The Oxford Handbook of Language Attrition 9780198793595, 0198793596
This volume is the first handbook dedicated to language attrition, the study of how a speaker's language may be aff
130 19 8MB
English Pages 657 Year 2019
Cover
The Oxford Handbook of Language Attrition
Copyright
Contents
Acknowledgements
List of Abbreviations
The Contributors
Chapter 1: Introduction
Part I: THEORETICAL IMPLICATIONS OF LANGUAGE ATTRITION
Chapter 2: Language attrition and the competition model
2.1 Four puzzles
2.1.1 Puzzle #1: permastore
2.1.2 Puzzle #2: variation across levels
2.1.3 Puzzle #3: childhood forgetting
2.1.4 Puzzle #4: catastrophic interference
2.2 The competition model
2.2.1 Entrenchment
2.2.2 Transfer
2.2.3 Participation, isolation, and identity
2.2.4 Resonance
2.2.5 Decoupling
2.3 Conclusions
Chapter 3: Language attrition and the feature reassembly hypothesis
3.1 Introduction
3.2 FRH in heritage grammars
3.3 Final remarks
Chapter 4: The interface hypothesis as a framework for studying L1 attrition
4.1 Introduction
4.2 The interface hypothesis in first generation L1 Arrtition
4.3 Selectivity of individual L1 attrition
4.4 Is indeterminacy in L1 attrition caudes by transfer or by bilingualism itself?
4.5 Reversibility of indivudual L1 attrition
Chapter 5: Implicatons of the bottleneck hypothesis for language attrition
5.1 Introduction
5.2 A fundamental division based on age: the maturational factor
5.2.1 The grammatical competence of early attriters
5.2.2 The grammatical competence of late attriters
5.3 Conclusion
Chapter 6: A complex dynamic systems perspective on personal background variables in L1 attrition
6.1 Personal background variables in L1 attrition research
6.2 Complex dynamic systems theory and language development
6.3 Complex dynamic systems theory and research methodology
6.4 Complex dymanic systems theory and personal background variables
6.5 Summary and outlook
Acknowledgements
Part II: PSYCHOLINGUISTIC AND NEUROLINGUISTIC APPROACHES TO LANGUAGE ATTRITION
Chapter 7: Introduction to psycholinguistic and neurolinguistic approaches to language attrition
7.1 Introduction and brief historical overview
7.2 Language Attrition as a processing mechanism
7.3 Language attrition as a memory mechanism
7.4 Language attrition as a brain mechanism
7.5 Section outlook
Chapter 8: Language attrition as a special case of processing change: A wider cognitive perspective
8.1 Introduction
8.2 Investigating attrition: research tradition
8.2.1 Representation, change, and the nature of processing
8.2.2 Interfaces
8.3 Unifying knowledge, performance, and development within one architecture
8.3.1 An overview
8.3.2 Modularity and collaboration: key elements of the framework
8.3.3 The developmental perspective
8.3.4 Representations and memory theory
8.3.5 Working memory
8.3.6 Resting levels of activation
8.4 Attrition and interface processing
8.4.1 Acquisition and growth
8.4.2 Interactions across interfaces in acquisition and attrition
8.4.3 Frequency effects
8.4.4 Degrees of forgetting
8.4.5 Inhibition and attrition
8.4.6 Metalinguistic knowledge and attrition
8.5 Concluding comments
Acknowledgements
Chapter 9: Memeory retrieval and language attrition
9.1 Introduction
9.2 Native language change in L2 learning and proficient bilingual language processing
9.3 Retrieval induced forgetting
9.4 The consequences of language immersion
9.5 Implications of retrieval induced forgetting for theories of attrition
9.6 Conclusions
Acknowledgements
Chapter 10: How billingualisn affects syntactic processing in the native language: Evidence from eye movements
10.1 The interactive nature of the bilingual system
10.2 Eye-mevement records as a premier behavioural tool to study L1 language change
10.3 Illustrative cases
10.3.1 Reconfiguration of the L1
10.3.2 Reactivation of the L1
10.4 Conclusion
Acknowledgements
Chapter 11: First language attrition and developmental language disorder
11.1 Introduction
11.2 Overlapping profiles: Monolingual developmental language disorder and bilingual typical development
11.3 Combined effects of developmental language disorder and bilingualism
11.3.1 Delay in the L2
11.3.2 L1 attrition
11.4 Factors influencing L1 attrition in bilingual chindren
11.4.1 Age of exposure to L2 and L1 development
11.4.2 Socio-economic status: parent–child interaction and L1 development
11.4.3 Language status and L development
11.5 Isolating the effect of developmental language disorder
11.5.1 Assessing the L1: comparing an individual child using appropriate norms
11.5.2 Assessing the L1: linguistic markers and valid tools
11.5.3 Possible alternatives: measures that areinsensitive to L1 attrition (or L2 delay)
11.6 Concluding remarks
Chapter 12: Ageing as a confound in language attrition research: Lexical retrieval, language use, and cognitive and neural changes
12.1 Introduction
12.2 Lexical attrition in bilingual adults
12.2.1 Causes of lexical attrition
12.3 Neurophysiological changes in language attrition
12.4 Lexical decline in monolingual older adults
12.4.1 Neurophysiological changes in cognitive ageing
12.4.2 The interaction between age-related declines and L1 attrition
12.5 Language attrition and ageing
12.6 Cognitive and neuropsychological contributions to language attrition and ageing
12.7 Conclusion
Acknowledgements
Chapter 13: Linguistical regression in bilingual patients with alzheiner's didease
13.1 Introduction
13.2 Alzheimer's disease, language, and bilingualism
13.2.1 Alzheimer’s Disease
13.2.2 Language deterioration in monolingual patients with Alzheimer’s Disease
13.2.3 Bilingualism and Alzheimer’s Disease: a topic of growing interest
13.3 Alzheimer's disease and bilingualism
13.3.1 The Babylonian benefit: a controversial issue
13.3.2 Second-language regression in Alzheimer’s Disease
13.3.3 Language mixing in bilingual individuals with Alzheimer’s Disease
13.4 Conclusion
Chapter 14: Electrophysiological approaches to L1 attrition
14.1 Event-related potential and language processing in real-time
14.2 How plastic is the adult brain? event-related potentialsin L2 learners and attriters
14.3 Morpho-syntactic processing in attriters' L1
14.3.1 Grammatical gender violations in German and Dutch
14.3.2 Number agreement violations at various positions in Italian sentences
14.3.3 Relative clauses in Italian immigrants: can one’s L2 render the L1 ungrammatical?
14.4 Lexical-semantic processing in attriters' L1 and L2
14.4.1 A study on confusable words in Italian sentences
14.4.2 Lexical processing in attriters’ L2
14.5 State of affairs, limitations, and future directions
Chapter 15: Neuroimaging perspectives on L1 attrition and language change
15.1 L1 attrition: the phenomenon
15.2 Introducing magnetic resonance imaging methods
15.3 Magnetic resonance imaging as a tool to tackle L1 attrition
15.4 Neural investigations of L1 change beyond extreme L1 attrition
15.5 Future directions in neuroimaging L1 attrition research
15.5.1 Longitudinal designs
15.5.2 Neural oscillations
15.6 Summary and conclusions
Part III: Linguistic factors in language attrition
Chapter 16: Introduction to linguistic factors in language attrition
16.1 Introduction
16.2 The Linguistic manifestations of language attrition
16.3 Knowledge gaps and future research agenda
16.4 Contributions to the current sections
16.5 Conclusion
Chapter 17: Phonetic drift
17.1 Introduction
17.2 Phonetic drift of consonants
17.3 Phonetic drift of vowels
17.4 Phonetic drift of suprasegmentals
17.5 Discussion and conclusion
Chapter 18: Phonetic attrition
18.1 Introduction
18.1.1 Definition of attrition
18.2 Aim of contribution
18.3 What is phonetics?
18.4 Speech learning model (SLM)
18.5 Segmental assimilation in phonetic attrition
18.6 Prosodic assimilation in phonetic attrition
18.7 Segmental dissimilation in phonetic attrition
18.8 Prosodic dissimilation in phonetic attrition
18.9 Global regional or foreign accent in a native language
18.10 Summary of research
18.11 Discussion of implications of phonetic attrition
Chapter 19: Phonological attrition
19.1 Introduction
19.2 Phonological contrasts
19.2.1 Contrasts in the vowel systems
19.2.2 Contrasts among obstruents
19.3 Phonotactic patterns
19.4 Future challenges in the study of phonological attrition
Chapter 20: Morphological attrition
20.1 Introduction
20.2 Background
20.3 Attrition of derivational morphology
20.4 Attrition of Inflectional morphlogy
20.5 Markedness theory
20.6 Regression hypothesis
20.7 Activation threshold hypothesis (ATH)
20.8: Interface hypothesis (IH)
20.9 The 4-M model
20.10 Conclusion
Chapter 21: Lexical attrition
21.1 What is lexical attrition and what does it look like?
21.2 Explanations of lexical attrition
21.3 Factors that affect lexical attrition
21.4 Methods for exploring lexical attrition
21.5 Final Thoughts
Chapter 22: Null and overt pronouns in language attrition
22.1 Introduction
22.2 The linguistic phenomena associated with null and overt pronominals
22.3 Adult L1 attrition in the pronominal domain
22.4 Summary and conclusion
Part IV: EXTRALINGUISTIC FACTORS IN LANGUAGE ATTRITION
Chapter 23: Introduction to extralinguistic factors in language attrition
23.1 Introduction and historical background
23.2 The european graduate network on language attrition
23.3 Attrition as a gradient phenomenon
23.4 Overview of the present section
23.5 Conclusion
Chapter 24: Age effects in language attrition
24.1 Introduction
24.2 Empirical findings on age effects in language attrition
24.2.1 Attrition in the absence of L contact
24.2.2 Attrition under reduced L contact
24.3 Theoretical accounts of age effects in language attritiion
24.3.1 Impediment accounts
24.3.2 A psychosocial account
24.3.3 The maturational account
24.4 Principles concerning age effects in attrition
Chapter 25: The impact of frequency of use and length of residence on L1 attrition
25.1 Introduction
25.2 Memory and the activation threshold hypothesis
25.3 What kind of L1 use?
25.4 The Empirical state of the art
25.4.1 Findings on L1 use
25.4.2 Findings on the impact of length of residence
25.5 Discussion and conclusion
Appendix
Chapter 26: L1 attrition, L2 development, and integration
26.1 Introduction
26.2 The bilingual mind
26.3 The Bicultural mind
26.4 The (assures) role of language in the integration outcomes
26.5 Conclusion
Chapter 27: Language contact and language attrition
27.1 Where does language contact take place?
27.2 Language contact in attrition processes
27.3 Key terms: borrowing, transference, convergence
27.4 Phenomena of language contact: types of transference and convergence
27.4.1 Lexical transference
27.4.2 Semantic transference/conceptual restructuring
27.4.3 Lexico-syntactic transference
27.4.4 Syntactic transference and convergence
27.4.4.1 Syntactic transference
27.4.4.2 Convergence/pivot matching
27.4.5 Phonological transference and convergence
27.4.6 Morphological transference
27.4.7 Pragmatic transference and convergence
27.5 External factors: language awareness and norm orientation in bilingual communities
Part V: Second language attrition
Chapter 28: Introduction to L2 attrition
28.1 Introduction
28.2 Terminological issues
28.3 Historical background
28.4 The baseline problem
28.5 State of the art
28.6 Delving into some of the studies
28.7 How we (try to) explain what we know
28.7.1 The Regression Hypothesis
28.7.2 The Savings paradigm
28.7.3 The Activation Threshold Hypothesis
28.7.4 Critical threshold
28.7.5 Dynamic Model of Multilingualism
28.8 Future research directions
28.9 Outline of part V
Chapter 29: Exploring the impact of extralinguistic factors on L2\FL Attrition
29.1 Introduction
29.2 From indivudual to cluster factors: A dynamic systems theory approach
29.3 Attained proficiency
29.4 The usual suspects: age, attitude and motivation, language contact, and use
29.4.1 Age
29.4.2 Attitude and motivation
29.4.3 Language contact and use
29.5 The great unknown: Context, length of residence/exposure
29.5.1 Context
29.5.2 Length of exposure/residence
29.6 Odds and ends: aptitude, level of education, literacy
29.6.1 Aptitude
26.6.2 Level of education
29.6.3 Literacy
29.7 Incubation
29.8 Discussion and conclusions
Acknowledgements
Chapter 30: Syntax and phonology in L2 attrition: Modularity and resilience
30.1 Introduction
30.2 Critical periods in acquisition and attrition
30.3 Patterns of retention in cases of childhood attrition
30.4 Resilience of phonology and syntax in L2 attrition
30.4.1 Attrition in L2 phonology
30.4.2 Attrition in L2 syntax
30.5 Discussion: disentangling syntax and phonology from lexical attrition
Chapter 31: L2 Lexical attrition
31.1 Introduction
31.2 Hypotheses concerning L2 Attrition
31.3 Memory loss and lexical attrition
31.4 Results from studies addressing L2 lexical attrition
31.4.1 Differences in L2 lexical attrition depending on age
31.4.2 Differences in L2 lexical attrition depending on incubation period
31.4.3 Proposing a definition for catastrophic loss andthe incubation time required to find such loss
31.4.4 Looking at studies with extremely long incubation periods, and finding an additional factor of target language
31.4.5 Which words are lost?
31.5 How Meara's constant decay hypothesis applied to the result of empirical L2 Attrition Research
31.6 Future research
Chapter 32: Attrition studies on japanese returnees
32.1 Introduction
32.2 Early investigatory stage
32.3 Hypothesis-testing stage
32.4 A new stage: Neurolinguistic approaches
Chapter 33: Event-related potentials as metrics of foreign language learning and loss
33.1 Language attrition
33.2 Event-related potentials as metrics of native-language comprehension
33.3 Event-related potentials as metric of foreign language learning and 'forgetting'
33.4 A longitudinal study of foreign language acquisition and attrition
33.4.1 Materials
33.4.2 Behavioural responses
33.4.3 ERPs elicited during the ‘learning phase’
33.4.4 Post-learning ‘attrition phase’
33.5 Discussion
Part VI: Heritage languages
Chapter 34: Introduction to heritage language development
34.1 Introduction
34.2 Bilingual acquisition: from infancy to adolescence
34.3 Language development and age
34.4 Other factors affecting the language development of heritage speakers
Chapter 35: Quantifying language experience in heritage language development
35.1 Introduction
35.2 The effect of specific properties of bilingual experience on (heritage) language development
35.3 How to qualtify bilingual/HL Experience
35.3.1 Conceptual issues
35.3.2 Practical issues
35.4 Survey of available parental questionnaires
35.4.1 Administration
35.4.2 Output
35.4.3 Population
35.4.4 Languages
35.5 A Final note on parental report
35.6 Conclusion
Chapter 36: Intra-generational attrition: Contributions to heritage speaker competence
36.1 Introduction
36.2 Attrition and its trickling down effects across generations
36.3 Empirical support
36.4 Conclusion
Chapter 37: 2L1 Simultaneous bilinguals as heritage speakers
37.1 Introduction
37.2 Language separation and CLI
37.3 AoO
37.4 Input and language dominance
37.5 Distance to the homeland
37.6 Conclusions
Acknowledgements
Chapter 38: Language loss and language learning in internationally adopted children: Evidence from behaviour and the brain
38.1 Introduction
38.1.1 Behavioural attrition
38.1.2 Neural attrition
38.1.3 Attrition summary
38.2 Implications of attrition for later language development/processing/learning
38.2.1 Implications for L1
38.2.2 Implications for L2
38.3 Conclusion
Acknowledgements
Chapter 39: Childhood language memory in adult heritage language (re)learners
39.1 Introduction
39.2 Heritage language loss among immigrant-background children
39.3 Language memory among childhood overhearers
39.4 Language menory amng childhood speakers of a heritage language
39.5 Evidence of childhood language menory among adult internationally adopted as infants
39.5.1 Re-accessing childhood language memory through relearning: the case of Korean-American adopted adults in Minnesota
39.5.1.1 Participants
39.5.1.2 Procedure
39.5.1.3 Language background assessment
39.5.1.4 Phoneme identification task
39.5.1.5 Phoneme production task
39.5.1.6 Results
39.5.1.6.I Language background assessment
39.5.1.6.II Phoneme perception
39.5.1.6.III Phoneme prouction
39.5.1.7 Relearning as one key to accessing international adoptees’ early childhood language memory
39.6 The long-lasting benefits of early childhood language experience
Chapter 40: Language development in bilingual returnees
40.1 Introduction
40.2 When the society language becomes the minority language
40.3 When the heritage language becomes the societal language
40.4 Concluding remarks
Concluding remarks
The coming of age of a research field
Defining language
Theoretical approaches: attrition and Universal Grammar
The balance hypothesis
The tidal waves of research on foreign language skills
Attrition and international adoptees
The unavoidable age factor
Some problematic areas
Units of analysis in attrition research
Incomplete acquisition and incomplete transmission between generations
Attrition and heritage languages
Suprasegmental aspects of language attrition
The attrition of gestures and sign languages
Attrition as a process rather than a state
Attrition at different time scales
To conclude
Acknowledgement
Annotated bibliography
References
Index
Recommend Papers
![Language Attrition in Progress [Reprint 2012 ed.]
9783110857863, 9783110130874](https://ebin.pub/img/200x200/language-attrition-in-progress-reprint-2012nbsped-9783110857863-9783110130874.jpg)
- Author / Uploaded
- Monika S. Schmid
- Barbara Köpke
File loading please wait...
Citation preview
OUP CORRECTED PROOF – FINAL, 25/6/2019, SPi
LANGUAGE ATTRITION
OUP CORRECTED PROOF – FINAL, 25/6/2019, SPi
OXFORD HANDBOOKS IN LINGUISTICS Recently published
THE OXFORD HANDBOOK OF PRAGMATICS Edited by Yan Huang
THE OXFORD HANDBOOK OF UNIVERSAL GRAMMAR Edited by Ian Roberts
THE OXFORD HANDBOOK OF LANGUAGE AND SOCIETY Edited by Ofelia García, Nelson Flores, and Massimiliano Spotti
THE OXFORD HANDBOOK OF ERGATIVITY Edited by Jessica Coon, Diane Massam, and Lisa deMena Travis
THE OXFORD HANDBOOK OF WORLD ENGLISHES Edited by Markku Filppula, Juhani Klemola, and Devyani Sharma
THE OXFORD HANDBOOK OF POLYSYNTHESIS Edited by Michael Fortescue, Marianne Mithun, and Nicholas Evans
THE OXFORD HANDBOOK OF EVIDENTIALITY Edited by Alexandra Y. Aikhenvald
THE OXFORD HANDBOOK OF LANGUAGE POLICY AND PLANNING Edited by James W. Tollefson and Miguel Pérez-Milans
THE OXFORD HANDBOOK OF PERSIAN LINGUISTICS Edited by Anousha Sedighi and Pouneh Shabani-Jadidi
THE OXFORD HANDBOOK OF ENDANGERED LANGUAGES Edited by Kenneth L. Rehg and Lyle Campbell
THE OXFORD HANDBOOK OF ELLIPSIS Edited by Jeroen Van Craenenbroeck and Tanja Temmerman
THE OXFORD HANDBOOK OF LYING Edited by Jörg Meibauer
THE OXFORD HANDBOOK OF TABOO WORDS AND LANGUAGE Edited by Keith Allan
THE OXFORD HANDBOOK OF MORPHOLOGICAL THEORY Edited by Jenny Audring and Francesca Masini
THE OXFORD HANDBOOK OF REFERENCE Edited by Jeanette Gundel and Barbara Abbott
THE OXFORD HANDBOOK OF EXPERIMENTAL SEMANTICS AND PRAGMATICS Edited by Chris Cummins and Napoleon Katsos
THE OXFORD HANDBOOK OF EVENT STRUCTURE Edited by Robert Truswell
THE OXFORD HANDBOOK OF LANGUAGE ATTRITION Edited by Monika S. Schmid and Barbara Köpke For a complete list of Oxford Handbooks in Linguistics please see pp. –.
OUP CORRECTED PROOF – FINAL, 25/6/2019, SPi
......................................................................................................................
LANGUAGE ATTRITION ...................................................................................................................... Edited by
MONIKA S. SCHMID and
BARBARA KÖPKE with
MIRELA CHERCIOV TUĞBA KARAYAYLA MEREL KEIJZER ESTHER DE LEEUW TEODORA H. MEHOTCHEVA SILVINA MONTRUL MARIA POLINSKY
1
OUP CORRECTED PROOF – FINAL, 25/6/2019, SPi
3
Great Clarendon Street, Oxford, , United Kingdom Oxford University Press is a department of the University of Oxford. It furthers the University’s objective of excellence in research, scholarship, and education by publishing worldwide. Oxford is a registered trade mark of Oxford University Press in the UK and in certain other countries © editorial matter and organization Monika S. Schmid and Barbara Köpke © the chapters their several authors The moral rights of the authors have been asserted First Edition published in Impression: All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, without the prior permission in writing of Oxford University Press, or as expressly permitted by law, by licence or under terms agreed with the appropriate reprographics rights organization. Enquiries concerning reproduction outside the scope of the above should be sent to the Rights Department, Oxford University Press, at the address above You must not circulate this work in any other form and you must impose this same condition on any acquirer Published in the United States of America by Oxford University Press Madison Avenue, New York, NY , United States of America British Library Cataloguing in Publication Data Data available Library of Congress Control Number: ISBN –––– Printed and bound by CPI Group (UK) Ltd, Croydon, Links to third party websites are provided by Oxford in good faith and for information only. Oxford disclaims any responsibility for the materials contained in any third party website referenced in this work.
OUP CORRECTED PROOF – FINAL, 25/6/2019, SPi
C
................................
Acknowledgements List of abbreviations The contributors
ix x xv
. Introduction
M S. S B Kö
PART I THEORETICAL IMPLICATIONS OF LANGUAGE ATTRITION . Language attrition and the Competition Model
B MW
. Language attrition and the Feature Reassembly Hypothesis
M T. P, S P-C, L Sˊ
. The Interface Hypothesis as a framework for studying L attrition
G C A S
. Implications of the Bottleneck Hypothesis for language attrition
R S
. A Complex Dynamic Systems perspective on personal background variables in L attrition
C O
PART II PSYCHOLINGUISTIC AND NEUROLINGUISTIC APPROACHES TO LANGUAGE ATTRITION . Introduction to psycholinguistic and neurolinguistic approaches to language attrition B K̈ M K
OUP CORRECTED PROOF – FINAL, 25/6/2019, SPi
vi
. Language attrition as a special case of processing change: A wider cognitive perspective
M S S
. Memory retrieval and language attrition: Language loss or manifestations of a dynamic system?
J A. L J F. K
. How bilingualism affects syntactic processing in the native language: Evidence from eye movements
P E. D, J R. Vˊ K, M J, Á V
. First language attrition and Developmental Language Disorder
E B, T B, J J
. Ageing as a confound in language attrition research: Lexical retrieval, language use, and cognitive and neural changes
E H, A L, M K, T M, L K. O
. Linguistic regression in bilingual patients with Alzheimer’s Disease
M B-D, Fˊ ˊ G, B Kö, L L
. Electrophysiological approaches to L attrition
K S K K
. Neuroimaging perspectives on L attrition and language change
E R, Y P, M T. D
PART III LINGUISTIC FACTORS IN LANGUAGE ATTRITION . Introduction to linguistic factors in language attrition
M S. S E L
. Phonetic drift
C B. C
. Phonetic attrition E L
OUP CORRECTED PROOF – FINAL, 25/6/2019, SPi
. Phonological attrition
vii
C C
. Morphological attrition
E S
. Lexical attrition
S J
. Null and overt pronouns in language attrition
A̧ G̈
PART IV EXTRALINGUISTIC FACTORS IN LANGUAGE ATTRITION . Introduction to extralinguistic factors in language attrition
M S. S M C
. Age effects in language attrition
E B
. The impact of frequency of use and length of residence on L attrition
M S. S
. L attrition, L development, and integration
G̈ Y
. Language contact and language attrition
C M R
PART V SECOND LANGUAGE ATTRITION . Introduction to L attrition
T H. M B Kö
. Exploring the impact of extralinguistic factors on L/FL attrition
T H. M K M
. Syntax and phonology in L attrition: Modularity and resilience
K B-H D S
. L lexical attrition J L-H
OUP CORRECTED PROOF – FINAL, 25/6/2019, SPi
viii
. Attrition studies on Japanese returnees
H T
. Event-Related Potentials as metrics of foreign language learning and loss
L O, I P̈ , J ML, M Z
PART VI HERITAGE LANGUAGES . Introduction to heritage language development
S M M P
. Quantifying language experience in heritage language development
S U
. Intra-generational attrition: Contributions to heritage speaker competence
F B, D P C, J R
. L simultaneous bilinguals as heritage speakers
T K
. Language loss and language learning in internationally adopted children: Evidence from behaviour and the brain
L J. P, F G, D K
. Childhood language memory in adult heritage language (re)learners
J S. O, T K- A, S-A J, R M. L
. Language development in bilingual returnees
C F
Concluding remarks K B
Annotated bibliography
M S. S
References Index
OUP CORRECTED PROOF – FINAL, 25/6/2019, SPi
A
......................... .......................................
We are very grateful to Asha Ali and Rachel Ting Yan Kan, who helped us put this volume together and keep track of too many things to list here. We are also deeply indebted to all of the anonymous reviewers whose input was vital in improving the quality of the chapters collected here. Joy Mellor, Kumar Anbazhagan, and Vicki Sunter from OUP were an absolute joy to work with and always responded promptly and kindly to our many (and often grumpy—apologies again, Joy, Kumar, and Vicki!) queries. Finally, we would like to express our gratitude and appreciation to the artist Bruno Catalano and his representative, Emilie Catalano, for kindly allowing us to reproduce a picture of one of his stunning sculptures on the cover of this volume. The field of language attrition has many different parents, but we would particularly like to remember two of the most influential ones here: Richard D. Lambert and Theo van Els, both of whom passed away in . They will be fondly remembered and the field is much poorer for their loss.
OUP CORRECTED PROOF – FINAL, 25/6/2019, SPi
L
........................................................................
AA
age of attrition
ACC
accusative
AD
Alzheimer’s Disease
Af S
affective structure
AGC
corpus of German immigrants in Australia
AL
Applied Linguistics
AMTB
attitude and motivational test battery
ANOVA
analysis of variance
AOA
age at onset of attrition
AoA
age of acquisition
AoO
age of onset
AOR
age of return
APT
Attrition by Processing Theory
ASL
American Sign Language
AaT
age at testing
AT
Activation Threshold
ATH
Activation Threshold Hypothesis
ATM
attitude and motivation
BAT
bilingual aphasia test
BH
Bottleneck Hypothesis
BIA
Bilingual Interactive Active Model
BIA+
Bilingual Interactive Active Model Plus
BOLD
blood-oxygen-level dependent
BP
Brazilian Portuguese
CC
Cognate Correct
CD
close duration/closure duration
CDST
Complex Dynamic Systems Theory
CELF
Clinical Evaluation of Language Fundamentals
CHILDES
Child Language Data Exchange System
CHL
computational procedure for human language
OUP CORRECTED PROOF – FINAL, 25/6/2019, SPi
CLI
cross-linguistic influence
CLLD
clitic left-dislocations
COMP
Complementizer
CP
complementizer phrase
CPH
Critical Period Hypothesis
CS
conceptual structure
CSF
cerebrospinal fluid
CT
Chaos/Complexity Theory
DAT
Dative
DLD
Developmental Language Disorder
DM
Distributed Morphology
DMM
Dynamic Model of Multilingualism
DOM
differential object marking (or marker)
DP
determiner phrase
D/P
declarative/procedural
DS
Dynamic System
DSM
Diagnostic and Statistical Manual of Mental Disorders
DST
Dynamic Systems Theory
DTI
Diffusion Tensor Imaging
DV
dependent variable
EC
English-correct sentences
EEG
electroencelphalogram
EFL
English as a Foreign Language
ELAN
early left anterior negativity
EP
European Portuguese
EPP
Extended Projections Principle
ERLAC
emotion-related language choice
ERP
Event-Related (brain) Potential
ESRC
Economic and Social Sciences Research Council
EVT
Ethnolinguistic Vitality Theory
FA
false alarms
FEM
feminine
FF
functional features
FL
foreign language
fMRI
functional Magnetic Resonance Imaging
fNIRS
functional Near Infrared Spectroscopy
xi
OUP CORRECTED PROOF – FINAL, 25/6/2019, SPi
xii
FRH
feature reassembly hypothesis
GAMM
generalized additive mixed models
GEN
genitive
GG
grammatical gender
HBP
Heritage Bilingual Paradox
HL
heritage language
HP
high proficiency
HS
heritage speaker
IA
International Adoptees/internationally adopted
IC
Inhibitory Control
ICFLA
International Conference on First Language Attrition
IH
Interface Hypothesis
INC
incubation period
IV
independent variable
JND
just-noticeable difference
L
language
LAD
language acquisition device
LAE
language acquisition effort
LAN
left anterior negativity
LATB
Language Attrition Test Battery
LC
linguistically competent user
LF
logical form
LI
Language Impairment
LID
Language Input Diary
LIFO
last in, first out
LME
language maintenance effort
LOR
length of residence
LTM
long-term memory
LTN
left temporal negativity
MANOVA
multiple analysis of variance
MASC
masculine
MCI
mild cognitive impairment
MEG
magnetoencephalography
MICD
Missing Input Competence Divergence
MOGUL
Modular Online Growth and Use of Language
OUP CORRECTED PROOF – FINAL, 25/6/2019, SPi
MP
Minimalist Program
MRI
Magnetic Resonance Imaging
MTLD
Measure of Textual Lexical Diversity
NIRS
Near Infrared Spectroscopy
NOM
nominative
NP
noun phrase
OA
onset age
OPC
overt pronoun constraint
PAS
Position of Antecedent Strategy
Pas
passive
PDP
parallel distributed processing
PET
positron emission tomography
PF
phonological form
PL
plural
PLI
Primary Language Impairment
PLU
plural
POp
Pars Opercularis
POSS
possessive
PRES
present
PRG
progressive
PrG
Precentral Gyrus
PRO
pronoun
PSA-NCAM
polysialic acid-neural cell adhesion molecule
PT
planum temporale
PTr
Pars Triangularis
PV
parvalbumin
PWM
phonological working memory
RH
Regression Hypothesis
RIF
retrieval induced forgetting
RL
recipient language
RP
Retrieval Practice
RP–
Retrieval Practice minus
RP+
Retrieval Practice plus
RT
reaction times
SA
study abroad
xiii
OUP CORRECTED PROOF – FINAL, 25/6/2019, SPi
xiv
SAI
short-latency afferent inhibition
SD
standard deviation
SES
socio-economic status
SG
singular
SING
singular
SL
source language
SLA
second language acquisition
SLD
second language development
SLI
Specific Language Impairment
SLM
Speech Learning Model
SLP
speech-language therapist
SOV
subject–object–verb
SPEC
specifier
SSBE
Standard Southern British English
STG
Superior Temporal Gyrus
STS
Superior Temporal Sulcus
SVO
subject–verb–object
TD
typical development/typically developing
TL
target language
TLU
target-like usage
TOEFL
Test of English as a Foreign Language
TOT
tip of the tongue
TTR
type–token ratio
UB
usage-based
UG
Universal Grammar
VFT
Verbal Fluency Task
VOCD
vocabulary diversity
VOT
voice onset time
WM
working memory
OUP CORRECTED PROOF – FINAL, 25/6/2019, SPi
T ............................................................
