our study is thus that the phonological systems of the two languages were interacting with each other. As discussed when presenting the materials, when considering the phonotactics of both Czech and English, only nonwords in the non-morphological condition are allowed strings of phonemes, since in Czech fi nal devoicing is compulsory. This might be the fundamental reason for the pattern observed in the early sequential bilingual group. Only nonwords that are allowed in both Czech and English are recognized quickly, while the rest are recognized with some delay.

Sensitivity to Morphophonological Cues in Monolingual and Bilingual Children

157

In other words, it might be the case that early sequential bilinguals are not performing this task looking only at the L2, i.e. English or English phonotactics. Instead, they might be using phonological representations active in both languages. This may especially be the case for children that have not had enough exposure to English, for whom the phonology of Czech may always be strongly active and possibly dominant in their phonologies. Thus, nonwords that are allowed in Czech are recognized more quickly than the other nonwords. In summary, one possible explanation for the performance of sequential bilinguals is that they are equally slow in the morphological and control conditions, because both of these are phonotactically odd in Czech. Despite its appeal, further studies are needed to ascertain the reliability of this interpretation, and the current data are not unequivocal. First, a solid prediction given the current proposition would be to obtain the same pattern in late sequential children as well. This is not, however, what we found, since the effect of condition was not significant in late sequential bilinguals. Secondly, our questionnaires did not address dominance in a direct and structured way, and as such it may be inappropriate to claim that the sequential bilinguals were less dominant in English, in comparison to the simultaneous bilinguals, even if our intuition suggests that this is the case. Novel purposely developed tasks and questionnaires may also assess effects of dominance and their interaction with effects of age of onset. So, for the moment our interpretation remains tentative. Given this premise, we may speculate what the implications would be if the pattern of early sequential bilinguals were actually due to the interaction of L1 and L2 phonologies. If this interpretation is correct, it would mean that the degree of separation of the two phonological systems in bilinguals (discussed by Paradis, 2001) can be modulated by the amount of exposure to each language. When exposure is comparable, the phonological systems will show a higher degree of separation, while when one of the two languages is clearly dominant, the phonological system of the non-dominant language will be more strongly influenced by the dominant language’s system. While there has been extensive work on the degree of separation and interaction in the phonological system of bilingual children and adults, little is known about the impact of exposure on this degree. Classic work by Ingram (1981) shows that bilingual children start creating separated phonological systems for the two languages very early on during acquisition. In his study, the author showed that an ItalianEnglish 2-year-old produced different syllabic inventories when using Italian and English: while multisyllabic sequences were preferred in Italian, the child opted for monosyllabic sequences in English. This differentiation suggests the existence of two separate systems from very early stages of development. In subsequent work, Paradis (2001) showed that bilingual children do indeed create separate phonological systems early on, but the two systems keep interacting with each other. Analyzing

158

An Anthology of Bilingual Child Phonology

truncation in nonword repetition with French-English 2-year-old bilinguals, the author demonstrated evidence for separation and evidence for interaction of the two phonological systems. On the one hand, the author showed that preservation of word-fi nal syllables in nonword repetition followed the patterns of each tested language. Bilinguals differentially preserved syllables in the two languages, suggesting sensitivity and separation in the two different phonological systems. On the other hand, the preservation of initial syllables differed in bilingual and monolingual English children, since the bilingual children did not preserve initial stressed syllables as much as the monolinguals; this suggests an effect of French onto the English representations. In short, the fi ndings of our study partially converge with the proposition in Paradis (2001), since the pattern of results from the early sequential bilinguals could be due to interaction between the Czech and the English phonological systems. Crucially, our study suggests that this interaction can also be strong in primary school aged children, but only in those children with low exposure to one of the two languages, and only if using a particularly challenging task. A closing fi nal remark regards the notion of exposure. In this study we used the terms input and exposure relatively interchangeably. However, attention is necessary, because input may refer to different notions, and varied kinds of input can have different effects on language acquisition. For example, maternal input may have a prominent role in the first stages of acquisition, while environmental input may have a prominent role later on. The quality of the input may have differential effects, and the level of interaction modulates the impact of exposure (for example, listening to TV is not the same as being actively engaged in an interaction with a person). In addition, the grouping of the participants in this study was based on age of onset in L2. Using questionnaires and background correlations, we concluded that age of onset correlated to amount of input. Children that were exposed to English earlier were those that received more input in the language, and this is mainly because they had more years of exposure to rely on. For this reason, age of onset, exposure and input are used in connection with each other in this study. This clearly does not always have to be the case; children may have extended exposure and high-quality input despite having a late age of onset. Further studies that can disentangle the effects of age of onset and input in participants will allow us to address the differential effects of exposure, input and age of onset in the acquisition of inflectional morphology. Conclusion

In conclusion, this study confirms previous fi ndings on the acquisition of inflectional morphology with a new pair of languages (Czech and English), and it does so by using a new task with nonwords. As with real

Sensitivity to Morphophonological Cues in Monolingual and Bilingual Children

159

words in previous studies, bilingual children were shown to be less sensitive to the presence of inflectional morphology than monolingual children. Crucially, simultaneous bilinguals were the only sub-group of bilinguals that partly replicated monolingual performance, while (early) sequential bilinguals performed according to the phonotactic constraints of their L1. This fi nding is in line with the claim of Paradis (2010) which reports that exposure has a crucial role in predicting the ability of bilingual children to detect the presence of inflectional morphology in English. Additionally, the pattern of performance of the sequential bilinguals is in line with the claim of Paradis (2001) which states that the phonological systems of bilinguals are separated but non-autonomous. The evident importance of input for the acquisition of this specific structure allows for some reflection on the relation between English inflectional morphology and other important milestones during bilingual language acquisition. Contrary to, for example, the order of heads and complements, the acquisition of English inflectional morphology in bilinguals appears to develop late and to rely massively on input. These traits are those identified by Tsimpli (2014) for the defi nition of late phenomena in bilingual acquisition. English inflectional morphology, thus, might be considered a late phenomenon in the sense of Tsimpli (2014), crucially contrasting with inflectional morphology in richer languages. Appendix Individual age matching in the bilingual groups Age simultaneous

Age early sequential

Age late sequential

9;03

9;03

9;02

9;07

9;06

9;02

9;07

9;10

10;02

10;01

10;00

10;04

10;06

10;01

10;06

10;06

10;03

10;06

10;06

10;03

10;08

10;06

10;04

10;10

10;07

10;06

10;10

10;08

10;10

10;10

11;06

11;00

11;00

11;08

11;00

11;01

11;08

11;06

11;01

12;06

11;08

11;06

11;10

11;08 11;08

160

An Anthology of Bilingual Child Phonology

List of items used in the nonword task

Description (from Cilibrasi, 2016: 116): ‘Although the productive rules used ensured the generation of a large number of nonwords, a few nonwords were created with different vowels because the productive rules led to the generation of real words. Specifically, in the block starting with /dʒ/, /ɑ/ was chosen instead of /ɪ/ because the use of /ɪ/ would have led to /dʒɪlz/, which is an existing word (the plural of Jill, or the possessive form for Jill). In the block starting with /v/, /ε/ was used instead of /aɪ/ because using /ε/ would have led to one of the nonwords having one of its values of biphone segment frequency equaling zero.’ Stem #

Morpho

Morpho

Nonmorpho

Nonmorpho

Control

1

1

vɪld

2

vɪlz

41

vɪls

81

vɪlb

82

vɪlm

2

3

vɛld

4

vɛlz

43 vɛlt

44 vɛls

83

vɛlb

84

vɛlm

3

5

væld

6

vælz

45 vælt

46 væls

85

vælb

86

vælm

4

7

vɔld

8

vɔlz

47

vɔlt

48

vɔls

87

vɔlb

88

vɔlm

5

9

vʌld

10

vʌlz

49

vʌlt

50

vʌls

89

vʌlb

90

vʌlm

vɪlt

42

Control

6

11 nɪld

12

nɪlz

51

nɪlt

52

nɪls

91

nɪlb

92

nɪlm

7

13 naɪld

14

naɪlz

53

naɪlt

54

naɪls

93

naɪlb

94

naɪlm

8

15 næld

16 nælz

55 nælt

56 næls

95

nælb

96

nælm

9

17 nɔld

18

nɔlz

57

nɔlt

58

nɔls

97

nɔlb

98

nɔlm

10

19 nʌld

20

nʌlz

59

nʌlt

60

nʌls

99

nʌlb

100

nʌlm

11

21 θɪld

22

θɪlz

61

θɪlt

62

θɪls

101

θɪlb

102

θɪlm

12

23 θaɪld

24

θaɪlz

63

θaɪlt

64

θaɪls

103

θaɪlb

104

θaɪlm

13

25 θæld

26 θælz

65 θælt

66 θæls

105 θælb

106 θælm

14

27 θɔld

28

θɔlz

67

θɔlt

68

θɔls

107

θɔlb

108

θɔlm

15

29 θʌld

30

θʌlz

69

θʌlt

70

θʌls

109

θʌlb

110

θʌlm

16

31 dʒald

32

dʒalz

71

dʒalt

72

dʒals

111

dʒalb

112

dʒalm

17

33 dʒaɪld

34

dʒaɪlz

73

dʒaɪlt

74

dʒaɪls

113

dʒaɪlb

114

dʒaɪlm

18

35 dʒæld

36

dʒælz

75

dʒælt

76

dʒæls

115

dʒælb

116

dʒælm

19

37 dʒɔld

38

dʒɔlz

77

dʒɔlt

78

dʒɔls

117

dʒɔlb

118

dʒɔlm

20

39 dʒʌld

40

dʒʌlz

79

dʒʌlt

80

dʒʌls

119

dʒʌlb

120

dʒʌlm

Acknowledgements

We would like to thank the children and the staff in the schools that participated in the project. In Prague: the Prague International School, the  Parklane International School and the Prague British School.

Sensitivity to Morphophonological Cues in Monolingual and Bilingual Children

161

In Cambridge: Jeavons Wood Primary Cambourne and St Albans Primary. We would like to thank the De Vincenzi Foundation for funding this project and Charles University – Progres Q10 for additional funding during the writing up phase. We would like to thank Alzbeta Brabcova for her precious help with data collection, and Megan Lee for her help in building a connection with the schools in Cambridge. We would like to thank Dana Tycova for her work on the reorganization of the data files. References Baayen, R.H., Davidson, D.J. and Bates, D.M. (2008) Mixed-eff ects modeling with crossed random effects for subjects and items. Journal of Memory and Language 59 (4), 390–412. Bates, D. (2005) Fitting linear mixed models in R. R News 5 (1), 27–30. Bialystok, E., Luk, G., Peets, K.F. and Yang, S. (2010) Receptive vocabulary differences in monolingual and bilingual children. Bilingualism: Language and Cognition 13 (4), 525–531. Bittner, D., Dressler, W.U. and Kilani-Schoch, M. (eds) (2011) Development of Verb Infl ection in First Language Acquisition: A Cross-linguistic Perspective. Studies on Language Acquisition 21. Berlin: Walter de Gruyter. Blom, E., Polisenska, D. and Weerman, F. (2008) Articles, adjectives, and age of onset: The acquisition of Dutch grammatical gender. Second Language Research 24, 297–332. Caramazza, A., Laudanna, A. and Romani, C. (1988) Lexical access and inflectional morphology. Cognition 28 (3), 297–332. Chomsky, N. (1993) Lectures on Government and Binding: The Pisa Lectures (No. 9). Berlin: Walter de Gruyter. Cilibrasi, L. (2016) Word position effects in speech perception. Unpublished PhD dissertation, University of Reading. Cilibrasi, L., Stojanovik, V., Riddell, P. and Saddy, D. (2019) Sensitivity to inflectional morphemes in the absence of meaning: Evidence from a novel task. Journal of Psycholinguistic Research 48, 747–767. doi:10.1007/s10936-019-09629-y. Clahsen, H., Eisenbeiss, S. and Sonnenstuhl-Henning, I. (1997) Morphological structure and the processing of inflected words. Theoretical Linguistics 23 (3), 201–250. Erdos, C., Genesee, F., Crago, M. and Debas, K. (2005) Does bilingual input decelerate the acquisition of grammatical schemas? Poster presented at the Boston University Conference on Language Development, Boston, MA. Field, A. (2013) Discovering Statistics Using IBM SPSS Statistics. London: SAGE. Flege, J.E. (2009) Give input a chance. In T. Piske and M. Young-Scholten (eds) Input Matters in SLA (pp. 175–190). Bristol: Multilingual Matters. Gathercole, V.M. (2002) Monolingual and bilingual acquisition: Learning different treatments of that-trace phenomena in English and Spanish. In K.D. Oller and R. Eilers (eds) Language and Literacy in Bilingual Children (pp. 220–254). Clevedon: Multilingual Matters. Gathercole, V.M. (2006) Miami and North Wales, so far and yet so near: Morphosyntactic development in bilingual children. Manuscript, University of Wales Bangor. Gathercole, V.M. and Hoff, E. (2007) Input and the acquisition of language: Three questions. In E. Hoff and M. Shatz (eds) Handbook of Language Development (pp. 107– 127). Chichester: John Wiley. Gathercole, S.E., Willis, C.S., Baddeley, A.D. and Emslie, H. (1994) The children’s test of nonword repetition: A test of phonological working memory. Memory 2 (2), 103–127. Guasti, M.T. (2017) Language Acquisition: The Growth of Grammar. Cambridge, MA: MIT Press.

162

An Anthology of Bilingual Child Phonology

Gutiérrez-Clellen, V.F., Restrepo, M.A. and Simón-Cereijido, G. (2006) Evaluating the discriminant accuracy of a grammatical measure with Spanish-speaking children. Journal of Speech, Language, and Hearing Research 49 (6), 1209–1223. Ingram, D. (1981) The emerging phonological system of an Italian-English bilingual child. Journal of Italian Linguistics 6 (2), 95–113. Johnson, V.E., de Villiers, J.G. and Seymour, H.N. (2005) Agreement without understanding? The case of third person singulars. First Language 25 (3), 317–330. Kovelman, I., Baker, S.A. and Petitto, L.A. (2008) Age of fi rst bilingual language exposure as a new window into bilingual reading development. Bilingualism: Language and Cognition 11 (2), 203–223. Möhring, A. and Meisel, J.M. (2003) The verb-object parameter in simultaneous and successive acquisition of bilingualism. In N. Müller (ed.) (In)vulnerable Domains in Multilingualism (pp. 295–334). Amsterdam: John Benjamins. Montrul, S.A. (2008) Incomplete Acquisition in Bilingualism: Re-examining the Age Factor. Studies in Bilingualism 39. Amsterdam: John Benjamins. Moreland, M.L. (2011) The status of the post-initial jod: Evidence from British English wordlikeness judgements. Unpublished PhD dissertation, University of Reading. Paradis, J. (2001) Do bilingual two-year-olds have separate phonological systems? International Journal of Bilingualism 5 (1), 19–38. Paradis, J. (2010) Bilingual children’s acquisition of English verb morphology: Effects of language exposure, structure complexity, and task type. Language Learning 60 (3), 651–680. Paradis, J., Nicoladis, E. and Crago, M. (2007) French-English bilingual children’s acquisition of the past tense. BUCLD 31 Proceedings (pp. 497–507). Boston, MA: Cascadilla Press. Pearson, B.Z. (2002) Narrative competence among monolingual and bilingual school children in Miami. In D.K. Oller and R.E. Eilers (eds) Language and Literacy in Bilingual Children (pp. 135–174). Clevedon: Multilingual Matters. Poeppel, D. and Wexler, K. (1993) The full competence hypothesis of clause structure in early German. Language 69 (1), 1–33. Post, B., Marslen-Wilson, W.D., Randall, B. and Tyler, L.K. (2008) The processing of English regular inflections: Phonological cues to morphological structure. Cognition 109 (1), 1–17. Raven, J.C. (1998) Raven’s Progressive Matrices and Vocabulary Scales. Oxford: Oxford Psychology Press. Tomasello, M. (2009) Constructing a Language. Cambridge, MA: Harvard University Press. Tsimpli, I.M. (2003) Features in L1 and L2 acquisition: Evidence from Greek clitics and determiners. Acquisition et Interaction en Langue Etrangere 20, 87–128. Tsimpli, I.M. (2014) Early, late or very late? Timing acquisition and bilingualism. Linguistic Approaches to Bilingualism 4 (3), 283–313. Tsimpli, I.M., Andreou, M., Agathopoulou, E. and Masoura, E. (2014) Narrative production, bilingualism and working memory capacity: A study of Greek-German bilingual children. Proceedings of the 11th International Conference on Greek Linguistics (pp. 1730–1742). Rhodes: University of the Aegean. Unsworth, S., Argyri, F., Cornips, L., Hulk, A., Sorace, A. and Tsimpli, I.M. (2014) On the role of age of onset and input in early child bilingualism in Greek and Dutch. Applied Psycholinguistics 35, 765–805. Vitevitch, M.S. and Luce, P.A. (2004) A web-based interface to calculate phonotactic probability for words and nonwords in English. Behavior Research Methods, Instruments, and Computers 36 (3), 481–487. Wexler, K. (1998) Very early parameter setting and the unique checking constraint: A new explanation of the optional infi nitive stage. Lingua 106 (1–4), 23–79.

7 Lexical-semantic Organization in Monolingual and Bilingual Hebrew Speaking Children: Evidence from a Word Association Task Atalia Hai Weiss

Introduction

Word association tasks are used as a means of inspecting lexical organization in children. Researchers have focused on syntagmatic versus paradigmatic responses in these tasks. They described a phenomenon, called ‘syntagmatic-paradigmatic shift’, based on fi ndings that 6-year-old children tend to respond with words related to a stimulus word in the discourse (‘syntagmatic relation’, e.g. dog-bone), whereas 9-year-old children tend to respond with a word from the same form class and semantic category (‘paradigmatic relation’, e.g. dog-cat). Younger children are reported to provide many phonological-related responses (e.g. dog-door). This distinction between phonological, syntagmatic and paradigmatic responses, however, does not take into account the fact that a major device organizing the mental lexicon in many languages is morphology. This might be especially relevant for morphologically rich languages such as Hebrew, where words from different grammatical word classes are related through a common consonantal root. In the current study, we aim to inspect the distribution of phonologic, syntagmatic and paradigmatic as well as morphological responses of Hebrew speaking children.

163

164

An Anthology of Bilingual Child Phonology

Background and rationale

The mental lexicon is a cognitive system in which lexical units are stored in an organized network (Dóczi & Kormos, 2016). With more developed lexical knowledge, a rich structured network of associations arises among the words known (Read, 2000). Accordingly, learning the meaning of new words involves more than the acquisition of isolated lexical units, but rather a gradual acquaintance with various kinds of connections related to the particular word (Meara & Wolter, 2004). Word association research is compatible with the metaphor of ‘network’ to describe the organization of the mental lexicon (Wilks, 2009). In a discrete word association task, the respondent is instructed to answer with the first word that comes to mind for each presented word. The associative links elicited in word association tasks are assumed to represent the strongest and most salient links in an individual’s lexical network (Albrechtsen et al., 2008). Thus, associative links allow identifying similarities and variations in these networks between individuals (Fitzpatrick & Munby, 2013). Word associations have been shown to be consistent across trials and make quantitative calculations easier than, for example, in discourse data (Fitzpatrick et al., 2015). Word association behaviour has often been assessed in terms of types of links between the prompt word and the response. Since early studies of this nature, analyses of the semantic links were based on the distinction between syntagmatic and paradigmatic relationships (Fitzpatrick et al., 2015) that relates to nouns as well as other grammatical word classes, i.e. verbs, adjectives, adverbs, pronouns and other function words (Cruse, 1986). Syntagmatic relations represent horizontal relations between words by supplying descriptive information (e.g. dog-guards, pours-milk, redtomato). On the other hand, pairs of words that can be substituted for one another without changing the grammatical word class are defined as paradigmatic (e.g. van-train) (Nelson, 1977). Paradigmatic relations refer to hierarchical systems (e.g. dog-animal, dog-poodle, dog-tail), as well as antonyms (e.g. close-open) and synonyms (shout-yell) (Cronin, 2002). Many developmental studies reveal that preschool and kindergarten children, when presented with a word association task, often generate responses that bear a phonological relationship to the cue words. These are sound-based responses, which include alliterations (e.g. candy-can) or rhymes (e.g. dig-fig). The intensified attention to sound properties during preschool and kindergarten may be, in part, due to the training concentrated on phonological awareness that occurs during this age period (Justice, 2006). This preference for phonologically influenced responses was found to be transient and to fade out during the fi rst years of schooling, while semantic associations show a sharp growth (Cronin, 2002). This shift from phonological to semantic dominance is accompanied by a possible subsequent shift termed ‘syntagmatic-paradigmatic shift’,

Lexical-semantic Organization in Monolingual and Bilingual Hebrew Speaking Children

165

which relates to a change from a dominance of syntagmatic responses to a dominance of paradigmatic ones (Brown & Berko, 1960; Danovitch & Keil, 2004; Entwisle, 1966; Ervin, 1961; Nelson, 1977). Studies show that at age 5 years most semantic responses of children are syntagmatic. However, by age 9 most responses are paradigmatic (Cole & Means, 1986). Importantly, some responses are paradigmatic even at younger ages, for some frequent nouns (e.g. dog-cat) or adjectives (e.g. cold-hot). Mechanisms involved in driving the shift from syntagmatic to paradigmatic dominance are various – from changes in an individual’s interpretation of the task, to formal schooling, acquisition of reading and cumulative academic experience (Anglin, 1985; Cronin, 2002; Ordonez et al., 2002; Verhallen & Schoonen, 1993, 1998). The current research aims at shedding light on the organization of the mental lexicon as expressed by a word association task among Hebrew speaking children, both monolingual and bilingual, who are growing up in Israel. Only a few studies have utilized the word association task in Hebrew, and these did not focus on children aged 6–9 years old, but rather on older school children or young adults (Prior, 2004; Seroussi, 2011). The Hebrew language is part of the Semitic branch of languages, a branch of the Afro-Asiatic language family originating in the Middle East, spoken by people across much of Western Asia and North Africa, as well as in immigrant communities in North America, Europe and Australia (Hammarström et al., 2017). Like other Semitic languages, Hebrew is a morphologically rich language in which many grammatical and lexical notions are encoded in word-internal structures. Thus, morphology is an important organizing principle of the lexicon of Hebrew speakers, and the traditional categorization described above into phonological, syntagmatic and paradigmatic responses does not take the morphological aspect into account (Berman, 2000, 2003; Ravid, 1990, 2003). One important characteristic of Hebrew is that all verbs and most nouns and adjectives contain a tri- or quadri-consonantal core, the Semitic root (C.C.C). This structural core appears discontinuously in the word, since it is interdigitated by vowels provided by the complementary vocalic structure of the pattern (RaTuV ‫‘ רטוב‬wet’, R.T.V. ‫ב‬.‫ט‬.‫)ר‬. In most cases, the root morpheme carries the basic meaning of the word, and the word pattern carries grammatical information, i.e. word class (noun/verb/adjective), grammatical gender and number, and in the verb system also tense and aspect (Berent et al., 2007). Although most base forms can be decomposed into a root and a word pattern, Hebrew also has a large set of morphologically simple words that do not have the typical Semitic structure, and in that respect they resemble Indo-European base words (Velan & Frost, 2011). For example, the base word mita ‫‘ מיטה‬bed’ cannot be decomposed into a root and a word pattern. Hebrew has a large set of such words, often names of common objects. These words are prevalent, and some of them are highly frequent.

166

An Anthology of Bilingual Child Phonology

Developmentally, Hebrew speaking children show awareness of morphology as early as age 3 years, by coining new words that adhere to the morphological principles of the language (Berman, 2000). Morphological knowledge also plays a central role in the acquisition of literacy and spelling in Hebrew (Ravid & Bar-On, 2005), and the literate lexicon includes a high ratio of derivationally complex words (Anglin, 1993; Nir-Sagiv et al., 2008). In general, studies conducted in Hebrew demonstrate that the developmental route of Hebrew derivational morphology belongs solidly in the domain of later language acquisition (Avivi Ben-Zvi, 2010; Berman, 2004). Furthermore, derivational morphology has a distinctive role in distinguishing between poor and proficient readers and between typically developing children and children with language impairment (Ravid et al., 1999). Lexical organization and activation in speakers of morphologically rich languages such as Hebrew are influenced by morphological structure and principles (Frost et al., 1997). However, an open question is whether the derivational knowledge of Hebrew speakers is expressed in a word association task. This question is especially interesting since Hebrew words that are related morphologically are often not of the same grammatical word class (e.g. GaDaL – GaDoL ‫גדול‬-‫‘ גדל‬grew up – big’), consequently related in a syntagmatic manner (adjective-verb) rather than paradigmatic. However, some words with the same consonant root are from the same word class (e.g. HiGDil – GaDal ‫גדל‬-‫‘ הגדיל‬made large – grew up’), and therefore can be considered related paradigmatically (verbverb). One of the aims of the current study was to explore whether the specific characteristics of the Hebrew morphology are reflected in a word association task, and whether they influence the expected phenomenon of syntagmatic-paradigmatic shift among children during the first years of schooling. As argued above, the classic categorization of responses into phonological, syntagmatic and paradigmatic does not take into account morphological aspects. A more recent categorization, suggested by Fitzpatrick (2006, 2007, 2009), makes some reference to morphological aspects, although not explicitly. Fitzpatrick suggests a renewed system, based on the classical three-way categorization that includes the following main categories: a category of ‘Meaning Based’ responses which parallels the classical paradigmatic category; a ‘Position Based’ category which parallels the classical syntagmatic category; and a ‘Form Based’ category which parallels the classical phonological category, but also including ‘change of affi x and similarity in form’. This last category, which includes not only phonological similarity but also morphological changes, relates to changes typical to English, such as changes of an affi x, suffi x or prefi x. This suggestion of placing phonological and morphological changes as part of the same category gave rise to some criticism. For example, Wilks (2009) argued that Form Based responses are given a priority over Meaning

Lexical-semantic Organization in Monolingual and Bilingual Hebrew Speaking Children

167

Based ones. Thus, a response with the word ‘disappear’ to the target word ‘appear’, could be regarded as a paradigmatically related response but, according to Fitzpatrick’s suggestion, it can be regarded as Form Based. Similarly, a response with the word ‘accidental’ to the word ‘accident’ could be regarded as syntagmatic, but according to the new system it should be regarded as Form Based. This kind of criticism is even more relevant for Hebrew, where derivational relations between words are less transparent and are related to subsequent language development, as described in detail above. The current study aims at shedding more light on the development of lexical organization in Hebrew speaking children, both monolingual and bilingual, as expressed in a word association task, while inspecting what proportion of semantic responses (whether paradigmatic or syntagmatic) are morphologically derived responses. In addition, since Hebrew has a rich inflectional morphology that relates to nouns, verbs and adjectives, special consideration was given to the status of inflected forms that are of the same part of speech (e.g. meTaPeS ‫(‘ מטפס‬he) climbs’, meTaPeSet ‫מטפסת‬ ‘(she) climbs’). A response to a prompt word with an inflection (e.g. dogdogs) could be considered an error in a word association task, in line with the suggestion that a word and its inflected forms should be considered one lexical unit (Francis & Kucera, 1982), although this statement is debated (Cilibrasi, 2016; Post et al., 2008). However, in the current research, morphologically inflected responses are considered to be part of sound-based responses. These responses might be less mature responses, similarly to the phonological ones, but they do not necessarily reflect phonological metalinguistic knowledge. The bilingual children inspected in the frame of the current study are English-Hebrew speakers exposed to English from birth and to Hebrew prior to age 3 years, thus considered simultaneous or early sequential bilingual children (Kohnert, 2013). It is well established that there is a persistent gap in vocabulary that prevails even after a number of years of exposure to the societal language, between native speakers and bilingual children (Farnia & Geva, 2011; Jean & Geva, 2009; Kieffer & Lesaux, 2012). Importantly, these studies focused on measures that relate to the size of mental lexicon by using receptive or expressive vocabulary tasks, i.e. breadth-related parameters (Meara, 2009). However, these measures often fail to differentiate monolingual children with language impairment from typically developing peers (Gray et al., 1999); thus, as an alternative, a research focusing on depth-related aspects of vocabulary has recently been put forward (Sheng & McGregor, 2010; Sheng et al., 2012). Few studies have used word association tasks in order to inspect the lexical organization of bilingual children. Sheng et al. (2006) investigated Mandarin-English bilingual children in the USA and examined their performance when presented with a repeated word association task (36 prompt words are repeated three times through the task) in comparison

168

An Anthology of Bilingual Child Phonology

to their English speaking peers. The authors reported on a subtle difference between the monolingual and bilingual groups, with bilingual children more frequently making paradigmatic responses as their fi rst associations compared to the monolingual group. In a more recent study conducted in Sweden among Arabic-Swedish bilingual children, Holmström et al. (2015) used a discrete word association test and found that lexical organization among typically developing 6- and 7-year-old bilingual children is similar to that among their monolingual peers. Aim of the Current Study

The current study aimed at providing knowledge on lexical-semantic organization in typically developing Hebrew speaking monolingual and bilingual 6–9 year-old children from medium–high socio-economic status (SES). Our means consisted of a discrete word association task which includes words from three grammatical word forms – nouns, verbs and adjectives – all familiar to and often used by young children. We aimed at inspecting the relative frequency of the three classical types of responses, i.e. phonological, syntagmatic and paradigmatic, as well as the unique types in Hebrew, i.e. morphologically derived and morphologically inflected responses. We also aimed at inspecting whether frequency of use of each type of associative response changed with age among both the monolingual and bilingual children. Naming proficiency in Hebrew was also examined among the two groups of children in order to gain some information regarding the breadth of the children’s lexicon, and thus to have a fuller picture regarding their lexical knowledge. We predicted that most responses by both the monolingual and bilingual children would be semantic. Moreover, in line with the known phenomenon of syntagmatic-paradigmatic shift, an advantage of paradigmatic over syntagmatic responses was expected, mainly for prompt words from the grammatical word classes of nouns and adjectives, and to a lesser extent from verbs (Cole & Means, 1986). This advantage should be more prominent in older children. We predicted that some semantic responses would be in a morphologically derived relation with the prompt words, influenced by the nature of Hebrew. However, we predicted that these responses might not appear too frequently, due to the structure of the presented list (most nouns are without a decomposed typical structure, as discussed in the section on ‘Method’ next), and might appear more often among older children. We also predicted that, due to reduced exposure to Hebrew, the group of bilingual children would on average achieve lower naming scores. However, it was expected that the amount of exposure to Hebrew that the bilingual children had would be sufficient to enable them to have comparable morphological competence to that of their monolingual peers (Marchman & Bates, 1994; Gathercole & Hoff, 2007), and thus, use of morphologically derived responses would be present to the same extent among the two groups.

Lexical-semantic Organization in Monolingual and Bilingual Hebrew Speaking Children

169

Method Materials General cognitive tests

General cognitive abilities were measured to match groups on this background variable using two subtests from the Hebrew version of the Wechsler Intelligence Scale for children (WISC-IV): Block Design and Digit Span (Wechsler, 2003). The Block Design task measures spatial reasoning abilities. The digit span test measures verbal working memory and comprises two parts: (i) Digit Forward – a series of orally presented digits should be repeated in the order of their presentation; and (ii) Digit Backward – a series of orally presented digits should be repeated in reverse order. In both conditions, the length of the sequence of digits is gradually increased. Picture naming test

Naming proficiency in Hebrew was assessed using a picture-naming test (SHEMESH; Biran & Friedmann, 2004) consisting of 100 colour pictures of concrete nouns from different semantic categories. Importantly, the Shemesh test is not a vocabulary measure test but rather a sensitive naming test that is used in the clinic for assessment of naming deficits in adults with brain damage (Biran & Friedmann, 2005) and school-aged children (Novogrodsky & Kreiser, 2015). Thus, this test focuses on the ability to retrieve words, similarly to the ability required in the association task. The instruction to the children was: ‘I will show you pictures of objects. Please tell me in one word what you see in the picture.’ Pictures for naming were presented with no time limit. All responses were transcribed. Each correct response received one point and each wrong response received zero. Naming errors were analyzed but, not being directly related to the goal of this study, they will be reported in a future article. Word association task

Lexical semantic organization was assessed by a discrete word association task. Thirty-six words were presented: 12 nouns, 12 verbs and 12 adjectives (Table 7.1). All words were presented in their masculine singular form, and verbs were presented in the present tense. The stimuli were semi-randomized so that words belonging to the same speech part (e.g. pushing, cutting) or the same theme (e.g. cutting, cake) did not occur consecutively. Stimulus words were based on lists of English words characterized as having high frequency and being acquired at an early age, as presented in Cronin (2002) and Sheng and McGregor (2010). These English words were initially translated into Hebrew by a native speaker. Then they were inspected for being acquired before age 6 by native Hebrew speakers, by looking at Hebrew diagnostic tests (Guralnik, 1995; Katzenberger & Meilijson, 2014), and by inspecting lists of words

170

An Anthology of Bilingual Child Phonology

Table 7.1 Word association task: List of presented words and their morphological characteristics Word (Hebrew)

Meaning

Grammatical word class

Identified consonantal root

1

matos

‫מטוס‬

airplane

noun

‫ס‬.‫ו‬.‫ט‬

T.U.S

2

ʔadom

‫אדום‬

red

adjective

‫מ‬.‫ד‬.‫א‬

ʔ.D.M L.K.K

3

melakek

‫מלקק‬

leaking

verb

‫ק‬.‫ק‬.‫ל‬

4

ʃovav

‫שובב‬

Naughty

Adjective

‫ב‬.‫ב‬.‫ש‬

S.V.V

5

doxef

‫דוחף‬

pushing

verb

‫ף‬.‫ח‬.‫ד‬

D.X.F

6

kelev

‫כלב‬

dog

noun

*

7

xalon

‫חלון‬

window

noun

*

8

gozer

‫גוזר‬

cutting

verb

‫ר‬.‫ז‬.‫ג‬

G.Z.R

9

gadol

‫גדול‬

big

adjective

‫ל‬.‫ד‬.‫ג‬

G.D.L

10

ʔuga

‫עוגה‬

Cake

Noun

*

11

ʃofex

‫שופך‬

Pouring

Verb

‫כ‬.‫פ‬.‫ש‬

12

xaʃux

‫חשוך‬

darkened

adjective

‫ך‬.‫ש‬.‫ח‬

X.S.X

13

metapes

‫מטפס‬

climbing

verb

‫ס‬.‫פ‬.‫ט‬

T.P.S

14

xultsa

‫חולצה‬

shirt

noun

*

S.F.X

15

ʔozen

‫אוזן‬

ear

noun

‫נ‬.‫ז‬.‫א‬

16

patuax

‫פתוח‬

open

adjective

‫ח‬.‫ת‬.‫פ‬

ʔ.Z.N P.T.X

17

kaved

‫כבד‬

heavy

adjective

‫ד‬.‫ב‬.‫כ‬

K.V.D

18

mexajex

‫מחייך‬

smiling

verb

‫כ‬.‫י‬.‫ח‬

X.J.X

19

para

‫פרה‬

cow

noun

*

20

xadaʃ

‫חדש‬

new

adjective

‫ש‬.‫ד‬.‫ח‬

21

mita

‫מיטה‬

bed

noun

*

22

katsar

‫קצר‬

short

adjective

‫ר‬.‫צ‬.‫ק‬

X.D.S

K.TS.R

23

matate

‫מטאטא‬

broom

noun

‫א‬.‫ט‬.‫א‬.‫ט‬

T.A.T.A

24

ʔole

‫עולה‬

rising

verb

‫ה‬.‫ל‬.‫ע‬

ʔ.L.H

25

paxdan

‫פחדן‬

fearful

adjective

‫ד‬.‫ח‬.‫פ‬

P.X.D

26

joʃev

‫יושב‬

sitting

verb

‫ב‬.‫ש‬.‫י‬

J.S.V

27

gezer

‫גזר‬

carrot

noun

*

28

mesader

‫מסדר‬

tiding

verb

‫ר‬.‫ד‬.‫ס‬

29

ʔets

‫עץ‬

tree

noun

*

30

jafe

‫יפה‬

beautiful

adjective

‫ה‬.‫פ‬.‫י‬

31

loxeʃ

‫לוחש‬

whispering

verb

‫ש‬.‫ח‬.‫ל‬

L.X.S

32

naki

‫נקי‬

clean

adjective

‫ה‬.‫ק‬.‫נ‬

N.K.H

S.D.R

J.P.H

33

tof

‫תוף‬

drum

noun

‫פ‬.‫ו‬.‫ת‬

T.F.F

34

mitlabeʃ

‫מתלבש‬

getting dressed

verb

‫ש‬.‫ב‬.‫ל‬

L.V.S

35

ratuv

‫רטוב‬

wet

adjective

‫ב‬.‫ט‬.‫ר‬

R.T.V

36

rats

‫רץ‬

running

verb

‫ץ‬.‫ו‬.‫ר‬

R.U.TS

Note: *No identified consonantal root.

Lexical-semantic Organization in Monolingual and Bilingual Hebrew Speaking Children

171

presented by Łuniewska et al. (2016). All chosen words had a single meaning and all of them, except for one, could clearly be identified as noun, verb or adjective. The exception was one word, katzar ‫קצר‬, presented as an adjective meaning ‘short’, which is familiar among young children, but it can also function as a infrequent verb in the past tense meaning ‘harvested’. Only one child among all the participants considered the latter meaning of the word. None of the words in the final list had a parallel cognate word in English. Following this initial stage of constructing a list of high-frequency words that are acquired early, the words were also inspected for their morphological structure and mainly for the presence or absence of an identified root. As shown in Table 7.1, eight of the 12 presented nouns do not have an identified consonantal root with a related meaning; thus a derived morphological response is not possible following them. On the other hand, all of the other four nouns and all verbs and adjectives have an identified root, and therefore a derived morphologic response is possible following them. Words were presented in the same order across the children, as presented in Table 7.1. The instruction to children was: ‘I will say a word and you should say the first word that comes to your mind following my word. Then I will say another word and you will say another one. We will start with three examples.’ The three practice items were a noun, an adjective and a verb. During practice, if the child responded with a syntagmatic association (e.g. car-drive), the examiner reinforced the child’s response and offered a model of a paradigmatic response by saying: ‘Yes, drive goes with car, and another word is truck.’ During the actual testing, the examiner provided non-contingent feedback to all responses. Analyses of responses

Word association responses were coded into the following categories: (1) Paradigmatic associations were defi ned as responses in the same grammatical word class, which could be a synonym (beautiful-handsome), an antonym (whisper-shout), a coordinate (red-green), a superordinate (bed-furniture) or a subordinate (tree-oak). Syntagmatic associations included responses that followed the prompts in the syntactic stream (smiling-baby, open-book) and shared functional, physical, causative, locative or thematic relations with the stimulus (e.g. bed-sleep, red-tomato, running-exhausted, airplane-Paris, cake-bakery). An exceptional case was observed with the responses given to the prompt word etz ‫‘ עץ‬tree’. A response given by three of the children could be considered as both paradigmatic and syntagmatic: tapuax ‫‘ תפוח‬apple’, perot ‫‘ פירות‬fruits’). A decision was made to consider these

172

An Anthology of Bilingual Child Phonology

responses not as collocational responses (syntagmatic) but rather as paradigmatic due to the specific choice of response words (apple and not apples, fruits and not fruit – options that are more common for the collocational expressions in Hebrew). (2) Morphologically derived associations were specified by words that shared a root with the prompt word tof-metofef ‫מתופף‬-‫‘ תוף‬drum – playing on a drum’, shared root: T.F.F, or addition of suffix kelevklavlav ‫כלבלב‬-‫‘ כלב‬dog – small cute dog’. All responses were initially coded as paradigmatic or syntagmatic and received an additional status of derived morphological associations when appropriate. For example, following the prompt word tof ‫‘ תוף‬drum’, the response metofef ‫‘ מתופף‬playing on a drum’ was initially coded as a syntagmatic response and additionally as a morphologically derived response. (3) Sound-based responses consisted of phonological responses such as rhyming and alliterative words (e.g. xalon-balon ‫בלון‬-‫‘ חלון‬windowballoon’), and also of morphological inflections of the prompt word (kelev-klavim ‫כלבים‬-‫‘ כלב‬dog-dogs’). An exceptional case was presented by one specific response, vilon ‫‘ וילון‬curtain’, given to the prompt word xalon ‫‘ חלון‬window’. The relation between these two words could be considered as both paradigmatic and phonological. Such a response was given by four children. A decision was made to consider this response as paradigmatic since these children gave no other phonological responses, and it could be the case of coincidence of a semantically related word that is also phonological. (4) Errors included repetitions of the prompts, or words that bore no identifiable relation to the prompts (e.g. dog-piano). No responses were defi ned when a ‘I don’t know’ answer was given following the prompt word.

Participants

The participants were 41 typically developing children born in Israel and attending fi rst, second or third grade in public elementary schools in Israel. Information about general development and about exposure to and use of each language for the bilingual children was collected through a parental questionnaire (BIPAQ; Abutbul-Oz et al., 2012). Participating children had no sensory-motor difficulties and no history of language delay or impairment. All participants’ general cognitive scores, as assessed by the Block Design and Digit Span subsets of the WAIS-IV (Wechsler, 2003), were within or above the normal range of the general population (scaled score of 7 or above, i.e. no less than a standard deviation (SD) of 2 below the average of the general population). The fi nal groups included 24 children (54% girls) reported as speaking Hebrew exclusively at home and classified as monolingual Hebrew

Lexical-semantic Organization in Monolingual and Bilingual Hebrew Speaking Children

173

Table 7.2 Means and SDs of general and cognitive related measures of the two groups of participants Monolingual Hebrew speaking children (n = 24)

Bilingual Hebrew speaking children (n = 17)

7;4 (0.75)

7;3 (1)

Wechsler – Block Design (Nonverbal Ability test)

12 (2)

13 (3)

Wechsler – Digit Span (Working Memory test)

11 (2)

12 (2)

Mother

16.4 (2.8)

17.2 (1.8)

Father

16.4 (3)

16.7 (2.3)

Age (years;months) General cognitive abilities (score)

Parental education (years)

speakers, and 17 children (47% girls) reported as having English as their dominant language and classified as English-Hebrew bilingual children. All monolingual participants scored within the norm expected for their age in the Hebrew naming test, Shemesh (Biran & Friedman, 2004). All bilingual children belong to families with at least one parent who immigrated to Israel from an English speaking country. According to parental reports, age of onset (AoO) of bilingualism ranged from 0 to 33 months; thus all children are simultaneous or early-sequential bilinguals (Kohnert, 2010). The two language groups are a result of convenience sampling, yet all participants were drawn from different public elementary schools in similar neighbourhoods with equivalent medium–high SES (see sample characteristics in Table 7.2). Parental consent was secured, and the study was approved by the ethics committee of the Hadassah Academic College. Procedure

The current study was part of larger project, which also investigated the bilingual children’s lexical skills in English, their home language. Speech-language pathologists administered tasks in one session lasting about 45 minutes in a quiet room in the children’s homes. During the tests, children were given general positive feedback regardless of performance. Results Hebrew naming proficiency

Naming proficiency in Hebrew was inspected among both the monolingual and bilingual children in the study. As seen in Table 7.3, the monolingual children named significantly more items correctly than did their

174

An Anthology of Bilingual Child Phonology

Table 7.3 Mean accuracy (SD) on productive vocabulary by group

Expressive vocabulary test

Monolingual children

Bilingual children

Mean (SD)

94 (0.5)

87.1 (8)***

Range

89–97

68–96

Note: *** = p < 0.001.

bilingual peers (t(39) = 3.9, p < 0.001), in agreement with previous fi ndings (e.g. Farnia & Geva, 2011). As shown in Table 7.3, there was large variability found among the bilingual children. Indeed, only six of all children (35%) performed within the range of scores of the monolingual peers (91–97), while the remaining 11 children (65%) performed within a lower range (68–86).

Patterns of responses in the word association task

Response- % from all responses

Each child was presented with 36 words and his responses were initially coded as either paradigmatic, syntagmatic, phonological, morphologically inflected, error, absent or in English (for the bilingual children). The percentage of use of each of these responses from all possible responses (36) was calculated for each child. Morphologically derived responses were coded as either paradigmatic or syntagmatic, and therefore they are not part of this initial calculation. Figure 7.1 shows the pattern of average responses among the two groups. As can be observed, semantic responses (i.e. paradigmatic and syntagmatic) entailed the majority of children’s responses, form-based responses (i.e. phonological and morphologically inflected) were most of all remaining responses, while very few of the responses were wrong or absent. The bilingual children did not produce any responses in English. 60 55 50 45 40 35 30 25 20 15 10 5 0

Monolingual

Paradigmatic

Syntagmatic

Phonologic

Morphologic-inflected

Error

Bilingual

No-response

Figure 7.1 Mean percentage (SEM) of use of each type of response as a function of group. One-way ANOVA revealed no difference between the two groups in relation to any of the responses

Lexical-semantic Organization in Monolingual and Bilingual Hebrew Speaking Children

175

Semantic responses: Paradigmatic versus syntagmatic

In order to investigate whether a syntagmatic-paradigmatic shift can be observed, a repeated measures ANOVA was performed, with group (monolingual and bilingual) as between-subjects factor, type of response (paradigmatic and syntagmatic) as within-subject factor, and age as a covariate. We found no effect of type of response (F (1, 38) = 0.292, p = 0.592) and no interaction between type of response and group (F (1, 38) = 0.001, p = 0.980), or type of response and age (F (1, 38) = 0.211, p = 0.649), meaning that the two groups had the same pattern of semantic responses that was not changing with age. In addition, percentage of use of paradigmatic and syntagmatic responses was compared in each child. Pairwise comparisons indicated no significant difference among the monolingual children (t(23) = −0.55, p  = 0.585), as well as the bilingual children (t(16) = −0.49, p = 0.629). Interestingly, an inverse relation between paradigmatic and syntagmatic response was found (Figure 7.2). However, as can be seen in Figure 7.2, large variability in both types of responses is seen among both groups of children. Morphologically derived responses

Semantic responses were inspected for being morphologic-derived in addition to being paradigmatic or syntagmatic. As mentioned in the Method section, all verbs and adjectives, but only four of the nouns presented in the list, have an identified root. These words are the most susceptible to being changed through a derivation. However, all words in Hebrew can be changed by adding a suffix (e.g. kelev-klavlav ‫כלבלב‬-‫‘ כלב‬dog – small cute dog’). Therefore, responses to all prompt words were inspected for being in a morphologic-derived relation with the prompt word. Monolingual Children

Bilingual children 100

Syntagmatic Responses% of all responses

Syntagmatic responses% of all responses

100 80 60

2: 40 20 0

80 60 40 20 0

0

20

40

60

80

100

Paradigmatic responses- % of ll responses

0

20

40

60

80

100

Paradigmatic Responses- % of all responses

Figure 7.2 Correlations between percentage of use of paradigmatic responses and use of syntagmatic responses among both groups of children. Significant correlation is seen in the monolingual group (r = −0.837, p < 0.001) as well as in the bilingual groups (r = −0.947, p < 0.001)

176

An Anthology of Bilingual Child Phonology

Table 7.4 Percentage (SD) of derived responses as part of syntagmatic and paradigmatic responses among both groups of children

Derived paradigmatic responses

Derived syntagmatic responses

Monolingual children

Bilingual children

Mean (SD)

0.46 (1.1)

0.33 (0.92)

Range

0–2.8

0–2.8

Mean (SD)

2.9 (6.6)

1.2 (2.3)

Range

0–30.6

0–8.3

Among the monolinguals only 2.8% (SD = 3.8) of all responses and among the bilinguals 1.8% (SD = 2.2) of responses were morphologically derived. The average percentage and range of morphologically derived responses (when separated into paradigmatic and syntagmatic) is shown in Table 7.4. As can be seen, very few responses were paradigmatic as well morphologically derived. Indeed, only four of the monolingual children and three of the bilingual children gave such responses, and these were very few. However, the use of syntagmatic responses that were also morphologically derived was slightly more frequent: nine of the monolingual children and five of the bilingual children gave such responses. These children gave between one and three syntagmatic responses that were also morphologically derived, except one of the monolingual children (at age 6;8) who gave such responses for as many as 11 of the prompt words (almost 30% of all responses).

Sound-based responses

As shown in Figure 7.1, sound-based responses (phonological and morphologically inflected) were, on average, given as a response for 10% of the prompt words among both groups of children. Although these fi ndings are in line with the expected decrease in use of the sound-based response among school children (Cronin, 2002), some individual variability regarding these responses was found among children in both groups. As expected, only a few of the children used such responses: only nine of the monolingual children and six of the bilingual children provided phonological responses, and they all provided up to four such responses. However, one bilingual child (at age 8;3) provided such responses for 17 of the 36 presented words (47.3%). Similarly, only six monolingual and seven bilingual children provided morphologically inflected responses, and while most monolingual children provided more than seven such responses, all bilingual children gave up to three such responses, except for one child (at age 6;5) who gave ten such responses. Importantly, although both phonological and morphologically inflected responses share similarity in sound with the prompt words, as expected, no correlation (using a Pearson r-test) between these

Lexical-semantic Organization in Monolingual and Bilingual Hebrew Speaking Children

177

responses was found (monolinguals: r = −0.033, p = 0.879; bilinguals: r = −0.137, p = 0.600). More information on the phonology-based results

As mentioned above, only 15 out of the 41 participating children (40%) gave phonological responses. All of these children, except for one, gave no more than four phonological responses out of the 36 possible responses (11%). These children, with no difference as to whether monolingual or bilingual, gave mainly responses of shared opening syllables (e.g. xalonxalav ‫ חלב‬-‫חלון‬, ‘window-milk’) or responses of shared closing syllables, which are rhymes (e.g. xultsa-bejtsa ‫ביצה‬-‫חולצה‬, ‘shirt-egg’). The exception is the case of one child, a 7;3-year-old monolingual girl. Her naming score was as expected for her age (95%). Only one of her naming mistakes was phonological and all the four other mistakes were semantic. However, this girl gave 17 phonological responses and two morphologically inclined responses. Three of the phonological responses were rhymes, and the other responses were mainly with the same opening syllable as the prompt word. The target words for which phonological responses were given were divided almost equally between nouns (eight) and adjectives (seven), although these targets enable a natural paradigmatic response (through semantic category, synonym or antonym), and only two of the phonological responses were provided following verbs. Indeed, her other 19 responses which were semantic, were mainly syntagmatic responses (14) and only five were paradigmatic. Thus, the extensive use of phonological responses by this girl does seem to go hand in hand with poor use of paradigmatic responses as expected from the ‘syntagmatic-paradigmatic’ shift. Correlation between age and pattern of associative responses

The above-mentioned results regarding no correlations with age were confi rmed using the Pearson correlation coefficient among each group separately. In addition, no correlation with age was found with the other type of responses – phonological, morphologically inflected and morphologically derived (Table 7.5). This was also found when looking at all 41 children from both groups together (paradigmatic responses: r = 0.103, p  = 0.522; syntagmatic responses: r = −0.041, p = 0.799; phonological responses: r = −0.061, p = 0.705; morphologically derived responses: r = −0.064, p = 0.693). The finding of no change with age in amount of paradigmatic responses is contradictory to the known phenomenon of syntagmatic-paradigmatic shift, according to which more paradigmatic responses are expected with older age. However, the finding of no change with age in sound-based responses is in accordance with previous fi ndings showing that these responses decrease towards age 6 years and, as mentioned above, the frequency of use of these responses was quite marginal.