Kathleen Bardovi-Harlig, Professor of Second Language Studies, Indiana University, teaches and conducts research on second language acquisition, temporal semantics, pragmatics, formulaic language as it relates to pragmatics, and attrition (in joint work with David Stringer). Bardovi-Harlig and Stringer’s work on attrition has appeared in SSLA, Linguistic Approaches to Bilingualism, and edited volumes. Melissa Barkat-Defradas received her PhD in Language Sciences in at University of Lyon and received the Young Researcher Award for her work. After a research fellowship at UC Berkeley she integrated the French National Centre for Scientific Research. She is now full-time researcher at The Institute of Evolutionary Sciences of Montpellier (France), where she develops interdisciplinary works that contribute to bridge the gap between linguistics and evolutionary biology. Her recent researches focus on psycholinguistic aspects of language processing in bilingual subjects suffering from Alzheimer’s disease. She is currently interested in the determination of biological and environmental factors that may explain senescence in humans. Fatih Bayram is a post-doctoral research fellow at UiT the Arctic University of Tromsø, funded by the European Union’s Horizon research and innovation programme under the Marie Skłodowska-Curie Individual Fellowship grant agreement No. . His research covers various types of child and adult bilingualism with a primary focus on heritage language bilinguals. Some of his recent research has appeared in International Journal of Bilingualism and Bilingual Education, Linguistic Approaches to Bilingualism, and Frontiers in Psychology. Elma Blom is Professor at the Department of Special Education at Utrecht University where she teaches about language development. Her research and publications are about developmental language disorder, the parallel development of multiple languages in immigrant children, and the relationship between language and cognition both in language-impaired and multilingual children. Besides theoretical issues, she works on the improvement of diagnostic tools for multilingual children. Tessel Boerma is a linguist with expertise on child language development and impairment. In , Tessel started her PhD at Utrecht University, investigating the linguistic and cognitive development of monolingual and bilingual children with developmental language disorder. Tessel defended her dissertation in October and now continues her work on language development in clinical child populations at Utrecht University. She is a postdoctoral researcher in the research programme ‘Language impairment in the q. deletion
OUP CORRECTED PROOF – FINAL, 25/6/2019, SPi
xvi
syndrome’. Moreover, she is involved in one of the interdisciplinary projects of Dynamics of Youth, entitled ‘The first critical days of a child’s life’. Kees de Bot graduated from the University of Nijmegen and became Professor of Applied Linguistics in Nijmegen and later in Groningen. His main interests are multilingualism, psycholinguistics, language policy, and more recently the philosophy of Science and the role of evidence in science and how that relates to notions of truth in society at large. He has published in most of the leading journals in applied linguistics. His most recent book is A History of Applied Linguistics –. (Routledge, ). He is now a full Professor of Applied Linguistics at the University of Pannonia in Hungary. Emanuel Bylund is Professor of General Linguistics (Stellenbosch University). His research concerns age effects in language acquisition and attrition, and conceptual representation in bilinguals. He has worked on languages such as Afrikaans, German, isiXhosa, Spanish, and Swedish. Chiara Celata is a researcher of linguistics and speech sciences at Scuola Normale Superiore of Pisa. She has a background in diachronic phonology and phonological acquisition, with particular reference to L Italian. She has taught applied linguistics at the University of Siena and L phonological attrition at CSIC and UIMP, Madrid. Her additional research interests include the sociophonetics of the Romance languages and the study of regional and individual variation according to multi-level acoustic and articulatory approaches. She has authored or co-authored more than fifty papers in scientific journals and volumes, including the editing of international books and journal special issues, in the domain of speech variation, sound change, and bilingual phonology. Gloria Chamorro is Lecturer in Applied Linguistics at the University of Kent. She obtained her PhD in Developmental Linguistics from the University of Edinburgh, and has researched and published on first language attrition, adult second language acquisition, and bilingual education. She has also developed the public engagement project English Language Support for Refugees, which helps unaccompanied refugee minors gain the English language skills they need to integrate into the community and access mainstream education. Charles B. Chang is Assistant Professor of Linguistics at Boston University, where he directs the Phonetics, Acquisition & Multilingualism Lab (PAMLab) and is Affiliated Faculty with the Center for the Study of Asia and the Center for Research in Sensory Communication & Emerging Neural Technology. His research concerns the processing, representation, and development of speech sounds in the context of multilingualism and language contact, the cognitive organization of dual phonological systems, and the cross-language interactions that occur during second-language learning and first-language attrition. Mirela Cherciov obtained her PhD in Linguistics (Department of French) in from the University of Toronto. Her thesis, titled Between Attrition and Acquisition: the Dynamics between Two Languages in Adult Migrants, was partly funded by a grant
OUP CORRECTED PROOF – FINAL, 25/6/2019, SPi
xvii
from the Social Sciences and Humanities Research Council. She has since held positions and taught at York University (Glendon Bilingual Campus) and Brock University. Her research focuses on linguistic and sociolinguistic factors linked to L attrition in adult migrants, bilingual language development across the lifespan, adult L acquisition, and pedagogy of FSL. Michele T. Diaz, PhD, is the Director of Human Imaging and an Associate Professor of Psychology at the Pennsylvania State University. Her research focuses on the neural representation of language and how this representation changes with age. In particular, she is interested in age-related differences in phonological and semantic aspects of language production. Her lab combines behavioural, neuropsychological, and neuroimaging measures to examine the relations between age, cognition, and the brain. She has also investigated contributions of the right hemisphere to language by studying figurative language, discourse, and novelty. Finally, she worked on the Biomedical Informatics Research Network focusing on best practices in neuroimaging and multi-site research. Paola E. Dussias is Professor of Spanish, Linguistics, and Psychology at Pennsylvania State University. The research that she and her students conduct employs methodological tools from linguistics, experimental psycholinguistics, and cognitive neuroscience to examine the way in which bilingual readers and speakers negotiate the presence of two languages in a single mind. They also examine the variables that incur processing costs during the comprehension and production of code-switched language. Their work, supported by grants from the National Science Foundation and the National Institutes of Health, has shown that the native language is open to influence from the second language, and that not all code-switches incur processing costs. Cristina Flores is Associate Professor of German Linguistics in the Department of German and Slavic Studies at the University of Minho/Portugal. Her research focuses on bilingual language development, especially on heritage language acquisition and on language attrition in returnees. She has coordinated several projects on Portuguese–German bilingualism. She is an integrated researcher at the Center for Humanistic Studies of the University of Minho, of which she is also vice-director. Frédérique Gayraud is Professor of Psycholinguistics at Lyon University. Her main research interests include early language acquisition and language decline in Alzheimer’s disease in a lifespan perspective. More specifically, the goal is to compare acquisition and desacquisition processes in order to test the retrogenesis hypothesis according to which degenerative mechanism reverse the order of acquisition in normal development. Another research interest focuses on bilingualism in normal and pathological ageing. Fred Genesee is Professor Emeritus in Psychology, McGill University. He has conducted extensive research on alternative forms of bilingual/immersion education for language minority and majority students, the academic development of at-risk students in bilingual programmes, language acquisition in typically-developing and at-risk pre-school bilingual children, and internationally adopted children. He has published numerous articles in
OUP CORRECTED PROOF – FINAL, 25/6/2019, SPi
xviii
scientific journals and magazines and has authored sixteen books on bilingualism. He is the recipient of the Canadian Psychology Association Gold Medal Award, Paul Pimsler Award for Research in Foreign Language Education, and the Canadian Psychological Association Award for Distinguished Contributions to Community or Public Service. Ayşe Gürel holds a PhD in Linguistics from McGill University. She is a full professor at the Department of Foreign Language Education of Boğaziçi University. Her research interests include first language attrition and second language acquisition of morpho-syntactic features as well as neurolinguistic and psycholinguistic aspects of bilingualism. She has publications in international journals and books in her fields of interest and she has been serving in the editorial boards of Second Language Research and Linguistic Approaches to Bilingualism. Among her current work is an edited volume on Second Language Acquisition of Turkish published by John Benjamins. Eve Higby is a postdoctoral researcher in Psychology at the University of California, Riverside. She received her PhD in Speech-Language-Hearing Sciences from the City University of New York Graduate Center. Her research focuses on cross-linguistic influences in bilingualism, cognitive contributions to language processing, and brain-behaviour relationships in bilingualism and ageing. Scott Jarvis is Professor and Chair of the Department of Linguistics at the University of Utah. His research concentrations include cross-linguistic influence and lexical diversity. In the former domain, he has developed frameworks for exploring conceptual transfer and confirming cross-linguistic influence in cases where it is difficult to detect or is confounded by competing variables. In the domain of lexical diversity, he has concentrated on developing a multidimensional construct definition of this phenomenon and an accompanying set of measures that account for how lexical diversity is perceived and how it contributes to the quality of both spoken and written language. Michael Johns is a PhD student in Spanish and Language Science at Penn State University. His research focuses on the integration of psycholinguistic methodologies, such as eyetracking and pupillometry, with insights from sociolinguistic data and bilingual corpora to investigate the processing of code-switched speech. A central theme to his research is the role of input and usage in shaping the processing system, and the interaction between external, social factors, and internal, cognitive factors. Jan de Jong is Associate Professor at the Department of Biological and Medical Psychology of the University of Bergen (Norway). Previously he worked at the University of Amsterdam where he is still a guest researcher. His past research has addressed grammatical symptoms of language impairment in Dutch, linguistic precursors of dyslexia and, most recently, language impairment in a bilingual context. He was Vice Chair of COST Action IS ‘Language impairment in a multilingual society’ and, together with Sharon Armon-Lotem and Natalia Meir, edited Assessing Multilingual Children (), a volume that summarizes the diagnostic tools created in the Action.
OUP CORRECTED PROOF – FINAL, 25/6/2019, SPi
xix
Sun-Ah Jun received her PhD in Linguistics from Ohio State University and is Professor at the Department of Linguistics at the University of California, Los Angeles (UCLA). Her research focuses on intonational phonology, prosodic typology, the interface between prosody and the various subareas of linguistics, and language acquisition. She has published the book, The Phonetics and Phonology of Korean Prosody: Intonational Phonology and Prosodic Structure (Garland ) and edited two volumes of Prosodic Typology: The Phonology of Intonation and Phrasing (Oxford University Press, & ). Tuğba Karayayla obtained her PhD in Applied Linguistics in from the University of Essex. Her thesis is on structural and lexical development of Turkish spoken as an immigrant and a heritage language in the UK, investigating attrition of Turkish and its development as a heritage language. Her research areas mainly include language acquisition, language attrition, bi- and multilingualism, and heritage language development. She is working as a lecturer at the Department of English Language and Literature, Yıldırım Beyazıt University, Ankara. Kristina Kasparian is a research associate at McGill University (Montreal, Canada) where she studies the neurocognition of language and multilingualism. Her PhD research used event-related-potentials to examine first language attrition in Italian native-speakers living in Canada, compared to Italian second-language learners. Kristina is especially interested in the factors that modulate native-like language processing, in cross-linguistic influence in different populations of multilinguals, and the parallels that can be drawn between L attrition, L learning, and L attrition. Kristina has also worked on the linguistic and sociolinguistic development of internationally adopted children and on cross-linguistic influences at play in adult bilinguals’ reading processes. Merel Keijzer is an associate professor of Applied Linguistics at the University of Groningen, where she also holds a position as Rosalind Franklin Research Fellow. She was the president of the Dutch association of Applied Linguistics and is currently the vice-president of the Young Academy Groningen and a member of the Young Academy (Jonge Akademie) of the Royal Dutch Academy of Arts and Sciences, where she chairs the interdisciplinarity working group. In her work on bilingualism across the lifespan, too, she adopts an interdisciplinarity framework. Focusing on the effects of long-term and short-term bilingual experiences in older adulthood, the research fields she actively tries to bring together are linguistics, cognitive science and psychology, neuroscience, and geriatrics. As part of the dynamics of bilingualism research, she is also active in language attrition research and wrote several papers and contributed to and edited various volumes on the topic. Her work has received funding from the Netherlands Organisation for Scientific Research (NWO). Terry Kit-fong Au obtained her PhD in Psychology from Stanford University. She was Associate Professor at Brown University and Professor at UCLA, and is currently Chair Professor of Psychology and holds the Karen Lo Eugene Chuang Professorship in Diversity and Equity at the University of Hong Kong. Professor Au is an elected Fellow of the Association of Psychological Science and has served as Associate Editor of the journal
OUP CORRECTED PROOF – FINAL, 25/6/2019, SPi
xx
Developmental Psychology, and research grant panel member for NIMH, NIH, and the Research Grants Council of Hong Kong. Denise Klein is Director of the Centre for Research on Brain, Language and Music, and Associate Professor in Neurology and Neurosurgery, McGill University. She has been in the Cognitive Neuroscience Unit at the Montreal Neurological Institute since where she did her postdoctoral fellowship under Brenda Milner after completing her PhD at the University of Witwatersrand in South Africa. Denise’s early work pioneered the use of brain imaging for the study of bilingualism and provided a springboard for current debates about bilingual brain organization. Her current work focuses on finding behavioural and neural biomarkers that predict language-learning success and difficulty. Barbara Köpke is Professor of Neuropsycholinguistics at the University of Toulouse and head of the Octogone-Lordat Laboratory. Her research involves neuro- and psycholinguistic aspects of language processing in bilingual subjects with specific attention to ‘extreme’ situations such as L attrition, simultaneous interpreting, and aphasia. Her work has appeared in journals such as Linguistic Approaches to Bilingualism and International Journal of Bilingualism, and she is the editor of a special issue of Journal of Neurolinguistics on first language attrition. Marta Korytkowska is currently a doctoral student in the Speech-Language-Hearing Sciences department at CUNY Graduate Center. She is a practising Speech-Language Pathologist in the acute care setting with experience in acute rehabilitation, outpatient, and home care settings. Her main areas of research interest include treatment approaches in bilingual populations with language disorders, bilingual aphasia, and the influence of cognition in recovery from aphasia. Judith F. Kroll is Distinguished Professor at the University of California, Riverside and the former director of the Center for Language Science at Pennsylvania State University. Her research uses the tools of cognitive neuroscience to understand how bilinguals juggle the presence of two languages in one mind and brain. Her work, supported by NSF and NIH, shows that bilingualism provides a lens for revealing the interplay between language and cognition that is otherwise obscure in speakers of one language alone. Tanja Kupisch is Full Professor at the Department of Linguistics, University of Konstanz and Professor II at UiT The Arctic University of Norway. She got her PhD from the University of Hamburg in . She has worked as a researcher and teacher at the universities of Calgary, McGill, Hamburg, Lund, Konstanz, and Tromsø, specializing in first, second and third language acquisition, heritage bilingualism, bilectal acquisition, and language attrition, including child and adult learners. Her work focuses on phenomena such as nominal syntax and foreign accent, including the languages French, Italian, English, German, Turkish, and Russian. Jenifer Larson-Hall is Associate Professor of Applied Linguistics at Kitakyushu University, Japan. Her interest in attrition stems from the effects that age has on the loss of a language as well as the acquisition of it. She undertook attrition research as a quixotic way to discover
OUP CORRECTED PROOF – FINAL, 25/6/2019, SPi
xxi
the ideal age at which a young person might be ‘dipped’ for several years into a different language and emerge as something close to a balanced bilingual. After raising four of her own bilinguals she now sees that learning a language is a much bigger and longer project than she had even imagined. Richard M. Lee, PhD is the Distinguished McKnight University Professor in the Department of Psychology at the University of Minnesota, Twin Cities Campus. His research focuses on the psychological aspects of culture, race, and ethnicity and its role in the development, well-being, and mental health of racial and ethnic minority youth and families. Since , he has been studying the transnational and transracial experiences of families and children who were adopted internationally from South Korea. Esther de Leeuw obtained her PhD in Speech and Hearing Sciences in from Queen Margaret University, Edinburgh. The title of her PhD thesis was ‘When your native language sounds foreign: A phonetic investigation into first language attrition’. She has since held positions at the Centre for Research on Bilingualism at Bangor University, Wales, and at the BBC Pronunciation Unit, in London, UK. She has also spent research sabbaticals at the Lifespan Cognition and Development Lab at York University, Canada, and at the Institute for Phonetics and Speech Processing at the LMU Munich, Germany. She is now Senior Lecturer and Director of the Phonetics Laboratory at Queen Mary University of London, and Associate Editor for the journal Second Language Research. She has published widely on phonetic and phonological dimensions of first language attrition, as well as on speech and bilingualism more generally. Laurent Lefebvre has acquired an expertise in the study of language and executive functioning deficits encountered by patients with neurodegenerative diseases (Alzheimer’s disease, Primary Progressive Aphasia, Vascular Dementia). He is at the head of the UMONS department of Cognitive Psychology and Neuropsychology since . He created some neuropsychological tests (e.g. GREMOTS, DTLA) and cognitive interventions (e.g. Logaatome) specifically dedicated for aged patients with cognitive deficits. Laurent Lefebvre is a member of several scientific and patients’ groups committees in neuropsychology and dementia. He is Full Professor at the University of Mons and Dean of the Faculty of Psychology and Education. Aviva Lerman is currently a doctoral student in the Speech-Language-Hearing Sciences department at CUNY Graduate Center. She is a qualified SLP and holds a Master’s degree specializing in bilingualism and biculturalism. Her research focuses on bilingualism, aphasia, healthy ageing, and dementia. Jared A. Linck holds a PhD in Cognitive Psychology from The Pennsylvania State University. He is a Research Scientist at the University of Maryland Center for Advanced Study of Language (CASL), where he conducts basic and applied research to improve foreign language capabilities. Dr Linck’s research interests include the psycholinguistics of multilingual language processing, second language acquisition, individual differences in learning processes and outcomes, and the impact of both technology use (e.g., web apps,
OUP CORRECTED PROOF – FINAL, 25/6/2019, SPi
xxii
virtual reality) and the context of learning (e.g., immersion, online, self-study) on second language learning outcomes. Brian MacWhinney is Professor of Psychology, Computational Linguistics, and Modern Languages at Carnegie Mellon University. His Unified Competition Model analyses first and second language learning as aspects of a single basic system. He has developed a series of thirteen TalkBank open access online databases for the study of language learning, multilingualism, and language disorders. His research topics include methods for online learning of second language vocabulary and grammar, neural network modelling of lexical development, fMRI studies of children with focal brain lesions, ERP studies of between-language competition, and the role of embodied perspectival imagery in sentence processing. Taryn Malcolm is currently a doctoral student and member of Loraine Obler’s Neurolinguistics lab in the Speech-Language-Hearing Science Department at The Graduate Center, CUNY. She has practised as a speech-language pathologist in acute and sub-acute rehabilitation with a focus on neurogenic disorders and respiratory/voice disorders. Her main areas of research interest include bilingualism, bilingual aphasia, and neurological processes underlying acquired language disorders. She is currently working on a research project investigating cross-linguistic influence in speakers of Jamaican Creole following immersion in the environment of their second language, English. Judith McLaughlin received her PhD in Psychology at the University of Washington. After receiving her degree, Dr McLaughlin served as a Research Scientist in the Cognitive Neuroscience of Language lab at the University of Washington for nearly two decades. Among her numerous contributions is the finding that foreign-language learners’ brain activity discriminates between foreign language words and ‘pseudowords’ after just fourteen hours of foreign instruction, even though the learners were at chance when making conscious word/non-word judgements about these stimuli. She is currently a Lecturer in the Department of Psychology at the University of Washington. Teodora H. Mehotcheva earned a double PhD in Applied Linguistics from Pompeu Fabra University, Barcelona, Spain, and University of Groningen, Groningen, the Netherlands. She is an Assistant Professor of Second Language Studies at New Bulgarian University, Sofia, Bulgaria. Her research interests include second and foreign language attrition, and foreign language acquisition and teaching of English. Silvina Montrul is Professor and former Head of the Department of Spanish and Portuguese and Professor in the Department of Linguistics at the University of Illinois at Urbana-Champaign. Her research focuses on linguistic and psycholinguistic approaches to second language acquisition and bilingualism, with particular emphasis on heritage speakers. She is co-editor of Second Language Research and author of The Acquisition of Spanish (Benjamins, ), Incomplete Acquisition in Bilingualism (Benjamins, ), El bilingüismo en el mundo hispanohablante (Bilingualism in the Spanish-speaking world)
OUP CORRECTED PROOF – FINAL, 25/6/2019, SPi
xxiii
(Wiley-Blackwell, ), and Heritage Language Acquisition (Cambridge University Press, ), in addition to over a hundred peer reviewed articles and chapters. Kleopatra Mytara (MA, University of Strasbourg) is a PhD student at the University of Toulouse , France and a member of the Interdisciplinary Research Unit Octogone-Lordat (EA ). Her research interests include second and foreign language attrition, second and foreign language acquisition and teaching, particularly French as a foreign language, as well as multilingualism and language control. Loraine K. Obler is Distinguished Professor at the CUNY Graduate Center in the programmes of Speech-Language-Hearing Sciences and Linguistics. Her first book, which opened up the field of neurolinguistics of bilingualism, was The Bilingual Brain: Neuropsychological and Neurolinguistic Approaches to Bilingualism (with M. Albert). More recently she has published articles and books on bilingual aphasia and cross-language study of aphasia. As part of her body of work on the language changes associated with ageing, she is now working on L learning in older adults. Janet S. Oh received her PhD in Psychology from UCLA. She is Professor of Psychology and Senior Director of Institutional Research at California State University, Northridge. Dr Oh’s work focuses on the experiences and development of children who come from linguistic minority backgrounds, including the development of children who are exposed to more than one language, as well as adults who are (re)acquiring their heritage language. Her current research focuses on the development of a three-language immersion model for preschoolers. Conny Opitz is Lecturer in German Studies at Cardiff University and holds a PhD from Trinity College Dublin, where she taught Russian, Linguistics, and Translation. Her main research interest concerns the multilingual development of varied populations of adult learners and users from the perspective of Complex Dynamic Systems Theory. Her publications include a co-edited special issue of the International Journal of Bilingualism entitled ‘Dynamics of first language attrition across the lifespan’ (, with E. de Leeuw and D. Lubińska) and a contribution to the volume Complexity Theory and Language Development: In Celebration of Diane Larsen-Freeman (, eds. L. Ortega and Z. Han). Lee Osterhout received his PhD in Psychology from Tufts University. Dr Osterhout uses electrophysiological methods to study language comprehension in native speakers and in foreign-language learners. His research with native speakers indicated that semantic and syntactic anomalies elicit distinct and independent brain responses. He also conducted some of the earliest work demonstrating changes in brain activity that occur during foreign-language learning. He is a Professor in the departments of Psychology and Linguistics the University of Washington, and a member of the university’s Graduate Program in Neuroscience. Diego Pascual y Cabo is Assistant Professor of Hispanic Linguistics and Director of the Spanish Heritage Language Program at the University of Florida. His primary research interest is heritage speaker bilingualism. Over the past few years, his work on this topic has
OUP CORRECTED PROOF – FINAL, 25/6/2019, SPi
xxiv
appeared in several scholarly journals such as Applied Linguistics, Linguistic Approaches to Bilingualism, Heritage Language Journal, and International Review of Applied Linguistics, among others. Silvia Perez-Cortes is Assistant Professor of Spanish at Rutgers University, Camden. She holds a PhD in Bilingualism and SLA from Rutgers University, New Brunswick, and MA in Hispanic Linguistics from the University of Massachusetts-Amherst () and a Bachelor of Arts in Spanish Philology from the University of Barcelona (). Her research interests lie in the areas of bilingual language acquisition in language contact situations. She is particularly interested in analysing the grammatical development of heritage populations (both children and adults) with the objective of exploring how syntax and the lexicon are accessed and represented in the bilingual mind. Lara J. Pierce is a postdoctoral research fellow at Harvard Medical School/Boston Children’s Hospital. She received her PhD in Psychology from McGill University. Her research centres on the effects of early experience on development. In particular, how deviation from ‘typical’ early experiences (e.g., variation in the early language environment, exposure to early stress/adversity) influence the developing brain, and the extent to which the nature and timing of early experiences have a lasting effect on the brain and behaviour. She has received awards from the Canadian Psychological Association, NSERC, Centre for Research on Brain, Language, and Music, and McGill University. Ilona Pitkänen received her PhD in Psychology at the University of Washington. She is currently on the teaching faculty in the Department of Psychology at Bellevue College in Bellevue, WA. Maria Polinsky is Professor of Linguistics at the University of Maryland. Yanina Prystauka received a BA in Translation Studies from Belarusian State University (Minsk, Belarus) in and an MS in Cognitive Science from the University of Trento (Rovereto, Italy) in . She is currently a PhD student in the Language and Cognition programme at the University of Connecticut. Her research interests include sentence processing, event comprehension, and bilingualism. Michael T. Putnam is Associate Professor of German and Linguistics at Penn State University. He has published widely on topics related to morpho-syntax, syntax, and the syntax–semantic interface in language attrition contexts. He has extensive research experience with moribund varieties of German (i.e., Sprachinseln) spoken throughout the world, and is interested in the role that cognitive ageing plays in language attrition in these speakers. He is broadly interested in how bi- and multilingualism impacts and enriches our understanding of the cognitive architecture underlying the language faculty. Claudia Maria Riehl is Professor for German Linguistics and German as a Foreign Language. She is chair of the Institute of German as a Foreign Language and director of the International Research Unit of Multilingualism (IFM) at the Ludwig-Maximilians-Universität
OUP CORRECTED PROOF – FINAL, 25/6/2019, SPi
xxv
München. Her research interests are sociolinguistic and cognitive aspects of multilingualism, language contact, minority languages and language policy, multiliteracy, and second language teaching. Professor Riehl is the author of many titles, among them the German-speaking introduction into language contact research, Sprachkontaktforschung. Eine Einführung (rd edn, Narr, ). Eleonora Rossi’s primary research interest focuses on language processing and bilingualism. She got her MA in Speech Pathology at the University of Padua (Italy) with a thesis on two case studies of bilingual aphasia. She followed this line of work with her PhD in Linguistics at the University of Groningen (The Netherlands), in which she investigated priming effects and language production in agrammatic aphasic speakers. Her postdoctoral work took her to Penn State (US) where she worked with Judith Kroll and Giuli Dussias on the neurocognition of bilingualism. Her recent work has utilized Event-Related Potentials (ERPs), functional and structural magnetic resonance imaging (MRI) data to unveil the neural bases of bilingual language control, code-switching, and first language attrition. Dr Rossi is currently Assistant Professor in the Linguistics Department at the University of Florida, and Research Associate at UC Riverside. Jason Rothman is Professor of Linguistics at UiT the Arctic University of Norway. He also has a fractional research appointment at the Universidad Nebrija in Madrid. He is the Executive Editor of the journal Linguistic Approaches to Bilingualism (LAB) as well as Editor of the book series ‘Studies in Bilingualism’ (SiBIL). Rothman’s research focuses on language acquisition and processing in children and adults in addition to the interface between domain general cognition and language. Some of his recent research has appeared in Bilingualism: Language and Cognition, Journal of Experimental Psychology: Learning, Memory and Cognition, Studies in Second Language Acquisition, Proceedings of the National Academia of Science (PNAS), Language, Cognition and Neuroscience, and Frontiers in Psychology. Liliana Sánchez is Professor of Spanish at Rutgers, The State University of New Jersey. She has published Bilingualism in the Spanish-speaking World with Jennifer Austin and Maria Blume (Cambridge University Press, ), The Morphology and Syntax of Topic and Focus: Minimalist Inquiries in the Quechua Periphery (John Benjamins, ), and QuechuaSpanish Bilingualism: Interference and Convergence in Functional Categories (John Benjamins, ) as well as articles in journals such as Linguistic Approaches to Bilingualism, Bilingualism: Language and Cognition, Lingua, International Journal of Bilingualism, Probus, Studies in Second Language Acquisition, among others. She has also co-edited the volumes Information Structure in Indigenous Languages of the Americas () and Romance Linguistics (). She is a co-founder of the PhD Program in Bilingualism and Second Language Acquisition at Rutgers. Monika S. Schmid is Professor of Linguistics at the University of Essex, having previously held positions at the Vrije Universiteit Amsterdam and the Rijksuniversiteit Groningen. Her research focuses on various aspects of first language attrition and has been published in journals such as Bilingualism: Language and Cognition and Applied Psycholinguistics. She is the author of several books including Language Attrition (CUP, ) and First
OUP CORRECTED PROOF – FINAL, 25/6/2019, SPi
xxvi
Language Attrition, Use, and Maintenance: The Case of German Jews in Anglophone Countries (Benjamins, ). Elena Schmitt is a Professor of Applied Linguistics and Director of the TESOL/Bilingual Education Program at Southern Connecticut State University in New Haven, Connecticut. A native of Russia, she was educated at Moscow State Linguistics University and received her PhD at the University of South Carolina. Her teaching focuses on linguistic theory and its practical applications to educating English learners. Her personal bilingual experiences continue to ignite her research interests in the fields of childhood bilingualism, bilingual education, language attrition/loss, classroom interaction, teacher education, and technology-assisted teaching. Michael Sharwood Smith is Emeritus Professor at Heriot-Watt University, Edinburgh, Honorary Professorial Fellow at Edinburgh University and a founding editor of Second Language Research. His main interests are what links language representation, language processing, and language development. To this end, he is working with John Truscott on the Modular Cognition Framework to integrate diverse explanations of language acquisition, attrition, and processing within a general account of human cognition as a whole drawing on current thinking in different branches of cognitive science. His more recent books include Introduction to Language and Cognition () and, with Truscott, The Multilingual Mind (). Roumyana Slabakova is Professor and Chair of Applied Linguistics at the University of Southampton, UK, and a Research Professor II at UiT The Arctic University of Norway. Her research interest is in the second language acquisition of meaning, more specifically phrasal-semantic, discourse, and pragmatic meanings. Her monographs include Telicity in the Second Language (Benjamins, ) and Meaning in the Second Language (Mouton de Gruyter, ). She co-edits the journal Second Language Research (SAGE) and is a founding editor of Linguistic Approaches to Bilingualism (Benjamins). Her textbook Second Language Acquisition was published by Oxford University Press in . Antonella Sorace is Professor of Developmental Linguistics at the University of Edinburgh. She has published widely on bilingualism and language learning over the lifespan, language typology, and gradience in natural language, bringing together methods from linguistics, experimental psychology, and cognitive science. She is also committed to public engagement and is the founding director of the research and information centre Bilingualism Matters. Karsten Steinhauer is Professor of Cognitive Neuroscience and Director of the Neurocognition of Language Lab at McGill University in Montreal. He and his students investigate the brain’s electrical activity (EEG) to uncover how the human mind/brain processes language in real-time, and what changes when we learn a second language or lose our first one. His original research has been published in high-impact journals such as Nature Neuroscience, PNAS, and the Journal of Cognitive Neuroscience, and has been supported by
OUP CORRECTED PROOF – FINAL, 25/6/2019, SPi
xxvii
all three federal agencies in Canada, by the Canada Research Chair programme, and by many prestigious studentships (e.g., Tomlinson, Vanier, PBEEE, and Fulbright). David Stringer, Associate Professor of Second Language Studies Indiana University, researches the acquisition of syntax and lexical semantics. He is interested in universal aspects of word meaning that play a role in grammar across languages. Other areas of research interest include World Englishes (especially in South Asia and Sub-Saharan Africa), language attrition (in joint work with Kathleen Bardovi-Harlig), and biocultural diversity (linking language revitalization in indigenous cultures to the conservation of ecosystems). His interests extend to all formal aspects of second language acquisition and multilingualism, and he works with students on a wide range of projects. Hideyuki Taura is Professor of Applied Linguistics at the Graduate School of Language Education and Information Science, Ritsumeikan University, Japan. He obtained his MA and PhD from Macquarie University, Sydney. His research interests include () bilingual language acquisition and attrition from both linguistic and neuro-linguistic perspectives, () early bilingual cognitive development, () bilingual narrative development, () EFL teaching in Japan, () novice English teachers’ professional development in China, and () translanguage teaching for heritage language education. Sharon Unsworth is Associate Professor in the Department of Linguistics, Department of English Language and Culture and the Centre for Language Studies at Radboud University in Nijmegen. Her research interests include child bilingualism and second language acquisition, both in naturalistic and instructed contexts, and she has authored articles in internationally renowned journals such as Applied Psycholinguistics, Bilingualism: Language and Cognition, and Language Acquisition, as well as numerous book chapters. Jorge R. Valdés Kroff is Assistant Professor of Spanish and Linguistics at the University of Florida. He is the director of the Bilingual Sentence Processing Lab which uses behavioural and eye-tracking methods to investigate how bilinguals adapt their parsing preferences to better anticipate upcoming code-switches. His recent work focuses on testing whether bilinguals tap into production asymmetries to guide comprehension and recruit greater engagement of cognitive control to rapidly integrate upcoming code-switches. Álvaro Villegas is Assistant Professor in the Department of Modern Languages and Literatures at the University of Central Florida, where he teaches Spanish Linguistics. He earned his PhD in Spanish Linguistics and Language Sciences at The Pennsylvania State University in . His research focuses on Psycholinguistics and Bilingualism, with a special interest in the production and comprehension of heritage speakers. He also works with bilingual communities in Orlando to inform parents and educators about the benefits of bilingualism and to promote bilingual education. Gülsen Yılmaz is currently a Post-Doc at the Humboldt University of Berlin, working with Prof. Artemis Alexiadou and her research group. Her current research focuses on the variability of the article system in heritage language grammar. She holds a PhD in Linguistics from Groningen University, the Netherlands. She did her MA in foreign
OUP CORRECTED PROOF – FINAL, 25/6/2019, SPi
xxviii
language education. Her research includes theoretical and experimental methods studying bilingual development with a focus on psycholinguistics aspects of language acquisition and attrition. She taught language attrition, contact linguistics and experimental methods at undergraduate and graduate levels. Margarita Zeitlin is a PhD candidate in the Department of Psychology at the University of Washington. Her research interests include language learning and the relationship between language and other cognitive domains.
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
......................................................................................................................
......................................................................................................................
. ö
I has become a time-honoured tradition for overviews and introductory chapters on language attrition to begin by pointing out the recency of the field. Most commonly this is done by referring to the time period elapsed since the publication of Lambert & Freed (eds.) in . We considered adhering to this tradition, or alternatively referring to the (much more impressive) eighty years since Einar Haugen published his seminal essay ‘Language and Immigration’ in , which not only—to the best of our knowledge—coined the term ‘attrition’ but also anticipated many future findings (for more details on this essay, see Schmid & de Leeuw, Chapter , this volume). We eventually decided, however, to use a much more personal milestone: as this volume is going into print we are coming up to the twentieth anniversary of a cold November afternoon in the small Dutch town of Veldhoven where the Third International Conference on Maintenance and Loss of Minority Languages () took place. At this conference two PhD students were presenting their first ever posters summarizing the initial findings of their respective studies on German L attrition. This was how the two authors of this introduction first met. We got to talking and it soon became evident that, while the scope, methodology, theoretical framework, and results of our two research projects were entirely different, we shared the same concerns regarding the field of attrition studies, which not long afterwards we summed up as follows: [ . . . ] research is still very much divided on the question of whether any such thing as adult L attrition exists in the first place. Such differences are likely to be the outcome of inconsistencies in the methodology used in the collection and interpretation of the data—and not necessarily of actual differences in the linguistic repertoire of the speakers analysed by these different studies [ . . . ]. It has been pointed out that there is great variance in the amount of attrition that individual studies have found, but it has not, so far, been investigated in what way these differences are linked to the methodology of collecting and analysing data. [We hope to] reconcile some of these differences, and develop a sound methodological framework for further studies in language attrition. For this, it appears crucial to reconcile the different disciplines involved, to take advantage of the strength of the linguistic, sociolinguistic and psycholinguistic approaches and to join them in an interdisciplinary frame able to account for all types of variation which render comparisons between different studies difficult or impossible. (Call for Papers for the First International Conference on First Language Attrition, which was held at the VU Amsterdam in 2002)
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
ö
This was the research agenda that we felt should guide language attrition studies into its next phase. We knew that it was an ambitious one. What we did not know at the time was to what extent we were fortunate to come to the topic at the right time and place, with circumstances as well as the personal generosity of many individuals fortuitously conspiring to support our efforts. In the first instance, the networking aspect of our agenda benefitted enormously from the fact that, at that time, it was still possible to bid for relatively small sums of money to support stand-alone scientific events. Today, in the rare cases where conference and workshop funding opportunities still exist, one must usually come up with series of linked events, preferably to be held in different countries—an endeavour that is simply not feasible for early career researchers with limited professional networks and experience. These funds—generously awarded by organizations such as the Netherlands Organisation for Scientific Research (NWO), the Dutch Royal Academy of Sciences, as well as our own institutions (the VU University Amsterdam, the University of Toulouse, and the University of Groningen) allowed us, between and , to not only organize two international conferences and five meetings of the European Graduate Network on Language Attrition but to provide financial support to a very substantial number of postgraduate students who would not otherwise have been able to attend these events. With the help of a Seminar Series grant from the British Economic and Social Sciences Research Council (ESRC), this was followed between and by a series of eight events focusing on interdisciplinary aspects of language attrition research, including the Third International Conference on Language Attrition (see https://languageattrition.org/attrition_seminars/). The focus of the initial series of meetings was on the development of a common methodology for language attrition studies. What we suggested was that each study, no matter what its eventual aim, theoretical framework, or specific linguistic interest, should adopt a common baseline methodology with respect to the collection of background features and language proficiency data, in order to make it possible to benchmark findings from different languages and different settings against each other. The tools we developed comprised a range of questionnaires on personal background, language use and attitudes, and self-evaluated proficiency, as well as the use of a C-Test, Verbal Fluency tasks, and a short passage of semi-structured discourse elicited through a film retelling task. The instruments, along with instructions on how to use, code, and analyse them, were made available in a number of languages on languageattrition.org, and are described in detail in Schmid (). They have since formed the basis for a large number of studies (see also Schmid & Cherciov, Chapter , this volume). While the exchange of ideas, methods, and results that took place at these events and the personal links and contacts thus came to form the backbone of language attrition research over the next years, the personal friendships and mentor/mentee relationships that developed from them became its heart. Kees de Bot, who himself has undoubtedly been the single most important figure in not only creating and shaping the field of L attrition research but leading it into the twenty-first century, remarks on this in the concluding chapter to this volume. Where previously attrition research had been characterised by the solitary existence of the PhD-student, spending a large part of her time doing fieldwork and, typically, having little opportunity and less money to go to conferences and make contacts in the international scientific community. (Köpke & Schmid, , p. )
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
PhD students, postdocs, and young researchers like ourselves began to feel part of the wider community of applied linguists, with the bonus of a distinctive identity as an attrition researcher. This is how not only the co-editors of the present volume but also many of its contributors began their career, and it is only partly due to the support from the abovenamed funding bodies, and in much greater measure to the generosity, kind-heartedness, openness, interest, and sound advice of established senior researchers such as Kees and too many others to name here. The second attendant favourable circumstance for our ambitions to advance language attrition from its niche status to an integral part of research on bilingual development were technological and practical innovations which over the past two decades have made it possible and feasible to conduct investigations on a previously unimaginable scale and level. Our efforts began around the time when the analysis of spoken data (an indispensable cornerstone to attrition research) made the quantum leap from painstaking (and, as we know from personal experience, unspeakably tedious) hand-coding to freeware tools such as the ones developed in the CHILDES-project or ELAN. The exponential increase in size and affordability of computer memory furthermore made it feasible to work with much larger audio, or even video, corpora, not to mention the availability of fully tagged corpora often comprising millions of words of the target languages, allowing much more precise benchmarking. Experimental methods, too, developed at breathtaking speed—it was not so long ago that even reaction-time software was exorbitantly expensive, while today many linguistics departments have their own eyetrackers and/or EEG scanners. Second language acquisition research had only just begun to appropriate all of these methodologies at the time that the authors of this chapter first met, so we are delighted that attrition research has since begun to avail itself of all of them, as many of the chapters in the present volume demonstrate. Alongside such methodological and practical issues, we felt that another key area in need of attention was the theoretical underpinnings of attrition research. This became the focus of the second attrition conference (held, again at the VU Amsterdam, in ), at which we attempted to bring together more theoretically-deductive and more empirical-inductive approaches to language attrition, and also to interest researchers from neighbouring areas of linguistics—such as contact linguistics or ethnolinguistic vitality theory—in applying their insights and theories to language attrition data. These efforts are ongoing, and we are delighted that some of the theoretical chapters in the first section of the current volume (such as the ones by MacWhinney, Putnam et al., and Slabakova, Chapters , , and , respectively) represent novel theoretical takes on language attrition. All of these developments have helped language attrition research to ‘reach its adulthood’ (de Bot, Concluding remarks, this volume). The increase in empirical work has demonstrated beyond any doubt that attrition is not, as had originally been assumed, an ‘extreme’ form of development, experienced by a small minority of bilinguals with decades of no or very little exposure, but a form of language development experienced from the early stages of second language acquisition and thus, in all likelihood, common to all bilinguals (Schmid & Köpke, a). The present Handbook is testimony to the fact that attrition is now very much part of the mainstream of applied linguistics. We mentioned above a second series of seminars (financed by the ESRC), in which the focus turned towards interdisciplinarity. We would like to suggest that this is indeed what should be on the agenda next for the field of language attrition: having consolidated itself as
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
ö
an area of investigation with a firm sense of identity, a solid theoretical and methodological foundation, and a wide base of empirical evidence, we feel that it is time to turn our attention again to the fact that language attrition is only ever one side of the coin of bilingual development. Limiting the perspective of investigations to what happens in the native language of bilinguals ignores at least half of the process. The same is true, we feel, for investigations of second language acquisition which, in turn, ignore language attrition. Our agenda, then, for the next phase of research in bilingual development should be to argue most strongly that, whether the object of an investigation is L acquisition or L attrition, ignoring the other language is to bias the results and limit the potential for understanding what actually happens when two languages interact within the same mind.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
.............................................................................................................
THEORETICAL IMPLICATIONS OF LANGUAGE ATTRITION ............................................................................................................. Edited by
. ö
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
......................................................................................................................
......................................................................................................................
T chapter explores ways in which the Competition Model can illuminate our understanding of language attrition. We begin by formulating some of the key puzzles that arise from empirical studies of language attrition. After this initial problem description, we will consider how the cognitive and social processes included in the Competition Model can help in the formulation of solutions to these various puzzles.
. F
..................................................................................................................................
.. Puzzle #1: permastore It is easy to imagine that language attrition arises simply from disuse of a speaker’s first language (L). The idea is that, for any given ability, you must ‘use it or lose it’ (Pinker, , p. ). Experimental psychology has provided abundant evidence indicating that unused memory traces fade over time (Ebbinghaus, ). In some accounts, forgetting arises through the weakening of a memory trace by changes at the level of the synapse. In other accounts, forgetting arises through interference of old memories with new memories. In either scenario, failure to repeatedly access and use the old memory is viewed as the cause of forgetting and problems with retrieval (Paradis, ). Problems of this type also arise in people suffering from dementia or anomia. However, there is also evidence that adults can maintain linguistic representations for decades without active use. For example, Bahrick (b) and Bahrick & Phelps () demonstrated that vocabulary learned in college Spanish classes was maintained after as much as fifty years without significant maintenance or reuse. This type of ‘permastore’ corresponds to what we find in people who continued using L through adolescence, but then ceased its use during adulthood. These speakers retain their ability to use their
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
L despite years of disuse (Schmid & Dusseldorp, ). We can refer to these speakers as ‘maintainers’. We can contrast people in this group with ‘attriters’ who show a greater loss of L forms. Within this group, we can contrast ‘strong attriters’ who have undergone the greatest amount of attrition with ‘weak attriters’ who show less attrition, but still not full maintenance. What is puzzling for standard psychological theories of forgetting is that even the strongest attriters maintain a large amount of their L (Steinkrauss & Schmid, ). We can also distinguish a third group of adult second language (L) learners that maintains use of L in various aspects of daily life. Because this group suffers minimally from attrition, we can refer to them as ‘late bilinguals’. There are no sharp cutoffs between these three groups; rather they represent various points along a continuum of degree of language attrition. Understanding why a speaker ends up at a given point along this continuum constitutes a further aspect of this puzzle.
.. Puzzle #2: variation across levels Both attriters and maintainers show the impact of an increasingly stronger L on a somewhat weakened L. These effects can reveal themselves through the inappropriate use of L words, expressions, and structures while speaking an L. In such cases, we can say that the L and the L are competing for activation and that the L form is used because the L form is weaker. There are at least five ways in which this competition can play out. . First, intrusion of an L form when speaking L may arise from problems retrieving the L form. If the activation of the L form is insufficient, the L form will win out in the horse race for activation and production (Ratcliff, ). This effect underscores the extent to which L remains active, if somewhat repressed, during production of L. . In a related case, the weaker activation of the L form can lead to a total failure to recall the required form. In such cases, we see omissions, hesitations, and other disfluencies without actual intrusion of an L form. . When L and L share words or constructions, it can happen that both sources of activation support the use of the same form (Tokowicz & MacWhinney, ). In such cases, intrusions may be scarcely noticed. This is particularly likely in closelyrelated languages, such as Norwegian and Danish. . In other cases, it may be that the L form is slightly more appropriate in the given context than the L form. This type of transfer from L resembles what we see in cases of lexical insertion during code-switching (Poplack, ; Myers-Scotton, ) in which a speaker will insert an L word into an L matrix sentence because it seems like the mot juste or most appropriate form in a particular context. It would be a mistake to think of this use of an L form as indicative of a weakened L alternative. . If a speaker’s L has marked dialect features, the contrast between those features and the features of the standard version of L can further complicate the competition between L variants and L. What may appear to be an error in L may simply be a reliance on a non-standard form. In some cases, the form in question may derive from an earlier standard L form that has now fallen out of use. Such competitions between alternative forms can lead to delays or errors in retrieval.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
What complicates matters further is that these alternative forms of competition and transfer work in different ways across the language levels of output phonology, input phonology, lexicon, collocation, syntax, and pragmatics. For output phonology, Schmid (a) has shown that maintainers are perceived as having the accent of native speakers, although strong attriters may eventually lose their native accent. We know that second language learners show a type of backward transfer in terms of shifts in the details of their control of phonetic features, such as voice onset time (VOT) for stop consonants (Major, ) during the process of L learning. The fact that this happens so quickly during early stages of L learning is surprising, given the relative resistance of L phonology to loss in language attrition. However, the variations that L learning forces on L phonetics are not so extreme as to push L sounds out of the normal range; rather, they represent deviations within the overall range of potential L sounds. Basic syntactic patterns also seem resistant to loss during attrition. However, attriters can begin to reduce their use of marked orders and formats, avoiding some of the more complex L forms they would have used in earlier years. These adjustments amount to minor losses of L forms and fluency (Schmid & Beers Fägersten, ). Complete abandonment of L syntax under the pressure of L learning seems to be minimal, even in strong attriters. The relative resistance of phonological and syntactic structures to attrition contrasts with greater attrition in vocabulary, idioms, and collocations. There are three theoretical approaches to explaining this contrast. A basic usage-based or frequentist account predicts that the least frequent L words should be the ones most subject to L intrusion. A further consequence of this would be that the first forms that are lost are also the ones that were learned last during L acquisition (Wixted, ). However, in a careful study of frequency effects on these intrusions, Schmid (a) finds only marginal effects for lexical or collocational frequency. A second account, based on the theory of between-language competition, suggests that intrusions arise from forces such as the appropriateness of the L form, contextual support for code-switching, and structural similarities between L and L. A third approach focuses on the contrast between procedural and declarative memory systems (Ullman, c). Here, the idea is that entrenchment operates most strongly for systems such as output phonology and word order patterning that have undergone proceduralization through basal ganglion and striate functioning (Stocco et al., ), whereas lexical and collocational patterns that are more responsive to hippocampal consolidation are more susceptible to intrusion. Figuring out which of these three accounts for the contrasting effects of linguistic levels is operative in a given case is a major challenge for studies of language attrition.
.. Puzzle #3: childhood forgetting Of the four puzzles related to language attrition, the third is the most perplexing. This puzzle arises from the observation that children who are adopted into a new family in a new country before the age of show a nearly complete loss of the ability to use their L (Pallier et al., ; Ventureyra et al., ). A basic psycholinguistic characterization of the process of entrenchment would predict that this should not happen. If a child is younger than age when coming to a new country, one could argue that their language had not yet fully consolidated and was therefore vulnerable to complete language loss. However, by age ,
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
a child will have spent six years making continual daily use of the L in fully appropriate social contexts. Based on Bahrick’s (b) findings regarding the entry of memories into a ‘permastore’, these children should have thoroughly consolidated or ‘overlearned’ their L by the age of . The type of natural learning that children experience is particularly well suited to the development of long-term consolidation (Pashleret al., ; Pavlik & Anderson, ). Moreover, studies of monolingual first language learners demonstrate entrenchment well before the age of . For example, Brooks & Tomasello () showed strong entrenchment effects for item-based patterns in causative constructions in English by age . These considerations indicate that late adoptees should retain a significant amount of L knowledge in permastore. At the very least, this knowledge should facilitate rapid relearning or recognition of L forms (MacLeod, ). However, studies of adoptee’s L relearning during adolescence have provided only weak evidence of any retention of L sensitivities, as revealed through somewhat more accurate relearning of some phonetic contrasts (Au et al., ; Oh, Au, & Jun, ; Oh, Jun, et al., ; Park, ; Singh et al., ). However, even these results may be due to motivational factors, rather than stored memories. There are two solutions that have been offered to this puzzle of childhood forgetting. The most popular solution is the Critical Period Hypothesis (CPH), which attributes childhood L loss to the expiration of a critical period for language acquisition. However, that theory was intended to account for language learning, not language loss. Nonetheless, it might be possible to relate early language loss to neurodevelopmental processes that impact both the ability to learn and the ability to lose a language. For example, declines in susceptibility to loss could be related to increases in the myelination of cortical layers. On this account, a high level of neuronal plasticity facilitates new learning, but also permits quick loss of new patterns. The theory of Universal Grammar offers a related viewpoint on this issue, particularly as elaborated in terms of ‘biolinguistics’ (Boeckx & Grohmann, ). On this account, L learning involves the setting of a series of parameters. Problems with resetting these parameters are supposed to be at the root of difficulties adult learners have acquiring L. In accord with the CPH, parameter setting by children is labile and incomplete, allowing for reversal and modification during childhood, but not afterwards. The actual neuronal implementation of this version of the CPH account could conceivably rely on the same neuronal mechanisms used to support other accounts based on entrenchment and plasticity. A second approach focuses on changes in the social and emotional supports needed for language maintenance. Pulvermüller & Schumann () argue that first language learning links in more directly to emotional roots than does second language learning. In addition, we need to consider the extent to which adoption and immigration can break off links between language and the original social and experiential contexts to which linguistic forms refer. Moreover, childhood forgetting may arise from the involvement of additional neuronal circuitry supporting changes in attention and motivation. We will consider these two accounts—critical periods versus abrupt contextual change—in further detail later.
.. Puzzle #4: catastrophic interference The fourth and final puzzle derives as a further complication from the third. This puzzle involves the fact that parallel distributed processing (PDP) connectionist models of second language acquisition demonstrate ‘catastrophic interference’ or complete loss of L at the
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
beginning of learning of a second language. This occurs because L knowledge overwrites the connection weights established during L learning (Zevin, ). The fact that adult maintainers demonstrate minimal language loss argues strongly against models predicting such catastrophic loss. At the same time, developers of these models could point to catastrophic loss in children as evidence in support of their account. There are also other neural network models, such as DevLex (Li et al., ) that avoid the problem of catastrophic interference through reliance on learning through self-organizing feature maps. However, those models then have no ready account of the phenomena of language loss by children.
. T C M
.................................................................................................................................. Taken together, these four puzzles indicate that no single-process model will be sufficient to account for the observed phenomena in language attrition. A model that combines the mechanism of entrenchment with a model of transfer would come close to explaining some of the puzzles, particularly if it incorporates a theory of differences between linguistic levels. However, a full account of these phenomena will require still further detail regarding social and motivational inputs, contextual changes, internal processing mechanisms for consolidation, and neurological foundations of plasticity and reorganization. The Competition Model (MacWhinney, b, ) is designed to provide an integrated approach to these issues, as well as to additional issues in online language processing, first and second language learning, bilingualism, code-switching, and language loss in aphasia. The model holds that language, as a fundamentally biological system, emerges from processes of competition. Like other emergent biological systems, language structure derives from properties and interactions unique to hierarchically organized levels of processing that interoperate across differential timescales. Implications of the emergentist constructs of competition, levels, and timescales are further explored in MacWhinney (a). This view of language structure as emerging from competing motivations (MacWhinney et al., ) is based on mechanisms that have been documented by experimental psychology, cognitive linguistics, sociolinguistics, typology, neuroscience, conversation analysis, and social research. However, we do not yet understand how these competing component processes fit together cooperatively into a functional system. This means that use of the model to account for patterns of language attrition requires extensive consideration of the ways in which competing motivations and forces are integrated. For the study of language attrition, the most important aspects of Competition Model analysis are those relating to the interplay between risk factors and protective factors. Risk factors are forces that operate to reduce smooth language learning and functioning. Support factors are forces that work to promote fluent language functioning. The model specifies the risk factors of entrenchment, transfer, over-analysis, and isolation, and the support factors of resonance, decoupling, chunking, and participation. The processes of over-analysis and chunking apply mostly to the first stages of language acquisition, so we will not consider them here. Let us consider how the other six factors (entrenchment, transfer, isolation, participation, decoupling, and resonance) operate in the case of language attrition.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
.. Entrenchment The basic unit of Competition Model analysis is the cue. Cues are relations between forms and functions (de Saussure /). Cues can involve units on any level of linguistic structure—phonology, lexicon, or grammar. If a cue correctly predicts a function during comprehension or selects out the correct expression during production, then it is high in reliability. Competition Model studies (MacWhinney, ) have shown that, during early stages of language learning, cue frequency or availability is more important than reliability. However, over time, cue reliability becomes the major determinant of cue strength. Cue availability and reliability are estimated through corpus-based counts (McDonald, ; Ellis et al., ), whereas cue strength is estimated through experimental studies that place different cues into competition (MacWhinney & Bates, ). A core finding of these studies is that cue reliability closely predicts cue strength. Like other models that are implemented in the neural network framework, the Competition Model assumes that cues are in continual competition and that this competition is determined through cue strength. Entrenchment arises from the continual use of a cue. When a cue is reliable, strength is increased; when it is unreliable, strength is decreased (McDonald & Heilenman, ). Entrenchment from this process of cue strengthening must play a central role in language maintenance and attrition. However, the four puzzles we have discussed suggest that entrenchment cannot work alone, and that its operation may be quite complex. Recent work in neurobiology has begun to uncover some of the mysteries of entrenchment in sensory areas. Examining the neuronal basis of amblyopia (poor visual acuity) arising from a lazy eye, researchers (Takesian & Hensch, ) have demonstrated an interplay between five neurobiological forces that lead to the entrenchment of a cortical region and those that can reopen that area for new learning. . Initial brakes. Before the onset of entrenchment, the polysialic acid-neural cell adhesion molecule (PSA-NCAM) prevents the opening of a cortical area to learning. . Triggers. A sensitive period opens when molecular triggers (such as orthodenticle homeobox 2, BDNF)—in response to sensory input—promote parvalbumin (PV) cell maturation, turning on GABA (gamma-aminobutyric acid) circuit functioning. GABA circuit maturation may operate by suppressing responses to intrinsic activity, allowing a focus on extrinsic input (Sale et al. ; Van Versendaal et al., ; Toyoizumi et al., ). . Learning. Learning then involves physical pruning and homeostatic regrowth of synapses by tPA, TNFα, and protein synthesis. . Terminal brakes. A consolidated or entrenched state is maintained by the production of functional molecular brakes, such as Lynx1 and NgR1, and physical brakes such as PNN and myelin, which may limit plasticity by preventing further structural changes. These brakes rely on the same processes that created changes during the plastic period, essentially preserving these changes. . Reopeners. The reopening of cortical areas to new inputs can be achieved by lifting the brakes through epigenetic manipulation (HDAC inhibitors) or heightened attention (neuromodulators such as 5-HT, Ach, NE). The operation of this system of excitation and inhibition across sensory levels varies based on the timing of maturation of the PV cells that trigger the opening of the learning
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
period. The variability in these distributions of PV cells could help us understand variations in entrenchment and maintenance across language levels. Because entrenchment can be reversed through attentional processes, we should think of this system as one supporting variable entrenchment, rather than an irreversible critical period. We should also note that research to date has focused on the operation of these brakes and reopeners in sensory systems. How they operate in motor areas and higher cortical regions is not yet clear, although the neurobiological components are common across areas.
.. Transfer Entrenchment constitutes a risk factor for L learning, because it restricts the availability of cortical areas for new learning, forcing the learner to engage in transfer of L forms to L. At the same time, entrenchment serves as a protective factor promoting L maintenance, as opposed to L loss. However, once L has become established as the primary means of daily communication, transfer from an increasingly stronger L to L begins to operate against the forces of L entrenchment. Because of the emphasis it places on interactive activation, the Competition Model holds that, for initial L learning, ‘everything that can transfer will’. In the domain of phonology, whenever an L phoneme can be mapped onto an L sound, learners will attempt to transfer that form and slowly correct for the inevitable mismatch. For lexicon, the meanings of L words will transfer, although the sounds typically will not. In morphology, transfer is seldom possible except between closely-related languages. In syntax, transfer will impact collocations and even basic word order. In each of these areas, what will transfer from L is the unmarked pattern. Studies of transfer in L learning largely support these predictions. There is evidence that they also apply to language attrition, as an increasingly stronger L begins to transfer patterns to L. For example, Tóth () compiled an inventory of transfer patterns in Hungarian and German L speakers in the San Francisco Bay Area. He could identify an L English source for each erroneous form he observed. The patterns of language transfer found in such studies also closely resemble those reported in cases of simultaneous childhood bilingualism (Döpke, , ; Yip & Matthews, ). Looking in greater detail at areas of particular susceptibility to transfer, some studies (Seliger, ; Ribbert & Kuiken, ) have suggested that transfer is particularly strong in areas where uses of an L form constitute a subset of uses of a similar, but more general, L form. This finding is in accord with the Competition Model expectation that the degree of interference from L to L will depend on the closeness of the mapping, as well as the markedness of forms in L and L (Eckman, , ). Because of its emphasis on competition, interaction, and variation, the Competition Model also predicts back transfer from L to L, as well as partial merger of grammar in certain areas. Such effects have been observed for syntactic structure (Liu et al., ; Dussias, ; Dussias & Sagarra, ), lexicon (Malt et al., ), and phonology (Major, ; Mennen, ), sometimes arising after only a few weeks of exposure to L (Chang, ). Thus, when studying speakers of two or more languages, we must always keep in mind the possibility that there will be dynamic interactions between the languages (Cook, a). However, such interactions do not, in themselves, constitute full language attrition.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
.. Participation, isolation, and identity For successful L learning, the single greatest protective factor is the ability to participate actively in L social groups without having to rely on switches to L. Conversely, the greatest risk factor for L learners is the threat of isolation from L groups. De Bot & Clyne (, ) have described situations in which immigrants give up on attempting to integrate with the L community and revert to reliance on L. However, there are also cases in which immigrants integrate so thoroughly with the L community that they cease daily use of L. This is the situation that gives rise to Puzzle # regarding the maintenance of an unused L in permastore. Schmid () conducted an examination of attrition effects in Jews who escaped to Anglophone countries from Germany between early adolescence and young adulthood at various times during the Fascist pogroms. If they escaped before the onset of the racial discrimination laws, these people tended to maintain a high level of functional use of German, along with a basic acceptance of their German heritage. However, people who escaped after the November pogrom and the beginnings of deportations tended to reject their German language and heritage. Schmid found that the single best predictor of degree of language attrition in this group was the time of leaving Germany. Although this is an extreme case, it still serves to reveal the importance of fundamental cultural identification as either a risk or protective factor for language maintenance. There are two ways in which cultural identity can play out on a daily level. For those who have decided to reject their L identity, there can be active suppression of L words and thoughts. This type of suppression of L can be accompanied by enthusiastic use of new L forms both in conversation and in inner speech, formulation, and planning. The other way in which cultural identity can play out in everyday life is through the organization of activities in L and L. Schmid (a) reports that ongoing use of L in a professional context is a good predictor of a minimal level of L attrition. For example, the need to use L for translation or writing requires ongoing attention to correct use of L forms. Conversely, if L usage is confined to less demanding environments such as the home or if it is being used in code-switched interactions, then relatively more attrition will occur. The link between identity and language maintenance may also help us understand Puzzle # regarding children’s forgetting of L. For many adoptees, life in the L setting may have been relatively unstable and harsh. Moreover, the new L environment provides little support for or even recognition of L. Because of this, adoptees may have also decided to actively repress their L. Burling () presents a rather poignant case of this type. During a two-year-long field trip in the hills of Myanmar, his son acquired Garo and English simultaneously. On the airplane flight home to the States, the -year-old boy attempted to speak to the flight attendant in Garo and she only replied in English. After that, he never used a word of Garo again.
.. Resonance In second language learning, the effects of L entrenchment can be countered through the process of resonance. This process is supported by resonant connections between the hippocampus and cortical areas, leading to the consolidation of memories and linkages
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
in the cortex. The hippocampus has connections to all cortical areas, each arranged in distinct subareas. When a new experience or episode is encountered, the hippocampus uses these connections to maintain an ongoing interaction or conversation with these cortical areas. This is not a one-shot activation of the cortex, but a process of ongoing maintenance of cortical activity (Nadel et al., ) through reentrant connections (Wittenberg et al., ). These patterns of activation go beyond mere consolidation of individual episodes. Relying on additional control from medial prefrontal areas, the consolidation process can link new experiences with old concepts into increasingly rich conceptual networks (Schlichting et al., ; Zeithamova et al., ). This method of building up new relational networks from new experiences can be particularly important for consolidation of language structures such as words and collocations, as specified in the Competition Model theory of learning from item-based patterns (MacWhinney, ). For people who are trying to maintain their L in an L environment, resonance can rely both on continued L input from conversation and reading, as well as internal use of L for formulation and planning. Resonance can operate both passively and actively. Its passive operation can be detected during rest (Karni & Sagi, ) and sleep (Lindsay & Gaskell, ). Its active function arises when learners use orthography to activate phonology (Share, ) or when they derive word meanings from analysis of derivational and inflectional patterns (Presson et al., ). There is also a second form of consolidation that is important for motor and sequential procedures. This system utilizes frontal-striatal and basal ganglion structures (thalamus, striatum, cingulate, precuneus, substantia nigra, globus pallidus) in coordination with the cerebellum to coordinate and proceduralize smoothly functioning action sequences for speech and syntax (Graybiel, ; Hinaut & Dominey, ). Although attriters may have developed these patterns in earlier years, they may need to reactivate these processes to maintain fluency. These two systems for resonant consolidation allow the language learner to avoid the potential problem of catastrophic interference. Although the L and L may share some phonological and semantic resources, the connections between linguistic elements are structured to maintain a separate pattern of connectivity, even if they reside in neighbouring areas of the cortex. The role of resonance in language learning helps us understand Puzzle # and the problems faced by international adoptees. During their first years of language learning, all children work intensely on the learning of systems for audition and motor planning. These systems rely on statistical learning (Thiessen & Erickson, ), rather than resonance, for consolidation. These are the systems that show the clearest entrenchment during L learning. On the other hand, resonance is centrally involved in the learning of the higher linguistic levels of lexicon, morphology, collocation, syntax, clause combination, narrative, dialogue, and discourse. Learning on these levels requires support from resonant connections based on episodic encounters with objects, events, and situations. For the youngest learners, these connections depend heavily on direct support from specific situations, people, and objects. Later, as children start to rely increasingly on inner speech, language forms resonate more and more with one another independent of the environmental context. For example, - and -year-olds can recite only a few short songs, but when they are a few years older they will repeat much longer songs with ease.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
If a child is removed from the original language context before being able to make extensive internal resonant connections, their L will be missing the crucial match to the situational context that it needs for support. The result of this rapid and forced environmental change is a type of linguistic amnesia that impacts systems depending on resonance more than systems depending on statistical learning. Resonance may also be involved in the phenomenon of permastore. Bahrick (b) emphasized the fact that his subjects made minimal use of Spanish after the end of their courses in college. However, because these forms were learned through internal resonant processes initially, rather than through direct contact with environmental experiences, there may have been enough ongoing resonant links between these forms and other L forms to show evidence of retention even decades later.
.. Decoupling Successful second language learning depends on the decoupling of L from its dependence on L (Kroll et al., ). To the degree that a speaker can use resonance to construct each language as a (partially) separately activated neural circuit, the effects of transfer can be minimized. To achieve decoupling, an L learner must combine inhibition of L with resonant activation of L. In addition to the underlying cognitive processing involved in resonance, learners and maintainers may need to establish social and attitudinal methods for decoupling their two languages. If use of L can be localized to contexts that require minimal competition with L, there can be greater decoupling and maintenance of the two systems. At this point, the second language learner is functioning in a more truly bilingual fashion. The fact that lexical semantics and idiomatic expressions are particularly subject to intrusion from L can be related to the fact that L learning relies so heavily on L resources in this area. In effect, the two languages derive their meanings from similar experiences in the world. However, even here, careful use of resonance and decoupling can pull apart forms based on fine-grained distinctions. Phonological processes must also involve shared resources between L and L. However, in this case, L has experienced earlier consolidation through the fronto-striatal proceduralization system and fine-tuning involves the development of recombinations of these gestures to support L.
. C
.................................................................................................................................. Language is a dynamic, emergent system (Van Geert & Verspoor, ), grounded on complex inputs from biology and society. As such, we should not be surprised that theories of language acquisition and loss must refer to a wide range of processes to understand language patterning across individuals, groups, language levels, and language change. To solve the four puzzles involved in language attrition, we need to refer to various combinations of these processes. The conclusions we reach here are very close to those developed in Steinkrauss & Schmid ().
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Entrenchment is certainly involved in the initial consolidation of L. However, as we now know, entrenchment is not irreversible. If life events force a person to abandon their L, they can avoid use of that language, emphasize decoupling and resonance in L, and actively suppress internal use of L. This type of abandonment of L is most easily achieved by young learners of L. For them, higher linguistic levels are particularly vulnerable to the loss of specific contextual supports for the L world. Speakers who do not seek to abandon their L will still need to deal with the pressures of transfer from an increasingly robust L. The exact nature of these transfer phenomena depends on the strength and accessibility of the L pattern, as well as the nature of the match between L and L. Because learners can make use of resonant connections to decouple L from L, they can avoid the problem of catastrophic interference. In effect, each language becomes a separate functional circuit, while still sharing common resources at the level of articulatory output and semantic reference. Because it is difficult to fully decouple languages at these levels, they remain susceptible to at least some transfer from L or L–L merger. By comparing the processes determining the shape of language attrition with those involved in L and L acquisition, as well as language loss in aphasia and dementia, we can gain an increasingly deep understanding of the emergent structure of human language.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
......................................................................................................................
......................................................................................................................
. , -, ´
. I
.................................................................................................................................. A bilingual acquisition phenomena, attrition, defined as the loss of access to linguistic features/structures previously acquired (Schmid, a), is one that forces theoretical linguists to reconsider the nature of the acquisition of atomic units of language and the way in which they are assembled and accessed throughout a bilingual individual’s lifespan (Paradis, ). In this chapter we extend to attrition contexts a particular attempt to understand the processes that may lead to syntactic restructuring previously proposed to account for second language acquisition and for heritage language development: the feature reassembly hypothesis (hereafter, FRH; Lardiere, , , , ). The FRH owes its origins to the initial theoretical developments in the Minimalist Program (MP; Chomsky, et seq.), where ‘a linguistic expression (π, λ) of L satisfies output conditions at the PF and LF interfaces’ (Chomsky, , p. ).1 In this framework, formal features are seen as indices on lexical items and larger syntactic objects that allow generated structures to be interpreted at external interfaces (i.e., PF and LF). Throughout the course of a derivational process, the operation known as feature checking takes place, which enables the legibility of linguistic expressions. Under the assumption that formal features and feature checking are universal elements of the CHL,2 a core
1
L = Language; PF = phonological form; and LF = logical form. CHL is the computational procedure for human language (Chomsky, ). This is roughly compatible with Hauser, Chomsky, & Fitch’s () Narrow Faculty of Language (FLN). 2
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
notion of the underlying linguistic competence of particular grammars (I-language) consists of knowing which features are assigned to which units and when and how they are interpreted and valued by external interfaces (commonly by means of the operation Spell-Out). According to these initial Minimalist assumptions, first language (L) acquisition proceeds through stages of acquiring the necessary functional projections, the formal features associated with lexical items and larger syntactic objects, and the additional steps of knowing when these features are to be valued with appropriate morphophonological reflexes at the PF-interface. The notion that the same processes take place in second language (L) acquisition has led to the postulation of the FRH. Seminal proposals by Lardiere (, , , ) and colleagues (see the contributions in Liceras, Zobl, & Goodluck, , for a general overview) laid the foundation for research based on the proposal of a feature reassembly operation, where the feature configuration of the L is adjusted to conform to the mapping of syntactic features onto morphological forms found in the L grammar. This research programme has been popular and successful in the generative L literature to date (Domínguez et al., ; Guijarro-Fuentes, ; Cho, ; Gil & Marsden, ; Spinner, ). As nuances have entered Minimalist theorizing since its inception, the FRH has also been updated to accommodate to these changes, with most contemporary studies adhering to the model of grammar commonly assumed in Distributed Morphology (DM; Embick & Noyer, ). Additionally, following initial proposals by Slabakova ( and Chapter , this volume), the question of whether or not the notion of feature is a more effective and appropriate atomic unit of linguistic competence than traditional parameters has also been raised. If language attrition is understood on a basic level to result in the loss of structural mappings, simpler structures—including stylistic variation—and paradigmatic levelling (Schmid, a), the FRH can also be extended to model these situations. Putnam & Sánchez () advance the proposal that the FRH can account for instances of incomplete acquisition3 and language attrition by appealing to the lack of activation (Levy et al., ; Linck et al., ; Seton & Schmid, ) of the recessive L grammar system over the lifespan of individuals. This approach differs from the notion of incomplete acquisition in that the latter assumes that at some point the process of acquisition is interrupted due to diminished input (Montrul, a). Incomplete acquisition has been argued to account for differences in ultimate attainment between heritage bilinguals living in a language contact situation and monolingual speakers with access to formal instruction (Montrul, ). More recently, the notion of incomplete mastery originally proposed by Berman () to characterize the variable nature of first language acquisition has been put forth as an explanation for the outcomes observed in heritage bilinguals (Montrul, b). Following Putnam & Sánchez (), we assume an alternative account to this view and propose to characterize variability across heritage speakers (HS), in particular those who experience attrition, as differences in activation for comprehension and production purposes rather than as incomplete acquisition.
3
For an alternative conceptualization, see Kupisch & Rothman ().