178

An Anthology of Bilingual Child Phonology

Table 7.5 Correlation (Pearson’s r) between age and percentage of use of the various types of responses among the two groups of children Paradigmatic Syntagmatic Morphologically Phonological Morphologically derived inflected Monolingual r = −0.076

Bilingual

r = 0.167

r = − 0.034

r = − 0.103

r = − 0.141

p = 0.722

p = 0.436

p = 0.873

p = 0.634

p = 0.511

r = 0.316

r = 0.167

r = 0.249

r = 0.012

r = 0.056

p = 0.216

p = 0.436

p = 0.336

p = 0.965

p = 0.831

Correlation between naming proficiency and pattern of associative responses

Since no correlation between type of response and age was found, correlations with naming proficiency were examined among each of the groups (Table 7.6). Among the monolingual children, no response type was significantly correlated with the naming score. However, among the bilingual children, percentage of use of morphologically inflected responses was negatively correlated with their naming score. These fi ndings suggest that semantic relations as expressed by a word association task are independent from lexical retrieval abilities expressed by a naming task. However, bilingual children, whose Hebrew naming abilities are lower, make use of the less mature morphologically inflected responses. Discussion

The current study aimed at shedding light on the lexical-semantic organization of 6- to 9-year-old monolingual and bilingual Hebrew speaking children using a word association task. Our goals were to compare the development of semantic responses, i.e. paradigmatic versus syntagmatic organization, while investigating the specific influence of Hebrew morphological characteristics on this process, as well as inspecting formbased responses, i.e. phonological and morphologically inflected responses. We further aimed at examining whether age and general naming abilities have an influence on the pattern of associative responses among children of both groups. Table 7.6 Correlation (Pearson’s r) between naming test scores and percentage of use of the various types of responses among the two groups of children Paradigmatic

Syntagmatic

Phonological

Morphologically inflected

Monolingual

r = − 0.355 p = 0.162

r = 0.290 p = 0.258

r = 0.035 p = 0.894

r = 0.211 p = 0.416

Bilingual

r = 0.051 p = 0.813

r = 0.170 p = 0.428

r = 0.020 p = 0.925

r = − 0.503* p = 0.01

Lexical-semantic Organization in Monolingual and Bilingual Hebrew Speaking Children

179

Semantic responses: Syntagmatic versus paradigmatic

The first interesting and unexpected result concerns the well-known ‘syntagmatic-paradigmatic shift’ relating to differences in the ratio of responses between younger and older children in word association tasks. As expected, we found that, among all children except one, most responses were related to prompt words through semantic relations and only a small proportion of responses were related to the prompt word through soundbased relations. However, among both monolingual and bilingual groups of children, the semantic responses were divided evenly into syntagmatic and paradigmatic. Thus, no advantage of paradigmatic over syntagmatic responses was detected. Moreover, no gradual change with age towards such an advantage was observed. These findings are not in full accordance with the assumption that the shift from syntagmatic to paradigmatic associations represents a conceptual change that takes place during the early school years (Anglin, 1985; Nelson, 1977). In the next paragraphs, some possible general explanations and some Hebrew-specific explanations of these seemingly unexpected fi ndings will be discussed. First, it is important to mention that although our fi ndings on a group level are not in line with the expected syntagmatic-paradigmatic shift, large variability was detected among the participants. In fact, half of the children, whether monolingual or bilingual and irrespective of their age, did show a paradigmatic over syntagmatic advantage. In addition, the observed pattern of the responses in the current study that show no change with age in paradigmatic responses could be related to a more gradual shift that occurs. In a study conducted among older Hebrew speaking children in the 6th and 10th grades, a gradual change in responses was observed (Seroussi, 2011). More specifically, as a response to noun prompt words (some familiar/frequent and other less familiar), almost three-quarters of the associations given by the 6th graders were paradigmatic, but this proportion increased to 85% in the 10th grade, going up to 87% in the adult group. Seroussi claimed that this fi nding ‘challenges the notion of a dramatic shift away from syntagmatic responses, suggesting a less radical, more gradual developmental pattern’ (Seroussi, 2011: 169). She further claims that the fi ndings provide ‘strong evidence that linguistic (and conceptual) development is by no means completed by age seven, and that the best way to describe this development is as a continuum rather than a shift’ (Seroussi, 2011: 169). The pattern of responses by the girl aged 7;3, who gave almost 50% phonological responses and very few paradigmatic ones, supports the claim for a possibly more gradual process than that described in the literature (Cronin, 2002). Interestingly, syntagmatic versus paradigmatic organization is a way of classification that relates to nouns as well as other grammatical word classes, i.e. verbs, adjectives, adverbs, pronouns and other function words (Cruse, 1986). Another well-established analogous developmental

180

An Anthology of Bilingual Child Phonology

phenomenon is the ‘thematic-taxonomic shift’, which relates only to objects, documented by the preference of younger children to select a thematically rather than a taxonomically related noun in various tasks (Brown & Berko, 1960; Cronin, 2002). However, some recent studies have challenged the equating of ‘thematic’ as ‘mature’ and that of ‘taxonomic’ as ‘immature’. On the one hand, there are reports that toddlers were able to attend to taxonomic relations in suitable experimental conditions (Waxman & Namy, 1997). On the other hand, reports concerning adults show that they could successfully adopt both types of relations, and a ‘cross-classification’ ability to navigate flexibly between words across various types of relations was described as a hallmark of mature linguisticconceptual proficiency (Liu et al., 2001). These fi ndings are in line with studies that claim that both syntagmatic and paradigmatic relations are essential for organizing lexical items among monolingual children (Walsh et al., 1993; Waxman & Namy, 1997) and bilingual children (Sheng et al., 2006). Morphologically derived responses

The focus on Hebrew relates to the growing body of evidence of typological considerations in language learning. More specifically, for Hebrew, morphological combination is an obligatory component of many surface words (Ravid, 1990). However, in the current study, in spite of this unique characteristic of the Hebrew language, only a small percentage of the associative responses had a morphologically derived relation. This might be explained by few factors. First, our participants were in the fi rst years of schooling, and thus at the initial stages of building their more mature lexicon which is related to their expanding reading experience (Berman, 2003). Thus, at this stage, concrete semantic relations probably take over the more abstract morphologically derived ones. In addition, previous studies that reported on the priority of morphological relations over semantic ones were mainly based on priming tasks in which pairs of words were presented (Deutsch et al., 1998; Velan & Frost, 2011). However, in the current study, a single prompt word and not a pair of words was presented in each trial. It could be the case that in such conditions semantic relations have priority over morphological ones. An additional explanation regarding the small percentage of morphologically derived responses could be related to the specific list of words presented in the current study. The words were chosen according to the criteria of being familiar and frequently used among young children who are not older than 6 years. It could be the case that words that are frequent, familiar and related to everyday situations are apt to provoke semantic responses. Such conclusions were extracted by Seroussi (2011), who presented both frequent as well as rare Hebrew words. She found that there were significantly more morphologically derived responses to

Lexical-semantic Organization in Monolingual and Bilingual Hebrew Speaking Children

181

low-frequency than to high-frequency nouns. She concluded that lowfrequency words would elicit more ‘structural associations’, while highfrequency words would elicit more semantic associations. Another factor related to the specific characteristic of the prompt words is their morphological characteristics. As mentioned in the Method section, two-thirds of the presented nouns in the current study were with no semantically related identified root. Consequently, for the nouns in the study, a morphologically derived response was often not possible. However, an overall small rate of morphologically derived responses was observed, even in reaction to words that do have an identified semantically related root. This fi nding might be interpreted in relation to the ongoing debate regarding the exact manner in which morphological information is realized in the lexicon of Hebrew speakers (Velan et al., 2005). Several theoretical accounts of Hebrew morphology question the necessity of the consonantal root as a separate constituent and explain a range of morphological phenomena by appealing only to stem-based representations (e.g. Bat-El, 1994; Ussishkin, 2005). These researchers propose that both structured (root-based) and unstructured (stem-based) representations exist within the Hebrew lexical system and even in the verbal system (Farhy et al., 2018). Bilingual children

The bilingual children in the current study exhibited the welldocumented impact of reduced exposure to the societal language on their productive vocabulary knowledge (Farnia & Geva, 2011; MancillaMartinez & Lesaux, 2011). Nevertheless, they revealed a similar pattern of associative responses to those among monolingual children, as was reflected in the proportion of semantic responses (paradigmatic as well as syntagmatic) and in sound-based responses (phonological and morphologically inflected). Importantly, the proportional quantity of all types of response was not found to be in correlation with the participants’ scores in the naming task, with the exception of the morphologically inflected responses that were negatively correlated with naming scores in Hebrew. Thus, bilingual children whose naming proficiency in Hebrew is low still have sufficient grammatical knowledge to enable them to respond with a morphologically inflected response. We also found that the bilingual children gave on average as many morphologically derived responses as the monolingual children. Usage-based models theorize that the acquisition of linguistic knowledge is input driven and is influenced by the frequency with which children encounter linguistic forms in the language they hear (Tomasello, 2003). These models suggest that two aspects of frequency are important: token frequency (the amount of exposure to a specific language unit) and type frequency (the exposure to a specific pattern in a given language). As

182

An Anthology of Bilingual Child Phonology

such, token frequency is a strong correlate driving vocabulary knowledge, whereas type frequency influences the acquisition of regularities including morphological structures (Bybee, 2007). Repeated exposure to and use of individual tokens leads to the accumulation of a ‘critical mass’ of exposure to their shared type (Nicoladis et al., 2007). This critical mass enables the child to generalize about morphological patterns by making associations among related words from his lexicon (Tomasello, 2003). Bilingual children are exposed to each language for less time than monolinguals on average, and they spend less time speaking any one language relative to respective monolinguals (Gathercole & Hoff, 2007; Gollan et al., 2008; Pearson et al., 1997). Thus, in agreement with the usage-based model, bilingual children may have, as demonstrated in the current study, on average less experience with the use of words in each of their languages, and therefore might experience some retrieval challenges when confronted with naming tasks. Nevertheless, although not tested directly, a ‘critical mass’ of exposure did enable them to use morphologically inflected and derived responses to the same extent as their monolingual peers. Conclusion and Future Directions

Overall, large variability can be observed among Hebrew speaking children in their fi rst years of schooling when confronted with a word association task. Although few of the children, whether monolingual or bilingual, provided many sound-based responses, phonological or morphologically inflected, most of them provided mainly semantic responses. Few of the children, regardless of being monolingual or bilingual, provided more paradigmatic than syntagmatic responses, but as a group the average distribution between these two types of responses was equal. The presence of morphologically derived responses was quite minimal on the group level, but few of the children used it as a strategy, and this led them to an advantage of syntagmatic responses over paradigmatic. Future studies, among older children, with the same list of words, and with a different list of words which includes less frequent words and more nouns with an identified root, would enable us to understand in more depth the patterns of responses among Hebrew speakers and their unique characteristics. References Abutbul-Oz, H., Armon-Lotem, S. and Walters, J. (2012) Bilingual Parents Questionnaire (BIPAQ). Ramat Gan: Bar-Ilan University. Albrechtsen, D., Haastrup, K. and Henriksen, B. (2008) Vocabulary and Writing in a First and Second Language: Processes and Development. New York: Palgrave Macmillan. Anglin, J.M. (1985) The child’s expressible knowledge of word concepts. In K.E. Nelson (ed.) Children’s Language (Vol. 5, pp. 77–127). Hillsdale, NJ: Erlbaum. Anglin, J.M. (1993) Vocabulary development: A morphological analysis. Monographs of the Society for Research in Child Development 58 (10), 1–166.

Lexical-semantic Organization in Monolingual and Bilingual Hebrew Speaking Children

183

Avivi Ben-Zvi, G. (2010) Lexical and morphological development in grade school. Doctoral dissertation, Tel Aviv University. [In Hebrew] Bat-El, O. (1994) Stem modification and cluster transfer in Modern Hebrew. Natural Language and Linguistic Theory 12, 571–596. Berent, I., Vaknin, V. and Marcus, G. (2007) Roots, stems, and the universality of lexical representations: Evidence from Hebrew. Cognition 104 (2), 254–286. Berman, R.A. (2000) Children’s innovative verbs vs nouns: Structured elicitations and spontaneous coinages. In L. Menn and N. Bernstein-Ratner (eds) Methods in Studying Language Production (pp. 69–93). Mahwah, NJ: Lawrence Erlbaum. Berman, R.A. (2003) Children’s lexical innovations: Developmental perspectives on Hebrew verb-structure. In J. Shimron (ed.) Language Processing and Language Acquisition in a Root-based Morphology (pp. 243–291). Amsterdam: John Benjamins. Berman, R.A. (2004) Between emergence and mastery: The long developmental route of language acquisition. In R.A. Berman (ed.) Language Development Across Childhood and Adolescence: Psycholinguistic and Crosslinguistic Perspectives (pp. 9–34). Trends in Language Acquisition Research (TILAR) 3. Amsterdam: John Benjamins. Biran, M. and Friedmann, N. (2004) SHEMESH: Naming a Hundred Objects. Tel Aviv: Tel Aviv University. Biran, M. and Friedmann, N. (2005) From phonological paraphasias to the structure of the phonological output lexicon. Language and Cognitive Processes 20, 589–616. Brown, R. and Berko, J. (1960) Word association and the acquisition of grammar. Child Development 31, 1–14. Bybee, J. (2007) Frequency of Use and the Organization of Language. Oxford: Oxford University Press. Cilibrasi, L. (2016) Word position eff ects in speech perception. PhD dissertation, University of Reading. Cole, M. and Means, B. (1986) Comparative Studies of How People Think. San Diego, CA: University of California Press. Cronin, V.S. (2002) The syntagmatic-paradigmatic shift and reading development. Journal of Child Language 29 (1), 189–204. Cruse, D.A. (1986) Lexical Semantics. Cambridge: Cambridge University Press. Danovich, J.H. and Keil, F.C. (2004) Should you ask a fi sherman or a biologist? Developmental shifts in ways of clustering knowledge. Child Development 75, 918–931. Deutsch, A., Frost, R. and Forster, K.I. (1998) Verbs and nouns are organized and accessed differently in the mental lexicon: Evidence from Hebrew. Journal of Experimental Psychology: Learning, Memory, and Cognition 24, 1238–1255. Dóczi, B. and Kormos, J. (2016) Longitudinal Developments in Vocabulary Knowledge and Lexical Organization. New York: Oxford University Press. Entwisle, D.R. (1966) The Word Association of Young Children. Baltimore, MD: Johns Hopkins University Press. Ervin, S.M. (1961) Changes with age in the verbal determinants of word association. American Journal of Psychology 74, 361–372. Farhy, Y., Veríssimo, J. and Clahsen, H. (2018) Universal and particular in morphological processing: Evidence from Hebrew. Quarterly Journal of Experimental Psychology 71, 1125–1133. Farnia, F. and Geva, E. (2011) Cognitive correlates of vocabulary growth in English language learners. Applied Psycholinguistics 32, 711–738. Fitzpatrick, T. (2006) Habits and rabbits: Word associations and the L2 lexicon, EUROSLA Yearbook 6, 121–145. Fitzpatrick, T. (2007) Word association patterns: Unpacking the assumptions. International Journal of Applied Linguistics 17, 319–331.

184

An Anthology of Bilingual Child Phonology

Fitzpatrick, T. (2009) Word association profi les in fi rst and second language: Puzzles and problems. In T. Fitzpatrick and A. Barfi eld (eds) Lexical Processing in Second Language Learners (pp. 38–52). Bristol: Multilingual Matters. Fitzpatrick, T. and Munby, I. (2013) Knowledge of word associations. In J. Milton and T. Fitzpatrick (ed.) Dimensions of Vocabulary Knowledge (pp. 92–105). Basingstoke: Palgrave Macmillan. Fitzpatrick, T., Playfoot, D., Wray, A. and Wright, M.J. (2015) Establishing the reliability of word association data for investigating individual and group differences. Applied Linguistics 36, 23–50. Francis, W.N. and Kucera, H. (1982) Frequency Analysis of English Usage: Lexicon and Grammar. Boston, MA: Houghton Miffl in. Frost, R., Forster, K.I. and Deutsch, A. (1997) What can we learn from the morphology of Hebrew: A masked priming investigation of morphological representation. Journal of Experimental Psychology: Learning, Memory, and Cognition 23, 829–856. Gathercole, V.C.M. and Hoff, E. (2007) Input and the acquisition of language: Three questions. In E. Hoff and M. Shatz (eds) The Handbook of Language Development (pp. 107–127). New York: Blackwell. Gollan, T.H., Montoya, R.I., Cera, C. and Sandoval, T.C. (2008) More use almost always means a smaller frequency effect: Aging, bilingualism, and the weaker links hypothesis. Journal of Memory and Language 58, 787–814. Gray, S., Plante, E., Vance, R. and Henrichsen, M. (1999) The diagnostic accuracy of four vocabulary tests administered to preschool-age children. Language, Speech, and Hearing Services in Schools 30 (2), 196–206. Guralnik, E. (1995) A Screening Test for Hebrew Speaking Preschoolers. Tel Aviv: Matan. Hammarström, H., Forkel, R. and Haspelmath, M. (2017) ‘Semitic’ Glottolog 3.0. Jena: Max Planck Institute for the Science of Human History. Holmström, K., Salameh, E.K., Nettelbladt, U. and Dahlgren Sandberg, A. (2015) A descriptive study of lexical organization in bilingual children with language impairment: Developmental changes. International Journal of Speech-Language Pathology 18 (2), 178–189. Jean, M. and Geva, E. (2009) The development of vocabulary in English as a second language children and its role in predicting word recognition ability. Applied Psycholinguistics 30, 153–185. Justice, L.M. (2006) Evidence-based practice, response to intervention, and the prevention of reading difficulties. Language, Speech, and Hearing Services in Schools 37, 284–297. Katzenberger, I. and Meilijson, S. (2014) Hebrew language assessment measure for preschool children: A comparison between typically developing children and children with specific language impairment. Language Testing 31, 19–38. Kieffer, M.J. and Lesaux, N.K. (2012) Knowledge of words, knowledge about words: Dimensions of vocabulary in fi rst and second language learners in sixth grade. Reading and Writing 25, 347–373. Kohnert, K. (2010) Bilingual children with primary language impairment: Issues, evidence and implications for clinical actions. Journal of Communication Disorders 43, 456–473. Kohnert, K. (2013) Language Disorders in Bilingual Children and Adults. San Diego, CA: Plural Press. Liu, J., Golinkoff, R.M. and Sak, K. (2001) One cow does not an animal make: Young children can extend novel words at the superordinate level. Child Development 72 (6), 1674–1694. Łuniewska, M., Haman, E., Armon-Lotem, S. et al. (2016) Ratings of age of acquisition of 299 words across 25 languages: Is there a cross-linguistic order of words? Behavior Research Methods 48, 1154–1177.

Lexical-semantic Organization in Monolingual and Bilingual Hebrew Speaking Children

185

Mancilla-Martinez, J. and Lesaux, N.K. (2011) Early home language use and later vocabulary development. Journal of Educational Psychology 103 (3), 535–546. Marchman, V. and Bates, E. (1994) Continuity in lexical and morphological development: A test of the critical mass hypothesis. Journal of Child Language 21, 339–366. Meara, P. (2009) Connected Words: Word Associations and Second Language Vocabulary Acquisition. Amsterdam: John Benjamins. Meara, P. and Wolter, B. (2004) V_Links: Beyond vocabulary depth. In D. Albrechtsen, K. Haastrup and B. Henriksen (eds) Angles on the English-speaking World 4 (pp. 85–96). Copenhagen: Museum Tusculanum Press. Nelson, K. (1977) The syntagmatic-paradigmatic shift revisited: A review of research and theory. Psychological Bulletin 84 (1), 93–116. Nicoladis, E., Palmer, A. and Marentette, P. (2007) The role of type and token frequency in using past tense morphemes correctly. Developmental Science 10, 237–254. Nir-Sagiv, B., Bar-Ilan, L. and Berman, R.A. (2008) Vocabulary development across adolescence: Text-based analyses. In A. Stavans and I. Kupferberg (eds) Studies in Language and Language Education: Essays in Honor of Elite Olshtain (pp. 47–74). Jerusalem: Magnes. Novogrodsky, R. and Kreiser, V. (2015) What can errors tell us about specific language impairment? Semantic and morphological cuing in a sentence completion task. Clinical Linguistics and Phonetics 29 (11), 812–825. Ordonez, C.L., Carlo, M.S., Snow, C.E. and McLaughlin, B. (2002) Depth and breadth of vocabulary in two languages: Which vocabulary skills transfer? Journal of Educational Psychology 94, 719–728. Pearson, B.Z., Fernandez, S.C., Lewedeg, V. and Oller, D.K. (1997) The relation of input factors to lexical learning by bilingual infants. Applied Psycholinguistics 18, 41–58. Post, B., Marslen-Wilson, W.D., Randall, B. and Tyler, L.K. (2008) The processing of English regular inflections: Phonological cues to morphological structure. Cognition 109 (1), 1–17. Prior, A. (2004) Exploring the nature of associations: Semantic factors in the formation of word associations. Doctoral dissertation, Hebrew University Jerusalem. Ravid, D. (1990) Internal structure constraints on new-word formation devices in Modern Hebrew. Folia Linguistica 24, 289–346. Ravid, D. (2003) A developmental perspective on root perception in Hebrew and Palestinian Arabic. In Y. Shimron (ed.) The Processing and Acquisition of Rootbased Morphology (pp. 293–319). Amsterdam: John Benjamins. Ravid, D. and Bar-On, A. (2005) Manipulating written Hebrew roots across development: The interface of semantic, phonological and orthographic factors. Reading and Writing 18, 231–256. Ravid, D., Avivi-Ben Zvi, G. and Levie, R. (1999) Derivational morphology in SLI children: Structure and semantics of Hebrew nouns. In M. Perkins and S. Howard (eds) New Directions in Language Development and Disorders (pp. 39–49). New York: Plenum Press. Read, J. (2000) Assessing Vocabulary. Cambridge: Cambridge University Press. Seroussi, B. (2011) The morphology-semantics interface in the mental lexicon: The case of Hebrew. PhD thesis, Tel Aviv University. Sheng, L. and McGregor, K.K. (2010) Lexical-semantic organization in children with specific language impairment. Journal of Speech, Language, and Hearing Research 53, 146–159. Sheng, L., McGregor, K.K. and Marian, V. (2006) Lexical-semantic organization in bilingual children: Evidence from a repeated word association task. Journal of Speech, Language, and Hearing Research 49 (3), 572–587. Sheng, L., Peña, E.D., Bedore, L.M. and Fiestas, C.E. (2012) Semantic deficits in SpanishEnglish bilingual children with language impairment. Journal of Speech, Language, and Hearing Research 55, 1–15.

186

An Anthology of Bilingual Child Phonology

Tomasello, M. (2003) Constructing a Language: A Usage-based Theory of Language Acquisition. Cambridge, MA: Harvard University Press. Ussishkin, A. (2005) A fi xed prosodic theory of nonconcatenative templatic morphology. Natural Language and Linguistic Theory 23, 169–218. Velan, H. and Frost, R. (2011) Words with and without internal structure: What determines the nature of orthographic and morphological processing? Cognition 118 (2), 141–156. Velan, H., Frost, R., Deutsch, A. and Plaut, D.C. (2005) The processing of root morphemes in Hebrew: Contrasting localist and distributed accounts. Language and Cognitive Processes 20, 169–206. Verhallen, M. and Schoonen, R. (1993) Lexical knowledge of monolingual and bilingual children. Applied Linguistics 14, 344–363. Verhallen, M. and Schoonen, R. (1998) Lexical knowledge in L1 and L2 of third and fi fth graders. Applied Linguistics 19 (4), 452–470. Walsh, M.E., Richardson, K. and Faulkner, D. (1993) Perceptual, thematic and taxonomic relations in children’s mental representations: Responses to triads. European Journal of Psychology of Education 8 (1), 85–102. Waxman, S.R. and Namy, L. (1997) Challenging the notion of a thematic preference in young children. Developmental Psychology 33 (3), 555–567. Wechsler, D. (2003) Wechsler Intelligence Scale for Children (WISC-IV), Administration and Scoring Manual. San Antonio, TX: Psychological Corporation. Wilks, C. (2009) Tangled webs: Complications in the exploration of L2 lexical networks. In T. Fitzpatrick and A. Barfi eld (eds) Lexical Processing in Second Language Learners (pp. 25–37). Bristol: Multilingual Matters.