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
, -, ´
A study that shows how this alternative view analyses differences in production and comprehension is Perez-Cortes (). This study found evidence of restructuring of subjunctive features and their corresponding morphological forms among HS of Spanish living in the US modulated by differential levels of activation of Spanish. The acceptability of indicative and infinitive forms in contexts where subjunctive was required, for instance, was more frequent among HS who, despite exhibiting high levels of proficiency, displayed lower levels of activation of Spanish. These effects extended not only to acceptability, but also to production and interpretation tasks, as seen in the following examples obtained from an Acceptability Judgement task () and a Sentence Completion task (): () El gato tiene hambre y quiere que los ratones *salen del agujero The cat has hunger and wants that the mice leave[ppIND] the hole ‘The cat is hungry and wants the mice to leave the hole’ (Perez-Cortes, , p. ) () Bob Esponja quiere que Patrick *se viajar a Hawaii antes Bob Esponja wants that Patrick CL to travel[INF] to Hawaii before ‘Bob Esponja wants Patrick to travel to Hawaii before him’ (Perez-Cortes, , p. ) In these examples, the use of subjunctive forms [salgan, viajen] should have been obligatory due to lexical selection and disjoint reference. However, participants with lower levels of Spanish activation accepted—and produced—indicative () and infinitival () forms in these contexts, suggesting difficulties in remapping the modal feature selected by the matrix verb onto the corresponding morphological form (subjunctive). The emergence of indicative and infinitive as alternatives to the subjunctive was also observed in heritage bilinguals with low proficiency in the minority language, and can be explained by adopting a feature reassembly approach. In order to successfully master obligatory mood selection in Spanish, HS who are dominant in English are required to remap the interpretable feature [M], typically hosted by (for)-to infinitival complements (‘Spongebob wants Patrick to travel to Hawaii before him’), onto an inflected subjunctive form (Iverson et al., ; Kempchinsky, ). The results obtained in Perez-Cortes’ () study indicate that the process of feature reassembly leads to a wide variety of systematic outcomes largely modulated by bilinguals’ frequency of language activation and proficiency. HS who experienced decreased activation of Spanish and/or low levels of proficiency had more difficulties remapping the interpretable modal feature onto an inflected form, accepting and producing infinitival forms such as the one illustrated in () over their indicative counterparts (). As their use of Spanish for comprehension and production purposes increased, so did their ability to interpret, accept, and produce subjunctive as a marker of directive force and disjoint reference. Additionally, and as initially proposed by Putnam & Sánchez (), this fluctuating process of restructuring affected HS’s productive and receptive abilities in a very specific manner. In bilinguals with higher levels of Spanish activation and proficiency, morphological optionality was limited to production where they exhibited difficulties in retrieving subjunctive morphology. Despite these difficulties, they were able to interpret the absence of subjunctive morphology as ungrammatical, and its presence as a marker of disjoint reference. Those with lower levels of activation and proficiency, however, presented additional complications associating subjunctive to disjoint reference contexts, suggesting that the effects of decreased activation had also affected their grammatical representations.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Although the empirical focus of the FRH in language attrition contexts generally involves heritage grammars (Domínguez, ; Putnam & Sánchez, ; Cuza & Pérez-Tattam, ), there is nothing a priori that would prevent its application to processes involving the second language acquisition of a majority language by heritage speakers, or to multilingual populations. When analysing the linguistic development of HS, the FRH offers one possible way to address what Bayram et al. (Chapter , this volume) call the Heritage Bilingual Paradox (HBP): ‘how is it possible that native child naturalistic acquisition, albeit in heritage language contexts, can result in large disparities in grammatical competence between HS and monolinguals as well as across subsets of heritage speakers?’ Here we explore and build upon the core inner-workings of the FRH in the context of the development of heritage grammars across the lifespan as initially proposed by Putnam & Sánchez (). Crucially, we wish to emphasize that we shift the common narrative of incomplete acquisition in HS to the alternative of restructuring as has been posited in attrition studies (Polinsky, ; Yager et al., ; Yager, ) under the FRH. In what follows we focus on how the FRH can account for heritage grammars and on the strengths and weaknesses of this model in the face of evidence adduced from additional shifts and debates in the theoretical literature as well as advances on bilingual grammars from the perspective of cognitive neuroscience.
. FRH
.................................................................................................................................. Putnam & Sánchez () propose a model of heritage grammar development that incorporates activation of features for comprehension and production (see Putnam & Schwarz, ; Yager et al., ; Cuza & Pérez-Tattam, ; Perez-Cortes, ; and Yager, , for extensions of this proposal). In this model, the mapping of syntactic features onto morphological forms may be differentially affected for comprehension versus production purposes. In top-down processes of integration for comprehension, some heritage speakers may be unable to access a representation that includes a mapping of syntactic features and morphology that resembles that of a monolingual or dominant speaker’s grammar. Others may be able to access such representations in comprehension tasks but the mapping would not be reflected in production. An example of such difficulties is provided in Perez-Cortes () in an utterance from an HS’s grammar with low proficiency in Spanish in which the modal feature is remapped to a periphrastic expression that can potentially be used in other structures to convey directive force instead of resorting to the subjunctive to do so: () El rey quiere que los príncipes *tien-en que dec-ir las The king wants COMP the princes have- that say- the noticias importantes. news important ‘The king wants the princes to have to say the important news.’ Those with the highest levels of proficiency may access such mappings in comprehension and also be able to show evidence of them in their production (Perez-Cortes, ). In this respect, HS show parallel developmental paths to those found in other groups of
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
, -, ´
attrited speakers who may similarly experience differential effects in comprehension and production (Seton & Schmid, ). This claim finds additional support in research by Gollan et al. (), which indicates that the activation of lexical items in comprehension shows different frequency effects than those for production. If, as suggested by Gollan et al. (), comprehension is primarily constrained by pragmatic and semantic factors, frequency effects related to the activation of lexical items and their related meanings will play an important role in this domain. Production, on the other hand, involves the integration of conceptual, syntactic, and phonological features (Bock & Levelt, ) and some level of automatization in accessing language units resulting from such integration. Given this complexity, production among heritage speakers is expected to result in more frequencysensitive stages. In this respect, comprehension contrasts with production, thus presenting significant challenges to bi/multilinguals leading to additional language dominance effects in this domain. What is more, if, following scholars such as Brysbaert & Duyck (), who claim that the competition of lexical items in the bilingual mind involves both languageinternal as well as cross-linguistic competition, information associated with lexical items exists in conflict with one another. We address this point in the final section (see also Kroll & Stewart’s () Revised Hierarchical Model). Putnam & Sánchez () present the sketch of a model of how language attrition would take place, making use of Minimalist tools. In their account, they assume that linguistic knowledge consists of sets of abstract features (Chomsky, , et seq.): (i) functional features (FFs); (ii) phonological form features (PF-features); and (iii) semantic features: [T]he acquisition of heritage language grammars involves a change in the frequency of the amount of input of the L, with extensive, if not exclusive, L input in the earliest stages of acquisition slowly shifting to L dominance at some point prior to ‘ultimate attainment’ of the L grammar. This process is thus characterized by changes in the frequency of exposure (i.e., from more to less) or by a limited exposure to lexical items that instantiate some values for FFs in language A (i.e., the L/heritage language). Increased exposure to lexical items in language B (i.e., the socially-dominant L) and lower frequency of exposure to lexical items in language A may result in lower levels of activation of certain FF values in language A. In turn, lower activation may result in their gradual replacement by the functional values in language B but in association with PF and semantic features of lexical items in language A. (Putnam & Sánchez, , p. )4
On this view, there is no ‘incomplete acquisition’ (pace Montrul, ) of the L/heritage grammar that takes place due to insufficient input, as the acquisition process is never truly interrupted. Rather, according to this proposal, heritage language acquisition and further developments (i.e., what is commonly referred to as language attrition) of the L/heritage grammar over the course of the lifespan are the result of the remapping of featural content (FF) to different semantic and PF-features. Abstracting away from age-related effects that have been appealed to in order to create a dichotomy between incomplete acquisition and language attrition, Putnam & Sánchez’s model interprets these effects as largely involving
4 This model can also subsume cross-linguistic effects (i.e., transfer) that are both facilitative and nonfacilitative in nature (see Jarvis & Pavlenko, ; Jarvis, ).
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
the same phenomenon (i.e., remapping and restructuring of linguistic information), thus viewing both of these processes as epiphenomenal. To account for the various degrees of interference and restructuring, Putnam & Sánchez establish different stages and their (potentially lasting) impact on the competence grammar of HS (adopted from Putnam & Sánchez, , pp. –): Stage : Transfer or re-assembly of some FFs from the L grammar to L PF and semantic features which may coincide with the activation of L lexical items on a more frequent basis from the standpoint of linguistic production; Stage : Transfer or re-assembly of massive sets of FFs from the L to the L PF and semantic features, while concurrently showing significantly higher rates of activation of L lexical items than L lexical items for production purposes (i.e., this group of bilinguals might code-switch more than those in the previous situation); Stage : Exhibit difficulties in activating PF and semantic features (as well as other FFs) in the L for production purposes but are able to do so for comprehension—especially in high frequency lexical items; and Stage : Have difficulties activating PF features and semantic features (as well as other FFs) in the L for both production and comprehension purposes.5 Attrition phenomena in adult learners represent an ideal area to test the descriptive adequacy of this model and its stages, because unlike young heritage speakers, adult attriters’ restructuring processes usually take place after having established specific feature values in their L and their mappings onto PF and semantic meaning. Furthermore, their level of activation of bundles of features is assumed to have been high prior to L acquisition. This can allow us to study those feature values that diverge in the L and the L and address the extent to which activation has an impact on the valuation of features.
. F
.................................................................................................................................. As we have demonstrated here, the FRH continues to serve as a viable option to capture language attrition and language restructuring generalizations (e.g., Lardiere, , ; Slabakova, ). Although the FRH was born under the auspices of the initial proposal of the MP, recent developments in this research programme call for a radical reduction of the number of formal features as well as the outright removal of the latter from the CHL (see, e.g., Boeckx, ; Eguren et al., ). Recent models that adopt the position that morpho-syntax is due to the realization of cues or features at the PF-interface (see, e.g., Grimstad et al., ; Alexiadou, Lohndal, et al., ; and Riksem et al., to appear, for exoskeletal approaches to language mixing) are also compatible with the core tenets of the FRH we outline above. Future applications of the FRH in attrition and restructuring 5
Although this original proposal focused exclusively on morpho-syntactic information and its link with semantic and pragmatic meaning, there is nothing in principle that would prevent an extension to the phonological domain.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
, -, ´
contexts can also be extended to proposals that adopt a parallel architecture of cognition underlying the language faculty (Jackendoff, , ; Mayer & Sánchez, ; Putnam, ). Under such assumptions, attrition and restructuring will be due to either the loss and/or reduction in connections among sub-levels or the introduction of new mappings. The possibility of uniting some version of the FRH with a parallel architecture is promising for a number of reasons: First, since the notion of ‘interface’ has been difficult to define in this research programme (Sorace, ; Montrul, a), a parallel architecture that focuses on the word as an interface rule would allow more fine-grained distinctions in grammatical descriptions. Second, this flexibility in the description of grammatical units is compatible both with the micro-cue approach (Lightfoot, ; Westergaard, ) and MacWhinney’s (Chapter , this volume) competition-based model. Third, the parallel architecture allows for a more detailed documentation of the two languages in conflict and the resulting grammatical representations of these duels. Evidence from cognitive neuroscience abounds showing that the conflict of competing forms is the norm rather than the exception (Kroll, Bobb, & Wodniecka, ; Kroll, Bobb, Misra, & Guo, ; Melinger et al., ). These ‘competing forms’ can be construed as correlates to features, which is the basic strategy adopted by Amaral & Roeper () in their proposal concerning the interaction of simultaneously conflicting linguistic information. Adopting this notion of conflicting grammatical representations interacting with executive and inhibitory psychological function results in a reclassification of language attrition proper, where attrition is perhaps more appropriately understood as the strengthening of neural activities and connections of the L/Ln grammars rather ‘than the actual degradation of the connections which represent the L’ (Seton & Schmid, , p. ; see also, e.g., Hulsen, ; Pallier et al., ; Schmid, ; Seidenberg, ). Additional evidence for this hypothesis can be adduced from the observation that receptive bilinguals (i.e., those who can comprehend a heritage grammar but encounter significant difficulties in producing forms in the same grammar or cannot produce them) maintain some degree of competence with respect to inflection morphology (see, e.g., Sherkina-Lieber, ). The application of a parallel model opens the door to gradient and probabilistic approaches to language acquisition, change, and attrition (Smolensky, ; Pater, ; Smolensky et al., ; Bousquette et al., ; Goldrick et al., a,b; Putnam & Klosinski, ). Irrespective of the formalism one chooses to model the restructuring of linguistic information in bilingual grammars, the FRH continues to serve as a viable option. As a scientific paradigm, the FRH is parasitic upon other core theoretical assumptions of a particular framework. Putnam & Sánchez’s () proposal to link the concept of formal feature with the Activation Threshold (Paradis, ; Ullman, b, ; see also Köpke & Keijzer, Chapter , this volume) has led to nascent research that also employs this version of the FRH. This research programme has yet to be fully tested in attrition studies; however, both the notion of feature reassembly and the stages proposed can be empirically investigated in the growing area of attrition research.
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
......................................................................................................................
......................................................................................................................
. I
.................................................................................................................................. T Interface Hypothesis (IH) was originally proposed by Sorace & Filiaci () to explain the non-convergence and optionality revealed in very advanced adult second language (L) learners in the comprehension and production of certain structures. The original proposal suggested that those language structures that are sensitive to conditions involving an interface between syntax and other cognitive domains, such as pragmatics, are more difficult to be acquired completely than structures that do not involve such an interface. The IH has evolved over time from assuming a dichotomy between ‘narrow’ syntax and ‘interface’ structures to a more fine-grained differentiation among types of interface conditions (‘internal’ versus ‘external’; Sorace, ; Tsimpli & Sorace, ), opening the way for research on the multiple cognitive mechanisms involved in acquiring and processing structures sensitive to different interface conditions. Specifically, ‘external’ interface conditions, such as those operating at the syntax–pragmatics interface, involve the integration of ever-changing contextual and pragmatic information while processing language; ‘internal’ interface conditions, on the other hand, involve the integration of semantic or morphological information, which is not sensitive to external contextual changes and relies on well-rehearsed processing mechanisms. The IH was later extended to early bilingual acquisition and individual first language (L) attrition, suggesting that interface structures are less likely to be acquired completely for the former and more likely to undergo attrition for the latter. In relation to L attrition, the current hypothesis further proposes that individual L attrition affects only the ability to
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
process interface structures but not knowledge representations themselves (Sorace, , ). It is important to emphasize that this prediction relates to first generation individual attrition in speakers who have acquired the L completely before the onset of attrition and not to second generation attrition in heritage speakers, for whom the acquisition of the L may be incomplete or divergent, depending on the quantity and quality of input received. The IH has been supported by a large body of research exploring cross-linguistic influence effects for different interface structures in diverse bilingual groups. For example, many studies have been conducted on the acquisition of null versus overt pronouns in bilingual children (e.g., Paradis & Navarro, ; Serratrice et al., ; Argyri & Sorace, ; Sorace et al., ; Serratrice et al., ), advanced adult L learners (e.g., Belletti et al., ; Lozano, ; Rothman, ) and L attriters (e.g., Chamorro, Sorace, & Sturt, ; Montrul, a; Tsimpli et al., ). The present chapter will focus on first generation individual attrition from the point of view of the IH, and particularly on recent research that provides evidence of the selectivity and reversibility of individual L attrition.
. T I H L
.................................................................................................................................. In line with findings in bilingual L acquisition and adult L acquisition, recent research on L attrition also supports the IH, revealing that the structures at the syntax–pragmatics interface are the most prone to undergo attrition, causing ‘emerging optionality’ in the attrited speakers. Sorace (a) tested anaphora resolution in Italian near-native speakers of English under L attrition and found that they overgeneralized overt pronouns in Italian to contexts in which native speakers of Italian would use a null pronoun due to the influence from English. Sorace further established a connection between L acquisition and L attrition given that both English near-native speakers of Italian and Italian L attriters overextended the use of overt pronouns in Italian as a result of the influence from English. Tsimpli et al. () also reported attrition effects in a group of Greek and Italian nearnative speakers of English in relation to subject pronouns. They tested the production and interpretation of null versus overt pronouns in Greek and Italian using a picture verification task to elicit the participants’ preference for the subject or the object antecedent with each pronoun. During the experiment, participants were presented with three pictures and an ambiguous sentence like the ones in (), and they were asked to choose the picture or pictures that correctly matched the meaning of the sentence. () a. Quando lei attraversa la strada, l’anziana signora saluta la ragazza. ‘While she crosses the street, the old woman greets the girl.’ b. Quando pro attraversa la strada, l’anziana signora saluta la ragazza. ‘While (she) crosses the street, the old woman greets the girl.’ c. L’anziana signora saluta la ragazza quando lei attraversa la strada ‘The old woman greets the girl when she crosses the street.’ d. L’anziana signora saluta la ragazza quando pro attraversa la strada ‘The old woman greets the girl when (she) crosses the street.’
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
Consistent with Sorace’s (a) results, attrition effects were found for Italian attriters on the interpretation of the overt pronoun, for which they showed more indeterminacy in their antecedent preferences than the control group of Italian monolinguals. In contrast, no attrition was revealed with the null pronoun, for which both groups of attriters preferred the subject as its antecedent, in line with both control groups of Greek and Italian monolinguals. Interestingly, the monolingual control group in this study did not perform categorically, showing a tendency towards more variable interpretation of the overt pronoun. Tsimpli et al. () also investigated the interpretation of preverbal versus post-verbal subjects in a group of native Greek L speakers of English. In this study, participants were also presented with three pictures and a pair of sentences like the ones in (), in order to compare the speakers’ interpretation of preverbal subjects, which are usually understood as ‘old’ information (i.e., topic), as in (a), versus post-verbal subjects, which can be ambiguously understood as ‘old’ or ‘new’ information, as in (b). () a. I gitonisa mu ston trito orofo apektise dhidhima. Xtes vradhi my neighbour on the third floor had twins last night ena moro ekleje. one baby cry-. ‘My neighbour on the third floor had twins. One of the twins was crying last night.’ b. I gitonisa mu ston trito orofo apektise dhidhima. Xtes vradhi my neighbour on the third floor had twins last night ekleje ena moro cry-. one baby. ‘My neighbour on the third floor had twins. A baby (one of the twins or some other baby) was crying last night.’ The results from this experiment also revealed attrition effects, given that Greek attriters showed significantly more indeterminacy when interpreting preverbal subjects in comparison with the control group of Greek monolinguals. Based on the results obtained from these two structures, the authors concluded that attrition affects properties at the syntax–pragmatics interface but not syntactic features obeying conditions internal to the grammar. Gürel (a) also investigated the L attrition of null and overt pronouns in Turkish L learners of English and found language attrition to be selective. Turkish has two overt pronouns, o ‘s/he’ and kendisi ‘self ’, and a null pronoun. Whereas all three pronouns can occur in subject position, as in (), and in object position, as in (), only the overt pronoun kendisi and the null pronoun, but not the overt pronoun o, can refer to the subject of the main clause when they occur in object position, as () illustrates. () O/kendi-si/pro Londra’ya git-ti (S)he/self-/pro London- go- ‘(S)he went to London.’ () Buraki o-nu*i/j/kendi-si-nii/j/proi/j begˇen-iyor Burak (s)he-/self--/pro like- ‘Buraki likes him*i/j/selfi/j/proi/j’
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
This study involved three tasks (a written interpretation task, a truth-value judgement task, and a picture identification listening task) that tested the participants’ interpretation of sentences like () and (). Gürel (a) reported that the interpretation of the overt pronoun o in Turkish was influenced by English because attriters appeared to treat this Turkish overt pronoun as if it was the English overt pronoun (i.e., co-referential with the subject antecedent). In contrast, the interpretation of the null pronoun and the overt pronoun kendisi did not show attrition effects. Gürel interpreted these results under Paradis’ () Activation Threshold Hypothesis (ATH) (see also Köpke & Keijzer, Chapter , this volume). The ATH establishes a correlation between the frequency of use of a language element and its availability (or activation) to the speaker. In particular, it proposes that when an item is not used, the threshold of activation would rise, and when it is used, the threshold of activation would be low. Therefore, a linguistic item that has not been frequently used would have a high activation threshold and it would be difficult to activate, which would lead to the attrition of the item. This suggests that different language elements, depending of their frequency of use, would have different thresholds of activation, so that some would be more likely than others to undergo attrition. More specifically, the ATH predicts that L attrition will occur when an element in the L with a high activation threshold (i.e., disused) has a corresponding ‘competing’ element in the L with a lower activation threshold (i.e., used more frequently). Gürel’s results are predictable under the ATH because it is the Turkish overt pronoun o, which is in competition with the English overt pronoun, the one that shows attrition due to its disuse in Turkish and frequent use in English, but the overt pronoun kendisi or the null pronoun in Turkish, which do not have a competing item in the L, do not show attrition effects. The ATH is also partly compatible with the most recent version of the IH, in that it provides a processing explanation, which is, however, only focused on L–L cross-linguistic effects. Otheguy et al. () investigated attrition effects in the use of Spanish pronouns by different Spanish-speaking communities in the United States. They analysed the use of pronouns using data from a corpus of , verbs extracted from interviews conducted with the six largest Spanish-speaking communities living in New York City and originated in six different Latin American countries. In order to analyse the use of overt pronouns, speakers were divided in terms of their dialect regions, ‘Caribbean’ (newcomers from the Dominican Republic, Puerto Rico, and Cuba) versus ‘Mainlanders’ (newcomers from Ecuador, Colombia, and Mexico), and in terms of their generation, ‘newcomers’ (those who arrived in New York after the age of and had lived there for a maximum of five years) versus ‘born and/or raised in New York’ (those who were born in New York or arrived before the age of ). The results from this study revealed that overt pronouns were used more frequently by the Caribbean speakers than by the Mainlanders. More importantly, the ‘born and/or raised in New York’ group showed a significantly higher rate of overt pronouns than the newcomers, which revealed the influence from English in the use of overt pronouns in the Spanish of both Caribbean and Mainlanders speakers living in New York City. L attrition effects have been found in languages with different inventories of anaphoric forms. Wilson () and Wilson et al. () used the visual world eye-tracking methodology to investigate the online processing of German anaphora with demonstratives and pronouns by English-speaking L learners of German and German-speaking L learners of English under L attrition. Participants were presented with a set of pictures while they
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
heard a pair of sentences like the ones in () and were asked to answer a yes/no question that revealed their antecedent preferences for the pronouns. Similar to the distribution of null and overt pronominals in null-subject languages, personal pronouns in German (er, sie, es) refer to the subject antecedent whereas demonstrative pronouns (der, die, das) refer to the object antecedent. () a. Der Kellner erkennt den Detektiv als das Bier umgekippt the- waiter recognizes the- detective as the beer tipped-over wird. Er ist offensichtlich sehr fleißig. is he- is clearly very hard-working ‘The waiter recognizes the detective as the beer is tipped over. He is clearly very hard working.’ b. Der Kellner erkennt den Detektiv als das Bier umgekippt the- waiter recognizes the- detective as the beer tipped-over wird. Der ist offensichtlich sehr fleißig. is he- is clearly very hard-working ‘The waiter recognizes the detective as the beer is tipped over. He is clearly very hard working.’ c. Den Kellner erkennt der Detektiv als das Bier umgekippt the- waiter recognizes the- detective as the beer tipped-over wird. Er ist offensichtlich sehr fleißig. is he- is clearly very hard-working ‘The waiter is recognized by the detective as the beer is tipped over. He is clearly very hard working.’ d. Den Kellner erkennt der Detektiv als das Bier umgekippt the- waiter recognizes the- detective as the beer tipped-over wird. Der ist offensichtlich sehr fleißig. is he- is clearly very hard-working ‘The waiter is recognized by the detective as the beer is tipped over. He is clearly very hard working.’ The results from these studies revealed that while L learners performed similarly to German native speakers with pronouns, they showed indeterminacy with demonstratives, revealing no clear preference for the object as their antecedent. Similarly, attriters showed more attrition effects with demonstratives than with pronouns in comparison to the control group of monolinguals, also revealing no clear preference for a specific antecedent. The results also showed that the degree of attrition effects depended on the attriters’ length of residence in the UK. The literature reviewed in this section supports the claims of the IH that structures at the syntax–pragmatics interface are likely to undergo attrition. There are also studies, mostly on L acquisition, exploring interface structures that have revealed mixed results with regards to the IH, with some findings being consistent and some being inconsistent with the IH (e.g., Slabakova & Montrul, ; Rothman & Iverson, ; Judy & Rothman, ; see Leal Méndez et al., , on heritage speakers). Crucially, many of these studies typically use offline measures involving metalinguistic tasks, which are not the most suitable to test the predictions of the IH concerning the processing of structures at the syntax–pragmatics
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
interface, rather than the mental representation of these structures (see Section . below). While a detailed review of these studies is beyond the scope of this chapter, they are a reminder that more research is needed—on a wider range of structures and based on multiple experimental methods—to precisely identify the boundaries on the applicability of the IH.
. S L
.................................................................................................................................. The previous section presented research on L attrition that supports the predictions of the IH, as they reveal attrition effects with structures at the syntax–pragmatics interface. This selectivity of L attrition as well as Sorace’s (, ) proposal that individual L attrition affects only the ability to process interface structures but not knowledge representations was tested in two recent studies: Chamorro, Sorace, & Sturt (), which investigated a structure involving an external interface, the syntax–pragmatics interface (i.e., subject pronouns), and Chamorro, Sturt, & Sorace (), which investigated a structure involving an internal interface, the syntax–semantics interface (i.e., the personal preposition). Chamorro, Sorace, & Sturt () investigated whether pronominal subjects, a structure at the syntax–pragmatics interface, would undergo attrition in L Spanish under prolonged exposure to the L English and, if so, whether these effects occur at the processing or representational level. Therefore, the interpretation and processing of overt versus null subject pronouns in Spanish was investigated using an offline naturalness judgement task and an online eye-tracking-while-reading task, where participants were presented with anaphora in which the antecedent preferences were predicted using Carminati’s () Position of Antecedent Strategy (PAS). Carminati proposed the PAS for Italian intra-sentential anaphora and it postulates that null pronouns are generally assigned to the subject antecedent, as in (a), whereas overt pronouns are generally assigned to the object antecedent, as in (b). The PAS was shown by Alonso-Ovalle et al. () to also apply to Iberian Spanish. () a. Quando Marioi ha telefonato a Giovannij, proi aveva appena finito di mangiare. ‘When Mario has telephoned Giovanni, (he) had just finished eating.’ b. Quando Marioi ha telefonato a Giovannij, luij aveva appena finito di mangiare. ‘When Mario has telephoned Giovanni, he had just finished eating.’ Based on Carminati’s PAS, participants in Chamorro, Sorace, & Sturt () were presented with intra-sentential semantically-neutral forward anaphora like the ones in (), where the grammatical number of the antecedents was manipulated such that the pronoun could only refer to either the subject or the object antecedent. This led to sentences where the pronoun agreed in number with the pragmatically infelicitous antecedent as predicted by the PAS, as in (a) and (d), and sentences where the pronoun agreed in number with the pragmatically felicitous antecedent, as in (b) and (c).
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
() a. La madre saludó a las chicas cuando ella cruzaba una calle the mother greeted- to the girls when she crossed- a street con mucho tráfico. with much traffic ‘The mother greeted the girls when she crossed a street with a lot of traffic.’ b. Las madres saludaron a la chica cuando ella cruzaba una calle the mothers greeted- to the girl when she crossed- a street con mucho tráfico. with much traffic ‘The mothers greeted the girl when she crossed a street with a lot of traffic.’ c. La madre saludó a las chicas cuando pro cruzaba una calle the mother greeted- to the girls when pro crossed- a street con mucho tráfico. with much traffic ‘The mother greeted the girls when she crossed a street with a lot of traffic.’ d. Las madres saludaron a la chica cuando pro cruzaba una calle the mothers greeted- to the girl when pro crossed- a street con mucho tráfico. with much traffic ‘The mothers greeted the girl when she crossed a street with a lot of traffic.’ On the other hand, Chamorro, Sturt, & Sorace () tested whether L attrition also affects structures involving internal interfaces or only those structures that involve external interfaces such as subject pronouns. In order to do so, they investigated the interpretation and processing of a syntax–semantics interface structure, the Spanish personal preposition a or differential object marking (DOM), to explore whether structures sensitive to semantic conditions also undergo L attrition. This structure differs from pronominal subjects in that its use does not depend on context, but is conditioned by semantic factors such as the animacy and/or specificity of the direct object. In line with Chamorro, Sorace, & Sturt (), this study also investigated whether any attrition effects revealed with this structure would be related to inconsistent or inefficient processing in real time or to a change in the attriters’ L knowledge representations (i.e., in their L grammatical competence) by implementing the same offline and eye-tracking measures as the pronoun study. The DOM, which occurs in Spanish but not in English, establishes that some direct objects must be introduced by a dative preposition, a ‘to’ in the case of Spanish. The presence or absence of this preposition depends on the type of direct object. Generally speaking, in Spanish, a direct object must be marked with the dative preposition if it is animate and specific, as () below shows, but inanimate direct objects, independently of the specificity, must not be preceded by the preposition, as in (). () a. María vio al1 niño esta mañana. María see- to-the kid this morning ‘María saw the kid this morning.’ 1
Note that al is the contraction of the preposition a and the masculine singular definite article el.
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
b. *María vio el niño esta mañana María see- the kid this morning ‘María saw the kid this morning.’
() a. María vio una película/la película esta mañana. María watch- a movie/ the movie this morning ‘María watched a movie/the movie this morning.’ b. *María vio a una película/la película esta mañana. María watch- to a movie/ the movie this morning ‘María watched a movie/the movie this morning.’ Based on this, participants in Chamorro, Sturt, & Sorace () were presented with sentences like the ones in (). Each item consisted of a simple sentence which contained a specific direct object, either animate or inanimate, which led to sentences where the animate object was correctly introduced by the preposition, as in (b), or ungrammatically lacked the preposition, as in (a), and sentences where the inanimate object correctly lacked the preposition, as in (c), or was ungrammatically introduced by it, as in (d). () a. *Juan defendió el conductor que fue despedido. Juan defend- the driver that was fired ‘Juan defended the driver that was fired.’ b. Juan defendió al conductor que fue despedido. Juan defend- to-the driver that was fired. ‘Juan defended the driver that was fired.’ c. Juan defendió el argumento de forma efusiva Juan defend- the argument in way effusive ‘Juan defended the argument in an effusive way.’ d. *Juan defendió al argumento de forma efusiva. Juan defend- to-the argument in way effusive ‘Juan defended to the argument in an effusive way.’ Unlike the experimental items in Chamorro, Sorace, & Sturt (), in these experimental items, whether the personal preposition should be used or not is completely clear and does not require participants to consult the discourse context: if there is an animate direct object, then the preposition must be used, and if the direct object is inanimate, the preposition must not be used, regardless of any context. The same group of Spanish native speakers under L attrition was tested in both studies to investigate two questions: () whether the IH would correctly predict attrition effects with the structure involving an external interface (i.e., pronominal subjects) but not with the structure involving an internal interface (i.e., the DOM); and () whether any attrition effects revealed occurred at the processing level (i.e., in the online eyetracking-while-reading task) but not at the representational level (i.e., in the offline naturalness judgement task). The group of attriters tested in both studies consisted of twenty-four Spanish native speakers from Spain who had been residing in the UK for a minimum of five years and were near-native speakers of English. This group was compared to a control group of monolinguals, which consisted of twenty-four Spanish
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
native speakers from Spain who had recently arrived in Edinburgh and had very little knowledge of English. Chamorro, Sorace, & Sturt’s () results on pronominal subjects reveal that attrition occurs with this structure at the syntax–pragmatics interface and that attrition affects online sensitivity rather than offline judgements. The offline judgement data shows no significant differences between the group of attriters and the group of monolinguals, with both groups showing equal sensitivity to the pronoun mismatch (i.e., anaphora containing an overt pronoun when a null pronoun is appropriate or anaphora containing a null pronoun where an overt pronoun is appropriate) when rating anaphora offline. Results from the eye-tracking experiment revealed that, in online reading, the group of monolinguals was reliably more sensitive than the attriters to the pronoun mismatch, as the monolinguals showed a significant mismatch sensitivity with pronominal subjects (i.e., significant Pronoun by Antecedent interaction effects in their reading times in the critical region, where the pronoun occurs, as well as in the post-critical region), whereas attriters did not reveal online sensitivity to the pronoun mismatch (i.e., no significant interaction effects were shown in any of the regions) and performed significantly differently from monolinguals. In contrast, Chamorro, Sturt, & Sorace’s () results on the DOM revealed no attrition effects with this structure at the syntax–semantics interface. The offline results showed an equal sensitivity to DOM violations for both groups, with all participants correctly differentiating the grammatical sentences in which the animate object was preceded by al and those in which the inanimate object was preceded by el from the ungrammatical sentences in which the animate object was preceded by el and those in which the inanimate object was preceded by al. Similarly, the eye-tracking results showed very early sensitivity to DOM violations (i.e., both groups showed significant interaction effects of Animacy by Article in first-pass reading time in the critical region, which is the earliest possible point), and this sensitivity was of an equal magnitude across both groups, with no significant differences revealed between the groups. This suggests that both groups were sensitive to the mismatching conditions when processing the DOM in real time (i.e., when the animate object was incorrectly preceded by el or when the inanimate object was incorrectly preceded by al). Considering the findings from both studies, Chamorro, Sorace, & Sturt () and Chamorro, Sturt, & Sorace () provide further support for the IH and the selectivity of individual L attrition, as it was shown that attrition affects structures at syntax–pragmatics interface, such as pronominal subjects, but that structures requiring the satisfaction of semantic conditions, such as the DOM, do not undergo attrition, either at the processing or the representational level. Crucially, these studies provide supporting evidence for Sorace’s (, ) proposal, as it was revealed that attrition affects online sensitivity with structures at the syntax–pragmatics interface but not knowledge representations.
. I L ?
.................................................................................................................................. In line with the findings discussed in Sections . and . for L attrition, research on L acquisition also reveal that advanced adult L learners show optionality in the production
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
and/or interpretation of anaphoric forms (e.g., Sorace & Filiaci, ; Belletti et al., ; Lozano, ; Rothman, ). Another aspect that these L studies have in common, as well as the ones on L attrition reviewed above, is that they all investigate speakers who have English as the other language. Therefore, one could argue that the optionality revealed with interface structures by these L learners and L attriters may be due to interference from the language that does not have a choice of anaphoric forms (i.e., English) on the language that has such a choice, as some of these studies have concluded. However, anaphoric forms have also been found to cause optionality in bilingual speakers of two typologically similar languages, not only in speakers of two null-subject languages (Bini, ; Lozano, ; Margaza & Bel, ; Mendes & Iribarren, ; de Prada Pérez, , ; Sorace et al., ), but also in speakers of two Germanic languages (Juvonen, ; Ellert, ), bimodal bilinguals (Bel et al., ), and unimodal bilinguals (Chen Pichler & Koulidobrova, ). These results suggest that the optionality revealed in L attrition and L acquisition with interface structures cannot be due only to interference effects from English, but may be also related to the cognitive effort of handling any two languages in real time (Sorace, , ). In addition, the convergences between L attrition and L acquisition suggest that both languages of late bilingual speakers may be affected by cognitive changes due to accommodating an L and that having two plausible anaphoric options may lead to processing difficulties when speakers compute pronoun-antecedent mappings, both in production and comprehension. It has been proposed that these difficulties could be attributed to bilingual speakers’ reduced efficiency when integrating information from different domains in real time and updating the mental discourse model when needed, as a side effect of the need to exercise inhibitory control to avoid interference from the language not in use (see Green, ; Costa et al., ; Green & Abutalebi, , on the role of inhibitory control; and Sorace, , on possible consequences for integration). Therefore, if the effects of attrition do not involve language change at the representational level, but rather affects the cognitive strategies to access and use this knowledge in real time, one may predict that these effects are not irreversible but may be sensitive to the frequency and recency of exposure to the L. The question of the reversibility of individual attrition was investigated by Chamorro, Sorace, & Sturt (), which will be discussed in the following section.
. R L
.................................................................................................................................. Another important issue in individual L attrition is the question of whether attrition effects are permanent or can be reversed. Chamorro, Sorace, & Sturt () also explored this question by investigating whether attrition effects with pronominal subjects can decrease or disappear with recent re-exposure to the attriters’ L. In order to investigate this, Chamorro, Sorace, & Sturt () tested a second group of Spanish L attriters using the same experimental items and the same offline and online tasks as the ones described in Section .. In line with the other group of attriters, this second group of attriters also consisted of twenty-four Spanish near-native speakers of
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
English who had been living in the UK for a minimum of five years, but they had been exposed exclusively to Spanish in Spain for a minimum of a week right before they were tested. This ‘exposed’ group was compared to the other group of attriters and the control group of monolinguals to explore whether attrition effects with structures at the syntax– pragmatics interface can decrease or disappear after recent exposure to L input. The results obtained from this exposed group revealed that attrition effects are not permanent but decrease as a result of L re-exposure. The offline judgement data revealed that this group was not significantly different from the attriters or the monolinguals, with all groups showing equal sensitivity to the pronoun mismatch when rating anaphora offline (i.e., all three groups revealed significant Pronoun by Antecedent interaction effects in their ratings and no significant differences were revealed when these groups were compared). Results from the eye-tracking experiment revealed that, in online reading, the monolingual and exposed groups are reliably more sensitive than the attriters to the pronoun mismatch, as they show significant Pronoun by Antecedent interaction effects in their reading times with these sentences in the critical region (i.e., the region where the pronoun occurs) as well as in the post-critical region, while the attriters do not show any significant interaction effects in any of the regions. These results can be explained following Paradis’ () ATH, which predicts that attrition diminishes with frequency and recency of exposure to the L. In line with this hypothesis, the results obtained for the exposed group did not reveal attrition effects with pronominal subjects, since this group, unlike the attriters, showed a reliable sensitivity to the pronoun mismatch when processing this interface structure in real time. Moreover, when they were compared to the monolinguals, no significant differences between the two groups were revealed, which suggests that attrition effects diminish after recent exposure to the L. Taking together all the results from Chamorro, Sorace, & Sturt () and Chamorro, Sturt, & Sorace (), these findings reveal that individual L attrition is selective and occurs with structures that involve an external interface (i.e., the syntax–pragmatics interface) but not with structures that involve an internal interface (i.e., the syntax– semantics interface). They also reveal that attrition effects on these structures are more likely to be revealed in tasks tapping real-time processing rather than offline metalinguistic tasks such as acceptability judgements. Crucially, Chamorro, Sorace, & Sturt () provide evidence for the question raised in this section, as it was revealed that individual L attrition affects online sensitivity with structures at the syntax–pragmatics interface rather than causing a permanent change in speakers’ L knowledge representations. This reveals that bilinguals are sensitive to input changes and provides further support for Sorace’s (, ) proposal that first generation individual attrition affects only the ability to process interface structures but not knowledge representations themselves.
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
......................................................................................................................
......................................................................................................................
. I
.................................................................................................................................. T Bottleneck Hypothesis (BH; Slabakova, , , ) was proposed to identify properties and constructions that are harder or easier for learners to acquire in their second language (L). It is predicated on the generative linguistics notion attributing a special place to functional morphology in grammar and acquisition. Functional morphemes, both bound (inflectional) as well as free, carry grammatical meanings that radically change the overall meaning of a sentence. No sentence meaning is complete without taking the contributions of functional morphology into account. The following example illustrates the division of labour between lexical items and functional categories. Suppose we want to combine the lexical items Jane, eat, and sandwich into a message with this meaning: ‘Jane is consuming a sandwich at the moment of speaking.’ It is a rare language that would just string the words together with no grammatical morphemes and without taking word order into account. In English, for example, the meanings of , , and noun are imparted by the indefinite determiner a; the aspectual meaning is reflected in the progressive -ing; the temporal meaning of and agreement with a third person singular subject are captured by the auxiliary is.1 The final makeup of the sentence Jane is eating a sandwich carries a complete meaning (its truth value) composed of lexical and grammatical meanings that are considerably removed from the raw assembly of lexical verb and nouns. These additional meanings that anchor the event in time and space are contributed by the functional morphemes.
1 I use small letters to set out grammatical meanings with predominantly semantic content, as opposed to purely grammatical meanings such as subject–verb agreement.
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
The grammatical and semantic features expressed by functional morphology, such as the ones mentioned above as well as , , etc., are universal in the sense that they have to be expressed in every language, in one way or another. However, languages differ with respect to how these universal meanings are expressed. For example, Mandarin Chinese does not express the past time of a state or an event with a dedicated morpheme, while English does. Korean and Russian are languages without a definite article. The lack of overt exponents does not mean that the respective languages cannot express those meanings, but that they express them by other means, including leaving them for the context to fix. An important feature of functional categories is their multi-functionality. In languages such as English and many others, functional morphemes can encode multiple grammatical meanings. Linguists express this state of affairs by postulating that a number of grammatical features reside (are checked) in the same functional category. In the sentence Gwen often eat-s fish, the agreement morpheme -s signals , and agreement with a third person subject. However, two other grammatical meanings are also captured by features in the same functional category Tense: the obligatory subject (* often eats fish) and the lack of verb movement over the adverb and negation (*Gwen eats often fish, *Gwen eats not fish). Thus, multi-functional Tense in English carries at least five features, three semantic features, and two formal syntactic features. In addition to being indispensable in carrying grammatical meanings, the functional morphemes are the main locus of linguistic variation. As per the Minimalist Program assumptions (Chomsky, ), the core syntactic operations (e.g., Merge, Agree, etc.) are the same in all natural languages, hence they are also universal just as the grammatical meanings. Parametric differences are no longer attributed to the syntax of two languages differing in constrained ways. If syntactic operations are universal, then variation among languages can only be located in the functional lexicon, the part of the lexicon where the functional morphology resides, and at the two interfaces with meaning and form (or expression). How about semantics? Just like syntax, semantics is postulated to contain only universal operations (e.g., functional application, lambda abstraction, etc.) which allow languages to translate sentence syntactic structure into sentence meaning, or truth value. As mentioned above, grammatical meanings such as , , and are reflected in the functional lexicon. We can think of each lexical item, including functional and open-class ones, as combining information of three types: semantic, phonological, and morpho-syntactic (Jackendoff, ). To illustrate with a verb such as drink, when we know the verb, we know that it means ‘to consume or imbibe a liquid’; we know it is made up of a unique sequence of sounds [drıŋk]; we also know that it is a transitive verb that most of the time takes an object, and that this object may be dropped but is implied (e.g., when used to characterize a person, He drinks does not mean He drinks water). This view of the language architecture (see Slabakova, , chs and ) informs views and predictions about language acquisition in general. In addition to learning the equivalents of the native open-class lexical items, acquiring the functional morphology with all its features secures the skeleton of the grammar. Potentially speaking, the three types of information encoded in lexical items can be decoupled in acquisition: learners may not
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
learn them all at the same time. For instance, one can imagine a speaker knowing the basic meaning of drink but not understanding the significance of the implied object (that it can only refer to alcohol). While formal syntactic features are universal, their UniversalGrammar-provided values are reflected in the specifications of the functional morphology and need language-specific input to be fixed. Grammatical meanings (, ) can also be uncoupled in acquisition, especially if they are complex ( and ). For example, the usage of the progressive tense with stative verbs to convey temporary events (He is being honest) may be acquired later than its primary meaning of ongoing event (He is playing). In summary, formal syntactic features and grammatical meanings reflected in a functional category can be dissociated in acquisition, as the features can be learned one by one, or in bundles. While this view of the language architecture is theory-specific, there is experimental evidence to support the dissociation between functional morphology and syntax. White (, pp. –), following the idea of morphological separation (Beard, , ; Halle & Marantz, ), provides convincing evidence for the so-called syntax-before-morphology view. She argues that L acquisition data from children and adult learners point to the following dissociation: The morphological expressions of the functional category Tense are being produced with accuracy between around % and % in obligatory contexts, while the syntactic reflexes of the same functional category (nominative subject, subject remains in VP) are observed with close to a % accuracy. It seems that the invisible syntactic knowledge is in place much earlier than the obligatory suppliance of the morphological expressions. Slabakova (, ) argued that universal semantic knowledge is also in place without much effort on the part of learners. To take a specific example, Slabakova () examined whether learners of L Chinese know which sentences are to be interpreted as encoding a past action, if there is no overt morphological expression of Tense. It turns out they do, and even at intermediate levels of proficiency. These and many other studies (see Slabakova, , for a review) supply evidence that syntactic and semantic knowledge come before fully accurate realization of the inflectional morphology. In addition, functional morphology is often expressed by complex paradigms (e.g., subject–verb agreement in Slavic and Romance languages), which have to be learned and accessed as separate lexical entries and assembled online both in comprehension and production. Since there is an element of effort-intensive lexical learning in acquiring paradigms, inexperienced learners (children, L learners) at the beginning of the acquisition process often assume one form to be the representative of the whole morphological paradigm, or a default form. Paradigmatic forms present more difficulty in processing language, as opposed to single-form expressions. While there are solid linguistic explanations of why a certain member of a paradigm can become a default, the phenomenon of default forms in principle suggests that not all members of a paradigm are equal, lexically learned at the same time and processed equally easily. When non-default forms are required by the linguistic context, errors are more likely. Finally, some functional morphology has only grammatical functions and not much communicative import. Take for example the agreement between a noun and its modifying adjective. Whether the agreement morpheme is present on the adjective or not, the noun phrase is still comprehensible. Such a situation might apply processing pressure for omissions or supplying default forms.
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
The BH, whose application to attrition I will explore in this chapter, is based on comparing degrees of success or difficulty in the L acquisition of three modules of the grammar: syntax, semantics, and morphology. Its main tenets, as specified in (i), (ii), and (iii), are both theoretically motivated and supported with acquisition facts. i. Functional morphology is more challenging for learners than syntax or semantics, which are universal. This claim is founded on the minimalist view that functional morphology encodes variation among languages, and hence is the locus of what has to be acquired. In addition, processing and learning considerations also point to heightened difficulty. ii. The syntactic representations attained by learners may be accurate even though the associated overt morphology may be lacking. iii. The compositional semantic representations employed by learners are accurate. In this chapter, I will apply this language acquisition hypothesis to a different population with the ensuing different outcomes. The purpose is trying to account for language loss, not language gain. Together with recent treatments (see Schmid, b, for an overview), I will assume that language attrition represents a reanalysis of the grammar over the course of a lifetime in the absence of abundant, consistent input, and in this reanalysis, the language architecture and inherent complexity of grammatical features are an essential factor that affects the process. When a fully acquired language is used under reduced input conditions and under pressure from another language in a bilingual environment, at issue is which areas of grammar remain resilient and likely to be spared, and which ones are vulnerable to attrition? The BH can provide a useful lens to these issues.
. A :
.................................................................................................................................. Recent research on language attrition in the lexicon as well as in grammar has led to the conclusion that we can no longer treat language attrition as a unitary phenomenon. In addition to the high amount of individual variation due to personal, motivational, and experiential factors, it appears that the age of onset of bilingualism is of utmost importance. With all the necessary caveats needed, when we average over a multitude of people’s linguistic experiences, we may be describing two distinct processes. One would be the heritage speaker experience: child learners growing up in homes where the native language is spoken, but in countries with another majority language. Very commonly, the heritage language is the individual’s first language chronologically, a mother tongue learned in a naturalistic setting in a family environment. The majority language comes in second. However, being used at school, with peers, and in the wider society, the majority language often becomes dominant and has a dramatic effect over the native language, setting off language attrition (Montrul, ; Benmamoun et al., a). Broadly defined, the exposure
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
to the dominant societal language comes either at school age or at the age of immigration, but crucially before the age of puberty (around ) (see Bylund, Chapter , this volume, as well as Kupisch & Rothman, , for more discussion). In our context, these speakers could be called ‘early attriters’. Summarizing recent research, Benmamoun et al. (a) argue that loss of functional structure is evident in this population. Another population exhibiting language attrition is adult immigrants. Very often, they are the parents of the heritage language speakers described above. Exposure to the L is sustained and intense, and may be accompanied with a sharp decrease in exposure to the native language. In our context, the defining characteristic of this group is that they have left the country where their mother tongue is spoken after the age of puberty, and sometimes much later. This ensures that their native language grammar was completely mature and stable before the onset of (potential) attrition. For the purposes of this chapter, I will call these ‘late attriters’. The current evidence suggests that changes in the core grammar, or in the functional category structure, is not much in evidence in this population. So, while we are pondering the issue of why is it the case that certain areas of the grammar are so resilient whereas other areas are malleable and subject to change throughout the life of a speaker, we should also keep in mind that the age of onset of attrition can change the answer dramatically (Schmid, ; Montrul et al., ).
.. The grammatical competence of early attriters Heritage speaker competence and performance are notoriously diverse and fluid, so generalizations are to be attempted with caution. Nevertheless, generalizations are indispensable in checking theoretical predictions. Benmamoun et al. (a), a keynote article digesting decades of research on heritage language speakers, summarizes and interprets data from several areas where heritage speakers appear to diverge from monolingual native speakers: phonology, lexical knowledge, morphology, syntax, case marking, and codeswitching. In order to check the predictions of the BH against these pooled findings, we shall examine the areas of functional morphology and syntax in some more detail. The following citations come from Benmamoun et al. (a). Inflectional morphology in languages that exhibit robust morphological systems, including regular and irregular paradigms, is particularly vulnerable in heritage languages. (Benmamoun et al., a, p. ) Morphological deficits in heritage languages are asymmetric; they seem to be more pronounced and pervasive in nominal morphology than in verbal morphology. (Benmamoun et al., a, p. ) Agreement is the functional feature that undergoes the highest erosion. (Benmamoun et al., a, p. )
A number of studies have demonstrated significant loss of subject–verb agreement in heritage speakers (as compared to monolinguals): Håkansson (); Anderson (); Polinsky (); Rothman (); Albirini et al. (). For example, Albirini et al. () recorded Egyptian and Palestinian heritage speakers with English as the dominant language and found up to .% of verb agreement errors and up to % nominal agreement errors.
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
Speakers experienced particular difficulty in creating agreement chains between the verb and plural or feminine subjects, as well as where more arcane rules of grammar dictate, for example, singular agreement with post-verbal conjoined subjects (p. ).2 Another example comes from Hungarian, where the possessed noun morphologically agrees with the possessor in person and number. Bolonyai () reports an error rate of % for this property among Hungarian speaking children in the US between the ages of and . She explicitly argues for the selective nature of morphological variability. Erosion of agreement is particularly striking given that other verbal categories seem to be more resilient, but is predicted by the BH. It is particularly striking that high percentages of agreement errors occur in null subject languages (Arabic in this case), where the agreement affix identifies the missing subject. The lack of semantic significance explanation presents a good account of the Hungarian agreement attrition. Since the possessed and the possessor are both present in the noun phrase, dropping the agreement morpheme does not incur communication loss. In addition to agreement, other functional categories morphologically expressed on the verb also create difficulties. For example, aspect vulnerability has been studied widely in heritage Russian, because aspect is such a prominent feature of the language and because it is intricately linked to lexical knowledge in a complex way (Slabakova, ). Russian grammatical aspect is a functional category encoding purely grammatical features but their expression is an amalgamation with derivational morphology. Perfective-marking prefixes comprise some eighteen to twenty derivational morphemes, and prefix–verb combinations have to be learned individually. Among many others, Laleko () described aspectual errors that heritage Russian speakers made in free oral production, and while they can be classified as aspectual errors, they also constitute wrong lexical choices. In order to understand the process, the reader must keep in mind that lexical aspect is a semantic property of verbal predicates including the telicity distinction. Unlike lexical aspect, grammatical aspect is expressed through inflectional morphology on the verb (Smith, ).3 Polinsky (, a, ) proposed the Lexical Aspect Hypothesis (see also Pereltsvaig, ), according to which verbal aspectual morphology (i.e., perfective prefixes) in heritage Russian encodes lexical, rather than grammatical aspect distinctions: ‘most verbs become either lexicalized perfectives or lexicalized imperfectives’ (Polinsky, , p. ). This functional lexicon reanalysis predicts that telic verbs will mostly appear in perfective forms while atelic verbs will surface as imperfectives. These predictions were supported by a vocabulary translation experiment (Polinsky, ) and by a forced production task (Laleko, ). The BH would explain this lexicalization process by the close link between the lexicon and functional morphology, and the need for these particular functional morphemes to be learned one by one with their appropriate verbs (Slabakova, ). Early attriters simplify their lexicon by fusing the prefix and the verbal root. Even if
2
It is also interesting to note that there were significant differences among the heritage Arabic speakers. Egyptian Arabic speakers showed a greater loss of agreement than Palestinian Arabic speakers. This divergence may have both structural and sociolinguistic explanations. 3 To eat a sandwich is a telic VP because it has a potential endpoint, while She was eating a sandwich is a clause encoding imperfective viewpoint aspect, since the speaker focuses attention on the process, not on the endpoint of the event. English sentences mark both lexical aspect (telicity) and viewpoint aspect.
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
attriters know that a prefixed verb signals a complete event and treat the prefix as separable from the root, they still might produce the wrong prefix for any verb. Staying among verbal inflectional distinctions, it is to be noted that some are much more vulnerable than others. As discussed above, the challenge of Russian aspect is based on its lexical rather than grammatical difficulty. Frequency of usage may be another factor explaining differential difficulty. Let us compare tense with subjunctive mood in Romance, which differ in frequency of occurrence. While both grammatical features are expressed with complex paradigms, tense occurs in every sentence whereas subjunctive mood is rarer and needs a specific context (Iverson et al., ). Not surprisingly, while mood distinctions are subject to significant attrition (e.g., subjunctive in heritage Spanish), tense distinctions are quite robust, even in a weak heritage language (Montrul, ). Next, we shall look at nominal morphology, which does not have much semantic effect on the interpretation of the sentence: gender, case, and differential object marking (DOM). In some languages, such as Slavic, Germanic, and Romance, all nouns are categorized in lexical classes of grammatical gender. Note that this categorization does not carry any informational function, because most nouns refer to inanimates, so gender marking is not linked to natural sex but is purely grammatical. Gender represents another vulnerable feature illustrating the lexical burden of functional morphology, both as noun categorization and as gender agreement within the noun phrase. Montrul et al. (a) found agerelated effects in the marking of heritage Spanish gender agreement, while Polinsky (b) discovered that heritage Russian speakers had restructured the formal cues used to assign Russian nouns to gender classes. In the latter study, this restructuring was related to general proficiency in Russian. Case marking also displays attrition and loss, especially in constructions that allow optionality, such as the genitive of negation in Russian (Leisiö, ). DOM in Spanish, a grammatical phenomenon with complex semantic distinctions (see examples in later sections), is frequently deficient among heritage speakers (Montrul & Bowles, ). In sum, early attriters are challenged by inflectional morphology when it is expressed by large paradigms involving affix learning and online assembly (agreement), when marking a grammatical category implicates a lexically fixed affix–verb combination (aspect) or individual exponence (gender) and when the distinction is rarely used (mood). In contrast to difficulties with functional morphology, early attriters rarely have issues with basic syntax, such as word order in simple clauses or verb movement. There is not much work on low-level syntactic properties because researchers focus their attention on much more complex syntax, but at least some evidence on accurate verb raising in heritage Spanish is available (Bruhn de Garavito, ). Among the complex syntactic characteristics most commonly considered in current research are A-bar movement regularities4 (O’Grady, Lee, & Choo, ; O’Grady, Kwak, et al., ; Lee-Ellis, ; Polinsky, ), wh-movement (Montrul et al., b), and binding (Kim et al., ; Keating et al., ). An interesting case in point is presented by Montrul et al. (b), which studied a range of interrogative sentences and extractions out of main and embedded clauses. The researchers tested early bilinguals at three proficiency levels and compared them to similar
4 A-bar movement involves extraction of a phrase from its position of merging and case marking to another position in the sentence, usually to the left periphery of the sentence.
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
proficiency adult-onset L learners of Spanish. Both the L learners and the heritage speakers demonstrated solid knowledge of constraints on wh-movement, even in constructions where transfer from English would lead the speakers astray. The only area where they differed from native speakers was subject and object extraction from embedded clauses. These findings were interpreted as pointing to correct underlying representations in all L learners and heritage speakers but difficulties in processing such long sentences. The general conclusion of this and comparable studies is that basic syntactic operations are not vulnerable to attrition even in early bilinguals, while their reduced practice of the heritage language may lead to superficial processing deficiencies. How about knowledge of semantics and appropriateness in discourse context? The BH predicts that comprehension resulting from semantic composition of already established grammatical features and constructions is not going to be vulnerable. Studies on such properties are rare, but they do exist. Leal Méndez et al. () tested whether heritage speakers at two proficiency levels and adult L-ers recognize the meaning and appropriate context of clitic left-dislocations (CLLD) and focus fronting in Spanish. The English equivalent to CLLD, topicalization, does not double the dislocated object with a clitic. CLLD is a very frequent feature of Spanish. The specific semantic constraint tested was the semantic freedom of the antecedent–dislocate connection: the dislocated clitic-doubled constituent can be identical, a subset or even a part of the discourse antecedent. This semantic freedom is universal for all topicalizations, hence the BH would predict that early attriters would not have any problem with comprehending it. The example below illustrates the part–whole relationship between the dislocated constituent (the table legs) and the antecedent (the table). () Context: What shall we do with the table? It is too big! Mira, las patas, las doblas así . . . look the legs Cl. fold..sg this-way ‘Look, you can fold the legs like this . . . ’
Part/whole
While attriters might make errors with the grammar of CLLD by not providing the cliticdoubling, they should not be less accurate on sentences such as () compared to sentences in which the dislocated element and its antecedent are identical. A felicity judgement task ascertained that the learners and heritage speakers were no different from the monolingual natives in their comprehension of this semantic freedom. The Interface Hypothesis (IH; Sorace, ) was initially proposed to address language attrition and later expanded to other bilingual conditions (see also Chamorro & Sorace, Chapter , this volume). The predictions of the IH and the BH differ: whereas the former hypothesis sees most challenge at the interface of grammar and external modules such as context, the latter hypothesis places the main burden of difficulty on the functional morphology. While there is substantial evidence for the IH coming predominantly from null and overt subject usage, a range of experiments on other interface constructions has not encountered heritage speaker deficiencies, challenging its predictions. For example, the same experiment cited above, Leal Méndez et al. (), in addition to checking a semantic constraint, also established that the heritage speakers had no problem with choosing the context-appropriate construction with the clitic-doubling. Furthermore Leal et al. () looked at a very rare construction, clitic right dislocation, which is only acceptable in highly
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
circumscribed contexts such as the affective context illustrated below. Note that the object is right-dislocated in (a), but is doubled by a pre-verbal clitic: () Context: Mariana and Omar are siblings and they inherited several pieces of furniture after their father died. Because Mariana has a larger house, she kept the majority of them. The problem is that many pieces of furniture were damaged, and Mariana initially didn’t know what to do with them. One day, Omar goes to visit her; he thinks he recognizes a chair in the living room. O: Veo que por fin decidiste arreglar la silla de papá. ‘I see that you finally decided to fix dad’s chair.’ (a) M: No. La regalé, la silla odiosa esa. Lo que arreglé fue el sofá que tanto me gustaba. no CL I-gave the chair ugly that. CL that fixed was the sofa that so me pleased (b) M: No. # Regalé, la silla odiosa esa. Lo que arreglé fue el sofá que tanto me gustaba. no I-gave the chair ugly that. CL that fixed was the sofa that so me pleased ‘No, I gave that hideous chair away. What I did repair was the sofa that I liked so much.’ The experiment used a judgement task where speakers had to rate the felicity of (a) or (b) in the given context. Results established that Spanish-dominant and heritage bilinguals were equally selective with the contexts in which they allowed clitic right dislocation. No group differences were in evidence. The researchers argued that these findings challenge the IH, at least in offline tasks where there are no time constraints. To bring all these disparate threads together, the comparison between early attriters’ knowledge of functional morphology, basic syntactic operations and semantic composition constitutes support for the BH, in the sense that functional morphology is much more vulnerable to attrition than the latter two areas of the grammar. However, in summarizing apparent knowledge of different properties, it also emerged that grouping linguistic properties and their acquisition on the basis of gross modules may be over-simplistic. Some functional morphology such as agreement seems to be more vulnerable than other functional morphology such as tense marking, with various features in between these two extremes. In other words, the functional morphology is not monolithic but can be selectively attrited. A tentative explanation based on the grammatical status, complexity of the grammatical features, and the evidence for their expression runs like this: If a grammatical feature does not affect meaning-making and comprehension, but captures purely syntactic effects (such as subject–verb agreement and DOM), then this feature is vulnerable. Seemingly redundant morphology is not retained in heritage grammars but simplified, possibly leading to a restructuring of the overall grammar (see also Seliger, ). If a piece of functional morphology has a semantic or discourse-pragmatic effect and the evidence for it is rarely seen in the language, then that feature may also be vulnerable (mood, DOM). If functional morphology is conflated with derivational morphology as in Russian aspect, it is likely to be restructured. One-to-one form–meaning mappings that are clearly and abundantly evidenced in the linguistic input (such as tense marking) endure the best. This explanation is compatible with the spirit of the Interpretability Hypothesis (Tsimpli & Dimitrakopoulou, ) because it elevates meaning-related features as more resilient than form-related ones.5 The Interpretability Hypothesis (Tsimpli & Dimitrakopoulou, ) has been proposed to explain the inability of L learners to reach native-like levels of competence. The hypothesis builds on the 5
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
.. The grammatical competence of late attriters Let us now turn our attention to later attriters, the immigrants who moved to the country of their L well after puberty. Studies documenting attrition of functional morphology among later attriters are rare, but do exist. For example, Ribbert & Kuiken () documented attrition of the infinitival complementizer among Germans who moved to the Netherlands as university students. There is both overlap and difference in the use of Dutch and German infinitival complementizer um/om: Dutch om is a superset of the contexts where German um is used. Furthermore, Dutch has a high number of contexts where om can be used optionally. The German speakers were asked to rate the grammaticality of pairs of sentences with and without a complementizer. Their preferences differed significantly from other German speakers who were not in contact with the Dutch language. Still staying within functional morphology and its interpretation, selective attrition is attested in late attriters, just like in early attriters. Gürel (a) provides a very interesting example. She looked at a rare morpho-syntactic phenomenon with semantic effects: the binding of overt, null, and reflexive pronouns in Turkish embedded clauses as exemplified in (a) below. The overt third person pronoun o, when it is embedded clause subject, cannot refer back to the main clause subject, while the null pronoun (pro) can. The English overt pronoun he in (b) has no such restriction. In the examples, the sub-index ‘i’ refers to the main subject while ‘j’ means that the pronoun has some other discourse-provided antecedent. (a)
Buraki [o-nun*i/j / kendi-si-nini/j / proi/j zeki ol-duğ-u]-nu Burak s/he- self-- pro intelligent be--- düşün-üyor think- ‘Buraki thinks that [he*i/j / selfi/j / proi/j is intelligent]’
(b)
Buraki thinks that hei/j is intelligent.
In two interpretation tasks where speakers had to choose between the pronoun antecedent being the main clause subject and a third person not mentioned in the sentence, native speakers overwhelmingly chose the third person, the only available interpretation. However, Turkish attriters living in the USA and Canada chose the wrong antecedent between % and % (on the different conditions). They still preferred the third person as antecedent, choosing it above %; however, their knowledge in this respect was evidently shaky. In contrast to the overt pronoun o, the attriters maintained their knowledge of the nominative reflexive kendisi and the null pronoun. It is also interesting to note that this selective attrition is parallelled exactly by selective acquisition among near-native L learners of Turkish (Gürel, ). It appears that the null pronoun and kendisi are acquired completely and resist attrition, while the antecedent choices of o in embedded subject position create problems for both near-native speakers and late attriters. Binding is among the most complex properties of language, relying on syntactic features captured by Minimalist Program interpretable–uninterpretable feature distinction and postulates that interpretable features are acquirable in the L while purely syntactic uninterpretable features are not, if they are not already established in the native language and thus transferable. Note also the explanation provided above is in contrast to that proposed by Tsimpli et al. (), which gives rise to the IH.
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
functional morphemes with semantic import, and frequently discourse-pragmatic import as well. The prohibition of main clause antecedents for the overt pronoun o, but not for the other two pronominal elements, is eroded under the influence of English where such prohibition does not exist. In one sense, this restructuring is a simplification of the grammar. In checking the predictions of the BH, it is informative to look at Schmid (), a study comparing a number of late attriters with very advanced L speakers, matched in proficiency and looking at a number of linguistic properties. The attriters were German native speakers living in Canada, while the advanced L learners and controls were living in Germany at the time of testing. The data comes from two spontaneous speech samples per subject, a film retelling task, and a semi-structured autobiographical interview, in which errors were calculated per , words. The range of functional morphemes tested included case, gender, plural, articles, pronouns, tense morphology (the weak/strong distinction), auxiliaries, and subject–verb agreement; while syntactic properties included verb-second and discontinuous word order, as well as word order in subordinate clauses. Schmid reported the error counts for each speech sample separately (her table , p. ). The late attriters were non-distinguishable from the monolingual native speakers for all of the properties tested, functional morphology and syntax alike. The BH cannot offer an explanation for their linguistic performance just because there is nothing to explain: these German immigrants to Canada did not demonstrate any attrition. In contrast to the findings in Schmid (), Montrul et al. () did attest attrition of functional morphology. This a large-scale study of DOM examined in heritage speakers and their parent generation, thus early and late attriters, of Spanish, Hindi/Urdu, and Romanian. DOM is among the most complicated aspects of functional morphology, where object marking depends mainly on animacy and specificity, but aspect, topicality, agentivity, and affectedness may also influence it. The example in () comes from Hindi. () Mira-ne ramesh-ko dekhaa. Mira-erg Ramesh- saw ‘Mira saw Ramesh.’ (*Mira-ne ramesh dekhaa.) The three languages under investigation work largely (although not entirely) in parallel with respect to DOM’s semantic restrictions, but while the Spanish marker is a (not audible after a vowel), Hindi and Romanian markers are syllabic (ko and pe, respectively), so better audible. The findings are complicated. The heritage speakers accepted ungrammatical DOM omission with animate and specific objects to higher degrees than they accepted omission of indirect object or locative marking. Thus, they were simplifying their grammar in the direction of redundancy elimination (compared to a locative, DOM is largely redundant, as context certainly distinguishes specific from non-specific objects). Thirteen (out of twenty-one; %) late attriters of Mexican Spanish had lost DOM in the USA, while Hindi and Romanian speakers were not attrited for this property at all. While these results are difficult to explain straightforwardly, the linguistic complexity of the phenomenon is the prime suspect. In addition to the functional morphology being largely redundant, the markers are multi-functional in the sense that they express other grammatical meanings as well, producing an ‘opaque form-meaning mapping’ (Slabakova, ; O’Grady et al., ). The phonological salience mentioned above cannot be the sole
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
factor explaining the group differences. The authors make a convincing argument that some language-internal structural factors such as clitic-doubling in Romanian and lack of definite articles in Hindi may be a cause of differences. Ultimately, DOM markers have to be considered as functioning in tandem with the rest of the grammar and being affected by the holistic linguistic experience. The authors conclude that ‘the structural properties of the DOM markers together with the syntax of definiteness and specificity in each language (more than external factors) seem to account for the degree of DOM erosion in each language’ (Montrul et al., , p. ). Turning to knowledge of syntax and of interpretation now, the findings of an influential study, Tsimpli et al. () are worth reviewing here. The study investigated production and comprehension of overt and null preverbal and post-verbal subjects among late attriters of Italian and Greek living in an English-speaking environment. A robust finding was that the syntax of subjects and interpretation of null subjects were not attrited, while overt subject interpretation choices were more vulnerable because they depended on consideration of discourse-pragmatics as well as syntax. These and other findings were conceptualized as the IH (see more on this hypothesis in Chapter ). Another recent study, Flores (), compared knowledge of syntax and discourse in four age groups of German-Portuguese bilinguals living in Portugal, for whom German was the minority language.6 Flores’ subject pool allowed her to make very precise distinctions based on age of loss of contact with the attriting language. She examined a narrow syntax property, verb placement in main and embedded clauses, and a syntax–discourse interface property, object expression under topichood. The speakers who had lost contact with German before the age of exhibited around % errors on both properties. Those who had moved to Portugal between the ages of and were markedly different: while their errors on topic object drop were in the % range, their word order errors were negligible. The third group of adult returnees (at ages –) and the group of children who were still in contact with German made no errors on either property. The author concluded that narrow syntax was more likely to suffer from attrition if contact with German was lost before age to , while object expression continues to be vulnerable to reduced input until the age of . Late attriters did not exhibit any attrition of these properties. Let us take stock of the main findings for late attriters. While attrition of grammatical features is much rarer among this population (Schmid, ), some functional morphology does suffer attrition (Ribert & Kuiken, ; Montrul et al., ). Purely syntactic properties such as verb placement (Flores, ) and the syntax of null and overt subjects (Tsimpli et al., ) are much more robust and do not appear to undergo attrition. Finally, interpretive knowledge seems to be vulnerable only if it is calculated at the syntax– discourse interface (Tsimpli et al., ). At the same time, Flores’ () adult attriter group did not demonstrate such vulnerability. The BH is largely borne out by such findings. If anything is vulnerable in the grammars of late attriters, it is the functional morphology and not pure syntax. The additional factors we saw at work in early attriters, namely linguistic complexity of the property under investigation, redundancy of the functional morphology marker, opaqueness of form–
6 Flores’ child control group was the exception, because these subjects were still living in Germanspeaking countries.
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
meaning mapping, as well as usage frequency hence activation, go a long way to explaining most of the findings we discussed in this section. At the same time, the hypothesis, as it is formulated now, may be too large-grained to provide a sufficient explanation of particular selective attrition, such as the one discovered by Gürel (a), where a restriction of one pronominal form is relaxed, or by Montrul et al. (), where one group of late attriters is affected but not others.