8 The Production of Marked Arabic Consonants by Arabic-English Bilingual Children Living in Canada Anwar Alkhudidi, Yasmeen Hakooz, Madeline Walker, Ryan A. Stevenson and Yasaman Rafat

Introduction

This study aims to investigate whether early bilingualism interferes with the linguistic development of Arabic as a heritage language during childhood. Specifically, we aim to determine whether Arabic-English speaking children who live in Canada, whose home language is Arabic and who are considered heritage speakers, are able to accurately produce some of the complex (marked) Arabic sounds. We will be testing the Speech Learning Model (Flege, 1995), which states that first language (L1) and second language (L2) sounds may merge in adult bilinguals. Specifically, we want to see whether, as a result of contact with English, early bilingual children may also exhibit some merged categories or influence from English. The phonological contrasts in this study consist of the following: (a) the emphatic-plain contrasts /t/-/t/ (e.g. /ta:ba/ ‘repented’ versus /taba/ ‘healed’), /d/-/d/ (e.g. /da:l/ ‘stray’ versus /dal/ ‘indexical’), /s/-/s/ (e.g. /sa:r/ ‘became’ versus /sa:r/ ‘walked’); (b) the pharyngeal-glottal contrast /ħ/-/h/ (e.g. /ħaram/ ‘sanctuary’ versus /haram/ ‘pyramid’); and (c) the uvular-velar contrast /q/-/k/ (e.g. /qalb/ ‘heart’ versus /kalb/ ‘dog’) (Al-Ani, 1970). These contrasts lend themselves well to a study on bilingual children because only /t, d, s, h, k/, which are less complex (unmarked) sounds, exist in English. Heritage speakers/early bilinguals

The participants in this study were heritage speakers/early bilinguals. There is currently no consensus on what constitutes a heritage speaker. The defi nition of heritage speakers and heritage language started to be 187

188

An Anthology of Bilingual Child Phonology

developed during the 1970s in Canada, but it gained the attention it deserves only 20 years later. Regardless of the defi nition chosen, it is important to remember that heritage speakers can have a first and a second language (order of acquisition) or may have been exposed to two or more languages from birth. Moreover, they may have a primary and a secondary language (functional dimension), or a majority and a minority language (sociopolitical dimension) (Montrul, 2012). According to Draper and Hicks (2000), a heritage speaker is someone who has been exposed to a non-English language outside the formal educational schooling setting. They also included speakers who grow up speaking a different language at home and speakers with in-depth exposure to another language. Nevertheless, this is a very broad description of heritage speakers. Similarly, Campbell and Peyton’s (1998) defi nition of a heritage speaker as an individual who speaks a fi rst non-English language at home or who is born in a different country is widely accepted. However, they seem to have forgotten an important connection made by Valdés (2001) regarding the individual’s personal connection with the language. Valdés (2001) suggests that heritage speakers are individuals who have a historical and/or personal connection to a language that is not normally taught at school. She adds that heritage speakers are individuals who grow up speaking a non-English language at home, perform at any level of proficiency and use any level of bilingualism. Montrul (2012) proposes that heritage speakers are early bilinguals, because of the exposure they have had to both the heritage and the majority language. She states that, as bilinguals, heritage speakers can be simultaneous when they grow up speaking both languages or sequential when they learn their second language after the age of 5 or 6 years. She suggests that, regardless of when a heritage speaker starts learning both languages, by the time they enter adulthood the heritage language is their weaker one. More recently, a new debate developed regarding whether heritage speakers should be considered native speakers (Kupisch & Rothman, 2018; Rothman, 2006, 2009; Rothman & Treffers-Daller, 2014). Rothman’s (2006, 2009) defi nition of heritage speakers starts where Montrul’s ended: heritage speakers are bilinguals, but not all bilinguals are heritage speakers. According to his studies (e.g. Rothman, 2006), heritage speakers are on the continuum between monolingual and bilingual, as they have been naturally exposed to a multilingual environment since childhood, if not even since birth, and have competence in both languages. Rothman (2009) suggests that the essential condition for heritage speakers is the naturalistic environment in which they need to learn their languages. Agreeing with Valdés (2001), Rothman (2006) argues that heritage speakers can present different levels of proficiency depending on personal and social factors. Moreover, Kupisch and Rothman (2018) divide heritage speakers into two categories: (i) early bilinguals who have been exposed to the language in a naturalistic environment and have strong personal

The Production of Marked Arabic Consonants by Arabic-English Bilingual Children

189

connection to the language via family, culture or intrinsic motivation; and (ii) late bilinguals who learned the language as adults and, therefore, their acquisition lacks the naturalistic environment. In their opinion, both the former and the latter can be considered as native speakers of that language (Kupisch & Rothman, 2018; Rothman & Treffers-Daller, 2014), regardless of the level of proficiency or the hypothesized incomplete acquisition that Valdés (2000) and Benmamoun et al. (2013) have proposed. The target population of this study is heritage speakers of Arabic, specifically children of Arab immigrants who established a community in Canada. Because they acquired English at a very young age, we will refer to them as early bilinguals. Shorrab (2009) investigated the language and lifestyle patterns of the Arabic bilingual community in North America, namely Buffalo, NY. He found that, since there has been an increase in immigration in this community, many children are being brought up in homes with two languages, one being the host language, English, and the other being the heritage language, Arabic. An acculturation process has taken place as the Arabic-English speaking community integrates into an English speaking society (Shorrab, 2009). A number of Arabic varieties are spoken in this community in Canada, which may include Lebanese, Palestinian, Jordanian, Syrian, Iraqi and Egyptian. The English-Arabic speaking community places a large emphasis on learning Arabic, mainly for religious reasons. Parents often choose to enroll their children in Sunday schools to learn Standard Arabic. Others enroll them in private schools, mostly Islamic schools, rather than the general public schools. In these private or Islamic schools, Arabic courses are taught more than once a week. Another characteristic of this community is that children have strong ties with relatives, especially grandparents, placing a constant demand on them to learn Arabic in order to communicate with those relatives that live outside Canada or do not speak English. Marked sounds in Arabic

In the previous section we provided a description of heritage speakers and mentioned that our participants consisted of Arabic-English early bilinguals who are heritage speakers of Arabic. In this section we will provide a description of the target Arabic sounds that we predict children in this study will have difficulty producing. Most Arabic varieties have features of emphatic, pharyngeal and uvular sounds (Al-Adam, 2015; Al-Ani, 1970). The emphatic sounds are also known as ‘pharyngealized’ consonants (e.g. /s, d, t, ð/ contrast with /t, d, s, ð/). Their production involves two types of articulations: primary articulation involving the dental/alveolar region in the vocal tract, and a secondary articulation where the back of the tongue is retracted towards the pharynx (Al-Ani, 1970). The pharyngeal /ħ/, contrasting with the glottal /h/, is a voiceless fricative consonant produced by forming a constriction with the dorsum

190

An Anthology of Bilingual Child Phonology

of the tongue against the posterior wall of the pharynx. The uvular /q/ is a voiceless stop articulated when the tongue makes contact with the uvula and contrasts with the voiceless velar stop /k/ (Al-Ani, 1970). A major acoustic correlate of the emphatic sounds, the uvular /q/ and the pharyngeal /ħ/, is the lowering of the second formant frequency (F2) in adjacent vowels. F2 correlates with the place of articulation in the oral cavity: the further back the articulation, the lower the F2 value. This lowering is reported to be 450–500 Hz for emphatics (/s, d, t/), 250–400 Hz for the pharyngeal /ħ/ and 400–600 Hz for the uvular /q/ (Al-Adam, 2015; Al-Masri & Jongman, 2004; Al-Ani, 1970; Alwan, 1986; Card, 1983; Norlin, 1987). Moreover, it has been reported that the F2 lowering effect is spread to all vowels in the word (Al-Ani, 1970; Al-Masri & Jongman, 2004), although first formant frequency (F1) and third formant frequency (F3) slightly increase in adjacent vowels. However, this study is only concerned with F2 values. Other acoustic cues (not investigated here) include shorter voice onset time (VOT) values and lower spectral means for emphatic consonants (Al-Adam, 2015; Jongman et  al., 2007). A number of studies have also reported that vowel quality interacts with the degree of F2 lowering (e.g. Al-Masri & Jongman, 2004; Jongman et al., 2007, 2011), where a higher degree of F2 lowering is observed in /æ/ than /i/ followed by /u/. Segmental acquisition in Arabic child phonology

Although most of the literature on early phonological development is conducted on the English language (e.g. Dyson & Paden, 1983; Khan & Lewis, 1986; Stoel-Gammon, 1985), there have been some studies on child phonology in Arabic (Amayreh & Dyson, 1998, 2000; Ammar & Morsi, 2006; Ayyad, 2011). Amayreh and Dyson (1998) examined the acquisition of Jordanian standard Arabic consonants and the results showed that most Arabic consonants are acquired during early (2;0–3;10) and intermediate (4;0–6;4) periods, while emphatic consonants (/s, d, t, ð/) are learned after age 6;4. Moreover, results showed that there was a higher degree of accuracy in production in consonants that are found in intervocalic positions rather than in word-initial and word-fi nal positions. This study showed support for universal phoneme acquisition; however, some differences arose due to language-specific effects. Universal phoneme acquisition was proposed by Jakobson (1968), suggesting that certain milestones are met before others. For instance, stops are typically acquired before fricatives in all languages. Amayreh and Dyson (1998) reported that all non-emphatic sounds are acquired before their emphatic counterparts, suggesting that there is universality in the acquisition of these groups of sounds. Another study by Ammar and Morsi (2006) reported that children acquired all Egyptian Arabic consonants by age 4;1 with the exception of /d, z, ɣ/, which are acquired between ages 4;1 and 5;0.

The Production of Marked Arabic Consonants by Arabic-English Bilingual Children

191

A study conducted on the phonological development of 80 Kuwaiti Arabic speaking children between the ages of 3;10 and 5;2 (Ayyad, 2011) revealed that in terms of segmental acquisition, children whose ages ranged between 3;10 and 4;5 acquired the following consonants, /b, t, d, k, ɡ, q, ʔ, m, n, r, f, .ˤ, s, x, ħ, ʧ, h, j, l, w/, whereas children aged between 4;6 and 5;2 acquired /t, d, f, ʃ, ɣ/. Moreover, while the younger group acquired the former set of sounds including the emphatic /s/, the older group did not meet the 90% criterion for the production of the following phonemes, /θ, s, z, ʕ, ʤ/, where the plain /s/ is included. This fi nding is interesting given the fact that the production of the emphatic /s/ requires a secondary articulation, where the root of the tongue is retracted towards the back wall of the pharynx making its production more complex than the basic /s/. The author attributed the reason to factors such as stress pattern and phonological salience which may have an effect on the production of emphatic consonants in some environments. Most of the above-mentioned studies indicate that Arabic speaking children, regardless of the dialect (e.g. Jordanian, Kuwaiti, Egyptian) they are exposed to, acquire all or some of the complex Arabic sounds such as the emphatics at a later age, usually around 5–6 years, in contrast with the plain consonants which are acquired around 4–5 years. This delay in production may be attributed to the difficult articulatory processes needed in order to produce complex Arabic sounds. Although there have been a number of studies on the L1 acquisition of Arabic marked sounds in monolingual speakers, we are not aware of any previous studies that have examined the production of these sounds by (early) bilingual children. Bilingual child phonological development

Until 10 years ago, the majority of research on child phonology was conducted on monolingual children (Johnson, 2018). However, more recently, the number of bilingual studies on both infants and children has increased, although further studies on child bilingualism in general are still needed (see Babatsouli & Ingram, 2015, 2018). One of the questions that informed studies initially was whether bilingual phonology develops as one or two systems. Earlier claims by Volterra and Taeschner (1978) suggest that bilingual children begin acquisition with a single linguistic system, while after some development they develop two different systems. However, current research shows that bilingual infants begin with two separate systems that interact with each other (Ingram et al., 2011; Kehoe et al., 2004). Ingram et al. (2011) found that 18- to 20-month-old bilingual SpanishEnglish speaking twins were able to acquire 92% of the phonological systems in each language, although these were not identical to those of respective monolingual speakers. They suggested that the bilinguals

192

An Anthology of Bilingual Child Phonology

acquire a system that is different from that of monolinguals in early childhood development. Goldstein et al. (2005) also studied Spanish-English bilingual children (aged 5;0–5;5 years), as well as children who predominantly used one of the two languages but were proficient in both. Their single-word assessment did not show that the children’s phonological skills in a predominant language differed significantly from those in the other. Another study on Spanish-English bilingual children by Fabiano-Smith and Barlow (2010) examined the complexity and completeness of phonetic inventories for both languages of bilingual children. Their results suggested that bilingual children are able to develop two comparable phonological inventories in the same time frame that a monolingual child develops one. However, this study also found evidence of English phonology in Spanish and vice versa. These results are similar to those found by Keshavarz and Ingram (2002) in a longitudinal study conducted on a Farsi-English bilingual child, finding that the child had developed two different phonological systems, but that there was also evidence of bi-directional transfer. Other studies have also shown evidence of transfer in bilinguals (e.g. Babatsouli, 2019; Heselwood & McChrystal, 2000; Paradis & Genesee, 1996). For example, Heselwood and McChrystal (2000) found effects of transfer in Punjabi-English speaking children aged 10. Their results showed that there was a higher rate of lead voicing in their production of English voiced stops which resulted from the influence of Punjabi (Heselwood & McChrystal, 2000). Interaction between the two languages can both accelerate and decelerate acquisition (e.g. Babatsouli, this volume; see Kehoe, this volume, 2018 for review). Regarding research on bilingual Arabic children, most of the literature has examined crosslinguistic effects on reading ability (Farran et al., 2012; Saiegh-Haddad & Geva, 2007). Khattab (2000) is one of the few studies that has investigated the acquisition of bilingual phonology of ArabicEnglish speaking children. In particular, she examined whether ArabicEnglish bilingual children acquire separate VOT values in each of their languages. Results showed that Arabic bilingual children were able to acquire separate VOT patterns; however, their values were different to some extent from those of monolinguals. The author ruled out age as a predictor of accurate acquisition since the older participant was unable to acquire complex features such as lead voicing, in contrast to the younger participant who showed gradual development towards the adult pattern. More importantly, the study reported some crosslinguistic effect found in these bilinguals’ VOT patterns. It also highlights the importance of sufficient input as an indicator of successful acquisition. Although case studies provide a good foundation for research, more cross-sectional work is required, also involving different language pairings, to contribute to the understanding of phonological development in bilingual children. Literature on bilingual acquisition is still limited, as

The Production of Marked Arabic Consonants by Arabic-English Bilingual Children

193

only a small number of studies on early bilingual phonological acquisition have been done (Babatsouli & Ingram, 2018; Ingram et al., 2011). In contrast, studies on the monolingual phonological development of children are numerous. In general, the majority of infants are being raised in multilingual environments so there is a need to focus on bilingual children in this vein of research (Johnson, 2018). In particular, there remains a gap in the literature on the phonological development of Arabic-English bilingual children. Although research exists on the L1 acquisition of Arabic phonology (Amayreh & Dyson, 1998, 2000; Ammar & Morsi, 2006; Ayyad, 2011; Saleh et al., 2007), little work has been done on the phonological development of Arabic-English bilingual children. Additionally, Omar (2017) recognizes that there is a need for more data on the acquisition of Arabic phonology. Previous studies tend to focus on the adult phonological system (e.g. Al-Ani, 1970). Lastly, future studies need to focus explicitly on young children (infancy to 5;6 years) in order to investigate what changes may occur in production later in childhood, because language dominance may change due to education and social factors. Flege’s Speech Learning Model and evidence of phonetic merger and phonological change in adult bilinguals

A number of L2 speech learning models have been introduced so far (Brown, 2000; Colantoni & Steele, 2008; Escudero, 2005; Flege, 1987, 1995). One of the most influential models that examines the potential effect of the sound categories of the L2 on the L1 is the Speech Learning Model (SLM) proposed by Flege (1995). The SLM addresses the potential effect of L2 on L1 category change. It proposes that L1 and L2 categories in L2 learners may interact with each other in different ways: influence from both languages can be explained in terms of assimilation or dissimilation. The SLM proposes that when the L1 and L2 sounds are similar, the two sounds may assimilate/merge in the learner’s phonological inventory. For instance, the emphatic /t̠ / may fall in the same phonological space as the plain /t/, a sound found in English, and merge with it over time instead of forming its own category. Flege (1997) was one of the fi rst studies to provide evidence of assimilation of the L1 and L2 phonetic categories in adult speakers. He found changes in the VOT of French-English and English-French adult bilinguals, where VOT values for French /t/ for both groups were longer than those of their monolingual counterparts. On the other hand, VOT values for English /t/ were shorter than their average native values, again for both groups. Likewise, the F2 for the vowel /u/ was lower than in their monolingual French counterparts for the French-dominant group but not for the English-dominant group. However, /y/ was produced in a native-like manner by the bilinguals. The results confi rmed the predictions that /u/

194

An Anthology of Bilingual Child Phonology

and /t/ would be classified as sounds in phonetic categories that already exist in the L1 and /y/ as a sound that is different from an existing category in the L1. Major (1992) also examined VOT values in Brazilian-English bilinguals in the US and, similarly to Flege (1997), found evidence of mutual L1–L2 interaction. VOT drifts in /p, t, k/ in the L1 of bilinguals have also been examined from a sociolinguistic point of view by Hrycyna et  al. (2011). A drift towards English VOT values was reported in successive generations (fi rst, second and third generations) of Italian-, Russian- and Ukrainian-English bilingual communities. They also reported differences between the language groups and suggested that social factors such as (i) the cohesiveness of a community that suggests opportunity for casual speech, (ii) the size of a community and (iii) attitude towards a particular variety of a language may be responsible for the between-group differences. De Leeuw et al. (2012) examined the change in the production of the lateral phoneme /l/ in the L1 German of late German-English bilingual speakers living in Canada. They found that the F1 and F2 values of the German /l/ of their bilinguals differed from their native German counterparts and showed a shift towards English. Furthermore, there was a high degree of variability both within and between bilinguals, and not all the participants exhibited this change. They proposed a dynamic system theory: that maturational constraints cannot be the only cause of attrition, and that various predictors which influence language development in individuals must be considered. Celata and Cancila (2010) investigated the perception of the geminate (long sound)–singleton contrast in native speakers of Lucchese Italian and among first-generation late Lucchese Italian-English bilinguals (those who emigrated to the US) and second-generation Lucchese Italian bilinguals (those who were born in the US). The results of a real word and a nonce word identification task revealed that bilingual speakers were significantly worse than the control Lucchese monolingual speakers at the perception of the geminate–singleton contrasts. In particular, the second-generation group exhibited a higher degree of phonological change than the fi rst generation group. Therefore, the authors concluded that the perception of the length contrast had become progressively impaired in their bilingual groups. Rafat et al. (2017) as well as Alkhuddid et al. (forthcoming) also found evidence of geminate duration reduction in first- and second- generation Farsi-English and Arabic-English bilinguals living in Canada. They attributed this change to contact with English. Although VOT remains one of the best-studied phenomena in studies that have examined the bi-directionality of language influence on speech production, recently there has been a growing interest in examining a phonetic shift and assimilation in other aspects of L1 in both adult and child bilinguals.

The Production of Marked Arabic Consonants by Arabic-English Bilingual Children

195

Research Questions

The research questions in this study are as follows: (1) Are bilingual children whose ages range between 8 to 13 years able to produce the Arabic complex consonants that are phonetically ‘similar’ (Flege, 1995) to English? Specifically, can they produce the following contrasts: /t-t̠ /, /d-d̠ /, /s-s̠ /, /h-ħ/, /k-q/? (2) Is there an effect of consonant type? (3) Is there an effect of position in the word? (4) Is there an age effect? Hypotheses

The predictions in this study are as follows: H1. The complex consonants in this study will be difficult for children to produce as they fall in the same phonological space as their English equivalents (Flege, 1995). H2. There will be an effect of consonant type (Rafat et al., 2017). The sounds listed below are ranked from most to least difficult from left to right: /t/, /d/, /s/, /ħ/, /q/. H3. Complex sounds in word-medial position will be easier than in other positions (Amayreh & Dyson, 1998). H4. Age will have an effect because older children will have received more exposure to Arabic over the years (Amayreh & Dyson, 1998; Ayyad, 2011). Specifically, as children age, the possibility of the acquisition of these contrasts will increase.

Methodology Participants

Participants consisted of five early English-Arabic bilingual children, who lived in either London or Toronto, Ontario, and were English dominant. They consisted of a 10-year-old female and four males. The four males comprised two 8-year-olds, one 11-year-old and one 13-year-old. Three males and one female were born in Canada. The other male participant was born in Saudi Arabia and came to Canada a year prior to this study. He was initially exposed to Arabic at birth and then English at the age of 3 years. The four participants born in Canada had all been exposed to English and Arabic simultaneously at birth. Participants spoke different varieties of Arabic, namely Palestinian, Jordanian and Saudi Arabian. According to the language background questionnaire completed by the participants’ parents, all participants received formal education in English and Arabic at some point. All children went to English schools and

196

An Anthology of Bilingual Child Phonology

attended Arabic language classes on Saturdays. In all cases, one or both parents spoke Arabic with the child at home. One participant, who was 8 years old, came to Canada at the age of 6, and had been exposed to English since he was 3 years old. Prior to coming to Canada he went to an English school and his parents reported that both Arabic and English were spoken at home. Three participants had been taught Arabic and English in school since the first year of school. The two remaining participants were taught in English all throughout school but were only taught Arabic in school for a total of four years. With regards to language input, all participants were addressed in English and in Arabic by their parents. Three out of five participants stated that Arabic was mostly used in their household, while the other two participants stated that their parents mainly used English. All participants were exposed to television and music in both languages; however, English was the main language used at school and in public settings outside the home. When speaking with friends, all participants used English. They were all able to read and write in both languages but they were more proficient in English. On occasion, the participants were read to in Arabic, but otherwise they were mainly read to in English. Two of five participants took part in activities that were run in Arabic outside of school. The remaining three only took part in activities conducted in English. In sum, participants received more input in English than Arabic in their everyday life. Tasks

The participants took part in a word-repetition task and completed a language background questionnaire. In the repetition task, participants were asked to repeat the target words inserted into a carrier phrase /anaː ʔasmaʕ ………. ʔalyaʊm/ ‘I hear _____ today’. The target word was shown to the participants in the carrier phrase via a slideshow on the computer. Additionally, the carrier phrase and target word were played in an audio-recording to the participant one to two times. This was recorded by a native speaker of Arabic who speaks a Saudi dialect. The participants were asked to repeat the phrase twice. Their production was recorded using an M-Audio Micro-track 24/96 professional two-channel mobile digital recorder and a lavaliere unidirectional microphone. The language background questionnaire consisted of 19 questions regarding the linguistic knowledge and language experience of the participant. These questions were designed to inform researchers on any previous exposure to different languages that might affect the acquisition of both English and Arabic, as well as to give the researcher details of the order of acquisition of the two languages (e.g. sequential bilingualism, balanced bilingualism, etc.).

The Production of Marked Arabic Consonants by Arabic-English Bilingual Children

197

Stimuli

The stimuli (see Appendix) consisted of 150 Arabic words. The following emphatic consonants /t̠ /, /d̠ /, /s̠ / and their plain counterparts /t/, /d/, /s/, the pharyngeal consonant /ħ/ and its plain counterpart /h/, and the uvular /q/ and the velar /k/ were included. For each consonant contrast there were five minimal or semi-minimal pairs (e.g. /t̠ aːbuːr/ ‘queue’ versus /taːbuːt/ ‘coffin’, /matiːn/ ‘robust’ versus /fat̠ iːn/ ‘clever’). The stimuli were controlled for stress and number of syllables and were randomized. Target sounds were tested in three different positions in the words: word-initial (e.g. /sa:m/ ‘poisonous’ versus /s̠ a:m/ ‘to fast’), intervocalic (e.g. /nasiːr/ ‘we walk’ versus /nas̠ iːr/ ‘we become’) and word-fi nal (e.g. /ʔʃbaːh/ ‘look alike’ plural versus /ʔʃbaːħ/ ‘ghosts’). Data analysis

The data were acoustically measured using Praat (Boersma & Weenink, 2007). The F2 measurements were specifically examined because this appears to be the common acoustic correlate of the target marked sounds in this study. Measurements were taken from the second vowel formant F2 in a wideband spectrogram. F2 values were measured in Hertz over a 20-millisecond (ms) window. For consonants in the wordinitial and intervocalic positions, the onset of F2 values (approximately 20 ms) were taken from the vowel following the target consonant. For consonants in the word-fi nal position, the offset of F2 values was taken from the vowel preceding the target consonant (Al-Masri & Jongman, 2004; Jongman et al., 2007). Figure 8.1 shows the F2 value for plain /s/ and Figure 8.2 shows the lowering for emphatic /s̠ / in the minimal pair / sa:ħ/ ‘ melted’ versus /s̠ a:ħ/ ‘shouted’ in our data. Following acoustic analysis, a number of statistical tests were used to address whether bilingual individuals produced the same drop in F2 associated with the production of the emphatics, pharyngeal and uvular, whether specific consonants impacted this effect, and whether position in the words impacted this effect. Results

There were 420 tokens recorded; however, only tokens preceding vowel /a/ were analyzed (200 tokens) as presentation of tokens for other vowels was too sparse for well-powered analysis (/a:/ n = 77), /i/ n = 67, /i:/ n = 48, /u/ n = 26 and /u:/ n = 2). Mean F2 values were calculated for each consonant, word position and contrast for each individual. The magnitude of F2 lowering was then calculated by subtracting the F2 associated with emphatic/pharyngeal/uvular productions from their respective plain production for each condition (Figures 8.3, 8.4 and 8.5).