. C
.................................................................................................................................. Examining findings from early and late attrited speakers invites fundamental questions on what the grammar of these speakers look like. In other words, what are the linguistic representations that allow not altogether erroneous, but ‘shaky’ or ‘rusty’ performance on some properties? The BH is useful in identifying the essential grammatical complexity of the functional morphology, especially with many-to-many features-to-form mappings, as the main reason. In comparison, the core syntactic operations and the compositional principles of putting the meaning of a message together are resilient in the face of suboptimal input, just because they are universal and are used in the dominant L as well as in the affected L. While the BH addresses the issue of why certain properties are differentially vulnerable in attrited grammars, a related important question is how this state of grammatical knowledge operates in individual speakers’ mind/brain. Reminiscent of Yang’s () variationist learning approach, a recent generative proposal, Amaral & Roeper (), specifically endorses and explains the mechanism of how all the grammars of a multilingual can arise and function in tandem. ‘[A]ny language contains properties of several recognizable language types, i.e. the grammar of a language L can have elements that form sub-grammars compatible with L, L, Ln’ (Amaral & Roeper, , p. ). According to this view, L/Ln acquisition is a natural proliferation of sub-grammars, a process of grammar expansion that is certain to affect the native language right back. Multilingual linguistic competence is an amalgamation of sub-grammars coming from the previously acquired languages, equipped with some sort of differentiation mechanism, such as a tagging device for the rules and lexicon of grammar A or grammar B. If one rule, in the attrition situation the L rule, is used much more intensively, its activation level is likely higher in the sense that it will be invoked much faster and with less effort. The L rule is not truly gone and it is still tagged as a L rule, but summoning it may be more effortful and costly. This is essentially a processing view of grammar usage (see also Sharwood Smith & Truscott, ; and Sharwood Smith, Chapter , this volume), which explains how grammatical rules, while they are in place, may appear ‘rusty’ and ‘shaky’ in performance.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
......................................................................................................................
......................................................................................................................
. P L
.................................................................................................................................. I is fair to say that it has become standard practice in L attrition research to collect information on background variables, for the purposes of describing and homogenizing research groups, on the one hand, and to assess the extent to which these variables may influence the outcome of attrition processes, on the other. While certain variables, such as gender and age of bilingualism, are reported in most studies, Köpke & Schmid () recommended that L attrition studies minimally take into account ‘age at onset of L acquisition (simultaneous/early/late bilingual); age at onset of L attrition (pre‑/post puberty); time since onset of attrition; level of education; attitudes; frequency, amount and settings of use of the attriting language’ (pp. –). This proposal was taken up in the work of the Graduate Research Network on First Language Attrition (henceforth the Network; Schmid, b). The Network sought to strengthen the empirical basis of L attrition research by developing a set of instruments, the Language Attrition Test Battery (LATB; Schmid, ; cf. https://languageattrition.org/), for use across diverse language contexts, theoretical frameworks, and research objectives. The LATB included a sociolinguistic questionnaire, which elicited data on the following variables and was deployed in a number of studies (Keijzer, ; de Leeuw, ; Dostert,
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
; Cherciov, ; Lubińska, ; Opitz, a; Varga, ; Yılmaz, ) (variables in italics were included in a subgroup of studies): • Personal background: gender; citizenship; marital status; number of children/grandchildren • Education and socio-economic status (SES): highest educational level; professional/ vocational training; previous and current jobs • Migration history: age at migration/arrival/onset of attrition (AOA), length of residence (LOR), age at testing (AaT) • Language learning history: number of languages (L1(s) +); age of acquisition/learning; variety of L1(s) spoken; motivation for acquiring/learning additional languages; past maximal and present proficiency (L1(s) +); language learning aptitude • Language use: frequency of use with family members (partner, children, parents) and friends; mode of use (face to face/telephone/Skype/letters; directionality); size of network; use in various domains (church, expat networks, shopping, pets, professional use), engagement with various media (TV/radio, newspaper, tapes/CDs/ videos/DVDs) • Language attitudes: towards intergenerational transmission (encourage children, correct children, regrets, relevance for contact with grandparents); towards language and culture of both communities (Attitude and Motivational Test Battery); towards own bi-/multilingualism The Network and related studies treated the elicited background variables as potential predictor, or independent, variables (IVs) for language proficiency outcomes (dependent variables, DVs), some taking a more exploratory approach, and others more specifically testing particular hypotheses concerning the relative importance of different variables. Thus, Bylund studied the impact of ‘age of reduced L contact’ (Bylund, , a,b) as well as language aptitude (Bylund et al., ; Bylund & Ramírez-Galan, ) in his maturational investigation of L attrition. Other studies focused on AOA, language use/ contact, aptitude, and attitude (e.g., Keijzer, ; Opitz, , a; de Leeuw et al., ; Lubińska, ; Cherciov, ; de Leeuw, Mennen, & Scobbie, ). Attempts to find meaningful correlations between IVs and DVs generally did not lead to conclusive results, with few variables emerging as definite predictors in their own right, and large parts of variability remaining unaccounted for (Schmid, Chapter , this volume; Schmid et al., ; and for L attrition, Mehotcheva & Mytara, Chapter , this volume). Schmid writes: ‘So far the findings from quantitative investigations of L attrition and external factors appear somewhat unsatisfying, and often contradictory’ (a, p. ). While maturational variables are sometimes viewed as the single indisputable predictor (Ammerlaan, , p. ; Köpke & Schmid, , p. ; Bylund, a, p. ; Schmid et al., , p. ), even they do not ‘predict’ attrition outcomes in the strict sense of the word, since participants with similar language biographies show substantial variability in the degree with which they succeed in maintaining their L (Schmid, a, a). Instead, variables that were assumed to be crucial, such as education level or contact, turned out to be considerably less influential than expected, or at least not to confirm simple cause–effect relationships (Schmid et al., ; Opitz, b). For example, Schmid (a) discusses two participants who despite a ‘nearly identical’ background showed ‘dramatic differences [ . . . ]
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
in their linguistic data’, and concludes that in order to account for these differences we ‘have to look elsewhere’ rather than the background variables (p. ). It is reasonable to assume that the elusiveness of strong correlations and significant effects may in some part be linked to methodological issues, such as the specifics of the samples drawn, and difficulties in operationalizing and reliably measuring certain variables (e.g., language use; Schmid, ). As mentioned above, the lack of comparable designs amongst the extant empirical studies had in fact been the motivation for elaborating a common methodology within the Network. Thus, in order to avoid the problematic distinction between L attrition and nonacquisition, many studies, including the Network studies, investigate adult migrants who are well past puberty, thereby limiting the potential impact of AOA. Other variables, for example SES and education level, are also often controlled in participant recruitment, to ensure the homogeneity of the population and in view of the particular elicitation tasks planned during an investigation. Indeed, Schmid (a) advises: It is therefore of vital importance that you try to eliminate any impact of those predictors which you do not want to investigate. That is, there should be as little variation as possible along those factors in your sample. (p. )
This, necessarily, constrains the potential explanatory power of these variables. With single factors failing to produce unanimous results, and some of the variables, for example SES, gender, and education level, apparently inextricably confounded, some authors suggested that it might be profitable to look at clusters of variables (e.g., Köpke, ; Opitz, b, a) or ‘compound factors’/‘composite determinants’ (Schmid & Dusseldorp, ) and covariation. As an additional consideration, clustering variables also has a practical benefit in that the required number of participants is lower than if the variables were studied individually. This line of research has been somewhat more fruitful, in that some variables do seem to show an effect in the presence of another variable. De Bot et al. () and Schmid (b), for example, studied the interplay of LOR and frequency of L use. In de Bot et al., LOR was a factor only for participants with little contact with the L. Schmid, on the other hand, found LOR to be a predictor both for participants with the least, and those with the highest amount of contact, which she interprets as evidence for different sources of attrition for the two sets of participants— that is, reduced access in the former, and exposure to a changed variety in the latter. Schmid & Dusseldorp (), in their multivariate study of extralinguistic factors, found evidence for ‘several reliable factor groups’ (p. )—that is, sets of variables with high internal consistency, categorized as: (a) general factors (biological age, sex, education level); (b) factors pertaining to bilingual populations (age at onset of bilingualism/at emigration, attitudes and (ethnic affiliation); and (c) factors specific to attrition (contact with and use of the L). However, these factor groups were shown to have relatively weak predictive power. While studying variable clusters is undoubtedly a promising avenue, the findings by de Bot et al. () and Schmid (b) point to an alternative explanation for the difficulties in establishing meaningful correlations between background variables and the outcomes of attrition. This proposal rests on the nature of language as a complex dynamic system, and, as this chapter argues, consequently on the properties and interactions of the background variables themselves (Opitz, b).
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
At a very basic level, some variables, such as age, education, or language background, are relatively fixed or change in a predictable measurable way, while others, such as language use and attitudes, are conceivably quite variable over time. This renders conventional analyses rather problematic. In addition, the findings by de Bot et al. and Schmid reveal non-linear interactions between variables, as theorized in Complex Dynamic Systems Theory (CDST), which also pose a challenge for traditional, means- or regression-based, statistical analyses. In the following subsections, I will present a CDST view of the challenges associated with studying background variables in L attrition.
. C D S T
.................................................................................................................................. CDST was initially introduced to the study of second language acquisition (SLA) in the guises of Chaos/Complexity Theory (CT) by Larsen-Freeman (e.g., ; Cameron & Larsen-Freeman, ), and Dynamic Systems Theory (DST) by Van Geert (e.g., , ) and de Bot and collaborators (e.g., de Bot, ; de Bot et al., a,b), respectively. It is now agreed that despite CT and DST’s differing origins and emphases, their tenets and implications are sufficiently compatible for the purposes of research into language development to refer to both theories jointly as CDST (Larsen-Freeman, ), a convention adopted in this chapter. The advancement of CDST to the study of language is directly related to the failure of traditional approaches to provide more convincing answers for how language develops and changes. As Larsen-Freeman et al. () put it, CDST responds to the ‘general unease with neat, straightforward, linear and monodimensional solutions’ which is discernible in much scientific endeavour in applied linguistics and beyond, and ‘the recognition that reality is usually messy and often unpredictable, since it is defined by multiple, diverse and interdependent factors’ (p. ). Later on in the text, the authors, in view of the available evidence from empirical research, argue that ‘[t]he traditional scientific method, which is based on analysis, isolation, and the gathering of complete information about a phenomenon, is incapable of dealing with [the] complex interdependencies’ present in language development (p. ) and that it ‘might be time to consider a radical change of scientific paradigm’ (p. ). The core of this changed perspective consists of moving away from the (natural and at times warranted) desire to reduce complexity (Larsen-Freeman et al., ; LarsenFreeman, ), towards embracing the interconnectedness, dynamism, and non-linear interaction of elements within the language system, and of the system with its environment. Complex dynamic systems are in constant flux, yet generally stable, continually adapting to and creating new conditions through self-organization and emergence (Cameron & Larsen-Freeman, ). Language development is viewed holistically as being shaped by ‘interrelated patterns of experience, social interaction, and cognitive processes’, and language acquisition, use, and change are considered different ‘facets of the same system’ (Becker et al., , p. ; see also Ellis, ). This view implies that L and other languages may be considered embedded sub-systems operating within the same context and under the same constraints. It also implies that
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
attrition is as much a feature of language development as is acquisition. Or put differently: ‘Linguists who convincingly want to account for the enigma that is human language have to develop models and explanations that are able to accommodate incomplete, transitory, unstable, changeable and fossilized languages’ (Larsen-Freeman et al., , p. ). A first formal description of language development including L attrition from a CDST perspective was presented in Herdina & Jessner’s () Dynamic Model of Multilingualism (DMM) (see also Jessner, ; Herdina & Jessner, ). But, having been ‘instrumental in placing [language attrition] on the bilingual map’ (Larsen-Freeman et al., , p. ), it was de Bot in particular who made the case for CDST’s applicability to attrition research, arguing that development entails both growth and reduction, which are affected by the same forces and constraints (e.g., de Bot, , ). In , a special issue of the International Journal of Bilingualism on ‘Dynamics of First Language Attrition across the Lifespan’ (Opitz et al., ) presented empirical research from a dynamic perspective. In their introduction, de Leeuw, Opitz, and Lubińska () discuss the propositions that ‘L attrition /as such/ is evidence for the view that L develops dynamically across the lifespan’ and that ‘L attrition itself is dynamic’ (p. ). In the same issue, Schmid et al. () discuss methodological issues arising from CDST for attrition research, while Herdina & Jessner () reflect on the implications of attrition for CDST. Attrition is thus ‘negative’ or inverse language growth (Jessner, , p. ), which becomes noticeable in the absence of conditions that would normally support the acquisition or maintenance of a language, such as input in that language. As a result of the interconnectedness of the components of the language system, adaptations to environmental or internal changes in one part of the system potentially have consequences in other parts and at whole-system level. For example, increased use of a language in a new linguistic environment may lead to retrieval difficulties in other languages due to their inhibition and lowered activation level. This strongly suggests a need to consider language attrition vis-à-vis the development of the overall system. However, the bulk of L attrition research to date, with the exception of Opitz (a, a), Cherciov (), and Yılmaz (, Chapter , this volume), has studied changes in the attriting language only, without much recourse to the other languages, which may be an additional source of the divergent findings. While it is recognized that the future behaviour of a complex dynamic system ‘sensitively’ depends on the initial conditions prevailing at a given time (de Bot et al., , p. ; de Bot & Larsen-Freeman, , p. ), such adaptations are neither linear (i.e., they may be disproportionate to an initial impact) nor directly causal since self-organization gives rise to emergent, qualitatively novel, properties (de Bot & Larsen-Freeman, , pp. –). Moreover, the variables that are associated with such adaptations are themselves subject to complex interactions with a multitude of other variables, and each variable’s impact is thought to change over time (Larsen-Freeman, , p. ). For language development this implies that, on the one hand, the proficiency level in a language matters for the subsequent speed and developmental trajectory of that particular language, and the language system overall. This is potentially observable in differential rates of attrition in first versus second language attriters, mature versus developing first-language attriters, though inter-individual variability even within groups is a common finding, as shown below. On the other hand, non-linear cause–effect relationships are at play when, for example, language learners experience some feature of that language suddenly and seemingly without effort ‘falling into place’. The non-linearity of changes in individual language systems, and
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
their vulnerability to minor changes, was also demonstrated in Meara’s () computer simulation of language attrition (see also Larson-Hall, Chapter , this volume). These characteristics of complex dynamic systems lead to ‘variation both in and among individuals’ (de Bot et al., ) and ‘at all levels of linguistic organization’ (Cameron & Larsen-Freeman, ; Becker et al., , p. ). Inter- and intra-individual variability are indeed a robust finding in both acquisition and attrition studies, with occasional extreme examples like the two participants described in Schmid’s (a) study. As discussed in the next subsection, it has been argued that it is this variability that should be the proper concern of researchers studying language development. Ultimately, these properties also lead to an inherent unpredictability of the outcomes of change processes, which means that research from a CDST perspective is more concerned with retrodiction—that is, the attempt to ‘predict’ a previous state from the present (Cameron & Larsen-Freeman, ; Verspoor, de Bot, & Lowie, ). Notwithstanding the intrinsic dynamism of complex systems, language systems do not develop entirely randomly, with different periods showing more or less stability and change. Moreover, there is a considerable amount of evidence of patterns in language development, such as U-shaped development, which is compatible with CDST tenets (Cameron & LarsenFreeman, ; Lenzing, ). A relevant concept in this context is the notion of ‘attractor states’. An attractor state has been defined as ‘a critical value, pattern, solution or outcome towards which a system settles down or approaches over time’ (Newman, , cited in Hiver, , p. ). Metaphorically speaking, they are ‘pockets’ of (relative) stability within the topography of ‘state space’, the hypothetical landscape of all the possible outcomes of a complex dynamic system (Larsen-Freeman & Cameron, ; Hiver, , p. ff ). Since language systems are open and dynamic, such stability is never absolute—there may be ‘stability but never stasis’ (Larsen-Freeman et al., , p. ). Importantly, these developmental patterns are neither brought about by external forces, nor by a central agency, nor do they in themselves exert any ‘attraction’ or bias for developmental outcomes. Instead, they result purely from the self-organization of the system (de Bot & Larsen-Freeman, , p. ), in response to inputs including feedback and perturbations, such as changes in linguistic environment, and increased or decreased opportunities to use a language. Developing systems overall are thought to follow positive and negative growth curves akin to those modelled for biological growth under constant conditions (Herdina & Jessner, ). Such growth curves entail slower growth at the extreme ends of the acquisition and attrition processes, where adaptions to external and internal variables are initially absorbed by the system, before the process ‘takes off ’, as it were. These adaptations become noticeable as ‘increased scatter’ or optionality in performance, before qualitative changes occur in the system and a different stage is reached (Herdina & Jessner, , p. ff; Larsen-Freeman & Cameron, ). Indeed, high variability has been associated with subsequent learning (Van Dijk et al., ). When viewed close up and over short intervals, the developmental trajectories for individual properties vary from one property to another, and can appear rather erratic. However, at larger timescales and higher levels of granularity, what may have looked like a negative trend (e.g., more errors on a particular feature) may in fact be part of a generally positive change (Van Geert, ; de Bot, , ). Conversely, a system that appears to be stable at a certain point may in fact be experiencing internal restructuring which ultimately becomes visible as changed proficiency.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
. C D S T
.................................................................................................................................. As we have seen, CDST emphasizes the interconnectedness of language subsystems within the overall system, and their context-dependence, the perpetually changing, variable, and emergent nature of these (sub-)systems, as well as the non-linearity and basic non-predictability of developmental processes. These properties present a considerable problem for theory and empirical research in language acquisition and attrition since they run counter to some long-standing traditions of doing research. These, in their quest for generalizability, have tended to ignore individual variability, traditionally seen as ‘noise’ (Thelen & Smith, ; de Bot et al., a; Cameron & Larsen-Freeman, ), and resorted to idealization. CDST in particular critiques traditional research designs and conventional statistics. The cornerstone of SLA research has been a reliance on (generally) cross-sectional investigations comparing (ideally large) groups of participants, while the statistical procedures applied to data serve to neutralize differences between individuals through averaging. However, as Larsen-Freeman (e.g., , , ) and others have pointed out, group averages do not allow researchers to make inferences about individuals and therefore have limited value in understanding actual language development (see also Verspoor & Van Dijk, ; Lowie & Verspoor, ). Moreover, most commonly used statistical tests rely on problematic assumptions concerning the properties of samples and data (normal distribution, independence, homoscedasticity) as well as the nature of the relations between variables (i.e., linear and potentially causal). If language is indeed a complex dynamic system, the assumptions of linearity and Gaussian distributions are fundamentally violated (e.g., Larsen-Freeman, ) making the application of conventional statistics inappropriate for many questions (Lowie & Verspoor, ; Larsen-Freeman, ). This does not mean that statistics has to be discarded entirely, since it may be used for questions for which the assumptions hold, but new statistical methods need to be developed that are capable of ‘examining the flexible, transient, dynamic aspects of learner language which emerge from use’ (Larsen-Freeman et al., , p. ). One route has been to improve statistical tests so that they are less affected by violations of assumptions, and to work with data in novel ways. This includes new non-parametric tests (e.g., Anderson, ; Terpstra & Magel, ), Bayesian and Paretian statistics (Andriani & McKelvey, , cited in Larsen-Freeman, , p. ), ‘robust statistics’ (Larson-Hall, ), estimation (e.g., ‘new statistics’; Cumming, ), resampling and simulation methods (e.g., bootstrapping, Monte Carlo), combined with data smoothing techniques (e.g., moving averages). (For overviews and applications in the context of second language development (SLD), see Verspoor, Lowie, & Van Dijk, ; and Verspoor, Lowie, Van Geert, et al., .) Another approach has been a move towards different types of research design and data, in which statistical tests are less relevant (e.g., Larsen-Freeman & Cameron, ). From a CDST perspective, ‘the study of variability as a factor in its own right’ is a paramount requirement (Larsen-Freeman et al., , p. ; Van Dijk et al., ; Verspoor & Van Dijk, ). This, in turn, revalidates the individual or small-scale case study, with change,
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
variability, and dense qualitative sampling at the centre of attention, and calls for longitudinal studies of language development: For [CDST] researchers, this change in perspective has implied a radical change in the way in which language is regarded, which implied a shift in focus from the internal structure of a closed module to change over time of an embedded and embodied self-organizing system. It also required a shift in the empirical approach, from a focus on the grand sweep of factors affecting the language module to more qualitative, fine-grained longitudinal studies. (Larsen-Freeman et al., , p. )
The current challenge lies in implementing these imperatives in empirical work. Studies into SLD conducted from a CDST perspective include Verspoor et al.’s () reanalysis of a longitudinal investigation of an advanced learner’s L negation; Larsen-Freeman’s () study of the writing and oral production of five Chinese learners; Caspi’s (; described in Verspoor & Van Dijk, ) study of four learners’ vocabulary learning; Polat & Kim’s () case study of untutored language development; and Opitz & Smyth’s () study of writing development in ab-initio learners (see also Opitz, ). An impressive example in the area of L motivation are the studies presented in Dörnyei et al.’s () volume. In the area of attrition, no studies have, to my knowledge, been designed in response to CDST requirements as yet, but a number have adopted a dynamic perspective in their analysis and discussion of results. These include two doctoral dissertations (Cherciov, ; Opitz, a), and empirical contributions by de Leeuw, Opitz, and Lubińska (); Opitz (); Cherciov (); and Ecke & Hall () to the aforementioned special issue of the International Journal of Bilingualism (Opitz et al., ). As Larsen-Freeman et al. () argue, ‘[d]ata-driven variability studies’ and computer simulations have already proven to be ‘powerful tools’; however, they are ‘still in their infancy compared to the wealth of research techniques and statistics available to “old school” approaches’ (p. ). Given that longitudinal research is associated with practical disincentives, such as the time it is necessary to invest and the possible loss of participants, it is unlikely that traditional ways of doing language research will be eclipsed by CDSTinformed research any time soon. Indeed, despite their criticism of reductionism (LarsenFreeman, ), and contrary to some researchers’ perceptions (e.g., Erdocia, ), the proponents of CDST are not intent on waging war, but argue for the complementarity of both approaches: ‘CT/DST researchers should acknowledge the value of more traditional research and traditional researchers should acknowledge the continuity of human cognition and the emerging complexity of language’ (Larsen-Freeman et al., , p. ).
. C D S T
.................................................................................................................................. The previous subsection has shown that CDST provides a challenge for research design as well as analysis. In view of the concern of this chapter, it also raises the question of what this perspective means for the study of background variables in first language attrition.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
According to CDST, personal background variables, like other system components, are characterized by non-linear trajectories and interactions, as well as unique initial conditions and specific combinations, which lead to the inherent unpredictability of outcomes for the language system. Therefore, viewing variables in isolation and treating them as ‘predictors’ becomes theoretically untenable. Byrne (, cited in Larsen-Freeman & Cameron, ) contends: ‘Let us understand clearly, once and for all, that variables don’t exist. They are not real’, thus declaring ‘death to the variable’ (p. ). The inadequacy of conventional approaches based on the identification of a single factor, such as age of onset or maturational constraints, to determine the outcome of language development within a multilingual system is due to the questionable assumption that single factors can be identified as the determining ones, that they can function as predictors of language development and that the growth and decay of languages within a multilingual system can be meaningfully studied in isolation (Herdina & Jessner, , p. ). In other words, adopting a CDST perspective to attrition requires a fundamental rethinking of the very notion of predictor variables. The problem is compounded by the ‘statistically possible’ (Van Geert & Steenbeck, , p. , cited in de Bot & Larsen-Freeman, , p. ), but theoretically untenable separation of learners (and their variables) from context (the ‘boundary problem’; Larsen-Freeman, , p. ; see also Cameron & Larsen-Freeman, ; King, ). Since every dynamic system is constantly interacting and interfacing with others and with the environment organically, the notions of ‘inside’ a system and ‘outside’ a system are never simple or uncontested. (Cilliers, , p. , cited in Larsen-Freeman et al., , p. )
A further practical problem arises from the fact that, despite the efforts and assurances of the traditional scientific method, it is impossible to anticipate all factors and collect complete information (de Bot & Larsen-Freeman, ). This is not to say that CDST is blind to the operation of factors; on the contrary, the study of the impact of ‘language-internal and language-external determinants’ is very much part of the changed research agenda, which is afforded by the methodologies outlined above (Larsen-Freeman et al., , p. ). However, these variables are not static, but emergent from interaction, with some acting as ‘precursors (features whose development is a necessary condition for the development of others)’, and others as ‘dependents (features that rely on precursors)’ (p. ). In other words, these variables cannot be divorced from the learner or the developmental process (Larsen-Freeman, , p. ). Far from simply ‘individualizing’ a universal acquisition/attrition process, they are constituent components of an essentially situated process. If language emerges from and in the intentional interaction with one’s environment, this perspective implies that, of the various variables listed in the first subsection, language use/ amount of contact is seen as a crucial variable (Herdina & Jessner, ; de Bot, Lowie, & Verspoor, a,b; Ellis, ). Larsen-Freeman & Cameron (a) argue that ‘a language at any point in time is the way it is because of the way it has been used’ and ‘would go so far as saying that every use of language changes it in some way’ (p. ). Thus, changes in natural input and opportunities to practise one’s languages should be reflected in the degree of acquisition/attrition experienced by individuals, albeit not in any deterministic, linearly causal way.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Moreover, language use is intuitively a prime example of a dynamic factor which interacts with multiple other internal and external resources and variables (all also potentially interacting with one another), such as age, language aptitude, memory capacity, and motivation on the one hand, and space and invested time on the other (de Bot & LarsenFreeman, , pp. –). These resources, modelled jointly as ‘language maintenance effort’ (LME) and ‘language acquisition effort’ (LAE) in DMM (Herdina & Jessner, , pp. and ), are required by the language system for development, that is growth or continued stability (de Bot et al., , p. ), and insufficient resources ‘will eventually result in language attrition’ (Herdina & Jessner, , p. ). They are also seen as potentially limited, and there may be competing demands from different parts (e.g., different languages) of a multilingual system for them, thus constraining potential development. To date, the importance of language use for attrition has more often been asserted than tested (Schmid, ). Empirical studies (e.g., Schmid, , b; Opitz, ) using traditional statistical means have found that language use is very hard to operationalize and measure; it is a multi-faceted factor in that it comprises active use and passive exposure, more or less personally relevant use, use in monolingual and bilingual modes, etc., which certainly has some part to play in the discovered effect being very moderate. It is as a result of the interplay between variables/resources that the tension between the inherent unpredictability of the language system, and the patterns discernible in language development may be reconciled. Larsen-Freeman () posed an important question and concern in CDST endeavours: ‘If we reject simple linear causality, do we have to give up on generalizability?’ (p. ). Given the disjuncture between individual trajectories and group averages highlighted above, and the fact that an exclusive focus on the particular is not very satisfactory, she offers three potential solutions: generalizing at system level, generalizing to theory, and finally, generalizing to potential (pp. –). With respect to the latter, she has suggested it may be possible to identify particular configurations of systems, ‘archetypes’ with ‘signature dynamics’ (Larsen-Freeman, ). Taking up this suggestion, a preliminary investigation of attrition profiles based on a comprehensive set of predictor variables found limited evidence for patterns in the attrition outcomes, amidst great variability between participants and counterintuitive patterning of variables for some participants (Opitz, a, b; see also Opitz, , for a first engagement with profiles based on a subset of variables). Thus, frequent L use was often associated with better L maintenance, knowledge of more foreign languages was associated with higher L attainment, the male participants on the whole were outperformed by the females; however, a number of individuals with similar profiles showed a divergent pattern for each of these variables. It appears that personal background variables do ‘conspire’ in the emergence of particular trajectories and profiles, but these outcomes are in no way predetermined.
. S
.................................................................................................................................. The present exposition of the relevance of CDST to attrition research has argued that the current stalemate in studying the impact of personal background variables, amongst others, can plausibly be connected to the nature of language as a complex dynamic system, and the
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
associated properties of the so-called predictor variables. This changed perspective has recently generated considerable interest, though the uptake has, perhaps tellingly, not happened in a linear fashion (Larsen-Freeman, ). Seen by some as potentially paradigm-shifting (Larsen-Freeman & Cameron, ; Verspoor, de Bot, & Lowie, ; Larsen-Freeman, ) or even ‘the quintessential future approach to human action, cognition and behavior, including language’ (Van Geert, , p. ), it has focused our attention on ‘developmental processes in which acquisition and attrition of all of a speaker’s languages are logically merged into the developing language system’ (Larsen-Freeman et al., , p. ). CDST has introduced a very necessary shift away from the one-size-fits-all endeavour to establish general, if not universal, dependencies between variables and outcomes, to a focus on systemic individual variability, a predisposition to look for connections and appreciating diversity, crucially restoring dynamism to language development (Larsen-Freeman et al., , p. ). This, in turn, has opened up many challenges and questions, be they theoretical, methodological, or analytical, and has set in train a process of expanding the tool kit of acquisition and attrition research. However, while it is to be anticipated that CT/DST approaches to language development will contribute significantly to the methodology of applied linguistics in the future, the research techniques available to investigate change over time are recent and have not yet been used to their full potential. (Larsen-Freeman et al., , p. )
CDST, despite the optimism of its proponents concerning CDST’s potential to transform research into language development, has also drawn a certain amount of criticism. Some researchers believe it is a fashion or bandwagon, and indeed there is a danger of notionally adopting the terminology of complex dynamic systems without truly engaging with CDST’s methodological and analytical imperatives. Another problem is connected to the fact that empirical evidence for language as a complex dynamic system is still fairly limited, and to the question of how research findings should relate to theory (the chicken-and-egg problem). While CDST grew out of theoretical positions that did not seem to match the available empirical evidence, and the inconclusiveness of many empirical findings, researchers nevertheless need to be mindful not to treat their findings as a priori evidence, but to consider alternative explanations. A final point of criticism concerns CDST’s status as theory, with many rejecting this notion, and others questioning the very applicability of a theory originating in the physical sciences to the study of humans (Al-Hoorie, , cited in Larsen-Freeman, , p. ). Indeed, Larsen-Freeman () writes: ‘In some ways, the fact that “theory” is in the name of Complex Dynamic Systems Theory (CDST) is unfortunate’ (p. ), and goes on to explain that CDST has more far-reaching consequences than would normally be expected with a new theory. Larsen-Freeman et al. () also note that the word ‘theory’ in CT/DST does not refer to one grand unified theory but, rather, represents a range of theoretical insights that call into question some of the assumptions that have been used to inform and to structure research for many years. (p. )
As Erdocia () noted, there are different uses of CDST in language development, ranging from ‘soft approaches’, or metaphorical use, through the notion of CDST as a
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
metatheory (‘medium approach’), to ‘hard’ and ‘radical approaches’, which respectively propose or accept language as a complex dynamic system and adapt their methodology accordingly. Larsen-Freeman () acknowledges the diverse uses of the concept (paradigm, metatheory, theoretical framework), preferring to see CDST as a transdisciplinary metatheory (Larsen-Freeman, , p. –), which is the view also adopted in this chapter. As such, CDST is particularly compatible with usage-based, emergentist, and constructivist approaches to language acquisition and attrition (Holland, ; de Bot & LarsenFreeman, ; Larsen-Freeman, ), while critiquing deterministic theories, as discussed above, and, in its rejection of the tenet of linear causality, ‘undoing’ the very notion of ‘good theory’ as one that ‘describes, explains and predicts’ (de Bot & Larsen-Freeman, , p. ). It is an understandable (and important) desire for humans to strive for simplification and objectivity (Larsen-Freeman et al., , p. ); nevertheless, this desire needs to be balanced with the need to take into account prevailing complexities. As this chapter has shown, the critique appears justified where the study of personal background variables in attrition is concerned: so far the notion of ‘predictor variables’ being in fact predictive has not been borne out in empirical research, and the available evidence seems to point to non-linear dynamic interactions and complex relationships between variables and other components. These findings, in turn, can be better accounted for from a CDST perspective.
A I would like to thank the editors Monika Schmid and Barbara Köpke, as well as Mirela Cherciov for their very constructive feedback on earlier versions of this chapter, and for their hard work in bringing this volume about.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
.............................................................................................................
PSYCHOLINGUISTIC AND NEUROLINGUISTIC APPROACHES TO LANGUAGE ATTRITION ............................................................................................................. Edited by
ö
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
......................................................................................................................
......................................................................................................................
ö
. I
.................................................................................................................................. T field of language attrition has long been dominated by linguistic, applied linguistic, and sociolinguistic approaches, and psycho- or even neurolinguistic approaches have had some difficulties in becoming established. However, the idea that language attrition is governed by processes and mechanisms that are fundamental in the human mind/brain has steadily become a pillar underlying most attrition research. One of the first questions to be raised from a psycholinguistic perspective was whether attrition is a competence or performance issue. In other words, does attrition affect linguistic ‘knowledge’ or rather ‘control of that knowledge’ (Bialystok & Sharwood Smith, ) or ‘online processing of knowledge’ (Sharwood Smith & Van Buren, )? While our present conception of competence and performance has radically changed since the early publications, one of the major discoveries of language attrition research has certainly been that the answer to this question seems to depend considerably on the age at which a modification of language use arises (commonly referred to as the ‘age of acquisition of the second language (L)’ or AoA, see also Bylund, Chapter , this volume). One of the most consistent findings in language attrition research is that there may be profound changes in the linguistic system of a speaker who reduces the use of a language as a child (i.e., pre-puberty migrants) or even stops using it altogether (e.g., in adoptees, see Pierce, Genesee, & Klein, Chapter , this volume), whereas in migrants who have grown up mostly monolingually (i.e., post-puberty migrants), language attrition, at least as far as the first
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
ö
language (L) is concerned, manifests itself mainly at the level of online processing of the L (see Köpke & Schmid, a; Schmid, ; Schmid & Köpke, a; Bylund, Chapter , this volume). Most of the early attrition studies were mainly interested in how attrition at the competence level affected the linguistic system, and attrition phenomena were examined mostly in order to test the predictions of different linguistic theories with respect to the vulnerability of specific structures. In a similar way, attempts have been made to compare attrition to other processes of language change and language development, as was proposed by de Bot & Weltens () in a seminal paper that invited researchers to consider attrition as one of numerous processes of language development through the lens of either the recapitulation or the regression hypothesis. Attrition studies based on the regression hypothesis, which in its original form stipulated mirror symmetries between child language acquisition and language loss in aphasia (Jakobson, ; see also Barkat-Defradas et al., Chapter , this volume) have yielded interesting findings with respect to the vulnerability of different linguistic levels at different times in non-pathological attrition (e.g., Keijzer, a). But however interesting and valuable these approaches are, documenting mirror symmetries between child language, attrited systems and/or pathological language loss itself does not have explanatory power over why these similarities occur. Hypotheses like regression have to be placed in a psycholinguistic framework taking into account cognitive factors such as competition in order to gain explanatory power. Research into cognitive processes and brain mechanisms underlying attrition has only slowly emerged. It was not before , more than twenty years after the first publications on language attrition, that a collection of papers focusing specifically on neurolinguistic aspects of attrition was published (Köpke, a). However, recent evolutions in theories of language development (e.g., Dynamic Systems Theory (DST); see also Opitz, Chapter , this volume) and intensification of research on bilingual language processing as an overarching field, to which attrition also belongs, reflect the current interest into psycho- and neurolinguistic processes and mechanisms involved in bilingualism, including language attrition. The aim of the present section is to provide the reader with an overview of research that takes into account the neurofunctional and cognitive bases of language and how they inform attrition research. Before presenting an outlook on the different chapters assembled in this section, we will outline the main concepts and hypotheses on the basis of psycho- and neurolinguistic studies of attrition. More specifically, we will present research considering language attrition either as a matter of language processing, of memory, or of brain mechanisms. While many of these issues still remain speculative, we will discuss what they may entail for attrition and how attrition data in turn could help to better understand bilingual language processing and brain mechanisms.
. L
.................................................................................................................................. The interaction of different languages during online processing is a central issue in research on bilingualism. The concept of cross-linguistic influence (CLI) was introduced by Sharwood Smith in , with the intention of replacing the notion of transfer—frequently
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
used in second language acquisition (SLA) research—by one with a less behaviourist connotation. While Sharwood Smith stated explicitly that CLI should be taken to refer to both influence of the L on the L and vice versa, the latter aspect has often been ignored in practice (see the discussion in Schmid & Köpke, a). But most importantly, the concept of CLI was meant to go well beyond the idea of interference between two language systems: CLI happens because different languages are processed through the same mechanisms, which leaves space for interactions of different types. Attrition, in this conception, is a direct product of the concomitant processing of two languages (Sharwood Smith, ). This idea was also developed in Green’s activation, resource, and control model. This model of bilingual language processing was very innovative at that time with respect to several aspects, two of which are particularly interesting for attrition. First of all, the model distinguished between selected, active, and dormant languages. The selected language is the one controlling ongoing verbal activity and the active language plays a role in verbal activity even though it is not selected. That languages which are not selected but are active may intervene in the processing of the selected language has, since Green’s model, been amply demonstrated for virtually every linguistic level (see Schmid & Köpke, a, for an overview). But what might have been the most interesting concept for attrition research was the idea that there may be dormant languages that are stored in long-term memory but do not intervene currently in ongoing language processing. Although Green’s model thus directly pertains to attrition, it has hardly ever been implemented in attrition research (but see Ammerlaan, ). It must, however, also be noted that no direct evidence has yet been found indicating that there might be a difference between active languages that do intervene in ongoing language processing and dormant languages that do not. In fact, most current bilingual processing models point to non-selectivity of all language systems regardless of their status (cf. Dijkstra & Van Heuven, ). The non-selectivity aspect was, however, also incorporated in another construct underlying Green’s () model, and has crucially shaped attrition research: the idea that there is constant competition between a speaker’s languages. Indeed, the idea of competition between languages clearly provided grist to the mills of attrition research and drew attention to notions such as activation and inhibition, both of which are now seen as pillars in bilingualism research and are perhaps most clearly attested in work on lexical access, for example in lexical decision tasks (for a review, see Van Assche et al., ). Green’s model and his later Inhibitory Control (IC) model (, see below) paved the way in this respect. The concept of activation in particular was further developed in the Activation Threshold Hypothesis (ATH) proposed by Paradis (e.g., , ), initially mostly to explain nonparallel language recovery patterns in bilingual aphasia. Based on the hypothesis that the two languages of a bilingual are organized in two neuro-functionally independent subsystems within the language system, the ATH holds that interaction between the languages is largely determined by the frequency of activation of each language and each item that is part of the language, and by the time elapsed since its last activation. Following Paradis (e.g., , p. ), the Activation Threshold (AT) of a given item or language is lowered each time the item (or the language) is activated. Lowering of the AT enhances facility of activation and lowers cognitive costs associated with activation. However, if the item (or the language) is not activated again, the AT will slowly be raised, and the item (or the language) will become more and more difficult (and costly) to activate. This framework clearly states that facility of activation is directly linked to frequency of use, and as such the
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
ö
ATH is of specific interest for attrition research, where one language is markedly less used in a speaker. The first study to test the predictions of the ATH with respect to first language attrition was Köpke (), who investigated two groups of thirty German migrants in France and in Anglophone Canada respectively. This study’s main focus was the application of the ATH for lexical versus grammatical processing. Köpke’s findings suggested that grammar is not affected by frequency of activation to the same degree as the lexicon and these results were also discussed in relation to the related declarative/procedural model (Paradis, ; see Section . below for a description of the model). Gürel (a), on the other hand, focused precisely on grammatical processing in a study of the binding strategies for overt and null pronouns in L Turkish of twenty-four Turkish-English bilinguals. Her findings showed, in line with her hypothesis, that attrition affected grammatical processing in a selective manner; following the ATH, Gürel’s prediction was that competition due to differences in frequency of use can only arise when there is a competing structure in the L. In the absence of such similarities, competition will not arise. This was borne out by Gürel’s data: Turkish pronouns that have a correspondence in English tended to be processed as in English, while Turkish pronouns without an English equivalent did not show signs of L influence or attrition. Later, Schmitt () investigated the prediction of the ATH using data from five early Russian-English bilingual speakers who migrated to the US between the ages of and years and were adults at the time of testing. Spontaneous speech data collected in informal sessions were analysed both from a language contact perspective inspired by models proposed by Myers-Scotton (e.g., ) and using the perspective of the ATH. Schmitt concluded that while theories on language contact were informative in predicting the vulnerability of different types of grammatical morphemes, the ATH allowed her to account for frequency effects in lexical items. This points to the possibility that competition may arise in a more generalized manner between lexical items, whereas, in the grammatical domain, competition remains restricted to very specific conditions, as demonstrated by Gürel (see also Dussias et al., Chapter , this volume). What is furthermore interesting in Schmitt’s findings is the implicit conclusion that attrition phenomena are hard to capture in one single framework. Rather, integrated combinations of different frameworks are needed in order to explain the complexity of attrition. The straightforward relationship between frequency of use and attrition, as suggested by the ATH, has been seriously questioned in recent years (see Schmid, Chapter , this volume, for an in-depth discussion). Instead, growing attention has been paid to competition in addition to frequency, as suggested by Gürel (a). Competition is, of course, not only operating in attrition, but is a force at work in all bilinguals through the parallel activation of the two languages at nearly all linguistic levels (for recent reviews, see Kroll, Dussias, et al., ; or Schmid & Köpke, a). This is why neurolinguistic models of bilingual language processing—with at their foundation the influential IC model (Green, )—focus on the resolution of competition as the main specificity of bilingual language processing. But the precise conditions for the occurrence of competition still need to be specified more precisely. A comparison of the results from recent studies on bilingualism and attrition suggests that attrition is not qualitatively different from other manifestations of interaction between the languages in bilinguals. Rather, where ‘normal’ competition between the bilingual’s
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
languages ends and attrition starts seems merely a matter of degree (e.g., Schmid & Köpke, b; Köpke, ). Indeed, that attrition is a phenomenon inherently tied to processing mechanisms is an option further developed in this volume by Sharwood Smith, Linck & Kroll, and Dussias et al. (Chapters to , respectively), for instance. Whether crosslinguistic interaction in language processing may, ultimately, lead to a restructuring of linguistic competence remains an open question. What is clear, however, is that such a question cannot be resolved without taking into account the memory structures that are supposed to underlie such linguistic competence.
. L
.................................................................................................................................. Extending language processing accounts, cognitive scientists have investigated how the human brain deals with incoming information: they have explored how we learn, process, and how and where new information is retained, including what we know about words, and grammatical rules. But whereas a substantial body of work has detailed how languages are learned and thus how new linguistic memories are formed, notably less attention has been paid to the other side of the coin: how languages are unlearned. Ecke () explored how established psychological theories of forgetting (such as distortion, decay, and cue dependency) can be used to explain linguistic attrition (see Schmid, Chapter , this volume). But attrition research can in turn also aid the understanding of the link between memory systems and language representation and processing. Specifically, the declarative/procedural (D/P) memory model of long-term memory has frequently been invoked in an attrition context (Paradis, ; Ullman, a). Ullman (a, p. ) defines declarative memory as being involved in the learning and use of factual knowledge (i.e., facts and events) that can typically be verbalized, so this knowledge is explicit. Procedural memory, by contrast, is involved in the learning and use—or rather control—of cognitive and motor skills that are sequenced in nature. This type of memory is typically automatized and therefore mostly implicit. In relation to language, lexical knowledge is typically labelled declarative, with grammar being assigned to procedural knowledge, but—interestingly in the context of attrition—there is a difference between earlier versus later learned languages in bilingual settings: whereas lexical knowledge is declarative in both the L and the L, grammar knowledge of the first language is mostly automatized and so procedural while L grammar mostly remains declarative (Clahsen & Felser, ; Köpke, ). It has even been claimed that declarative knowledge can never become procedural, not even in very advanced L learners (Paradis, ). This, of course, has consequences for language attrition. It has been hypothesized that especially in postpuberty migrants not much grammatical interference from the L to the L will be witnessed, as both rely on distinct memory systems, but that L and L vocabulary items are likely to show considerable interference (Köpke, ), a claim that has been substantiated by empirical attrition research. Attrition itself can also feed back into theorizing within the D/P model: How does declarative versus procedural knowledge interact in attriters? Is it possible that procedural grammatical knowledge becomes weakened in
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
ö
attriters to the point where declarative knowledge takes over? And what role does age at emigration play in this? This taps into the evidence presented above that post-puberty migrants mostly show language processing difficulties in their L but no changes in their linguistic repertoires per se. Earlier work on attrition in the context of memory systems has tended to rely on memory metaphors of misplacing language structures or words, ranging from file cabinets to storage bins (cf. Ecke, ). Discussions regarding the D/P model, too, presuppose different storage facilities, although this has never been empirically substantiated. While neuropsychological evidence does point to different neural substrates subserving procedural and declarative memory, the question of storage is not without controversy. Underlyingly, all memory metaphors view the mind as a processor, much like a computer, where files can be stored, replaced, and deleted. This idea has recently been completely rejected, framed by a Radical Embodied Cognition framework (cf. Chemero, ). Instead, changes in language skills within that framework are best captured as stemming from a direct interaction between organisms and the world they live in: depending on a speaker’s environmental input, language learning, unlearning, or even relearning occurs (Keijzer & de Bot, ). Although language use and environment are generally seen as important external predictors of attrition (cf. Köpke, ; Schmid, a), they are not commonly viewed as inherent determiners of the attrition process. But from a Radical Embodied Cognition perspective, languages (words and structures) are never truly lost but are rather associated with or embedded in a given context and can be recalled by triggers pertaining to that environment. There is empirical attrition work to substantiate this claim, most iconically work that has been done on the savings paradigm. First introduced by MacLeod (), the savings account makes predictions about which type of linguistic information is retained after periods of non-use of newly learned words. More specifically, MacLeod found meaning to be retained to the detriment of word forms. In other words, semantic information rather than phonological information tends to be ‘saved’. Contrary to other memory metaphors, the savings paradigm works on the basis of input and associations. It was first applied to language attrition research by de Bot & Stoessel (), who investigated the retention of L Dutch after years of non-use by two German adults who had spoken Dutch for decades following a move to the Netherlands but at the time of testing had not used the language for a long time. Crucially, de Bot & Stoessel also added a relearning task and found that the experimental participants scored significantly better on the relearning task than a control group who learned Dutch for the first time. This and a series of studies following it (e.g., Hansen et al., ; Schneider et al., ) not only confirmed that residual (especially lexical) L linguistic knowledge is present in memory, but that it can be retriggered based on input. In other words, associations are reactivated upon changing input patterns. Retrieving linguistic information in memory in the context of attrition is, of course, also closely linked to the concepts of activation and inhibition outlined above. The theory of retrieval induced forgetting (RIF; cf. Anderson & Spellman, ), basing language retrieval and failure on associations and input, combines both perspectives. It has its roots in the cognitive psychological theory of catastrophic interference (McClosky & Cohen, ) and has relatively recently come to be applied to attrition (see also Larson-Hall and Linck & Kroll, Chapters and , this volume, respectively). Under this theory, it is assumed that when one retrieves a piece of information, related knowledge is inhibited to avoid
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
interference and—with time—such local inhibition can spread to global inhibition of a language system. A classic RIF paradigm consists of three phases: familiarization, practice, and testing. Levy et al. () looked at attrition data invoking RIF. Their goal was to test the theory’s prediction with a sample of L American students taking a semester of Spanish at college. They found that these students became progressively slower to name an L English item after the same item had been named ten times in Spanish. Interestingly, when the word prompt was semantic in nature (e.g., naming the word snake upon seeing venom and cued with the initial word letter s . . . ), only a facilitation effect occurred; the RIF effect instead characterized phonological prompts (i.e., presenting a prompt that rhymed with the target word; the same word snake but this time with the phonological prompt break— s . . . ). Although the numerically small effect of this study has been critiqued (Runnqvist & Costa, ), this research has offered an important step in understanding the nature of forgetting linguistic information and most notably the nature of language storage and retrieval without having to resort to memory metaphors. In terms of theorizing, the (ir)reversible nature of forgetting can be examined more closely: within the RIF framework, L will be reactivated upon more L input. Combining insights from the savings account and RIF, it is evidently crucial to not only examine language decline that results from inhibition, but also to relate this to how that language developed in the first place and also whether it can be relearned and the circumstances that facilitate such relearning, or reactivation.
. L
.................................................................................................................................. In recent years, the progress made in neurosciences has allowed us to reconsider a number of ideas about language and the brain that had been taken for granted for decades. While notions based on neurophysiological foundations, such as activation and inhibition for instance, have underpinned psycholinguistic research on attrition from the beginning, current insights provided by the neurosciences have hardly been taken into account yet in applied linguistics and specifically in research on language attrition. Research with monolinguals has allowed the development of very detailed models of the neural substrate and the time course of language processing (e.g., Hickok & Poeppel, ; Indefrey, ; Friederici, ; Price, ; Hickok, ). However, these are not, or only marginally, referred to in research on the processing of multiple languages in the bi- or multilingual speaker. Similarly, growing evidence from the neurosciences strongly supports the idea of neural multifunctionality, defined as the constant and dynamic interaction of nonlinguistic functions—that is, cognitive, affective and praxic activity, with neural networks specialized for language (Cahana-Amitay & Albert, ). Such multifunctionality has specifically been investigated with respect to the role of executive functions in language processing in speakers with and without language disorders and these functions have been shown to not only interact closely in cognition, but also to rely largely on common neural networks that are shared by language processing and executive functions (Zappalà et al., ). Despite the extensive debate on executive function benefits in bilinguals
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
ö
(e.g., Kroll & Bialystok, ), little to no reference is made to such insights in research on bilingualism or attrition. The most important concept in this context is that of neural plasticity. Recent research in neuro-rehabilitation, for instance, has challenged the notion of brain plasticity, which in turn questions the foundation of the critical period hypothesis (or related frameworks to account for age effects in language development). Evidence from neuroscience clearly demonstrates that brain pathways are adapted when new skills are learned and that rapid functional and structural changes may occur not only in children but also in adults. Studies have shown that increased grey-matter density, associated with skilled behaviour, can be measured after short periods of training in people learning to juggle (Draganski et al., ; Driemeyer et al., ), receiving training for working memory (Takeuchi et al., ) or music (Osterhout et al., ). But most importantly for our purpose, structural changes attributed to neural plasticity are also observed when people are involved in learning new speech sounds (Golestani et al., ) or exposed to a new language (Mecelli et al., ; White et al., ). Findings from Osterhout et al. () suggest that Event-Related Potentials (ERPs) allow us to see changes arising as a function of classroom L instruction in the brain’s electrical activity. Recent findings based on analyses of functional and effective connectivity in brain imaging data suggest that it is also possible to capture successful L learning using such techniques (Li & Grant, ). As also suggested by these neurofunctional data, the early stages of learning a new task are critical and involve the greatest changes, a challenging finding for our conception of attrition, generally based on the idea that attrition is a long term process. It is not known yet whether such neurofunctional impact on the L also has consequences on the existing L at these early stages of L immersion. One can only speculate what the consequences of attrition on brain structure might be since studies on brain plasticity have for the moment focused exclusively on the skills to be learned (e.g., language acquisition) and not on the reorganization of previously mastered skills (e.g., L processing). The picture is probably a complex one since it has also been suggested that expertise in multilingual language and control functions may reduce grey-matter density (Elmer, Hänggi, & Jäncke, ). For the moment, thus, we have only behavioural evidence for (temporary?) changes in processing the L during the first stages of learning a new language (e.g., Chang, ; Chapter , this volume) or in enhanced immersion contexts (Dussias & Sagarra, ; Dussias et al., Chapter , this volume), in the future, this should be explored with neurophysiological and imaging data, as the contributions to this section suggest. These new insights about neural plasticity underscore the huge capacity of adaptation and flexibility of the human mind. These appear also very clearly in studies on attrition. Attrition studies have already contributed significantly to major questions in SLA (e.g., the age factor). But much remains to be done and we believe that attrition research is in a unique position to shed light on bilingual theorizing. However, in order to do so, substantial work still needs to be done on psycho- and specifically neurolinguistic aspects of attrition: Attrition has to take into account insights from the neurosciences, but models of the neural networks of language also need to be able to account for the specification of bilingual language processing and development as in attrition. The links between attrition, as non-pathological language loss, and pathological language loss and disorders have to be elucidated. This is probably most urgent with respect to developmental language disorders where diagnosis of bilingual children is still difficult because we do not know enough on
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
bilingual language development and attrition in young children (see Blom, Boerma, & de Jong, Chapter , this volume). Finally, the debate on cognitive advantages in bilinguals should also take into account attrition and explain how the notion of cognitive advantages gained through the continuous need to inhibit one language is compatible with the observation that skilled bilinguals are not always able to inhibit the stronger, interfering language.
. S
.................................................................................................................................. Adopting different psycholinguistic and neurolinguistic frameworks, the contributions in the present section not only provide a state of the art of psycho-/neurolinguistic approaches to attrition, but also uniquely show how attrition data can be used to inform psycholinguistic and neurolinguistics theories of language representation and processing. Chapters – explore the origins of attrition within the specificities of more general bilingual language processing. Sharwood Smith (Chapter ) reviews how language processing theories have been invoked in acquisition and attrition research. Treating both as instances of language development allows for a broader discussion of knowledge representations over time with language processing as the main driver underlying such changes. Within this framework, attrition is treated as a manifestation of general cognitive processing limitations stemming from competition between two language systems. Also departing from bilingual language control more generally, Linck & Kroll (Chapter ) posit that attrition is not language loss per se but rather reflects the dynamic outcomes of two constantly interacting language systems in flux. Focusing on the attrition of lexical retrieval, they build on the foundations of retrieval induced forgetting as a general memory mechanism that can have explanatory value regarding attrition. In this process, they also present more details regarding the constructs of inhibition and activation to show that even after very brief exposure times to the L, the L can be impacted. In Chapter , Dussias et al. elaborate on this hypothesis, but do so in the context of morpho-syntactic processing. Their perspective can be pivotal in elucidating how the construct of competition differentially impacts the lexicon versus the grammatical system, as discussed above. For the morphosyntactic domain too, Dussias et al. do not find evidence for true loss occurring as a function of L immersion, but rather that attrition is shaped by language processing changes. In this process, the authors show the usefulness of eyetracking as a device to show even fine-grained language processing changes in attriters. The fact that attrition is governed by changes in bilingual language processing is further taken up by Chapters –, which collectively address the links between pathological language dysfunction, which we know is governed by general cognitive abnormalities, and attrition. Blom, Boerma, and de Jong (Chapter ) explore how the complementing nature of attrition and bilingual processing literature can help elucidate the diagnosis of bilingual children whose language (learning) abilities raise concern. Blom et al. discuss the difficulties in ascribing the deviances in language abilities to either bilingual processing or Developmental Language Disorders (DLD). As the children concerned often have a migrant language background, insights from attrition could help in reaching a diagnosis. Higby, Lerman, Korytkowska, Malcolm, and Obler in their chapter (Chapter ) also explore methodological concerns, but in a fundamentally different way: they do not use
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
ö
attrition research to solve methodological issues in bilingual processing but rather point to methodological confounds within attrition research. Moreover, they move from child populations to the other end of the lifespan, namely older adults. Past attrition work has—often incorrectly as Chapters to demonstrate—adopted a minimum of ten years of residence in the L environment as a subject recruitment criterion. Often coupled with a post-puberty age at emigration criterion, this has meant that many attrition findings are based on older adults. In their chapter, Higby et al. describe the problem this poses: what is now ascribed to attrition could be more accurately labelled as ageing effects. Focusing on the lexical domain, the authors show that ageing and attrition exert different influences on lexical retrieval and ageing therefore needs to be taken into account in any attrition endeavour. Staying in the older adult lifespan range, Barkat-Defradas et al. (Chapter ) also emphasize the need to take ageing into consideration, when they suggest to compare non-pathological language attrition with language changes in late-life acquired disorders, most notably Alzheimer’s Disease (AD). It is especially the interaction between memory and language processing, which has been amply studied in AD studies, that may lead to important insights for attrition research, the authors claim. The chapters in this section present the state of the art of neuro-psycho-linguistic approaches to attrition and pointing to applications of attrition data in bilingual language processing theorizing, but also point to future directions that could further advance the field of attrition. Chapter –, then, introduce two innovative neuroimaging methods to the field of attrition, which until now has relied predominantly on behavioural evidence. Steinhauer and Kasparian (Chapter ) present the ERP technique as particularly suited to advance attrition research due to the method’s power to detect even subtle cognitive processing changes with a considerable level of detail. Importantly, ERPs cannot only detail L lexical processing changes, but can also tap into changes in L morpho-syntactic processing, shedding unique light on the role of interlanguage competition within both domains. Rossi et al. (Chapter ) present another method to chart the neural oscillation changes in language attrition: functional Magnetic Resonance Imaging (f MRI). The method is described in detail as well as its application in studies examining L attrition in international adoptees (see also Pierce, Genesee, & Klein, Chapter , this volume). Bringing the section full circle with the initial Chapters –, Rossi et al. suggest that functional and structural patterns of brain changes tapped through (f )MRI can be used as a test bed for the early stages of attrition, detailing the neural consequences of inhibiting an L.
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
......................................................................................................................
A wider cognitive perspective ......................................................................................................................
. I
.................................................................................................................................. T main aim of this chapter is to set language attrition research within a wider frame of reference. This should enable integrating research carried out so far into explanations of how the mind works in general (see Jackendoff, , for a related discussion). This, at the very least, should give a firmer foundation and greater plausibility to those explanations of attrition that can be aligned with research in other areas of cognitive science. While hypotheses and research findings that are focused exclusively on attrition can contribute greatly to our understanding within a narrow frame of reference, they still leave open to speculation a number of basic questions, especially with regard to processing mechanisms and their role in how language ability changes over time. The title of the chapter might also suggest an exclusive concern with how linguistic representations are processed in real time. This would not really tell the whole story so some elaboration on what is entailed by the ‘wider cognitive perspective’ is necessary. First, although it covers more than just language cognition, the theoretical framework to be used for this particular discussion falls within the realm of more formal approaches to language development. This entails a concern with the nature of linguistic representations themselves, their abstract properties, as well as how they operate during actual performance. Second, in line with much current thinking, attrition is treated as part of a general developmental account that includes acquisition. That said, online processing in real time does take centre stage. This is because changes in linguistic properties, that is, changes in the way in which the mind represents a particular language system, are viewed as being driven by processing mechanisms (see also Sharwood Smith, ; cf. Carroll, ; O’Grady, a,b).
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
More complete accounts and implementations of the framework reported elsewhere make clear the distinction between a theoretical ‘framework’ and a theory within a given subdomain of cognitive science such as language attrition (see Truscott & Sharwood Smith, ; Truscott, a; and Sharwood Smith, , for fuller accounts). The framework might conceivably be treated as a basic model of the mind but it is not a model of attrition per se, although it should test current explanations of attrition for their plausibility within a broader view of how the mind works as a whole and help resolve at least some of the important specific theoretical and methodological issues raised within the field of attrition itself. This chapter will accordingly explore the phenomenon of language attrition using the framework to broaden the explanatory frame of reference using the Modular Cognition Framework or, to use its better known title as applied to language research, the Modular Online Growth and Use of Language (MOGUL) framework. The ensuing discussion will first explain the need for new ways of making sense of the empirical data that has already been amassed so far, other chapters in this volume providing numerous examples. This will be followed by a detailed overview of the framework itself. Only then will it be possible to show how attrition data can be both investigated and explained using the framework, thereby sketching a possible future for attrition research (see also Sharwood Smith, ).
. I :
..................................................................................................................................
.. Representation, change, and the nature of processing Both acquisition and attrition are types of change. Change is a dynamic notion and so mental processing of some kind must be involved, the kind that results in various kinds of alteration of whatever happens to be the current knowledge state. It is important to appreciate not only what has been done so far but also what has not yet been achieved. In particular, the study of processing in psycholinguistics has normally meant looking at the millisecond-by-millisecond online manipulation of knowledge structures that are already part of the language user’s system, such as particular syllables, words, and larger configurations exactly as they have been defined in advance by given descriptive and theoretical linguists. This, in many cases, has been carried out without acknowledging any need for supplementary explanations of how the relevant linguistic properties and relationships might be redescribed within a real time psychological architecture. A case in point would be the exact nature, in psycholinguistic terms, of an ‘interface’ between two linguistic systems like syntax and discourse (to be discussed in Section ..). Another one would be rule ordering where some fixed sequence in the application of a set of grammatical rules in a given abstract theory of grammar may not necessarily reflect any comparable sequence of operations in real time processing. When it comes to looking at how an individual’s representation of a language develops over time either in a direction best termed ‘attrition’ or in a direction best called ‘acquisition’, periods much longer than milliseconds obviously have to be involved.
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
This separate focus on transition as opposed to rapid online processing might also require new ways of thinking about the interaction of processing mechanisms with linguistic representational properties. In sum, developmental linguists, including those who study language attrition, cannot by definition ignore the real time dimension; while they base any of their analyses on theoretical linguistic research, processing issues just have to be important for an understanding of what they have chosen to investigate (see, e.g., Köpke & Schmid, ; Schmid, a, for an overview of attrition research including psycholinguistic studies involving processing issues). This need to take real time processing into account has already been generally recognized and a number of studies have appeared in recent times that at least make explicit reference to the role of processing in making sense of attrition data (e.g., Tsimpli & Sorace, ; Tsimpli, ; Chamorro, Sorace, & Sturt, ). Neurolinguists deal with a quite different level of description, their domain being the physical brain. They too address the same basic questions: how languages are stored, how what is stored becomes accessible during actual performance, what changes over time occur, and what psychological processes can account for this (see, e.g., Köpke, b; Keijzer, ; Kasparian et al., ). Neurolinguists also need to relate their observations of neural activity and changes in the physical brain to patterns of human behaviour. For them, as for developmental psycholinguists, what is needed is some kind of common ground, a well-worked out, usable—that is, sharable—conceptual framework that can assist the mutual exploration of relationships between overt behaviour, brain, and mind (Sharwood Smith, ). Nevertheless, most research into attrition and acquisition has been going on for such a long time focusing solely, and indeed for understandable reasons, on precisely defined and limited hypotheses within an explicitly limited domain. One particular illustration of the lack of conceptual clarity in accounting for changes in an individual’s representation and use of language is the issue of whether developmental data should best be viewed as reflecting discrete stages or gradual continua because data can be analysed to show both (see related discussion in Sharwood Smith & Truscott, ). By using theoretical linguistic frameworks, but also theoretical linguistic concepts and methodologies in order to compare and contrast knowledge states over an extended period of time, it is certainly possible to analyse observed changes and make insightful, linguisticallybased statements about discrete stages of development on that basis. At the same time, only looking at putative stages, however elegant and explicit the analysis, cannot possibly be the whole story. Not everything proceeds in one direction and in clearly defined hops and jumps. The key questions are exactly why and exactly how one stage develops into another, gradually or in discrete steps, and this problem needs more than linguistic theory by itself. It needs theories of acquisition and theories of attrition and possibly a developmental theory that actually encompasses both kinds of change as part of a unified explanation. References to online processing may sometimes be used as a convenient way of accounting for recalcitrant data that do not fit with some current dominant trend as predicted by a given hypothesis. An interesting example of this is the debate in second language (L) acquisition research between two sides: first, the ‘representational deficit’ side claims that older L acquirers constructing the mental grammar of their new language show imperfect access to ‘universal grammar’ that children can rely on to abstract away from imperfections in the input and arrive at a language system with remarkably few grammatical errors. On the other side, the opponents of this representational deficit hypothesis argue that such
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
access is intact in L acquirers and that the apparent failure to produce the relevant grammatical structures in observable performance is actually due to processing difficulty (see, e.g., Hawkins & Liszka, ; Lardiere, , ). This represents a theoretically important difference in the interpretation of the same data: in one case the relevant grammatical principle is no longer available, whereas in the second case the observed absence is misleading and the relevant principle is present in the learner’s mind, but it is processing difficulty that obstructs its appearance in actual performance; closer analysis and other ways of eliciting data should show that there are in fact observable effects in performance that indirectly signal the presence of the presumed missing representation (Juffs, ). The question of what ‘processing difficulty’ precisely entails is left open because there seems to be no way of expressing with any rigour and precision what that concept actually means. Exactly the same holds for attrition: one must again ask whether divergent performance necessarily reflects divergent representations and, if not, how this should be explained (see Sharwood Smith, a, pp. – for an early discussion of this issue). It should be clear by now that research so far has been throwing up a steady flow of unanswered questions, hence the attraction of a framework that combines both acquisition and attrition under one heading of development and makes explicit the nature of developmental processes. This attraction has been with us ever since Jakobson’s intriguing claim, for phonology at least, that stages in attrition might be the same as stages of acquisition but in reverse order (Jakobson, ; see also Olshtain, ; and Barkat-Defradas, Gayraud, Köpke, & Lefebvre, Chapter , this volume). Keijzer () rightly observes that this ‘regression hypothesis’ is not properly explanatory by itself and states that the larger context of linguistic theory is required for explanatory adequacy to be achieved. The claim in this chapter is an extension of this argument, namely that an even larger, ‘mind-wide’ framework is required. Even if full explanations do not immediately fall out by widening the perspective, it should enable a broader and more comprehensive understanding upfront of what has been achieved so far as attrition research is concerned and what still needs to be explained.
.. Interfaces In an attempt to improve the quality of explanations, recent work has concentrated on interfaces as the possible sources of difficulty either in forming new representations in the first place or in accessibility—that is, processing (Tsimpli & Sorace, ; Sorace, ; see also Chamorro & Sorace, Chapter , this volume). An interface is, as the name suggests, a place where two different systems interact. The existence of different systems that may have to collaborate in some way assumes, as is also assumed in this chapter, some type of modularization in the mind: mental modules are systems that have a specific function not shared by others and their own set of principles for constructing representations. Examples include the visual system and the auditory system for example, or—in linguistics—syntax and semantics. The visual system creates visual representations that are structured and organized in ways appropriate for managing visual experience: these ways will not work for other kinds of representations. At the same time, visual experience and other types of experience (or rather those mental structures that have been formed to represent them) often need to be coordinated. Coordination between structures of different types, for example visual and auditory representations, or syntactic and conceptual representations,
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
is managed via interfaces. In other words, however functionally unique, modular systems may be defined by a given preferred theory, and they must in some way be connected by interfaces (see, e.g., Goad & White, ; White, ). Interfaces may not always work in an optimal fashion. For instance, whereas the core or ‘narrow’ syntactic system of an L learner may reach native-speaker levels, the semantic or discourse aspects of related structures may not. This, by hypothesis, is because of an assumed problem at the interface between syntax and other modular systems, the ones that handle semantics and pragmatics: the right associations are not being made by the learner (Tsimpli & Sorace, ). Similarly, the core syntactic system has been claimed to be relatively resistant to attrition effects compared to its interface with semantic and pragmatic systems, these connections being more vulnerable to change and resulting in the inappropriate use of syntactic structures such as the non-nativelike inclusion of overt grammatical subjects in languages where omission of these is permissible and, in some contexts, required or at least overwhelmingly preferred (Tsimpli et al., ). At this point in the whole explanation of interfaces, the picture becomes, for the psycholinguist at least, more than a little vague. The way the interfaces function is typically not specified in any detail and there sometimes, as previously mentioned, appears to be a prevarication between a purely representational account of interfaces and one which involves processing in real time. The representational account might go something like this: the interface between syntax and semantics, or syntax and pragmatics, presents learners with the formidable challenge of isolating the relevant semantic or pragmatic properties from a mass of contextual data made available in the input and then making the right principled associations between those properties and particular syntactic structures. This account focuses more on structural complexity at the levels of representation without making claims about processing complexity, perhaps simply implying that the two types of complexity might be related. This is not a necessary relationship, of course: what is structurally complex may still be easy to process given the existence of appropriate processing mechanisms that can deal effortlessly with such complexity. Particular theoretical commitments are required to determine how exactly any kind of complexity may be defined but certainly ever since Chomsky & Miller’s classic paper that appeared in , the relationship between syntactic complexity and what happens in performance has been generally acknowledged as not being a straightforward one. More recently, just to cite one example of this, Gayraud & Martinie () looked at various kinds of embedded sentences using pause length as an indicator of processing cost, and were not able to equate structural complexity with processing difficulty. This kind of research highlights the need for separate explanations of the processing mechanisms involved, including the nature of working memory (WM), and the nature of processing complexity (see Sorace, , for a more nuanced analysis of complexity and how interfaces might work specifically in a processing context). Sometimes the interface problem is defined as relating to both representation and processing, distinguishing between (a) the principled relationship between structures in one system and structures in another—that is, ‘interface structures’ (see Contreras-García, , for a useful survey of the syntax-semantics interface in different theoretical linguistic perspectives); and (b) the way in which these interface structures are processed online, a recent example of this being Chamorro, Sorace, & Sturt, (), a study also focusing on attrition. The framework used in this chapter at least attempts to provide a degree of coherence and granularity to accounts of processing concepts such WM,
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
competition, inhibition, etc., and how they interact with representational properties and proposed mechanisms such as interfaces without any claim to be a comprehensive theory of processing.