198

An Anthology of Bilingual Child Phonology

Figure 8.1 F2 value for plain /s/: 1485 Hz

To address whether bilingual children exhibited typical drops in F2 preceding emphatic consonants, the difference between F2 was averaged across all utterances for each individual, for each consonant. Lowering of F2 following emphatics /d/, /s/ and /t/ are typically in the range of 450– 500 Hz (Al-Masri & Jongman, 2004). Participants exhibited average drops of 498 ± 78 Hz (mean ± standard error) with the consonant /d/, placing the typical lowering well within one standard error (SE) of the current data. Further, three of five participants were within the typical range. With the consonant /s/, the mean F2 lowering was 383 ± 70 Hz, placing the typical lowering approximately 1 SE away from the current data’s mean. Two of five participants were within the typical range. With the fi nal emphatic /t/, F2 lowering was 250 ± 42 Hz, well outside the typical lowering, and none of the five participants was in the typical range. F2 lowering with the pharyngealized /ħ/ is typically between 250 and 400 Hz (Al-Ani, 1970). Participants exhibited an average lowering of 248 ± 94 Hz, placing the typical lowering well within the SE of the

The Production of Marked Arabic Consonants by Arabic-English Bilingual Children

199

Figure 8.2 F2 value for emphatic /s/: 1408 Hz

current data. Further, two of five participants were within the typical range. F2 lowering with the uvular /q/ is typically between 400 and 600 Hz (Al-Ani, 1970). Participants exhibited an average lowering of 593 ± 107 Hz, within the typical range. Further, four of five participants were within the typical range. To explore the role that consonant and position within the word play in F2 lowering following emphatics, one would typically conduct an omnibus two-way repeated-measures ANOVA (consonant x position). Without data from the medial position with the consonants /h/ and /k/, however, this analysis cannot be conducted. To address this, we conducted three distinct analyses. First, we conducted a linear regression testing of whether consonant, position and individual were significant predictors of F2 lowering. The overall regression was significant (F(3,61) = 5.03, p = 0.004, R 2 = 0.45). Individually, consonant was marginally predictive (t = − 1.72, p = 0.09, rpr = − 0.20), position was significantly predictive (t = − 3.48, p = 0.001, rpr = − 0.40), but individual was not significantly predictive (t = 0.15, p = 0.88, rpr = 0.02).

200

An Anthology of Bilingual Child Phonology

Figure 8.3 F2 lowering for emphatics /t/, /d/, /s/, pharyngeal /ħ/ and uvular /q/ by word position. On the left panel, light grey bars represent F2 values for plain consonants and black bars represent F2 values for marked consonants. The right panel (dark grey bars) represents the F2 lowering, measured as the difference between mean F2 values associated with plain and marked consonants. Error bars represent the standard error of the mean

Secondly, a two-way, repeated-measures ANOVA was conducted excluding the consonants with missing data, /k/ and /h/. There was a significant main effect of consonant (F(2,8) = 5.63, p = 0.03, ƞp2 = 0.59), but no main effect of position (F(2,8) = 2.49, p = 0.14, ƞp2 = 0.38). There was no two-way interaction between consonant and position (F(4,16) = 0.73, p = 0.58, ƞp2 = 0.16).

The Production of Marked Arabic Consonants by Arabic-English Bilingual Children

201

Figure 8.4 F2 lowering for emphatics /t/,/d/,/s/, pharyngeal /ħ/ and /q/ collapsed across word position. On the left panel, light grey bars represent F2 values for plain consonants and black bars represent F2 values for marked consonants. The right panel (dark grey bars) represents the F2 lowering, measured as the diff erence between mean F2 values associated with plain and marked consonants. Light grey shading represents the typical range of F2 lowering for each consonant. Error bars represent the standard error of the mean

Thirdly, a two-way, repeated-measures ANOVA was conducted excluding the position with missing data, word medial. In agreement with the previous analysis, the main effect of consonant (F(4,16) = 3.36, p = 0.04, ƞp2 = 0.46) was significant, but here the main effect of position was also significant (F(1,4) = 7.54, p = 0.05, ƞp2 = 0.65). There was no two-way interaction between consonant and position (F(4,16) = 0.49, p = 0.75, ƞp2 = 0.11). To explore the effect of consonant, follow-up t-tests were conducted for F2 lowering for each consonant: emphatics /d /, /s/ and /t/ showed significant F2 lowering (t(4) = 6.89, p = 0.002, d = 3.07; t(4) = 7.92, p = 0.001, d = 2.47; and t(4) = 5.50, p = 0.005, d = 4.66, respectively); pharyngeal /h/ showed marginally significant F2 lowering (t(4) = 2.65, p = 0.057, d = 1.19); and uvular /q/ showed significant F2 lowering (t(4) = 10.41, p = 0.001, d = 3.55). Further, non-parametric related-samples Wilcoxon signed rank tests confirmed these results, with emphatics /d/, /s/ and /t/ and uvular /q/ showing significant F2 lowering (p = 0.04), and uvular /q/ showing marginally significant F2 lowering (p = 0.08). To explore effect of position, follow-up t-tests were conducted for each position. In all word positions, emphatics showed significant F2 lowering (t(4) = 5.63, p = 0.005, d = 2.52; t(4) = 6.38, p = 0.03, d = 2.85; and t(4) = 3.83, p = 0.02, d = 1.71, respectively). Further, non-parametric related-samples Wilcoxon signed rank tests confirmed these results, with emphatics showing significant F2 lowering in all word positions (p = 0.04). Finally, to test whether age impacted F2 lowering, we conducted a linear regression including participant age as a predictor in addition to consonant and position. While the overall regression was significant (F(3,61) = 5.02, p = 0.004, R2 = 0.45), age was not predictive (t = − 0.05, p = 0.96, rpr = − 0.01).

202

An Anthology of Bilingual Child Phonology

Figure 8.5 The effect of position on F2 lowering of marked consonants, collapsed across consonants. On the left panel, light grey bars represent F2 values for plain consonants and black bars represent F2 values for marked consonants. The right panel (dark grey bars) represents the F2 lowering, measured as the diff erence between mean F2 values associated with plain and marked consonants. Error bars represent the standard error of the mean

Discussion

The main research question in this study was whether early bilingual children are able to acquire the F2 lowering associated with the marked Arabic consonants tested in this study. We also asked whether there was an effect of consonant, position in the word and age. The results showed that early Arabic-English bilingual children are indeed able to exhibit F2 lowering, suggesting that different categories have been established for these sounds in Arabic. However, our results indicate that the degree of F2 lowering was not always target-like/within the expected norm. Moreover, the variation in the degree of F2 lowering in the bilingual children in this study was constrained by the type of consonant and individual variation. Specifically, the results suggested the following hierarchy, beginning with the greatest degree of lowering: /q, ħ, d/ were within the standard error of normal; /s, t/ were not. This suggests that /q/, /ħ/ and /d/ were easier than /s/ and /t/ to acquire. Easier acquisition of /q/ and /ħ/ was expected, because F2 lowering is a characteristic of the primary place of articulation for these sounds. For the emphatics, F2 lowering is a secondary feature of articulation, making this feature more difficult to produce from an articulatory point of view. However, we are not sure why /d/ was easier than /s/ and /t/ since they are all emphatic consonants in Arabic. With respect to position, our results demonstrated that F2 lowering was observed in all word positions, and that the magnitude of this effect did not significantly differ across position. With that said, and given the relatively small sample size, these results numerically suggest the possibility that word-initial position may be more conducive to F2 lowering followed by word-medial and word-final positions. If confirmed with a larger sample, this would not be consistent with previous literature on the acquisition of

The Production of Marked Arabic Consonants by Arabic-English Bilingual Children

203

Arabic consonants by monolingual Arabicspeaking children (e.g. Amayreh & Dyson, 1998) and as such it needs to be investigated further. We tested Flege’s SLM model which predicts that merged categories may form in highly proficient adult bilingual speakers. Overall, our data show that similarity and markedness/articulatory difficulty may play a role in the production of L1 sounds. Our fi ndings provide a more gradient view of L1 bilingual production, indicating that, in a bilingual child, F2 values are at times native-like, at times drift away from the native values, and at other times approximate native-like values, suggesting that early bilingualism may lead to the establishment of L1 categories that differ from those of monolingual children. Previous studies on bilingual child phonology have shown that bilingual children are either able to establish two separate phonological systems, although there might be transfer (e.g. Keshavarz & Ingram, 2002), or they might form two categories that are distinct from those of the monolingual children (e.g. Khattab, 2000). With regard to adult bilingual production in previous studies, VOT values have been shown to change over time in highly proficient bilinguals (see Flege, 1987, 1995), geminate (long sound) duration has been shown to decrease across generations (e.g. Alkhuddidi et al., forthcoming; Rafat et al., 2017), and merged intonation patterns have been reported for bilingual adults and heritage speakers (Aziz et al., 2018). Our data also showed no age effects. This fi nding is similar to that reported by Khattab (2000). In the initial analysis for her subjects, she ruled out the correlation between age and the successful acquisition of voicing lead; however, in the follow-up study, Khattab (2000) reported the younger participant as having acquired voicing lead. She attributed this to an increase in exposure to Arabic formally (taking an Arabic course) and informally (socializing with Arabic neighbors). Similarly, we predict that our participants might become successful in their production of complex Arabic consonants if they have more exposure to Arabic. That input is ‘compromised’ in child bilingualism is also mentioned in Babatsouli and Ingram (2018). It is also possible that the lack of age effect is due to the small sample size. It is worth noting that we have considered F2 lowering, which is one of many acoustic correlates of the marked contrasts tested in this study. In the future, the production of other acoustic features of these consonants needs to be investigated. For example, future studies need to look into whether /q/ was produced as a stop or a fricative, before they can claim that this sound was fully acquired or not. Moreover, it would be important to see whether the non-native-like realizations of F2 lowering are perceivable by native speakers of Arabic. For example, future studies could investigate whether the emphatic sounds that were produced with a lower degree of F2 lowering than expected are perceived as emphatics, accented emphatics, plain consonants or something else.

204

An Anthology of Bilingual Child Phonology

Conclusion

The present study aimed at investigating the production of Arabic consonant contrasts by early bilingual child heritage speakers of Arabic whose dominant language is English. Specifically, we looked at the production of the following contrasts:/t/-/t/, /d̠ /-/d/, /s̠ /- /s/, /ħ/- /h/ and /q/-/k/. The results indicate that the bilingual children were able to produce most of these segments in Arabic. However, secondary articulation was not fully acquired for all segments. We attribute this to influence from English and articulatory difficulty. In our study, age was not a predictor of successful acquisition. This study adds to the existing literature on child bilingualism by supporting the notion that there may be two language systems for bilingual children. Future studies need to examine bilingual speech production in both languages of the participants. Given the small sample size in this study, we recommend that the hypotheses put forth in this study be tested with a larger sample size and other Arabic dialects in the future. It must be noted that emphatics in different varieties have different characteristics and the correlates are not completely identical (Al-Masri & Jongman, 2004). It would be beneficial to test bilingual children of the same variety of Arabic only and, in addition, to compare these results with an acoustic analysis of the phonological development of Arabic monolingual children with the same age range, as the bilingual group would provide more insight on the developmental pattern of these sounds. In considering other factors, an investigation of the effect of vowel quality on the degree of F2 lowering would be invaluable (e.g. Al-Masri & Jongman, 2004; Jongman et al., 2007) to see if vowels that show greater F2 lowering facilitate production. Likewise, future studies would benefit from incorporating a more sociolinguistic perspective. Lastly, an examination as to whether or not bilingualism has an impact on young children’s production of other segments and prosodic aspects of Arabic speech would be interesting. The effect of early bilingualism could also be examined with respect to the perception of the children’s heritage language or non-dominant language. Appendix: Stimuli

/t – t̠/

Word initial

Intervocalic

Word final

/t̠aːbuːr – taːbuːt/ ‘queue – coffin’

/futuːɾ – fut̠uːr/ ‘coldness – breaks’

/qaːnit – qaːnit̠/ ‘obedient – despondent’

/taːba – t̠aːba/ ‘repented – healed’

/ʕataː – ʕat̠aː/ ‘rebuffed – gave’

/saqat – saqat̠/ ‘she watered – fell’

/taɾak – t̠aɾaq/ ‘left – knock’

/qitaːl – qit̠aːɹ/ ‘fight – train’

/kuruːt – ʃuɾuːt̠/ ‘cards – conditions’

The Production of Marked Arabic Consonants by Arabic-English Bilingual Children

/d – d̠/

/s – s̠/

/q – k/

205

/taɾaf – t̠aɾaf/ ‘luxury – end side’

/χitaːm – χit̠aːb/ ‘the end – speech’

/mumiːt – muɦiːt̠/ ‘deadly – ocean’

/taːbiʕ – t̠aːbiʕ/ ‘subordinate – stamp’

/matiːn – fat̠iːn/ ‘robust – clever’

/baːʔit – ħaːʔit̠/ ‘stale – wall’

/dal – d̠al/ ‘indicate – strayed’

/hadam – had̠am/ ‘destroyed – digested’

/qeɪd – beɪd̠/ ‘chain – egg’

/daːɾ – d̠aːr/ ‘home – harmful’

/mada – mad̠a/ ‘distance – went’

/fard – Fard̠/ ‘individual – imposing’

/daɾb – d̠aɾb/ ‘path – hit’

/qadiːm – qad̠iːm/ ‘old – vellum’

/ʕaqd – ʕard̠/ ‘contract – show’

/damm – d̠amm/ ‘blood – join’

/nadir – nad̠ir/ ‘rare – flourishing’

/s̠eɪd – feɪd̠/ ‘fishing – abundance’

/durr – d̠urr/ ‘pearls – harm’

/ʕadaː – qad̠aː/ ‘except for – spent’

/ħeɪd – ħeɪd̠/ ‘deviation – menstruation’

/seɪf – s̠eɪf/ ‘sword – summer’

/fasiːħ – fas̠iːħ/ ‘spacious – fluent’

/ʕurs – qurs̠/ ‘wedding – disk’

/saːr – s̠aːr/ ‘walked – pleasant’

/ʕasiːr – ʕas̠iːr/ ‘difficult – juice’

/raʔs – raqs̠/ ‘head – dance’

/saːm – s̠aːm/ ‘poisonous – fast’

/nasiːr – nas̠iːr/ ‘we walk – supporter’

/qaʊs – ɣaʊs̠/ ‘bow – diving’

/saːħ – s̠aːħ/ ‘traveled – screamed’

/nasab – nas̠ab/ ‘kinship – set up’

/nafs – naqs̠/ ‘soul – lack’

/sahm – s̠ahr/ ‘arrow – melting’

/ħasad – ħas̠ad/ ‘envy – reaped’

/naːs – baːs̠/ ‘people – bus’

/kalb – qalb/ ‘dog – heart’

/sakit – saqit/ ‘quite – falling’

/falak – falaq/ ‘orbit – split’

/kabl – qabl/ ‘shackle – before’

/ʃakiː – saqiː/ ‘complainer – barmaid’

/malik – maliq/ ‘king – flatterer’

/kabħ – qamħ/ ‘restraining – wheat’

/nakir – naqil/ ‘denier – transporter’

/salak – salaq/ ‘took path – boil’

/kahl – qahr/ ‘elderly – defeating’

/raːkib – ɵaːqib/ ‘passenger – farsighted’

/ʕiraːq – ʕirak/ ‘Iraq – fighting’

/kaːl – qaːl/ ‘measured – said’

/yakiːs – yaqiːs/ ‘put in a bag – measure’

/taːrik – saːriq/ ‘leaving – thief’

References Al-Adam, H. (2015) Acoustic correlates of emphatic sounds in Palestinian Arabic speaking persons with aphasia. Language in Focus 1 (2), 1–16. Al-Ani, S.H. (1970) Arabic Phonology: An Acoustical and Physiological Investigation. The Hague: Mouton. Alkhudidi, A., Rafat, Y. and Stevenson, R. (forthcoming) Geminate attrition in the speech of Arabic–English bilinguals living in Canada. Heritage Language. Al-Masri, M. and Jongman, A. (2004) Acoustic correlates of emphasis in Jordanian Arabic: Preliminary results. In Proceedings of the 2003 Texas Linguistics Society Conference (pp. 96–106). Somerville, MA: Cascadilla Proceedings Project. Alwan, A. (1986) Acoustic and perceptual correlates of pharyngeal and uvular consonants. Unpublished Doctoral dissertation, MIT. See http://hdl.handle. net/1721.1/34302.

206

An Anthology of Bilingual Child Phonology

Amayreh, M.M. and Dyson, A.T. (1998) The acquisition of Arabic consonants. Journal of Speech, Language, and Hearing Research 41, 642–653. Amayreh, M.M. and Dyson, A.T. (2000) Phonetic inventories of young Arabic-speaking children. Clinical Linguistics and Phonetics 14, 193–215. Ammar, W. and Morsi, R. (2006) Phonological development and disorders: Colloquial Egyptian Arabic. In Z. Hua (ed.) Phonological Development and Disorders in Children (pp. 204–232). Clevedon: Multilingual Matters. Ayyad, H. (2011) Phonological development of typically developing Kuwaiti Arabicspeaking preschoolers. Unpublished Doctoral thesis, University of British Columbia. Aziz, J., Machado, V., Swiderski, N., Valdivia, C., Rafat, Y., Stevenson, R. and Rao, R. (2018) Investigating the sources of nuclear intonation in Argentinian-Canadian heritage speakers of Spanish: Evidence of parental and English influences. 48th Linguistic Symposium on Romance Languages, Toronto, ON, April. Babatsouli, E. (2019) Melodies of child Greek-English phonological interference. Revista de Estudos da Línguagem 17 (2), 209–236. doi:10.22481/el.v17i2.5347 Babatsouli, E. and Ingram, D. (2015) What bilingualism tells us about phonological acquisition. In R.H. Bahr and E.R. Silliman (eds) Routledge Handbook of Communication Disorders (pp. 173–182). London/New York: Routledge/Taylor & Francis. Babatsouli, E. and Ingram, D. (2018) Prologue. In E. Babatsouli and D. Ingram (eds) Phonology in Protolanguage and Interlanguage (pp. 1–23). Sheffield: Equinox. Benmamoun, E., Montrul, S. and Polinsky, M. (2013) Heritage languages and their speakers: Opportunities and challenges for linguistics. Theoretical Linguistics 39 (3/4), 129–181. Boersma, P. and Weenink, D. (2007) Praat: Doing Phonetics by Computer [Computer program]. Version 4.6.09. See http://www.praat.org/S (accessed 19 July 2007). Brown, C. (2000) The interrelation between speech perception and phonological acquisition from infant to adult. In J. Archibald (ed.) Second Language Acquisition and Linguistic Theory (pp. 4–63). Malden, MA: Blackwell. Campbell, R. and Peyton, J. (1998) Heritage language students: A valuable language resource. ERIC Review 6 (1), 38–39. Card, E. (1983) A phonetic and phonological study of Arabic emphatics. Unpublished Doctoral dissertation, Cornell University. Celata, C. and Cancila, J. (2010) Phonological attrition and the perception of geminate consonants in the Lucchese community of San Francisco (CA). International Journal of Bilingualism 14, 1–25. Colantoni, L. and Steele, J. (2008) Integrating articulatory constraints into models of second language phonological acquisition. Applied Psycholinguistics 29, 489–534. De Leeuw, E., Mennen, I. and Scobbie, J.M. (2012) Dynamic systems, maturational constraints and L1 phonetic attrition. International Journal of Bilingualism 17, 683–700. Draper, J.B. and Hicks, J.H. (2000) Where we’ve been; what we’ve learned. In J.B. Webb and L. Miller (eds) Teaching Heritage Language Learners: Voices from the Classroom (pp. 15–35). Yonkers, NY: American Council on the Teaching of Foreign Languages. Dyson, A.T. and Paden, E.P. (1983) Some phonological acquisition strategies used by twoyear-olds. Journal of Childhood Communication Disorders 7, 6–18. Escudero, P. (2005) Linguistic perception and second language acquisition. LOT Dissertation Series 113, Utrecht University. Fabiano-Smith, L. and Barlow, J.A. (2010) Interaction in bilingual phonological acquisition: Evidence from phonetic inventories. International Journal of Bilingual Education and Bilingualism 13 (1), 81–97. Farran, L.K., Bingham, G.E. and Matthews, M.W. (2012) The relationship between language and reading in bilingual English-Arabic children. Reading and Writing: An Interdisciplinary Journal 25 (9), 2153–2181. doi:10.1007/s11145-011-9352-5 Flege, J.E. (1987) The production of ‘new’ and ‘similar’ phones in a foreign language: Evidence for the effect of equivalence classification. Journal of Phonetics 15 (1), 47–65.

The Production of Marked Arabic Consonants by Arabic-English Bilingual Children

207

Flege, J.E. (1995) Second language speech learning: Theory, fi ndings, and problems. In W. Strange (ed.) Speech Perception and Linguistic Experience: Issues in Crosslinguistic Research (pp. 233–277). Timonium, MD: York Press. Flege, J.E. (1997) Amount of native-language (L1) use affects the pronunciation of an L2. Journal of Phonetics 25, 169–186. Goldstein, B.A., Fabiano, L. and Washington, P.S. (2005) Phonological skills in predominantly English-speaking, predominantly Spanish-speaking, and Spanish-English bilingual children. Language, Speech, and Hearing Services in Schools 36 (3), 201–218. Heselwood, B. and McChrystal, L. (2000) Gender, accent features and voicing in PunjabiEnglish bilingual children. In D. Nelson and P. Foulkes (eds) Leeds Working Papers in Linguistics and Phonetics 8, 45–70. Hrycyna, M., Lapinskaya, N., Kochetov, A. and Nagy, N. (2011) VOT drift in 3 generations of heritage language speakers in Toronto. Canadian Acoustics 39, 166–167. Ingram, D., Dubasik, V., Liceras, J. and Fernández Fuentes, R. (2011) Early phonological acquisition in a set of English-Spanish bilingual twins. In C. Sanz and R.P. Leow (eds) Implicit and Explicit Language Learning: Conditions, Processes, and Knowledge in SLA and Bilingualism (pp. 195–205). Washington, DC: Georgetown University. Jakobson, R. (2012) Child Language, Aphasia and Phonological Universals. Janua Linguarum Series Minor 72. New York: Walter de Gruyter. Johnson, E.K. (2018) Putting the terms ‘monolingual’ and ‘bilingual’ under the microscope. Applied Psycholinguistics 39 (4), 753–756. Jongman, A., Herd, W. and Al-Masri, M. (2007) Acoustic correlates of emphasis in Arabic. In Proceedings of the XVIth International Congress of Phonetic Sciences: ICPhS XVI (pp. 913–916). Saarbrücken: Saarland University. Jongman, A., Herd, W., Al-Masri, M., Sereno, J. and Combest, S. (2011) Acoustics and perception of emphasis in Urban Jordanian Arabic. Journal of Phonetics 39 (1), 85–95. Kehoe, M. (2018) Crosslinguistic interaction in early bilingual phonology: A critical review. In E. Babatsouli and D. Ingram (eds) Phonology in Protolanguage and Interlanguage (pp. 49–75). Sheffield: Equinox. Kehoe, M.M., Lleó, C. and Rakow, M. (2004) Voice onset time in bilingual GermanSpanish children. Bilingualism: Language and Cognition 7 (1), 71–88. Keshavarz, H. and Ingram, D. (2002) The early phonological development of a FarsiEnglish bilingual child. International Journal of Bilingualism 6 (3), 255–269. Khan, M.L. and Lewis, N.P. (1986) Khan-Lewis Phonological Analysis. Circle Pines, MN: American Guidance Service. Khattab, G. (2000) VOT production in English and Arabic bilingual and monolingual children. In D. Nelson and P. Foulkes (eds) Leeds Working Papers in Linguistics 8, 95–122. Khattab, G., Al-Tamimi, F. and Heselwood, B. (2006) Acoustic and auditory differences in the /t/- /T/ opposition in male and female speakers of Jordanian Arabic. In S. Boudelaa (ed.) Perspectives on Arabic Linguistics XVI: Papers from the Sixteenth Annual Symposium on Arabic Linguistics (pp. 131–160). Cambridge: John Benjamins. Kupisch, T. and Rothman, J. (2018) Terminology matters! Why difference is not incompleteness and how early child bilinguals are heritage speakers. International Journal of Bilingualism 22 (5), 564–582. Major, R.C. (1992) Losing English as a fi rst language. The Modern Language Journal 76 (2), 190–208. Montrul, S. (2012) Is the heritage language like a second language? In L. Roberts, C. Lindqvist, C. Bardel and N. Abrahamsson (eds) EUROSLA (pp. 1–29). Amsterdam: John Benjamins. Mousa, M., Amayreh, A. and Dyson, T. (2000) Phonetic inventories of young Arabicspeaking children. Clinical Linguistics and Phonetics 14 (3), 193–215.

208

An Anthology of Bilingual Child Phonology

Omar, M.K. (2017) The Acquisition of Egyptian Arabic as a Native Language. Berlin: Walter de Gruyter. Paradis, J. and Genesee, F. (1996) Syntactic acquisition in bilingual children: Autonomous or interdependent? Studies in Second Language Acquisition 18, 1–25. Rafat, Y., Mohaghegh, M. and Stevenson, R.A. (2017) Geminate attrition across three generation of Farsi-English bilinguals living in Canada: An acoustic study. Ilha do Desterro – Investigating Second Language Speech Learning 71 (3), 151–168. See https:// periodicos.ufsc.br/index.php/desterro/article/view/2175-8026.2017v70n3p151. Rothman, J. (2006) Heritage speaker competence differences, language change, and input type: Inflected infi nitives in heritage Brazilian Portuguese. International Journal of Bilingualism 11 (4), 359–389. Rothman, J. (2009) Understanding the nature and outcomes of early bilingualism: Romance languages as heritage languages. International Journal of Bilingualism 13 (2), 155–163. Rothman, J. and Treffers-Daller, J. (2014) A prolegomenon to the construct of the native speaker: Heritage speaker bilinguals are natives too! Applied Linguistics 35 (1), 93–98. Saiegh-Hadadd, E. and Geva, E. (2007) Morphological awareness, phonological awareness, and reading in English-Arabic bilingual children. Reading and Writing: An Interdisciplinary Journal 21, 481–504. Saleh, M., Shoeib, R., Hegazi, M. and Ali, P. (2007) Early phonological development in Arabic Egyptian children: 12–30 months. Folia Phoniatrica Logopedica 59 (5), 234–240. Shorrab, G. (2009) Bilingual patterns of an Arabic-English speech community. International Journal of the Sociology of Language 61 (1), 79–88. Stoel-Gammon, C. (1985) Phonetic inventories, 15 ± 24 months: A longitudinal study. Journal of Speech and Hearing Research 28, 505–512. Valdés, G. (2000) Introduction. In Spanish for Native Speakers (Vol. I, AATSP Professional Development Series Handbook for Teachers K-16). New York: Harcourt College. Volterra, V. and Taeschner, T. (1978) The acquisition and development of language by bilingual children. Journal of Child Language 5 (2), 311–326. Wahba, K. (1993) Linguistic variation in Alexandrian Arabic: The feature of emphasis. Unpublished PhD dissertation, Alexandria University.

9 On Heritage Accents: Insights from Voice Onset Time Production by Trilingual Heritage Speakers of Spanish Raquel Llama and Luz Patricia López-Morelos

Introduction

Interest in the acquisition and use of heritage languages (HLs; understood in this chapter as the languages of immigrants learned by their offspring at home) has long existed, although they may have been referred to under various labels over time (Montrul, 2015). Whereas the topic received some consideration in the previous century, it was not until the 2000s, and especially during the last 10 years, that the field of heritage language acquisition (HLA) started drawing unprecedented attention, as highlighted by Montrul (2015) and Pascual y Cabo and De la Rosa-Prada (2015). In a broad sense, HLs ‘are culturally or ethnolinguistically minority languages that develop in a bilingual setting where another sociopolitically majority language is spoken’ (Montrul, 2015: 2). Hence, we could generally describe heritage speakers (HSs) as bilinguals who are exposed to some degree to a minority language at home and to a majority language in the community. While some HSs could also be exposed to the majority language in the family environment, it is usually when they enter the educational system that these speakers are increasingly exposed to the socially dominant language and start using it more, to the potential detriment of the HL. Over time, HSs tend to become more proficient in the majority language, which initially was their second language (L2) or their second first language (2L1).1 Here, this L2/2L1 will be referred to as the dominant language (DL). It is usually taken for granted that HSs speak the majority language of their community in a monolingual-like fashion, and that it stays virtually unaffected by potential influence from the HL (Kupisch, 2019). Conversely, 209

210

An Anthology of Bilingual Child Phonology

their competence in the HL does not always go unquestioned. In this regard, it is generally expected that they outperform late L2 learners but still differ from their monolingually raised peers. In fact, studies within HLA research are often devoted to exploring how HSs compare to other speakers, either L2 learners or their monolingual counterparts, in different areas of the HL. As stated earlier, this field of investigation has experienced a considerable upsurge over the last two to three decades. According to Rao and Ronquest (2015), the bulk of such research has centered around these speakers’ (morpho)syntactic knowledge, ‘or has been aimed at devising pedagogy tailored specifically to meet their needs’ (Rao & Ronquest, 2015: 403). In contrast, the sub-domain of heritage phonetics and phonology (HPP) remains underexplored (Benmamoun et al., 2013; Rao & Ronquest, 2015). It has been suggested that the scarcity of studies on HPP may be due in part to the fact that HSs, thanks to an early exposure to the language at home, are believed to have an advantage over other types of L2 learners regarding pronunciation. More precisely, pronunciation is the area in which they have the most noticeable advantage (Kim, 2016). In spite of that, their speech still tends to be labeled as ‘funny’ or ‘off ’ (Polinsky & Kagan, 2007), and they can be described as ‘not real speakers of the language’ (Polinsky & Kagan, 2007: 378) by monolingually raised speakers. The fact that their accent tends to be less pronounced than that of late L2 learners but still distinguishable from that of monolingual native speakers makes their speech unique (Polinsky & Kagan, 2007) and warrants further investigation into its distinct characteristics. In this chapter we investigate HL acquisition but target trilingual HSs, rather than bilingual HSs as is common in this type of studies. This aligns with an emerging strand of research which is making the focus of investigation shift to HSs who are learning a third language (L3), and to the relative effects their HL and their DL may have on the acquisition of the L3. Studies of this sort are still scarce, as pointed out by Lloyd-Smith et al. (2018). Our main objective is to examine the extent to which these trilinguals approximate native-like production in the HL. A secondary aim is to gauge the impact of their HL on their learning of an L3, in an attempt to contribute to the emerging strand of research we have just mentioned. To do so, we focus on the production of voice onset time (VOT) in the Spanish (HL), English (DL) and French (L3) of our participants.

Voice Onset Time and its Acquisition by Heritage Trilingual Learners VOT patterns in Spanish, English and French: An overview

VOT is measured in milliseconds (ms) and corresponds to ‘the time that elapses between the release of the articulators for a stop and the onset

On Heritage Accents

211

of vocal fold vibration for the following segment’ (Yavaş, 2009: 244). Traditionally, three categories of stops have been established: (1) prevoiced, with lead or negative VOT (i.e. under 0 ms); (2) short-lagged, with positive VOT between 0 and 30 ms; and (3) long-lagged, with positive VOT over 30 ms. On the one hand, Spanish and French voiceless stops (/p t k/) are classified as short-lagged, while their voiced counterparts (/b d g/) are implemented with lead VOT. On the other hand, English /p t k/ fall under the long-lagged category, and are usually produced with values over 60 ms when placed in stressed, word-initial position (Lisker & Abramson, 1964), whereas their voiced counterparts are implemented with short lag. A simplified comparison of stop categories across Spanish, English and French is depicted in Figure 9.1. Worth noting is the overlap between English voiced stops and French/Spanish voiceless stops. VOT is an appealing candidate for investigation in bilingual and, increasingly, in multilingual populations for various reasons. One of them is that it can be used to contrast voiced and voiceless stops within a given language and, crucially, in two different languages (Splendido, 2017), as explained above and shown in Figure 9.1. Another reason lies in the difference resulting from the presence in long-lagged stops versus the absence in short-lagged stops of aspiration. Often described as a burst of air that accompanies the release of /p t k/ in stressed onset position in English (Avery & Ehrlich, 1995), aspiration has been claimed to be audible when VOT exceeds 40 ms (Lindblad, 1998, in Johansson et al., 2001). For our language combination, the production of French- and Spanish-like (i.e. short-lagged, unaspirated) /p t k/ in English can cause native speakers of English to perceive them as /b d g/ instead. Conversely, the production of English-like (i.e. long-lagged, aspirated) voiceless stops in Spanish/French can result in accentedness (Lord, 2005). A common claim regarding VOT is that it correlates with degree of foreign accent (Riney & Takagi, 1999). Crucial to this study, it has been proven that measurable differences in VOT values between young adult L1 French speakers and adolescent L2/L3 learners of French can be perceived by L1 French raters during an accent rating task that involved listening to individual words containing /t/ (Gabriel et al., 2016). More details about this study will be provided later in this section. Release (0 ms)

Spanish / French English

Lead

Short lag (unaspirated)

/b, d, g/

/p, t, k/ /b, d, g/

Long lag (aspirated)

/p, t, k/

Figure 9.1 Simplified comparison of stop categories in Spanish, English and French Source: Adapted from Splendido (2017).

212

An Anthology of Bilingual Child Phonology

VOT in bilingual and trilingual heritage speakers

As hinted earlier, the acquisition of VOT patterns ‘presents a unique challenge to many bilingual individuals’ (Fabiano-Smith & Bunta, 2012: 148), and the sizeable body of research investigating the perception and/ or production of voiced and/or voiceless stops on the part of speakers of English and a Romance language, both children and adults, continues to grow. A common goal those studies often pursue is to establish whether bilingual speakers develop separate or merged categories for voiced and voiceless stops in their two languages. It is not our intention to present a comprehensive overview of this issue, given that bilingual speakers are not the targeted population in the present study. Instead, we will highlight the most salient findings reported for early bilinguals, and then turn to a more detailed review of studies investigating the acquisition of this phenomenon by trilingual HSs. On the one hand, numerous studies have shown that it is possible for simultaneous or successive bilinguals to separate stops or at least one set of stops (usually the voiceless set) in the two languages (Deuchar & Clark, 1996; Kim, 2011; MacLeod & Stoel-Gammon, 2009; Sundara et  al., 2006). On the other hand, we fi nd claims of the creation of a hybrid or merged system; that is, the VOT values are somewhat intermediate between those produced by monolinguals of their L1 and monolinguals of their L2 (Caramazza et al., 1973; Flege, 1991; Fowler et al., 2008; Kupisch et  al., 2014). We should keep in mind that even when bilinguals may closely resemble respective monolinguals of their languages, they are not equivalent to two monolinguals (Flege, 1999; Grosjean, 1989; Paradis, 2001) and represent instead a unique combination (Cook, 1992). Moreover, there is ample evidence that the L1 and L2 sound systems are not fully independent, and that interaction between them is rather inevitable (Flege, 1995). Unlike in the case of bilingual HSs, research exploring the acquisition of VOT patterns by trilingual or multilingual HSs is still very scarce. To the best of our knowledge, only five studies have addressed this issue to date (Bandeira & Zimmer, 2012; Bondarenko, 2018; Dittmers et al., 2018; Gabriel et al., 2016; Llama & López-Morelos, 2016). In the earliest of such studies, Bandeira and Zimmer tested VOT production with regard to voiceless stops in L1 Pomeranian, 2 L2 Brazilian Portuguese (BP) and L3 English. The participants were 40 children (8–10 years old): 20 monolingual (BP) and 20 bilingual (Pomeranian-BP) beginner learners of L3 English. Taking into account that /p, t, k/ are aspirated in Pomeranian (Bandeira & Zimmer, 2012) and English, whereas in BP they are not, it is reasonable to expect transfer to English from the L1, provided that those children produced native-like VOT values in their two fi rst languages. The results showed high VOT values across all languages, which was unpredicted for BP. Overall, the authors claimed L1 to L2 and L2 to L3

On Heritage Accents

213

transfer patterns. This can be interpreted as negative transfer from the L1/ HL to the L2, since it resulted in significant differences between the monolingual and bilingual groups in BP. Interestingly, the fact that the L2 was already influenced by the HL resulted in positive transfer from this language to the L3. This is evidenced by the advantage (i.e. closer to Englishlike values) bilingual participants had over their monolingual counterparts regarding the production of /t/ and /k/. Of note, the authors report differences in performance across the three voiceless stops, a fi nding that must not be overlooked, as it points to highly complex interactions among a multilingual’s various languages. In a more comprehensive study, Gabriel and colleagues (2016) investigated not only production but also perception of the stop dyad /t/-/d/. Their participants were adolescent speakers of Mandarin as an HL attending high school in Germany. They spoke German as the majority/ DL and were learning French as an L3. The authors also collected control data from three different groups: (1) L1 German speakers without knowledge of Mandarin who were learning L2 French; (2) L1 Mandarin speakers without knowledge of German who were also learning L2 French; and (3) L1 French speakers. Out of the various research questions the authors sought to answer, we focus on whether the trilingual speakers had an advantage over their bilingual peers, and if any potential differences in L3 VOT values could be perceived by French native speakers. In regard to the production data in Mandarin and German, the authors claim no noticeable differences between Mandarin-German bilinguals and monolinguals of both languages; that is, these HSs seem to behave overall like monolinguals of both their heritage and dominant languages. Regarding French production data, Mandarin-German bilinguals and L1 German speakers pattern together and differ from L1 French speakers in the case of /d/ (due to potential influence from both native languages), and also in the case of /t/ (due to potential influence from German). Gabriel and colleagues interpret these fi ndings as a lack of (dis)advantange for the HSs when compared to their L1 German counterparts. In respect to the accent ratings, a large group of L1 French raters from three Francophone countries were presented with stimuli from the production task. Of interest, the L1 German and the Mandarin-German bilinguals received similar judgements and, together, were perceived as more accented than the L1 Mandarin learners of French. Also in Germany, Dittmers and colleagues (2018) recruited two groups of adolescent HSs of Turkish and Russian as HLs, who spoke German as a DL and were learning English and French as L3s or additional foreign languages. Their control data came from monolingually raised German speakers who were also learning two foreign languages at school (either English and French or English and Russian). Their aim was to explore crosslinguistic influence in the production of /p t k/. Recall from earlier

214

An Anthology of Bilingual Child Phonology

discussion that voiceless stops in stressed onset position are implemented with short lag in French and with long lag in English (see Figure 9.1 and the related explanations). For the sake of simplification, we will describe German as largely patterning with English (as per fi ndings from a recent study by Van de Weijer & Kupisch, 2015) and Turkish and Russian as patterning with French (for a more detailed language comparison, see Dittmers et al., 2018: 42). Using such a complex set of groups, the authors expected to show that (1) HSs would produce shorter VOT means in their DL, German, as well as less target-like values in English than their monolingually raised German peers, but (2) would also perform better than them in Russian and French. In short, it was hypothesized that the HL would have a negative effect on VOT production in the DL and any foreign language that was typologically distant (i.e. English). In contrast, it would have a positive effect on the learning of typologically similar foreign languages (in this case, French and Russian). All their hypotheses were largely or partially confi rmed. While HSs did in fact produce lower VOTs in German than the L1 German controls, all learners seemed to approximate English-native norms in a similar way. In French as a foreign language, HSs did indeed outperform the monolingually raised group. Interestingly, in Russian, L1 German learners clearly showed negative transfer from the L1, whereas HSs of Russian who attended Russian classes produced voiceless stops in a more target-like manner than those who spoke the HL merely at home. In an ambitious study, Bondarenko (2018) examined the effects of the HL and the DL on the acquisition of three L3 segments (vowels, voiced and voiceless stops) across several types of tasks (narrative, picturenaming, sentence-reading, and nonce words). To this aim, she recruited two groups of adult HSs (11 Polish and six Ukrainian speakers) who grew up in the US (English as a DL), and who were learning L3 Spanish at university (beginner and intermediate levels). In addition, there were two bilingual control groups: L1 English-L2 Spanish (n = 5) and L2 Spanish-L1 English (n = 5) speakers, all of them residing in the US. Regarding voiceless stops, Polish and Ukrainian are similar to Spanish and differ from English. Overall, Bondarenko reports native-like VOT in Ukrainian (short-lagged), Polish (short-lagged) and English (longlagged, although shorter but not significantly different from that of L1 English speakers), which could indicate minimal impact of the HL on the DL and vice versa. In L3 Spanish, Ukrainian HSs produced short targetlike VOTs, whereas Polish HSs produced a mix of short- and long-lag voiceless stops. It is argued that this discrepancy could be due to a lower level of Spanish proficiency on the part of the Polish speakers. Even then, this group outperformed the L1 English-L2 Spanish bilinguals. Therefore, speaking Polish as a HL seems to still put them at an advantage over L1 speakers of their DL. It is noteworthy that the reliance on

On Heritage Accents

215

the HL, the DL or both varies across segments, with more influence from English in the case of vowels and more influence from Ukrainian/Polish in the case of voiceless stops. Lastly, in an exploratory investigation conducted in Canada, Llama and López-Morelos (2016) examined VOT production involving voiceless stops in Spanish as a HL, English as a DL and French as an L3 by a small group of adolescents of diverse Hispanic origin (El Salvador, Mexico and the Dominican Republic). The study’s main objective was to uncover the extent to which the trilingual participants had achieved native-like production in all three languages. Their mean VOT values were found to be closely in tune with those of an age-matched group of Spanish monolinguals recruited in Mexico and Venezuela. In addition, no significant differences were reported in relation to English between them and a control group of English-French bilinguals. Surprisingly, they failed to rely on their HL in the case of /t/ and /k/, which they produced with English-like (thus, non-target) values instead. For /p/, they seemed to create a hybrid category, that is, with a mean somewhat intermediate between those expected for English-like and Spanish-like /p/. To summarize this section, we have tried to uncover common trends and point to contrasting fi ndings from the five studies we have just outlined. Some of them seem to have found a merged system for all three languages (Bandeira & Zimmer, 2012), while others report separate ones, at least for the two strongest languages (Gabriel et al., 2016; Llama & López-Morelos, 2016). In some cases there is evidence of influence from the HL on the DL as, for instance, lower VOTs in German (Dittmers et  al., 2018) or higher VOTs in BP (Bandeira & Zimmer, 2012). With regard to the impact of speaking an HL on the acquisition of L3 VOT patterns, there are confl icting results that cover the whole spectrum of possibilities: positive (Bandeira & Zimmer, 2012; Bondarenko, 2018; Dittmers et al., 2018); negative (Dittmers et al., 2018); and neutral, that is, HSs perform similarly to native speakers of their DL (Gabriel et al., 2016; Llama & López-Morelos, 2016). Research Questions and Hypotheses

The present study addresses the following questions: (1) Do these early trilinguals produce monolingual-like VOT patterns in Spanish as an HL? If not, are their VOT patterns suggestive of a heritage accent? (2) Do these early trilinguals separate VOT categories for their different languages? If so, do they display monolingual-like patterns in English as a DL and French as an L3? The review of previous studies presented earlier left us with somewhat disparate fi ndings. Accordingly, formulating solid predictions on the basis

216

An Anthology of Bilingual Child Phonology

of that review is not a straightforward matter. Nonetheless, and based on previous fi ndings, we put forward the following hypotheses: H1. In Spanish, these trilinguals will for the most part produce native-like (i.e. short lag) VOT, but their mean values will be longer than those produced by their monolingually raised counterparts. H2. Those higher means will in some cases be caused by off-target (i.e. out of the short-lag range) productions. Based on reports of accentedness traceable to VOT measurement differences in French (Gabriel et al., 2016), higher means could be suggestive of accentedness. H3. These trilinguals would have created separate categories for English and Spanish, their strongest languages. In English, these HSs will produce native-like (i.e. long lag) VOT. Counter-intuitively, but based on Llama and López-Morelos’ (2016) results, we predict that French will align with English; that is, we expect influence from the DL to the L3. This means that their French VOT values will be overly long and therefore non-native-like. Methodology Participants

We tested a total of 36 participants, divided into four groups, which were initially six as we explain in the Results section. The experimental group, henceforth Group H (trilingual HSs), consists of five children and seven (pre-)adolescents 3 who speak Mexican Spanish (MSp) as a HL, Canadian English (CEn) as a DL and Canadian French (CFr) as an L3. Their linguistic profi le is summarized in Table 9.1a. The five children’s ages ranged from 6 to 8 years (mean age: 7), whereas the seven (pre-)adolescents’ ages ranged from 12 to 16 years (mean age: 14). Spanish was their home language. They all belonged to Mexican families, except for one of the children whose mother was Mexican but whose father was English Canadian. Having spent an extended period of time in Mexico, he spoke Spanish at a near-native level. Participants were born in Canada but were talked to mainly in Spanish until they started school at around the age of 3 years. They had regular exposure to monolingual Spanish through extended visits from relatives and during holidays in Mexico. This, of course, varied greatly from participant to participant. They started their schooling in the community language, English, but Table 9.1a Language profile of participants in the experimental group Group

Sub-groups

H

Children (Pre-)adolescents

Age range

n

HL

DL

L3

6–9

5

Mexican Spanish

Canadian English

Canadian French

12–16

7

On Heritage Accents

217

were soon enrolled in a French immersion program. Exposure to French started at around the age of 5, and consisted of 50 minutes per school day during the fi rst year. It then increased to different degrees in different grades including occasional full school years, as was the case in Grade 2, for example. In addition, we collected data from three control groups: one group of six L1 English/L2 French bilinguals, henceforth Group B; and two groups of monolinguals made up of 12 native speakers of Mexican Spanish and six native speakers of Canadian French, more specifically Quebec French. These two groups will be referred to as Group M and Group Q, respectively. A summary of their language background is provided in Table 9.1b. The bilinguals’ ages ranged from 12 to 17 years (mean age: 15.5). They were recruited among our HSs’ classmates. The main aim behind including this control group was to obtain baseline data for Canadian English, on the one hand and, on the other hand, to have VOT reference values for immersion French, i.e. the French our HSs are mostly, if not exclusively, exposed to. Ideally, we would have had a larger group and a sub-group of children as well, in order to be a better fit for comparison with our experimental group. However, many of the students enrolled in French immersion classes were also HSs of various languages and were not included in the study. In Group M, the five children’s ages ranged from 6 to 9 years (mean age: 7.4), whereas the seven (pre-)adolescents’ ages ranged from 12 to 16 years (mean age: 13.7). They were selected from a larger pool of participants (10 children and 10 adolescents) to age-match those in the experimental group as closely as possible. They were all being raised in Spanish speaking homes by monolingual parents in Mexico City. Although they had English classes at school, they considered their proficiency in English to be low and reported not using any languages other than Spanish in their daily lives. Lastly, in Group Q, the participants’ ages ranged from 15 to 19 years (mean age: 16.5). They were recruited from two small towns close to Quebec City. Much like the Spanish speakers in Group M, they reported using English mainly in the school setting during English classes just a few hours a week, and French for all other purposes in their daily lives. However, unlike them, they self-assessed their English proficiency as intermediate. Table 9.1b Language profile of participants in the control groups Group

Sub-groups

Age range

n

L1

L2

L3

B

(Pre-)adolescents

12–17

6

Canadian English

Canadian French

–

M

Children

6–9

5

Mexican Spanish

English

–

(Pre-)adolescents

12–16

7

(Pre-)adolescents

15–19

6

Quebec French

Canadian English

–

Q

218

An Anthology of Bilingual Child Phonology

Data collection: Instruments and procedure

Participants performed one experimental task that consisted of naming 23 objects portrayed in separate slides and presented via a slideshow on a laptop computer. We opted for a visual task instead of presenting them with word lists to read because we wanted to avoid any potential issues related to the children’s reading skills. However, this introduced a different complication: it was unclear whether all the participants would know the words we wanted to elicit. Moreover, the same image could be named differently by different participants. To ensure they produced the intended target words, participants were asked to engage in a game-like activity prior to naming the images. This practice exercise proceeded as follows: a set of 23 printed images were displayed on a table, and the participants had to listen to a recording of the 23 target words read by a monolingual of the language in question. They were instructed to match the objects in the cards to the words they heard. It became obvious at this point that the bilingual and trilingual participants knew most of the words in English and Spanish, but were unsure about some French words. This could be due to two reasons: first, French was their weakest language and, secondly, the French words used as stimuli could be less frequent than the English/Spanish ones. Nonetheless, this game ensured that all participants produced the intended words in the end. Bilingual and trilingual participants were tested on separate occasions (one per language).

Data analysis

Out of the 23 images, 17 depicted objects whose names were used as stimuli (see Appendix A). These target words were: (1) either disyllabic or monosyllabic, with a (2) ‘voiceless stop + low/mid vowel’ sequence, in (3) stressed, word-initial position. Although a total of 1128 tokens were collected for analysis, 24 tokens needed to be excluded because they were mispronounced. They corresponded mostly to French words produced either by the HSs or the bilingual controls. The software Praat 6.0.12 (Boersma & Weenink, 2016) was used to perform the acoustic analysis of the remaining 1104 tokens. VOT was measured from the point of consonant release to the onset of voicing for vowel production, and it is reported in milliseconds (ms). It is worth mentioning that, initially, we intended to analyze children’s and (pre-)adolescents’ data separately. There seems to be evidence that children attain adult-like VOT in different languages at varying ages. For instance, Splendido (2017: 58) mentions that French monolinguals produce longer VOT values than adults ‘probably even after the age of 10’. Moreover, according to Lowenstein and Nittrouer (2008), when it comes to VOT production for voiceless stops and at least until the age of 7 years, children display greater intra-speaker variability than adults. We were

On Heritage Accents

219

then concerned about causing an unnecessary confound by mixing children and (pre-)adolescents. However, there were several reasons that made us decide against dividing them. First, we were unable to recruit monolingual French and English-French bilingual children, which clearly prevented us from creating sub-groups in some cases. Secondly, our sample size was already small, and this division would have made our groups even smaller, while increasing the number of comparisons to be made. Lastly, if we wanted to take into account the fact that our trilinguals started learning each of their languages at different ages, calculating a cut-off that would be appropriate for all languages alike seemed extremely complicated, to say the least. Fortunately, when compared by the means of a Mann–Whitney–Wilcoxon test, neither the trilingual nor the monolingual children differed significantly from the (pre-)adolescents. This reassured us in our decision to collapse these sub-groups. Note that these fi ndings are briefly touched upon in the next section. For the sake of completeness, all means per stop and group, along with the p-values obtained for each comparison, are provided in Appendix B. Results

We fi rst report the VOT means produced by our monolingual and bilingual controls. They constitute the baseline against which our trilingual participants will be compared. In turn, we present the means produced by our trilinguals in each of their languages. Lastly, in order to answer our second research question, we compare our trilinguals’ three languages. To uncover whether or which between- and within-group differences were significant, we performed non-parametric tests, due to the small sample size. More specifically, we chose Kruskal–Wallis tests (for three-way comparisons), followed by Mann–Whitney–Wilcoxon tests (for paired comparisons) as required. An alpha value of 0.05 was assumed. Monolingual controls

As can be seen in Table 9.1b, the Mexican monolinguals were initially divided into (pre-)adolescents and children. Once it was determined, through a Mann–Whitney–Wilcoxon test, that there were no significant diff erences between them regarding the mean values for either /p/ (p = 0.432) or /t/ (p = 0.343) or /k/ (p = 0.202), both groups were collapsed into one. Together, their means per stop are 13.69 ms for /p/, 13.89 ms for /t/ and 28.85 ms for /k/ (separate values are provided in Appendix B), which agree with native values usually reported in the literature for adult, monolingual speakers of Spanish (for example, 13.10 ms for /p/, 14 ms for /t/ and 26.5 ms for /k/ in Rosner et al., 2000). These values are portrayed in Figure 9.2 (left-most bars).

220

An Anthology of Bilingual Child Phonology

100

Mean VOT (in ms)

80

60 /p/ /t/ 40

/k/

20

0 Group M

Group Q

Figure 9.2 Mean VOT values produced in Spanish (Group M) and French (Group Q) by the monolingual controls

Also in Figure 9.2 (right-most bars), we provide a graphical representation of the mean VOT values obtained for all three stops (/p/ = 19.33 ms, /t/ = 24.61 ms and /k/ = 45.67 ms) from Group Q, the monolingual Francophones. For the most part, these means also agree with monolingual values reported in previous studies (for example, 18 ms for /p/, 23 ms for /t/ and 32 ms for /k/ in Caramazza et al., 1973), although the mean for /k/ is slightly higher than expected. Bilingual controls

It was not possible to recruit monolingual English speakers.4 In the absence of such a group, our bilingual participants’ data serve as the baseline for comparison regarding the DL. Their mean VOT values, portrayed in Figure 9.3 (left-most bars), are the following: /p/ = 63.28 ms, /t/ = 82.19 ms and /k/ = 84.65 ms. They nicely agree with the 62 ms for /p/ and the 90 ms for /k/ reported by Caramazza et al. (1973) for monolingual adult speakers of Canadian English. Although the value for /t/, 70 ms, is slightly lower, this difference is not important. All means without exception fall within the range of 60–100 ms proposed by Lisker and Abramson (1964) for this set of stops. While these bilinguals produce native-like values in their L1, they fail to do so in their L2, as is evident from the following means: 54.81 ms for /p/, 57.47 ms for /t/ and 71.06 ms for /k/ (see right-most bars in Figure 9.3). Not only do they depart from target-like norms in the literature (Caramazza et al., 1973), but they also differ from our monolingual controls. More importantly, these values exceed the 40 ms mark after which aspiration

On Heritage Accents

221

100

Mean VOT (in ms)

80

60 /p/ /t/ 40

/k/

20

0 L1 English

L2 French

Figure 9.3 Mean VOT values produced in L1 English and L2 French by the bilingual controls

could start to be audible. In addition, we performed a cross-language comparison for this group. While non-significant differences were found for /p/ (U = 12.5, p = 0.378) and /k/ (U = 10, p = 0.200), the statistical test uncovered a significant difference for /t/ (U = 3, p = 0.016). In general, these findings point to shared rather than separate VOT systems for these bilinguals’ two languages (but note the significant difference reported for /t/). Experimental group Spanish as a heritage language

As a group, our trilinguals produced means of 18.28 ms for /p/, 20.42 ms for /t/ and 33.18 ms for /k/, as shown in Figure 9.4 (left-most bars). The Mann–Whitney test performed to compare them to the monolinguals raised in Mexico (see Figure 9.2) indicated a significant difference only for /t/ (U = 21.5, p = 0.004), and non-significant differences neither for /p/ (U = 44, p = 0.106) nor for /k/ (U = 58.5, p = 0.436). For the most part, they behave like and should be perceived as Spanish monolinguals. However, the significant difference uncovered for /t/, as well as the slightly higher means in general (if compared to the monolinguals), leave room to speculate about a potential ‘heritage’ accent, as we do in the Discussion section. English as a dominant language

In the DL, the acoustic measurements yielded means of 65.65 ms for /p/, 74.86 ms for /t/ and 86.03 ms for /k/, as shown in Figure 9.4 (middle bars). When compared to their bilingual classmates (see Figure 9.3), the Mann– Whitney test failed to reveal any significant differences for any of the stops:

222

An Anthology of Bilingual Child Phonology

100

Mean VOT (in ms)

80

60 /p/ /t/ 40

/k/

20

0 Spanish

English

French

Figure 9.4 Mean VOT values produced in Spanish (HL), English (DL), and French (L3) by the experimental group

/p/ (U = 29, p = 0.512), /t/ (U = 26, p = 0.349), /k/ (U = 52, p = 0.640). These trilinguals behave like other L1 speakers of their DL (see English means reported in Appendix C). French as a third language

In their weakest language, the acoustic measurements yielded means of 52.52 ms for /p/, 62.23 ms for /t/ and 68.92 ms for /k/, as shown in Figure 9.4 (right-most bars). At a first glance, we observe that they pattern with their bilingual classmates (see Figure 9.3), and clearly depart from the monolingual norm (see Figure 9.2). The Kruskal–Wallis test performed on these results determined that there were significant differences regarding all stops: χ2 (2) = 10.526, p = 0.005 in the case of /p/; χ2 (2) = 9.125, p = 0.010 in the case of /t/; and χ2 (2) = 10.560, p = 0.005 in the case of /k/. The Mann–Whitney–Wilcoxon tests that ensued revealed that the bilinguals and trilinguals did not differ significantly from each other, while the monolinguals differed significantly from the two other groups. For ease of presentation, the p-values obtained for each voiceless stop and pair of groups are summarized in Table 9.2, which also contains the relevant means with the corresponding standard deviations. All means and related standard deviations and confidence intervals for this and all other groups are provided in Appendix C. Cross-language comparison

Figure 9.4 provides a visual representation of all means for all three voiceless stops across the three languages of our trilingual participants. On fi rst impression, one could suppose that they have created separate

On Heritage Accents

223

Table 9.2 Mean VOT comparison for /p t k/ in French as an L1, L2, and L3 Stop

Mean (SD)

p

L1 Fr

L2 Fr

L3 Fr

L1–L2

L1–L3

L2–L3

/p/

19.33 (8.05)

54.81 (18.63)

52.52 (26.16)

0.004*

0.004*

0.596

/t/

24.61 (7.54)

57.47 (25.57)

62.23 (21.81)

0.048*

0.003*

0.472

/k/

45.67 (19.06)

71.06 (32.14)

68.92 (21.92)

0.003*

0.005*

0.572

Notes: L1 Fr: monolingual values (group Q); L2 Fr: bilingual values (Group B); L3 Fr: trilingual values (Group H). Asterisks indicate statistical significance (p < 0.05).

Table 9.3 Mean VOT comparison for /p t k/ in Spanish (HL), English (DL) and French (L3) Stop

Mean (SD)

p

Sp

En

Fr

Sp-En

Sp-Fr

En-Fr

/p/

18.28 (10.45)

65.65 (22.72)

52.52 (26.16)

0.000*

0.001*

0.146

/t/

20.42 (9.97)

74.86 (21.81)

62.23 (21.89)

0.000*

0.001*

0.712

/k/

33.18 (14.11)

86.03 (21.46)

68.92 (21.92)

0.000*

0.004*

0.063

Asterisks indicate statistical significance (p < 0.05).

categories for their strongest languages, English and Spanish. Less clear is whether they have created a separate category for French as well, or if this language patterns with English. In order to get a better understanding of what these means indicate, we can turn to our statistical results. Again, a Kruskal–Wallis test established that there were significant differences regarding all stops: χ2 (2) = 22.635, p = 0.000 in the case of /p/; χ2 (2) = 23.788, p = 0.000 in the case of /t/; and χ2 (2) = 23.084, p = 0.000 in the case of /k/. This was followed by post hoc Mann–Whitney– Wilcoxon tests, which allowed us to determine differences across language pairs in order to fi nd out where the significant difference lay. For ease of presentation, the p-values obtained for each voiceless stop and pair of languages are summarized in Table 9.3, which also contains the relevant means with the corresponding standard deviations. As is evident from the means and statistical results reported in Table 9.3, these trilinguals have created separate categories for their two stronger languages, while their L3 patterns with their DL. Discussion

In this study we set out to inquire into the extent to which our early trilinguals behaved as monolinguals in their HL regarding the production

224

An Anthology of Bilingual Child Phonology

of VOT patterns. We also wanted to know whether our results would suggest the presence of a heritage accent. Having found non-significant differences for two out of the three stops under investigation, we believe our H1 has been mostly confi rmed. Overall, our trilingual HSs produced native-like (i.e. short lag) VOT. The slightly higher values we observed can be explained as mild influence of the DL on the HL, as we will argue. Answering the second part of the first research question based on our fi ndings is more complicated. Ideally, we would have set up a foreign accent rating task, as Gabriel et al. (2016) did. Alternatively, we have scrutinized our data and identified tokens produced with values of over 40 ms. In the case of our monolinguals, these instances account for 2.5% of the data, never exceed 45 ms and are limited to /k/. It is rather safe to argue that, even if aspiration were audible at exactly 40 ms, these monolinguals’ voiceless stops would not sound accented to other monolinguals. In the case of our trilinguals, these ‘deviant’ instances account for 10% of the data, go from 42 to 81 ms, and are present in /p/, /t/ and /k/. We have reason to believe that these measurement differences would result in occasional production of accented voiceless stops on the part of our HSs, as they did in the case of Gabriel et al. (2016). Given that the percentage of potentially accented stops is rather low, we argue that this agrees with general claims that HSs sound ‘off ’ or have a slight accent, and may not be consistently taken for non-native speakers. Intuitively, and generally speaking, one of the characteristics that could set heritage accents apart from foreign accents might be a rather high production of target-like speech, with infrequent instances of non-target-like production, as we show here. Interestingly, in a germane study (Llama & Cardoso, 2018), late learners of L3 Spanish produce voiceless stops with less variability than these early trilinguals do. What we mean is that while late learners produce VOT values that are consistently higher than the norm (range of 41–74 ms, mean: 55.17 ms for /t/), 5 these HSs display more variability (range of 10–81 ms, mean: 23.17 ms for /t/), 6 without generally producing intermediate values. Based on this observation, chances are that VOT causes late L3 learners to be consistently perceived as accented, while it only causes HSs to sound accented intermittently. Admittedly, our data do not allow us to claim that our trilinguals would be perceived as accented, nor to further elaborate on this issue. Our second research question was whether they had created separate categories for their different languages, and if they displayed monolingual-like patterns in English and French. Based on a related study (Llama & López-Morelos, 2016), we had hypothesized the creation of separate categories for Spanish and English, and that French would align with English. Moreover, Gabriel and colleagues (2016) reported that, for French /t/, their Mandarin-German dominant bilinguals were only slightly more target-like than German monolinguals, which suggests stronger transfer from the DL to the L3. Based on this, we would predict stronger

On Heritage Accents

225

influence from English than from Spanish. We acknowledge that, in the absence of these studies, we would have predicted positive transfer from the HL, given the typological similarity between these two Romance languages with regard to stops. Our results reveal the creation of mostly separate systems for Spanish and English although traces of influence (mainly from the DL on the HL) are found, consistent with the claim that L1 and L2 systems are never fully independent (Flege, 1995). Much like in the case of Llama and López-Morelos (2016), our fi ndings also indicate transfer from the DL to the L3. The similar values and lack of significant differences between the French and English stops suggest that our trilinguals treat these stops as belonging to a shared system. Creating a shared system between French and one of the previously acquired languages (rather than creating a hybrid one consisting of values intermediate between one of the Romance languages and English) would have been the right strategy, had they merged French with their HL. In that case, their production of French voiceless stops would have very likely matched that of our monolingual controls (Group Q). It is worth noting, however, that our participants are enrolled in an immersion program, created for Anglophones (and English-dominant speakers of other languages) living in Canadian regions where French is a community minority language (Poljak, 2015). They learn the language mostly from their classmates and their (often English-dominant) teachers, and use it to interact among themselves. In this kind of scenario, students tend to develop a very particular kind of French, which certainly displays transfer from English. This could explain our results and raises the question of whether they would care to produce non-accented French voiceless stops in this context. Consciously, or most likely not, it may be more desirable for them to sound like their peers and be perceived as members of the group (for a discussion regarding a phenomenon related to language and membership in social groups, namely ‘[ethnic] group affiliation’, and its potential effect on L2 pronunciation accuracy, see Gatbonton et al., 2005). Concluding Remarks

This study makes a contribution to the sub-field of heritage phonetics and phonology by presenting results from an understudied group: HSs who are also L3 learners. Our fi ndings provide further support to claims that successive bilinguals can create separate categories for voiceless stops, and that HSs perform close to monolinguals in their HL. They also point to slight deviations from the monolingual norm of the HL, which we argue could reveal traces of a heritage accent. Due to the small size of our sample and, crucially, to the lack of accent judgements, we cannot offer solid conclusions in this respect. This topic should be further investigated in future studies that combine production data and accent ratings in larger groups, as well as in more than one segment.

226

An Anthology of Bilingual Child Phonology

In addition, we have shown that, although the HL would have been the right source of transfer towards the L3, participants relied on the DL instead. Regarding production of French voiceless stops, they were indistinguishable from their L1 English classmates. This could point to a trend to transfer mainly from the DL to the L3 for VOT (Gabriel et al., 2016; but see Bondarenko, 2018, for counter evidence7 ), even when it is not the most advantageous option. In the case of our trilinguals, however, another explanation is plausible: they reproduce the VOT patterns they are exposed to. Nonetheless, we should be cautious about this interpretation. In a study on global foreign accent, Lloyd-Smith et al. (2017) found that, despite having learnt L3 English in Germany (and having been exposed to the Germanaccented English spoken by their teachers and peers), not all of the Turkish HSs were identified as L1 German learners of English. The higher their proficiency and reported use of Turkish, the less likely they were to be taken for L1 German speakers. The authors concluded that there was no common source of accent towards the L3 among their HSs. Our results, together with the review of previous work, raise interesting questions regarding the interplay of the heritage and dominant languages and their role during L3 acquisition, which deserve further attention in future studies.

Notes (1) We do not use the labels L1 for Spanish and 2L1 or L2 for English for reasons we are about to explain. These HSs started being exposed to English in earnest at the age of 3 years (see Methodology section on participants). According to McLaughin’s (1978) age criterion, acquiring two languages before the age of 3 can be referred to as simultaneous acquisition, whereas learning the second language after that age can be described as successive bilingualism. In that case, our HSs could be considered as successive bilinguals and, by extension, successive trilinguals. Should we want to make a fi ner distinction between the HL and the DL versus the L3, which they started learning at the age of 5, we could use Meisel’s (2008: 59) tentative age ranges to say that our trilingual participants are acquiring their two main languages as two L1s (2L1 ≤ 3) and French as a child L2 (≥ 4; for a more detailed explanation of what this difference entails, see Meisel, 2008). However, these defi nitions and cut-offs are arbitrary, and alternative criteria have been proposed. Our HSs could be straddling the line between categories. To somewhat circumvent this terminological issue, we refer to them using a broad term, early trilinguals, and to their languages as HL, DL and L3. (2) According to Bandeira and Zimmer (2012), Pomeranian (and/or German) is the L1 of the majority of the population in the town where their study was conducted. Interestingly, this variety of Low German has been preserved in that isolated area of Southern Brazil, but it is no longer spoken in its territory of origin. Based on their explanation regarding the learning and use of Pomeranian and BP in the region, we consider Pomeranian to be a HL and their participants as HSs. Nonetheless, they differ from other HS populations mentioned in our study in that they are not secondgeneration speakers, and that Pomeranian may at times compete with the majority language in Brazil, BP, as the community language. (3) Five of these speakers, as well as four participants in Group B, three participants in Group Q and four participants in Group M, were taken from Llama and LópezMorelos (2016).

On Heritage Accents

227

(4) As a reviewer rightly pointed out, none of our control groups is monolingual in a strict sense. Instead, they display different levels of L2 mastery. Although it is common to compare bi/multilinguals to monolinguals, this practice has been questioned in recent studies since ‘monolingual productions are not comparable to those of speakers with multiple systems that yield bidirectional influence’ (Cabrelli Amaro, 2013: 103). Cabrelli Amaro (2013) proposes using the participants’ L1, L2 and L3 data over time (i.e. they act as their own controls). However, this is not always possible. An alternative is to compare them to controls who speak the same language combinations and can display influence from the other known languages (see control groups in Cabrelli Amaro, 2017). In this sense, our bilingual control group is an adequate baseline for comparison. (5) These sample values are taken from a germane study (Llama & Cardoso, 2018) which employs a comparable stimuli list. Words were also produced in isolation. The participant who produced them is an adult L1 English-L2 French speaker who started learning Spanish in high school. (6) These sample values are taken from one of the children in our experimental group. It has to be noted, however, that this child displays the largest range of all of our participants. (7) Although Bondarenko’s participants seem to rely more on the HL for voiceless stop production (thus providing counter-evidence to this trend), they exhibit more influence from English during vowel production, which is supportive of a potential trend to transfer from the DL. These fi ndings highlight the importance of comparing various segments from the same group of speakers, as they may reveal different transfer patterns.

Appendix A Spanish list

pato ‘duck’, pala ‘shovel’, pera ‘pear’, perro ‘dog’, pollo ‘chicken’, pozo ‘well’, taco ‘taco’, taza ‘cup’, tela ‘fabric’, tecla ‘key’, toro ‘bull’, torre ‘tower’, casa ‘house’, cama ‘bed’, queso ‘cheese’, coco ‘coconut’, copa ‘wineglass’ French list

patte ‘paw’, pas ‘step’, pêche ‘peach’, père ‘father’, pomme ‘apple’, porte ‘door’, tarte ‘pie’, tasse ‘cup’, tête ‘head’, terre ‘earth’, thon ‘tuna’, toque ‘hat’, carte ‘map’, casque ‘helmet’, quai ‘dock’, corde ‘rope’, coq ‘rooster’ English list

pants, parrot, pencil, penny, pocket, potty, tag, tank, ten, tent, tongue, toy, car, candy, kettle, corn, comb

228

An Anthology of Bilingual Child Phonology

Appendix B: Mean VOT comparison between (pre-)adolescents and children in Groups H (trilingual HSs) and M (monolingual speakers of Mexican Spanish)

Group M

/p/

p

/t/

p

/k/

p

p-ados

14.95

0.432

15.02

0.343

30.43

0.202

children

11.93

Spanish (L1)

Group H

12.3

26.64

Spanish (HL) p-ados

18.29

children

18.07

0.876

21.09

0.639

19.21

33.66

0.876

32.55

English (DL) p-ados

61.81

children

71.03

0.372

73.85

0.935

76.27

79.51

0.167

95.16

French (L3) p-ados

49.48

children

56.73

0.372

62.83

0.808

61.78

68.66

0.935

64.88

Appendix C: VOT means, standard deviation (SD) and confidence intervals (CI) for all groups and languages Mean

SD

CI

Group H Fr

CI

Group B

Mean

SD

CI

Group Q

52.52

26.16

46.33 58.71

54.81

18.63

48.47 61.15

19.33

8.05

16.59 22.07

/t/

62.23

21.89

56.58 67.88

57.47

25.57

47.94 67.00

24.61

7.54

22.04 27.18

/k/

68.92

21.92

63.21 74.63

71.06

32.14

59.08 83.04

45.67

19.06

38.56 52.78

Group B

/p/

65.65

22.72

60.31 70.99

63.28

18.25

57.07 69.49

/t/

74.86

21.81

70.09 80.41

82.19

16.84

76.46 87.49

/k/

86.03

21.46

80.49 91.57

84.65

36.39

70.85 98.45

Group H Sp

SD

/p/

Group H En

Mean

Group M

/p/

18.28

10.45

15.81 20.75

13.69

6.49

12.17 15.21

/t/

20.42

9.97

18.06 22.78

13.88

4.88

12.83 14.93

/k/

33.18

14.11

29.54 36.82

28.85

12.82

25.56 32.14

On Heritage Accents

229

References Avery, P. and Ehrlich, S. (1995) Teaching American English Pronunciation. Oxford: Oxford University Press. Bandeira, M.T. and Zimmer, M.C. (2012) The dynamics of interlanguage transfer of VOT patterns in multilingual children. Linguagem & Ensino 15, 341–364. Benmamoun, A., Montrul, S. and Polinsky, M. (2013) Heritage languages and their speakers: Opportunities and challenges for linguistics. Theoretical Linguistics 39 (3–4), 129–181. Boersma, P. and Weenink, D. (2016) Praat: Doing Phonetics by Computer [Computer program]. Version 6.0.12. See http://www.praat.org (accessed 28 January 2016). Bondarenko, M. (2018) Acquisition of Spanish by heritage speakers of Ukrainian and Polish: A phonetic and phonological account. ProQuest Dissertations & Theses A & I (UMI No. 20363 04246). Cabrelli Amaro, J. (2013) Methodological issues in L3 phonology. Studies in Hispanic and Lusophone Linguistics 6 (1), 101–117. Cabrelli Amaro, J. (2017) Testing the phonological permeability hypothesis: L3 phonological effects on L1 versus L2 systems. International Journal of Bilingualism 21 (6), 698–717. Caramazza, A., Yeni-Komshiam, G.H., Zurif, E.B. and Carbone, E. (1973) The acquisition of a new phonological contrast: The case of stop consonants in French-English bilinguals. Journal of the Acoustical Society of America 54, 421–428. Cook, V.J. (1992) Evidence for multicompetence. Language Learning 42 (4), 557–591. Deuchar, M. and Clark, A. (1996) Early bilingual acquisition of the voicing contrast in English and Spanish. Journal of Phonetics 24 (3), 351–365. Dittmers, T., Gabriel, C., Krause, M. and Topal, S. (2018) The production of voiceless stops in multilingual learners of English, French, and Russian: Positive transfer from the heritage languages? In M. Belz, C. Mooshammer, S. Fuchs, S. Jannedy, O. Rasskazova and M. Zygis (eds) Proceedings of the 13th Conference on Phonetics and Phonology in the German-speaking Countries (pp. 41–44). Berlin: ZAS. Fabiano-Smith, L. and Bunta, F. (2012) Voice onset time of voiceless bilabial and velar stops in three-year-old bilingual children and their age-matched monolingual peers. Clinical Linguistics and Phonetics 26 (2), 148–163. Flege, J.E. (1991) Age of learning affects the authenticity of voice-onset time (VOT) in stop consonants produced in a second language. Journal of the Acoustical Society of America 89, 395–411. Flege, J.E. (1995) Second language speech learning: Theory, fi ndings, and problems. In W. Strange (ed.) Speech Perception and Linguistic Experience: Issues in Crosslanguage Research (pp. 233–277). Timonium, MD: York Press. Flege, J.E. (1999) Age of learning and second language research. In D.P. Birdsong (ed.) Second Language Acquisition and the Critical Period Hypothesis (pp. 101–131). Hillsdale, NJ: Lawrence Erlbaum. Fowler, C.A., Sramko, V., Ostry, D.J., Rowland, S.A. and Hallé, P. (2008) Crosslanguage phonetic influences on the speech of French-English bilinguals. Journal of Phonetics 36, 649–663. Gabriel, C., Kupisch, T. and Seoudy, J. (2016) VOT in French as a foreign language: A production and perception study with mono- and multilingual learners (German/ Mandarin-Chinese). In F. Neveu, G. Bergounioux, M.-H. Côté, J.-M. Fournier, S. Osu, P. Planchon, L. Hriba and S. Prévost (eds) Actes du 5e Congrès Mondial de Linguistique Française (pp. 1–14). Paris: ILP/EDP Sciences. Gatbonton, E., Trofi movich, P. and Magid, M. (2005) Learners’ ethnic group affi liation and L2 pronunciation accuracy: A sociolinguistic investigation. TESOL Quarterly 39, 489–511. Grosjean, F. (1989) Neurolinguists, beware! The bilingual is not two monolinguals in one person. Brain and Language 36, 3–15.

230

An Anthology of Bilingual Child Phonology

Johansson, V., Horne, M. and Strömqvist, S. (2001) Final aspiration as a phase boundary cue in Swedish: The case of ‘that’. Working Papers, Lund University, Department of Linguistics 49, 78–81. Kim, J.Y. (2011) Discrepancy between perception and production of stop consonants by Spanish Heritage speakers in the US. Unpublished MA thesis, Korea University. Kim, J.Y. (2016) The perception and production of prominence in Spanish by heritage speakers and L2 learners. Unpublished Doctoral dissertation, University of Illinois at Urbana-Champaign. Kupisch, T. (2019) 2L1 simultaneous bilinguals as heritage speakers. In M. Schmid and B. Köpcke (eds) The Oxford Handbook of Language Attrition. Oxford: Oxford University Press. Kupisch, T., Lein, T., Barton, D., Schröder, D.J., Stangen, I and Stoehr, A. (2014) Acquisition outcomes across domains in adult simultaneous bilinguals with French as weaker and stronger language. Journal of French Language Studies 24 (3), 1–30. Lisker, L. and Abramson, A.S. (1964) A cross-language study of voicing in initial stops: Acoustical measurements. Word 20, 384–422. Llama, R. and Cardoso, W. (2018) Revisiting (non-)native influence in VOT production: Insights from advanced L3 Spanish. Languages 3, 30. Llama, R. and López-Morelos, L.P. (2016) VOT production by Spanish heritage speakers in a trilingual context. International Journal of Multilingualism 13, 444–458. Lloyd-Smith, A., Gyllstad, H. and Kupisch, T. (2017) Transfer into L3 English. Linguistic Approaches to Bilingualism 7, 131–162. Lloyd-Smith, A., Gyllstad, H., Kupisch, T. and Quaglia, S. (2018) Heritage language proficiency does not predict syntactic CLI into L3 English. International Journal of Bilingual Education and Bilingualism. doi:10.1080/13670050.2018.1472208 Lord, G. (2005) (How) can we teach foreign language pronunciation? On the effects of a Spanish phonetics course. Hispania 88, 557–567. Lowenstein, J.H. and Nittrouer, S. (2008) Patterns of acquisition of native voice onset time in English-learning children. Journal of the Acoustical Society of America 124, 1180–1191. MacLeod, A.A.N. and Stoel-Gammon, C. (2009) The use of voice onset time by early bilinguals to distinguish homorganic stops in Canadian English and Canadian French. Applied Psycholinguistics 30, 53–77. McLaughlin, B. (1978) Second Language Acquisition in Childhood. Hillsdale, NJ: Lawrence Erlbaum. Meisel, J.M. (2008) Child second language acquisition or successive fi rst language acquisition? In B. Haznedar and E. Gavruseva (eds) Current Trends in Child Second Language Acquisition (pp. 55–82). Amsterdam: John Benjamins. Montrul, S. (2015) The Acquisition of Heritage Languages. Cambridge: Cambridge University Press. Paradis, J. (2001) Do bilingual two-year-olds have separate phonological systems? International Journal of Bilingualism 5, 19–38. Pascual y Cabo, D. and De la Rosa-Prada, J. (2015) Understanding the Spanish heritage language speaker/learner. EuroAmerican Journal of Applied Linguistics and Languages 2, 1–10. Polinsky, M. and Kagan, O. (2007) Heritage languages: In the ‘wild’ and in the classroom. Language and Linguistics Compass 1 (5), 368–395. Poljak, L. (2015) A search for immersionese: Identifying French immersion accents in BC. Unpublished MA thesis, Simon Fraser University. Rao, R. and Ronquest, R. (2015) The heritage Spanish phonetic/phonological system: Looking back and moving forward. Studies in Hispanic and Lusophone Linguistics 8, 403–414. Riney, T.J. and Tagaki, N. (1999) Global foreign accent and voice onset time among Japanese EFL speakers. Language Learning 49, 275–302.

On Heritage Accents

231

Rosner, B.S., López-Bascuas, L.E., García-Albea, J.E. and Fahey, R.P. (2000) Voice-onset times for Castilian Spanish initial stops. Journal of Phonetics 28, 217–224. Splendido, F. (2017) The initial development of voice onset time in early successive FrenchSwedish bilinguals. In M. Yavaş, M. Kehoe and W. Cardoso (eds) RomanceGermanic Bilingual Phonology (pp. 56–78). Sheffield: Equinox. Sundara, M., Polka, L. and Baum, S. (2006) Production of coronal stops by simultaneous bilingual adults. Bilingualism: Language and Cognition 9, 97–114. Van de Weijer, J. and Kupisch, T. (2015) Voice onset time in heritage speakers and second language learners of German. In E. Babatsouli and D. Ingram (eds) Proceedings of the International Symposium on Monolingual and Bilingual Speech (pp. 414–420). Chania: ISMBS. Yavaş, M. (2009) Factors influencing the VOT of English long lag stops and interlanguage phonology. In M.A. Watkins, A.S. Rauber and O. Baptista (eds) Recent Research in Second Language Phonetics/Phonology: Perception and Production (pp. 244–256). Newcastle upon Tyne: Cambridge Scholars.

10 Linguistic Outcomes and the Role of Phonology in Typically Developing and Late Talking Toddlers Kakia Petinou and Loukia Taxitari

Introduction

Language acquisition occurs fast, effortlessly and without any formal instruction (Chomsky, 1965; Lenneberg, 1967). It constitutes a complex and dynamic process encompassing all linguistic subsystems which interact in a synergistic manner. The ability for speech production as well as a cognitive-linguistic system for mapping sound onto meaning develops within the fi rst years of life (Locke, 1993). Despite significant variation in developmental trajectories, pivotal language milestones are expected to appear during the first and second years of life (see Crystal, 1997, for an extensive overview). These include the production of fi rst words around the age of 12 months, the emergence of a critical lexical mass followed by lexical acceleration between the ages of 18 and 24 months and grammatical emergence around the age of 24 months (Marchman & Thal, 2005). Research on interconnectedness among linguistic subsystems provides evidence for the existence of mutual influences which operate synergistically (e.g. lexicon vis-à-vis grammar), in the sense that one linguistic subsystem facilitates the emergence of other comparable components (e.g. segments provide the basis for lexical development) (Bates & Goodman, 2001; Marchman & Bates, 1994; Wiethan et al., 2016). Such a theoretical stance derives from the continuity hypothesis, traced back to infancy, proposing a link between early phonetic skills and later phonological and lexical diversity. Specifically, studies on pre-linguistic infant vocalizations provide strong evidence for the interaction between babbling and later phonological development (Morgan & Wren, 2018; Vihman et al., 1985). That is, children’s phonological preferences in early meaningful word targets are directly linked to the phonetic forms produced during the babbling period (Vihman et  al., 1985). It has been proposed that these 232

Linguistic Outcomes and the Role of Phonology in TD and LT Toddlers

233

frequently practiced phonetic and vocal schemes allow the creation of productive templates that enhance the ‘articulatory-auditory loop’. Consequently, such practices enhance the motor-kinesthetic feedback mechanism responsible for the ‘strengthening’ and ‘sharpening’ of phonological and lexical representations (Locke, 1993; Stoel-Gammon & Sosa, 2007; Vihman, 1992). On parallel grounds, Chapman et al. (2003) found that true canonical babbling at the age of 13 months was correlated with speech and language measures at the age of 21 months. Detailed analyses of the phonological and lexical characteristics of the fi rst 100 words uttered by English speaking toddlers supported a synergistic, bidirectional interaction between phonetic and lexical developmental patterns during the first two years of life (Sosa & Stoel-Gammon, 2012). It has been suggested that certain language subsystems develop in tandem and appear to influence cognitive-linguistic representations (Beckman et al., 2007). For example, expressive vocabulary growth and productive ‘practice’ contribute to the child’s gradual development and sharpening of the ‘fi ne-grained’ phonological patterns inherent to the words the child produces over time (Metsala & Walley, 1998). These interactions between early phonetic and later lexical skills during the first two years of life are based on phonetic segments serving as the ‘building blocks’ for subsequent lexical development (Stoel-Gammon, 1989; for an extensive review, see Stoel-Gammon, 2011). As such, phonetic development may act as a harness point for the development of expressive vocabulary and the surfacing of word combinations. As children gain a ‘critical mass’ of lexical items (Marchman & Bates, 1994), phonological organization shifts from a ‘gestalt’ to a more analytic form which may impact on the phonological/lexical interphase in the form of phonological bootstrapping (Vihman et al., 1994). The latter has not been limited only to the acquisition of words but has also been studied in relation to the acquisition of syntax. The processing and representation of the speech signal in early language development is suggested to lie at the core of later developments, such as the learning of morphosyntax, by acting as the basis on which infants decipher the language puzzle (Morgan & Demuth, 2014). There has been keen interest in investigating the relationships between lexical and grammatical skills during the second and third years of the child’s life; fi ndings corroborate the existence of significant correlations between early expressive vocabulary and morphosyntactic development, with the former constituting a strong predictor of later morphological and syntactic achievements (Bates et al., 1988). It has also been shown that children start to combine words, and thus their grammatical/syntactic development begins, after they have accumulated a lexicon of around 50 words (e.g. Rescorla, 1989). On parallel lines, expressive vocabulary at the age of 20 months has also been found to be a strong predictor of grammatical/syntactic abilities in the form of mean length of utterance (hereafter MLU) at age 3 years, supporting the

234

An Anthology of Bilingual Child Phonology

existence of linguistic interdependencies between subsystems in line with single-mechanism approaches to language development (McGregor et al., 2005). Further studies, mainly from English corpora, provide converging evidence related to interdependencies between morphosyntactic and lexical abilities. Specifically, Bates and Goodman (2001) investigated the relationship between morphosyntactic abilities in the form of MLU and semantic abilities in the form of expressive vocabulary size (hereafter EVS), as these surface at different points in language development (e.g. from age 2 to 5 years). The fi ndings revealed significant relationships between lexical and grammatical (type of word combinations) skills, again supporting a highly interconnected language system (Weithan et al., 2016). Along these lines, Moyle et al. (2007) investigated longitudinal interactions between lexical and grammatical skills in typically developing (hereafter TD) and late talking (hereafter LT) children across the ages of 2–5 years. They reported strong relationships between lexical and grammatical skills for all children throughout the span of the study, with LTs, however, exhibiting fewer such linguistic interdependencies. Overall, the current literature adheres to the existence of linguistic interdependencies within and across subsystems regardless of causality and directionality in development (Moyle et al., 2007; Stoel-Gammon, 2011; Stoel-Gammon & Sosa, 2007), rendering language acquisition a complex dynamic process (Vihman et al., 1985) which is characterized by a high degree of variability. This variability is manifested between children of different ages but also between children of the same age; although children of a certain age are expected to reach certain milestones, variability is now considered the norm and children exhibit large differences, albeit all following a typical development. This is considered to be the case with different language skills, such as the lexicon (Fenson et  al., 1994), the acquisition of segments (McLeod, 2007) or morphosyntax (Moyle et al., 2007). Nevertheless, the ways in which linguistic subsystems interact during the course of language learning have mainly been examined through the study of interactions between two subsystems, and mainly in English. A simultaneous interaction among all three skills (sounds, lexicon, grammar) warrants further investigation, thus setting the main goal and novelty of the current investigation. Given the acceleration of language development during the second and third years of life, one of the central research questions addressed by the current study revolves around language interconnectedness and variability beyond the age of 2 years. Furthermore, the study investigated the operation of linguistic interconnectedness beyond the 50-word language landmark, allowing us to observe interaction among more advanced language skills during a developmental period that is supposed to be characterized by higher developmental stability (Moyle et al., 2007). Additionally, the children’s level of performance in their phonetic, lexical and grammatical skills at a certain point was used to differentiate them in groups

Linguistic Outcomes and the Role of Phonology in TD and LT Toddlers

235

of high and low performers in determining whether linguistic interconnections between subsystems are affected and how skilled a child is at a specific language skill. Furthermore, the study addressed the interconnectedness between phonetic, lexical and grammatical links in Greek. Despite the plethora of data from English corpora, fi ndings do not necessarily generalize to other languages and/or dialects due to differences in their respective phonetic/ phonological, semantic and grammatical properties (Marton et al., 2006). In the current investigation the focus was on Greek, and more specifically the Cypriot Greek dialect (hereafter CG). Children growing up in Cyprus are exposed to two varieties of the same language, the Standard Modern Greek variety (e.g. Babatsouli, 2019) used mainly in the media, formal interactions and the educational system and regarded as the high variety, and the local dialect, CG, used in everyday interactions and regarded as the low variety. Thus, children in Cyprus are considered bi-dialectal, or bilectal, and are thought to develop at a middle ground on the continuum between monolingualism and bilingualism, sharing linguistic and cognitive traits with both groups (Grohmann et al., 2016). Additionally, CG is a variety with a rich morphological system. Phonetically and morphologically rich languages provide the benchmark for examining crosslinguistic patterns in early language development and shed light onto the existence of language-specific and language-universal patterns of development. Even though developmental language research in CG has received a remarkable amount of attention during the past 10 years (e.g. Tsimpli et al., 2017), the different linguistic skills have been examined independently. Studies on phonological development focused on the age and order of acquisition of different phonetic segments (e.g. Okalidou et al., 2010; Petinou & Okalidou, 2006; Petinou & Theodorou, 2015), while studies on the development of syntax focused on MLU at different ages (e.g. Voniati, 2016) or the acquisition of dialect-specific morphosyntactic elements (Tsimpli et al., 2017). Finally, the development of the lexicon has recently been studied extensively through parental feedback on the CG CDI by Taxitari et al. (2017), while Taxitari (2019) studied the CG morphosyntactic development using the same tool. Petinou and her colleagues (2011) studied all three language skills (phonological, lexical and grammatical) longitudinally in TDs and LTs in their third year. However, the focus of the study was the comparison between TDs and LTs, so the three language skills were not compared directly to one another. The only study examining developmental interdependencies between linguistic subsystems was conducted by Taxitari et al. (2017) using the CG version of the CDI, showing an increase in grammatical elements as the size of the lexicon increased. Apart from this latter study, no other studies have directly examined linguistic interactions in CG-speaking TD youngsters, a fact that adds novelty to the present investigation. Given the sparse data that exist on interactive developmental patterns of CG-speaking toddlers, the study

236

An Anthology of Bilingual Child Phonology

attempted to shed light on the mechanisms that might govern the early stages of the language acquisition process. Thus, examining such relationships might offer an insight into different operations that underlie linguistic interconnectedness among different language skills during the early stages of language acquisition and provide crosslinguistic support regarding the existence of the continuity hypothesis and the dynamic systems approach beyond the age of 2 years and in a bilectal population. In particular, the study examined synchronic links among language domains (within 28 and within 36 months) as well as predictive links among language domains (across 28 and 36 months). Note that due to the variability of development as a function of age, the examiners analyzed the data on the basis of performance according to cluster analyses (low versus high language performers). Specifically, the study set to investigate: • •

interdependencies among early developing linguistic skills; interaction of linguistic subsystems as a function of time and language variability marked through group-cluster analysis, including the phonetic inventory size (PIS), lexical diversity through expressive vocabulary size (EVS) and grammatical development based on mean length of utterance in words (MLUW).

Method Participants

The participants in this study included 31 toddlers (12 girls, 19 boys). The children’s language abilities were studied at two ages: (a) intake at age 28 months (M = 28.6 months, range = 27.6–29.4, SD = 0.47) and (b) reassessment at age 36 months (M = 36.4 months, range = 35.7–36.9, SD = 0.27). These children were a subset of participants from a larger cohort of subjects in a longitudinal investigation regarding linguistic milestones in CG. Note that these participants were those who completed all experimental language protocols at the two testing time points. They were all recruited through different private and state preschools, advertisements in local newspapers and newsletters, as well as through flyers posted to local public places, the Cyprus Association of Speech and Language Therapists, pediatric offices and personal contacts. Parents who expressed willingness to participate signed a written consent form. The study and its protocol received approval from the Council Committee of the Cyprus Institute of Neurology and Genetics. All participants came from CG speaking households and presented with a typical course of language development based on linguistic, hearing and cognitive screenings conducted at intake. All participants were matched for socio-economic status (mid-high) on the basis of maternal education and family income (Cyprus Statistical Service, 2005).

Linguistic Outcomes and the Role of Phonology in TD and LT Toddlers

237

Cognitive non-verbal ability was assessed via clinical observations and with the use of a checklist adapted from the Bayley (1969) Scales of Infant Development on the mental development index (MDI) administered at the time of intake. All participants showed typical non-verbal ability according to this criterion. Expressive and receptive language screening was performed with the use of the Preschool Language Scale-3 (PLS-3: Zimmerman et al., 1992) adapted for GC (for details see Petinou & Spanoudis, 2014). That is, a number of items from the PLS-3 were adapted to fit each child’s language characteristics and parameters of the CG dialect. Case histories regarding medical issues and developmental milestones were also gathered at intake through developmental questionnaires completed by each child’s caregiver. In addition, all children passed hearing screenings at 500, 1000, 2000 and 4000 (6000 and 8000) Hz presented at 25 dB HL according to the guidelines suggested by the American Speech-Language-Hearing Association (2005) using a GSI-38 portable audiometer. Procedures

Two trained research assistants carried out language screening and data collection. Each session lasted approximately 45 minutes and was audio-recorded using a Marantz PMD-222 digital recorder. Recordings utilized an Audio-Technica flat unidirectional microphone placed on the table directly in front of the child. During the sessions children interacted with the examiner and/or the caregiver (usually the mother) while playing with various sets of toys (plastic food items, dolls, plastic cups and plates, plastic tractors, puzzles, books and pictures). The toys remained constant across all participants across both screening sessions. Measures

Broad phonetic transcriptions were performed by trained researchers for all consecutive utterances, produced 10 minutes within the experimental session. The number of utterances ranged from 70 to 220 and the number of tokens analyzed ranged from 120 to 670. At each screening session, three language measures were calculated based on the spontaneous language samples collected: Phonetic inventory size (PIS)

Individual phonetic profi les for each child were constructed on the basis of an independent analysis process during which each production was coded irrespective of its reference to the adult target (Amayreh & Dyson, 1998; Olswang et al., 1987; Petinou & Okalidou, 2006). Established consonants were considered to be those occurring in at least two different words at respective word positions. The analysis incorporated all words produced including multiple productions of the same target. For example,

238

An Anthology of Bilingual Child Phonology

if the child produced an allophonic variation, she was credited with the actual segment produced within the given word target (e.g. if the word ‘chocolate’ was produced as [t͡ ʃokoˈlata], [sokoˈlata] and/or [ʃokoˈlata], credit was given for [t͡ ʃ], [s] and [ʃ], respectively, depending on the child’s productive preference for that given target). In addition, instances in which the child substituted the target segments (e.g. [s] was substituted with [θ], and [ɾ] with [l]) the child was credited with the segment produced regardless of non-target realization (e.g. /s/ → [θ] in [‘supa] ‘soup’ produced as [‘θupa]). The number of established sounds was used to construct each child’s PIS index. Expressive vocabulary size (EVS)

EVS has been a widely used measure of lexical diversity (Fenson et al., 1994). In the current study, each child’s EVS was constructed by following the recommended process presented by Olswang and colleagues (1987) in their assessment of early linguistic behaviors in toddlers. The EVS counts included function words, as well as content words (verbs, nouns, articles, adverbs and prepositions). Imitated utterances, rituals, songs, as well as rote memorized and unintelligible, non-identifi able productions were excluded from the counting process. Each word was counted only once irrespective of the number of times it was produced. Mean length of utterance in words (MLUW)

MLU has been used to examine morphosyntactic development. In the current data set, MLU was calculated by dividing the total number of words produced by the total number of all consecutive utterances. Since CG is a highly inflected linguistic variety, for the current investigation counting words rather than morphemes was considered to be the most appropriate method of data analysis. This regime has been used successfully by investigators with a focus on CG (Petinou & Spanoudis, 2014; Voniati, 2016). The counting of words as opposed to morphemes has been strongly recommended by Leonard (2014) in examining syntactic growth in children who speak a highly inflected language. The rationale for this methodological procedure stems from the fact that in languages with rich inflectional systems, including CG, grammatical morphemes such as suffixes and prefixes cannot be produced in isolation since they are encapsulated in the whole word and are used to modulate gender, number, case and tense (Leonard, 2014). Utterance boundaries were determined on the basis of factors including falling intonation contour and pauses of a second’s duration, as suggested by Petinou et al. (1999). Furthermore, identification and segmentation of an utterance was performed in reference to communication units as suggested by Moyle et al. (2007), in which a unit consists of an independent intelligible sentence containing subject + predicate and all its subordinating clauses. Partially intelligible productions, elliptical responses to questions,

Linguistic Outcomes and the Role of Phonology in TD and LT Toddlers

239

memorized rote speech productions, incomplete sentences and dysfluent unintelligible productions were excluded from the counting process. Reliability

Approximately 10% of recorded samples were randomly selected for the purpose of phonetic transcription reliability, EVS and MLU counts. The samples were phonetically transcribed by the fi rst author and were checked against comparable coding from an independent transcriber (a speech-language pathologist trained in phonetic transcription) who was unfamiliar with the purpose of the study. Reliability on the relevant linguistic categories was based on the number of agreements divided by the sum of agreements and disagreements after the two transcribers had jointly listened to the targets and had compared their transcriptions with regard to place and manner of articulation. Inter-rater transcription reliability for manner and place of articulation was approximately 90% and 84%, respectively. For EVS and MLU agreement, percentages were 95% and 85% respectively. Analysis

The relationship among the three language measures, phonetic inventory size (PIS), expressive vocabulary size (EVS) and mean length of utterance in words (MLU), and for the two age levels (intake – 28 months; reassessment – 36 months) was examined using a Pearson r-correlation coefficient and based on Cohen’s (1992) guidelines. The analysis was conducted synchronically to investigate how these measures behave at the same time, and predictively to investigate whether a language measure at intake level can predict other language measures at reassessment level. In addition, using a K-means cluster analysis, we grouped the participants in three different ways based on their performance (low/high) in each language measure at intake level separately. This resulted in a total of six groupings: High/Low PIS28, High/Low EVS28 and High/Low MLU28. The reason for the division of participants in these groups was to examine whether the relationships found in the whole sample apply uniformly to the children at a certain age or whether low and high performers show different profiles. Descriptive statistics for all language measures are provided in Table 10.1. Results Relationship of language skills within ages

Table 10.2 presents all correlations between the three measures at age of intake and age of reassessment for all participants. We observe that the

240

An Anthology of Bilingual Child Phonology

Table 10.1 Descriptive statistics for each participant in the three language measures ID

PIS28

EVS28

MLU28

PIS36

EVS36

MLU36 2.12

1

12

63

2.15

21

221

2

17

87

2.87

23

353

2.53

3

17

60

1.4

18

85

1.55

4

29

118

1.92

22

176

1.63

5

28

158

2.47

27

423

2.62

6

25

50

2.1

30

91

2.92

7

11

40

0.78

16

223

1.69

8

31

119

3.12

20

363

3.33

9

24

85

1.24

34

260

1.74

10

24

168

2.11

23

127

2.1

11

20

154

1.45

28

392

2.05

12

9

32

1.45

18

208

1.7

13

16

42

1.9

28

173

2.7

14

28

128

2.62

26

214

2.46

15

9

9

0.89

6

33

1.19

16

25

62

1.4

28

96

1.86

17

24

69

1.54

29

422

2.26

18

14

71

1.16

22

243

2.47

19

18

74

1.48

26

80

1.6

20

25

89

2.31

26

224

2.55

21

25

89

1.94

25

193

1.91

22

25

82

1.2

27

95

2.8

23

22

68

1.73

24

222

2.46

24

4

12

1

16

117

1.8

25

8

15

1

7

25

1.2

26

16

200

1.78

18

100

2

27

20

120

2.9

23

128

3.2

28

14

180

1.77

22

190

2.2

29

8

40

1

18

140

2.5

30

20

100

2.55

24

190

3.35

31

19

118

2

20

142

2.2

Mean (SD)

18.94 (7.14)

87.16 (49.73)

1.78 (.63)

22.42 (6.07)

191.90 (107.60)

2.21 (0.57)

three language skills for all participants at intake level are highly positively correlated with each other (p < 0.01). At reassessment, significant moderately positive correlations exist between PIS36 and NWD36 (r(31) = 0.44, p < 0.05) and between PIS36 and MLU36 (r(31) = 0.46, p < 0.05), while the relationship between MLU36 and EVS36 is marginally significant (r(31) = 0.46, p = 0.05).

Linguistic Outcomes and the Role of Phonology in TD and LT Toddlers

241

Table 10.2 Correlations between the three measures at age of intake and age of reassessment for all participants

PIS28

PIS28

EVS28

MLU28

PIS36

EVS36

MLU36

–

0.51 (