. U , ,
..................................................................................................................................
.. An overview MOGUL is an attempt to resolve issues such as the ones mentioned above (see http:// .../~mogulfra/ for links to more detailed accounts). It permits an integration of independent research into linguistic properties—that is, ‘representations’, grammatical or otherwise—with other psycholinguistic research focusing on real time processes—that is, ‘performance’. In addition, it permits an account of linguistic change (‘acquisition’ or ‘attrition’) that obviates the need for special grammatical learning mechanisms (cf. O’Grady, b). In fact, the framework does even more than that because it encompasses all kinds of cognitive phenomena and in this way allows a simultaneous study of the nonlinguistic forces that shape the way language is used and the way it changes in the individual (see, e.g., its use in studying code-switching in Truscott & Sharwood Smith, ). These claims would be speculative but for the existence of two important characteristics that the framework possesses. First, the inspiration the framework draws depends on established theories and research findings across relevant disciplines within cognitive science as a whole, especially psychology, neuroscience, and linguistics. Second, it is a framework—not a theory in its own right—that transcends any single subdiscipline of cognitive science, unlike, for example, the Minimalist Program in linguistics which, although it has clear implications for cognition in general, is nevertheless designed to direct research within a particular branch of theoretical linguistics (Chomsky, ). The MOGUL framework provides a skeleton model or ‘map’ of how the mind works and how ‘language’ in both its broadest and narrowest definitions operates within the mind: it still crucially depends on research carried out by linguists and researchers in other areas of cognitive science to specify how all the various aspects of this model should be fleshed out (Sharwood Smith & Truscott, ). For example, a full specification of, say, a visual processor or a syntactic processor requires the principles and properties of, respectively, vision and morpho-syntax, to be spelled out in detail. The framework, whatever its inbuilt theoretical commitments such as to the modularity of mind and a particular view of the language faculty, does not do that. The incorporation of, in the case of the above examples, vision or syntactic research into a common framework should facilitate interdisciplinary synergies and collaboration. In any case, the framework does not provide these detailed specifications, this being the job of the researcher with appropriate expertise in these two fields. Current descriptions of MOGUL do in fact use existing theoretical linguistic, psycholinguistic, and language acquisition and attrition research to provide specific implementations of the framework (e.g., Sharwood Smith & Truscott, ; Truscott &
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
Sharwood Smith, ). But although the framework’s architecture is theoretically committed to some extent, there are still many alternative ways to instantiate it in detail: (a) when planning a given investigation or (b) when dealing with existing research findings that require richer, or just different interpretations.
.. Modularity and collaboration: key elements of the framework MOGUL architecture, although inspired by a range of views across cognitive science, owes much to Ray Jackendoff and can certainly be seen as an extrapolation and adaptation of his views although there are some significant differences (see Jackendoff, , , ; Truscott & Sharwood Smith, ; Sharwood Smith, ). Briefly, the basics of the framework are as follows. The mind is composed of expert systems such as the various perceptual systems (vision, audition, etc.), that is, the mind is ‘modular’. The expert systems are specialists, each of them unique in the type of task they carry out. These systems, very generally, reflect the basic organization of the brain as currently conceived. At the same time, it needs to be emphasized that accounts of the brain represent different levels of description and, while exploring relationships between the two is important, the two levels, as implied earlier, should nevertheless be kept distinct, the one abstracting away from physical details, the other not. For example, the functional map of the brain’s perceptual systems might show details of neural pathways and/or neural locations, neither of which need figure in a more abstract display that describes perceptual processing from a psychological perspective (see Sharwood Smith, , p. ). The modules in the mind are not so independent that they exist in a vacuum. On the contrary, they form a network of often highly interconnected systems allowing them to collaborate in many different ways. The fixed, stable, systematic nature of the modules’ internal mechanisms thus contrasts with the flexibility and dynamism of their interactions. In this way a perceptual experience can be built up through an interaction between the various perceptual modules and, for example, modules associating respectively meanings and values to the various representations that have been activated. This collaboration goes on all the time, activating or building ways of representing all kinds of external and internal input, incrementally and in parallel and virtually always without conscious intervention. At this level of abstraction, the modularity of mind and brain and the notion of collaboration are for the most part practically incontrovertible, controversy focusing on the precise nature and number of modules and the ways in which they can collaborate.
.. The developmental perspective In addition to the unification of the two above perspectives, there is a further consideration that is crucial for the study of attrition, namely the developmental perspective: this is the dynamic ‘change’ dimension mentioned above (in Section ..), which covers both attrition and acquisition. As already suggested, psycholinguists typically deal with the way existing knowledge is processed at a given moment but not processes that lead an individual to change over time. Without this third, transitional, perspective, however, language attrition can only be studied by relying on synchronic slices at a given moment in time and
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
comparing these slices with slices taken at a later time. This may lead to some principled speculation about the way change has or has not occurred in the intervening period. However, there is no substitute for describing the processes that actually drive this change. In short, this is why we need a developmental (i.e., transitional) account of how and why representations change over time and a crucial part of that account, in the framework under discussion, is the role of processing. Attrition by Processing Theory (APT) is a reformulation of Acquisition by Processing Theory (Sharwood Smith, ). It is a general principle that characterizes attrition in all kinds of areas including mental arithmetic, piano playing, or knowledge of history, as the outcome of the individual’s processing history. Part of the explanation has to do with the straightforward absence of processing, which has negative effects on the relative accessibility of a given (linguistic or other) cognitive structure (as explained below in Section ..). A language that is not used for an extended period or, say, lexical aspects of that language, will tend, as a direct result of this lack of use, to undergo effects that we call attrition and are experienced as loss or forgetting. However, the way APT as a general cognitive processing principle is actually instantiated will also depend on local intramodular principles for the way it operates within a given modular memory store. Thus, syntactic attrition as manifested in comprehension and production performance will depend not only upon the activation history of an individual’s syntactic representations, complex or otherwise—that is, on a ‘use it or lose it’ principle—but also on the presence of other syntactic representations which compete with them. The connections across interfaces with other system will also have a role to play so that ‘valued’ representations that have strong affective associations may survive a long period of non-use and remain relatively accessible. In this way APT as a general principle is instantiated in the syntax module via local principles and effects: these will be observed, elicited, described, and explained by those working on syntactic attrition and applying the particular theoretical and methodological tools that they favour. More details follow in subsequent sections. It is perhaps worth noting in passing that dynamic change and dynamic effects in online processing might suggest an application of Dynamical Systems Theory (see also Opitz, Chapter , this volume). This possibility certainly exists but only within the constraints posed by the framework’s modular architecture and not as the sole explanans (see discussion in Sharwood Smith & Truscott, ; Sharwood Smith, ).
.. Representations and memory theory If people want to do responsible research in attrition, they must have more than a vague understanding of human memory. They need to commit to some currently plausible view of what it is and how it operates. In the MOGUL framework, the elements—call them features, representations, or simply structures—are located in a store. For the purposes of talking about processing, this store can indeed be viewed as a memory store. There is no common repository where all kinds of cognitive structures are housed. Rather, each expert system—that is—module, has its own dedicated store, for example a visual memory store or a syntactic memory store. The respective module’s processor that manipulates them only deals, and can only deal, with the contents of its own store. Thus the visual processor is only equipped to handle visual representations and the syntactic processor can only deal with syntactic structures. By analogy with, say, JPEG or MP files, the representations of separate stores are written in different, mutually incompatible codes. From this it may be correctly
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
inferred that the basic make up of any module is a processor plus a (memory) store where neither the principles on which the processor operates nor the structures that it manipulates are shared by any other module. There is also no central processor and no central memory facility. Everything works by collaboration between particular combinations of these modules at any given moment. The only sense in which processing comes together as a single unified phenomenon is as conscious experience—recall that the vast majority of the mind’s operations are subconscious—and the framework has its own account of how that happens which will be very briefly touched upon at the end of this chapter (see Sharwood Smith & Truscott, ; Truscott, a,b, for more elaborated discussions of conscious awareness in this framework).
.. Working memory The question that now arises is the nature of so-called long-term memory (LTM) and WM, a shared temporary storage facility used during online processing. It would in principle be possible for there to be a set of separate stores, one for each expert system, but only one shared WM where separate types of structures originally activated in the stores of different mental systems, vision, audition, syntax, the conceptual system, etc., are manipulated all in one place during online processing. Establishing more precisely how WM operates is particularly important since it plays a crucial role in the current explanation of attritional change. In fact, within the framework, there can be no shared WM since, as implicit in the basic design of a module just described, the different types of structure are completely incompatible with one another and can only be handled according to the unique principles of their own dedicated processor.1 Second, WM, in line with Cowan’s (rather than Baddeley’s) ideas is viewed not as a separate buffer, located separately, but as a change of state within a memory store (Baddeley & Hitch, ; Cowan, , ; Baddeley, , ; Truscott, ).2 In other words, syntactic WM, for example, is a change of state of representations specifically within the syntactic memory store alone. The basic set-up of a/ the modular system in the framework is therefore not a tripartite one: Processor + LTM + WM, but rather a bi-partite one: Processor + M, where ‘M’ indicates an undifferentiated memory store. Thus, when a given element is accessed in a given module during online processing, it is activated within its own store and forms part of current WM along with any other member of the store that happens to be activated at the same time. In MOGUL, the metaphor chosen for representing activation is of rising and falling within the store, so the activated element ‘rises up’. This means that the WM state is conceived of as the upper area of the store. There are other possible metaphors but it is just that showing activation as points along a vertical plane is easy to represent graphically. Although there are differences, for example with regard to the treatment of the declarative/procedural dichotomy, Paradis has a similar approach to activation except that he talks in terms of an activation ‘threshold’ (Paradis, ). For the likelihood of something to be selected, a ‘low’ activation threshold is required, which is equivalent to a ‘high’ resting level in MOGUL. In each case the representation in question is effectively in a position that gives it easy access to WM. 1
However, Truscott makes an exception for highly synchronized perceptual WMs which effectively operate as a shared workspace (personal communication; see also Truscott, 2015, p.120). 2 Cowan nonetheless does see WM as a single shared space.
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
.. Resting levels of activation Another crucial related notion for understanding attrition is the resting level of activation. This may be compared to the neural level of description with the notion of ‘resting potential’ in quiescent neurons (see, e.g., Daignan-Fornier & Sagot, ). The absence, especially in the longer term, of activation results in a progressive ‘lowering’ of the structure within the store. This means that, in order to make it into WM and participate in a current act of processing, if and when it gets activated, the structure (representation) has further to rise, again using the same metaphor. A decline in its resting level makes it less accessible. This crucially affects its chances of participating effectively in online processing since one basic feature of processing that most if not everyone assumes is competition with other potential candidates (see, e.g., Dijkstra et al., ; Schwartz & Kroll, ). Both monolinguals and multilinguals become aware of this when they are struggling to produce a word and the ‘wrong’ word pops into their minds, seemingly blocking the one they are sure they do actually know. In contrast to this attrition-inducing scenario, when a structure is activated, and especially when it is activated regularly, it undergoes gains in resting level, making it more, rather than less, accessible and therefore more competitive. For this reason, acquisition is viewed as the lingering effect of processing. With attrition, there are therefore two related forces at work. First there is the gradual decline in the resting levels (and therefore accessibility) of structures that are not frequently activated, or activated enough to maintain their resting level. Second, there is the gradual rise in resting levels (and increase in accessibility) of structures associated with another language that now begin to compete successfully in the scramble for selection in WM, thereby changing the dominance of the languages concerned. In general, there has recently been quite a lot of work on activation and inhibition levels in bilinguals including studies on how even early stages of L learning can lead to (local) attrition effects in the first language (see, e.g., Bice & Kroll, ; Linck & Kroll, Chapter , this volume). Alongside the specific findings of this kind of research, a more structured understanding of the underlying psychological mechanisms such as is provided by a mind-wide cognitive framework ought to contribute substantially to our understanding of the phenomena in question.
. A
..................................................................................................................................
.. Acquisition and growth The further specification of what an ‘interface’ is (see Section .. above) and what interfaces do is seen as a very important requirement for any approach that uses this notion. In the MOGUL framework they are explained as follows: interfaces may be viewed as simple processors that manage associations of structures across adjacent modules. This involves both forming an association in the first place and, once that association is formed and one of the structures is activated, co-activating its matching structure in the other module. ‘Associating’ only means matching and co-activating. No information flows from
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
one module into another. All interfaces work the same way, so they effectively enable collaboration of different systems in the mind and together make the mind into a network of interconnected modules. Where there are only two modules involved, the result is a structural chain of two structures, for example a visual representation paired with an auditory representation, thereby giving a sound a meaning, or a meaning a sound (since processing interfaces work in both directions). The mechanism for identifying structures is an index, which is a shared identity tag, not the kind of index used in classic generative linguistics (see Jackendoff, , p. ). Indices are assigned once the initial association is made: a structure in module A (e.g., an image) and a structure in Module B (e.g., a sound) are then tagged with the same index. In this sense, acquisition, as the initial formation of an association, is instant. However, attrition can also set in quickly if the indexed association is not reactivated sufficiently, which means that acquisition can be both instant and ephemeral. What is far from instant is growth. Growth is the result of changes over time in the resting level of activation of the structures concerned, as described above, and a continual lack of activation will have as its consequence the lowering (and possible extinction) of a structure. Extinction might be empirically established with the increasing sophistication of psychological elicitation techniques, but ultimately, reference to neuroscientific evidence might have to be a way of deciding whether structures that consistently fail to affect performance in any sense, easily observable or more subtly observable, can be treated as ‘actually extinct’ or just ‘very dormant’. An index may be assigned to any structures in any module that has been associated by means of the relevant interfaces, so associations result typically in a network of many structures each of which may have many indices (i.e., cross-modular connections) but amongst them share one index. The conceptual structure TREE and the syntactic structure N(oun) will share an index, for example. Any individual structure like N will, of course, have many different indices, reflecting its multiple associations with given structures in adjacent phonological and conceptual systems.
.. Interactions across interfaces in acquisition and attrition The role of the interface, as simple as it is as a mechanism, is crucial in accounting for how the mind creates complexes of different types of mental structures to make sense of the inner and outer world of an individual. The way the mind’s connections work is, in essence, very simple, with only two ways of linking different structures: first, there are the external connections, associating and co-activating structures written in different codes—that is, across modules; and, second, there are the module-internal associations. Whereas external associations are managed by interfaces which can be seen as very simple processors, these internal associations are managed by the much more complex processor belonging to the module in question. Structures within a module’s store are associated in various ways according the principles unique to that processor and so, for example, visual representations, both simple and complex, will be structured differently from auditory, conceptual, or syntactic ones. Primitive visual elements and particular combinations of these primitives may, when co-activated together, form a representation of some simple cube-shaped object in view, but, as the visual scene is built up further, larger visual structures may be combined
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
to form ever more complex constructions. At the same time whatever is activated inside the visual store will be triggering, via the relevant interfaces, associated complex representations in other stores. Take a simple example of activity within and across two modules during utterance production, involving the stores of respectively the syntactic and conceptual systems and take also the two syntactic structures: Adj(ective) and N. As alternative combinations of structures are acquired by developing sequential bilinguals (L acquirers, heritage speakers), competition between these alternative combinations will inevitably arise in the course of forming a message. On receiving conceptual input consisting of two conceptual structures, say RED and BOAT, the conceptual–syntactic interface will activate both an Adj-N sequence and a competing N-Adj sequence in the syntactic store. This is because two alternative associations across the conceptual–syntactic interface have been created during exposure to two languages. What happens then will depend on the relative strength of the activation of the participating structures. All other things being equal, the dominant language system will win out. For example, the English acquirer of French wishing to talk about a boat that is red will be prompted, by virtue of the dominance of the relevant English syntactic construction, to produce in French, *‘rouge bateau’ (red boat) rather than ‘bateau rouge’ (boat red). The only chance for the French sequence to win the competition is if the resting levels of the N-Adj construction are given a sufficient boost over the longer term by the involvement of other interface connections as will now be illustrated. Apart from the basic driver of change—continued exposure to and processing of French input—it is important to appreciate the important supporting role of the affective—that is ‘value-assigning’ system. For example, where the situational context is French-oriented— French is being spoken—an association with the concept FRENCH in the conceptual system is co-activated; it has probably already been co-activated to some extent because the French construction was originally acquired in a French context with characteristically French sounds and French text, but now, given the immediate French-speaking situation, the conceptual structure (CS) FRENCH is activated together with a very strong positive value: this value is an affective structure (AfS) which is co-activated with the conceptual structure FRENCH via the affective–conceptual interface. This means in very simple terms that French is being treated as ‘currently important’.3 This positive loading will have the effect of giving a boost to the activation level of any other associated structure currently co-activated. This now, at least temporarily, gives the N-Adj construction more of a fighting chance in the current race for selection in WM4 against its normally dominant rival.5
3
In such a situation, the evaluation may be further complicated by such factors as language anxiety reducing the value of French possibly promoting only passive participation in the exchange or the special importance of the conversational exchange in question for some reason increasing the positive value. 4 ‘Selection’ is a metaphor since there is no special selection mechanism per se. As processing builds up a complex representation, incrementally but also with a back-and-forth search for coherence between structures, it is just a matter of what ultimately comes out on top in the competition. In the assembling of competing representational networks across the various interfaces, the closest we come to the idea of selection the best-fit principle (Truscott & Sharwood Smith, ). 5 This would be the framework’s instantiation of Grosjean’s bilingual language mode (Grosjean, ).
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
The boost it gets may still not be sufficient for selection to take place on this occasion. However, as a general principle, each bit of activation, however small, gradually builds up the default resting level of a structure, making it ever more competitive over time. Note in passing that the ‘default’ resting level is an abstraction from what in effect is a level that is continually changing. Also, the way structural properties gradually grow together or grow apart along with the interface connections with other structures can be described as ‘emergent’, although the local organizing principles of a module remain predefined so linguistic systems are not self-organizing in the sense intended by emergentists who generally do not accept the need for an account of grammatical growth that requires module-specific construction principles (MacWhinney & Bates, ; O’Grady, a). Rather, they claim that complex systems like grammars emerge only as a result of the interaction of simple principles of a general nature, although there is no wide agreement as to what these principles are (see O’Grady, b, for a concise overview) Now consider a later situation where the same learner has become an attriter. Infrequently activated constructions associated with English consequently suffer a steady decline in their default resting levels. In terms of the vertical metaphor, the structures involved have a longer way to go to make it up into the WM of their store and are therefore much less competitive as a result. The L, French, on the other hand, by virtue of the very frequent activation of French-associated structures, gradually becomes the default dominant language. Without any supportive situational context such as described above, the ‘neglected’ English Adj-N sequence in the syntactic system is therefore less and less likely to win out: the resting levels of activation of those structural features that determine the sequence accordingly sink further and further down and the Adj-N construction becomes correspondingly less accessible. This is a very basic account of how attrition might work in processing terms. On the one hand, a reduction in the frequency of activation makes L structures less accessible. This in itself has nothing to do with the presence of an L. On the other, the newly dominant L will start to compete successfully and cross-linguistic influence will manifest itself in various simple and complex ways, the precise manner being the focus of attrition researchers—in observable performance. In other words, this basic account definitely needs to be supplemented. The framework provides an appropriate background set of processing principles and assumptions but certainly not a comprehensive attrition theory in itself. If it did it would be a cognitive ‘theory of everything’. However, there are other aspects of attrition that the framework does provide, that is, over and above the most basic account, particularly in connection with interaction between different systems and notably the special role of affective and conceptual systems as touched on above.
.. Frequency effects It is worthy of note that frequency effects, in the framework, are limited to the internal frequency of activation within given memory stores. In development—as acquisition— there may quite often be a discrepancy between the high external frequency of structures perceived by the hearer in the immediate environment and the actual processing subsequently carried out (or not) within particular modules (Gass & Mackey, ). Here, then, it is clear that frequency of this external input is not to be equated with the real driver of growth, namely the frequency with which structures are activated within modules. In the
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
case of development—as attrition—minimal frequency in the input from the environment or the complete absence of such input would appear, on the face of it, to be the main driver of change but again it is first and foremost a matter of the frequency of internal input that either brings about activation or no activation within modules that brings about the relevant changes. External input frequency encountered in speech, texts, and signing is still a secondary factor until it is processed by one or other of the internal systems; also, purely internally generated language use as in writing—even thoughts and dreams—in a given language can play a supportive role in countering or slowing the decline in resting levels of otherwise unused representations.
.. Degrees of forgetting It is really not important in such an account of attrition whether complete forgetting takes place. Once a structure’s resting level of activation has sunk down far enough it will not figure at all in spontaneous language performance. As suggested earlier, it is an empirical issue to be resolved using techniques perhaps more sensitive than are currently available as to whether, after many years or as a result of some pathological condition, an attrited structure can reappear without any reacquisition taking place.
.. Inhibition and attrition The notion of an activated structure that is regularly outcompeted and accordingly drops back to a lower resting level may indicate the operation of inhibitory processes, as also the idea that this resting level will steadily decline over time through lack of activation reducing accessibility (Green, , ). Note, however, that inhibition may in principle occur as a post-facto response to non-selection, that is, only after a rival candidate of selection has been outcompeted and not be the force that actually brings non-selection about (see Berg & Schade, , p. ; and also Köpke, , for related discussion).
.. Metalinguistic knowledge and attrition The knowledge that such and such a construction is anomalous or ‘correct’ in a given language may exist independently as explicit ‘metalinguistic’ knowledge. In terms of the current framework, such knowledge is represented separately, in the conceptual system, along with any other factual and consciously accessible knowledge of the world. As such, it has no direct bearing on what is happening to the phonological and syntactic representations in other modules (Sharwood Smith & Truscott, ; Truscott, ). Where it can be projected into consciousness it is more or less equivalent to what elsewhere is termed ‘declarative’ knowledge (Ullmann, a). Such conceptually-coded knowledge of language can, of course, also suffer attrition in its own right and this happens in precisely the same manner as earlier—that is, on a use-it-or-lose-it basis. It is just that there is no direct relationship between the two types of linguistic knowledge.6 Sophisticated 6
The precise role of metalinguistic knowledge in attrition may well go beyond self-correction and have subtle indirect effects on activation levels beyond the conceptual system. This merits a fuller discussion than is possible here (see Truscott, 2015).
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
metagrammatical knowledge can even exist with very little grammatical ability at all, as is exhibited by many theoretical linguists who certainly do not have to be in any way fluent in a language to talk coherently about its structure. In cases where attrited speakers have both types, metagrammatical knowledge can survive longer than core syntactic or phonological knowledge; it can also play some corrective role in their performance and hence delay some of the outwardly observable effects of attrition.
. C
.................................................................................................................................. The framework used for this discussion may be described as incorporating notions of emergent and dynamic systems, albeit constrained within a particular view of the modular mind. This allows many different observed characteristics of cognitive systems to be investigated as part of unified explanation of how the mind works. This certainly holds for the study of language attrition, which is driven in the first instance by the lack of stimulation of a range of linguistic and non-linguistic structures (also known as representations) involved in language use and located with various modular systems. These representations are all located within one or other of the specialized modules and their composition is governed by the local principles pertaining to the module in question. Their growth and attrition, however, is also driven by cross-modular principles, that is, ones that are common to all specialized mental systems. What we then see is a mind that is both very simple and extraordinarily complex. As certain representational properties and interconnections begin to weaken, then those that remain may undergo all kinds of changes as a result. Studying these changes is the task of attrition researchers, but the more explicit and detailed nature of an overarching cognitive framework should facilitate fruitful discussion of their results and their collaboration with other researchers whose own research interests are quite different. This indeed appears to be a necessity, not a luxury, if any major progress is to be made. Readers are invited to take findings reported elsewhere, including those in this volume, and build broader explanations of the data using the architecture of the processing framework that has been discussed in this chapter or some worthy substitute. These explanations need not replace already existing, more specialized accounts, but they should allow an embedding of those accounts within a more general explanation of how the mind works and what is ‘behind’ the data, asking crucial ‘why’ and ‘how’ questions about the nature of memory, representational change over time, and what is really involved in learning and forgetting.
A Many thanks are due to my anonymous reviewers for their useful comments and suggestions which I have in most cases taken account of in this final version.
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
......................................................................................................................
Language loss or manifestations of a dynamic system? ......................................................................................................................
. .
. I
.................................................................................................................................. F extended use of a second language (L), many bilinguals experience significant difficulty in accessing their native language (L). For some, it feels as though they have lost parts of their L—a feeling that fits with theoretical claims of language attrition. In this chapter, we situate this phenomenon within the context of memory retrieval processes that may shed light on the nature of the mechanisms underlying language attrition. Our premise is that basic memory processes may provide mechanistic accounts of the reduced access reflected in language attrition. In particular, we consider the hypothesis that L attrition reflects the normal dynamics of the bilingual language system. We begin with a review of the evidence suggesting that developing proficiency in an L has implications for the accessibility and use of the L, with a focus on psycholinguistic laboratory studies. In particular, inhibitory mechanisms have been suggested to play an important role in managing any potential cross-language competition. We consider how normal L-inhibition effects shown in everyday language use may be able to explain larger-scale changes found in immersion settings. We then briefly summarize the literature on retrieval induced forgetting (RIF) that implicates inhibitory mechanisms in the management of representational conflict within models of (monolingual) memory retrieval. We conclude by considering the implications of these two bodies of research for theories of language attrition.
. N L
.................................................................................................................................. Traditional accounts of late L learning assumed that the native L is stable and largely autonomous, providing a source of knowledge and transfer to the newly emerging L, but
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
benefiting or suffering little from the presence of the L itself. This can be seen, for example, in Clahsen & Felser’s () ‘shallow structure hypothesis’ by which the L carries a privileged status but is not impacted by the presence of an L, or in more modest form in findings that many aspects of the L remain relatively unchanged even after ten years of immersion (e.g., de Bot et al., ). The emerging psycholinguistic evidence suggests a situation that is far more dynamic, with both languages changing over the course of L learning and with immediate consequences of exposure to the L, whether it be in the context of the classroom, language immersion, or the laboratory. In this section of the chapter we illustrate the nature of this evidence and consider how these changes reflect the typical use of two languages and also potentially require a revision of the traditional views about the stability of the L. Three different research strategies have been used to investigate the way that the L changes in the face of L use. One is to examine performance in the L when it is used alone to determine whether L learners and bilinguals differ from monolingual speakers of the L and whether language processing in the L reflects the influence of the L (e.g., Dussias, ). Another strategy is to require L speakers to use the L and then to assess whether there are spillover effects from the L to the L when L use is subsequently required (e.g., Guo et al., ). A third strategy is to compare L speakers who are otherwise similar to one another and to immerse them in contexts that differ in the degree to which the L is required (e.g., Linck et al., ). Here we review the first two of these strategies and later in the chapter we consider the consequences of language immersion. If L speakers retained monolingual-like use of the L, then we would expect that language processing in the L would resemble that of monolingual speakers. L speakers would be as fast and as accurate as L speakers in both comprehension and production tasks and they would be relatively insensitive to the presence of converging or conflicting information across the two languages. A very long line of research on word recognition and production has shown that bilinguals are exquisitely sensitive to cross-language interactions when they are processing their L, suggesting that the activation of the L comes to influence the L (see Kroll, Dussias, Bice, & Perrotti, , for a recent review). But are there similar effects from the L to the L? At the lexical level, an early study by Van Hell & Dijkstra () showed that lexical decision performance in L for Dutch-English bilinguals who differed in whether they were also proficient in French as a third language (L), was not only faster in Dutch when the word was a cognate with English, the L, but also when it was a cognate with French, the L, when the speakers were sufficiently skilled in French. The findings of this study were interpreted to suggest that with adequate proficiency, activation can spread from the L to influence processing in the L, supporting the claims of models such as the Bilingual Interactive Active Model Plus (BIA+; Dijkstra & Van Heuven, ) that the lexicon is fundamentally open across all of the codes that are shared between the two languages. A recent study, using electrophysiological methods in addition to behavioural indices of response time and accuracy, suggests that a high level of L proficiency is not a requirement to observe changes to the L. Bice & Kroll () asked English learners of Spanish at relatively early stages of L study to perform a similar lexical decision task as the one used by Van Hell & Dijkstra (). They recorded via EEG (electroencephalogram) while the lexical decision task was performed and asked whether there was evidence that performance in English, the L of these learners, was affected by Spanish, even for learners at this early stage. Like Van Hell & Dijkstra, they included cognates, translations that share lexical
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
form across the two languages. Bice & Kroll found that response times and accuracy were no different for cognates relative to non-cognate control words but that the Event-Related Potential (ERP) data revealed an emerging cognate effect for L learners relative to monolingual English speakers who showed no effects in either measure. In particular, these learners showed a reduced N for cognates relative to non-cognates, suggesting that L processing is already being impacted at this early stage of developing L skill. These results suggest that the brain may outpace behaviour in showing sensitivity to new learning. Like the findings of studies that use ERPs to track L development in typical university classroom learners (e.g., McLaughlin et al., ), the Bice & Kroll results show that not only is learning taking place at very early points in L acquisition, but that cross-language activation can be seen in L as well as L language processing. The two studies we have reviewed on L effects on the L each examined comprehension in the L and demonstrated a pattern of facilitation when the L converges with the L in the case of cognate translations. But what about production? Contrary to the facilitation in L seen for cognate processing, the evidence from production suggests that there may be a cost to speaking the L as a function of using an L actively. In a long line of studies, Gollan and colleagues have shown that bilingual speakers are slower to produce words in their L than monolingual speakers of the same L and also have more tip of the tongue (TOT) experiences than monolinguals (e.g., Gollan & Acenas, ; Gollan, Montoya, Cera, & Sandoval, ; Gollan, Montoya, Fennema-Notestine, & Morris, ). One interpretation of the reduced accessibility of the L in bilingual speakers is that the opportunities to speak each language are necessarily shared over time, thereby reducing the functional frequency of each language relative to that of a monolingual speaker. That process is hypothesized to have particular consequences for retrieving low frequency words. At issue is whether changes to frequency alone can account for the reduced accessibility in production or whether there is also interference from the co-activation of the two languages (see Kroll & Gollan, , for a discussion of these alternative accounts for the observed reduced accessibility of L words). The consequences of L use for the L have also been examined in studies that have monitored bilingual speech in switching contexts. An initial demonstration by Meuter & Allport () in a language production paradigm revealed larger switch costs for speaking words in the L following the L than the reverse. This switch cost asymmetry across the two languages has been studied extensively since the Meuter & Allport study, with a number of different interpretations proposed (see Bobb & Wodniecka, , for a recent review). The account that is most congenial with an attrition perspective is given by Green’s () Inhibitory Control (IC) model. According to the IC model, bilinguals control language production through a combination of top-down inhibition at the level of task schemas (e.g., activation of the schema ‘speak in the L’ would generate inhibition of the schema ‘speak in the L’) and reactive, lateral inhibition between specific representations (e.g., activation of L ‘gato’ generates lateral inhibition of L ‘cat’). In this model, the amount of representational inhibition is directly proportional to the asymmetry in activation—that is, to the amount of competition between the two representations. When bilinguals plan speech in the weaker L, the stronger L is assumed to be inhibited and to a greater extent than the L. If L production is required immediately following speech in the L, then the hypothesized inhibition must be overcome. Slower production in the L following L production has been interpreted to reflect the process of resolving inhibition
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
(e.g., Green, ). From the perspective of attrition research, the language switching data show that even a single word-to-word switch may be sufficient to create momentary suppression of the dominant or native language. Although there has been debate about the naturalness of the forced switching task, the point is that selective changes to the L can be imposed relatively easily, even by a simple switch at the lexical level. An early hypothesis about the switch cost asymmetry in the lexical switching paradigm was that it was a feature only of less proficient bilingual use, for those whose L was far more dominant than their L. Costa & Santesteban () found that highly proficient and closely balanced Spanish-Catalan bilinguals produced a pattern of switch costs that was symmetric, in comparison to relatively unbalanced bilinguals whose pattern matched the one reported initially by Meuter & Allport (). They interpreted the different patterns as showing that only learners and less proficient L speakers require IC. However, even the balanced bilingual speakers were slower overall to speak the L than the L, suggesting global inhibition across conditions that require mixed use of the two languages, as is encountered in the language switching paradigm. The importance of this study was to suggest that there may be different components and indices of IC that are related to proficiency and language context. In the case of language attrition, we can speculate that the switching asymmetry might be flipped (with larger switch costs in the L than the L) based on the dominance of the L over the L. In another variant of the lexical switching task, bilinguals are again asked to produce single words but the opportunity to speak each language extends over a longer block of time. In an electrophysiological study, Misra et al. () used this paradigm and found an inhibitory pattern in L speech following naming in the L. Misra et al. asked relatively proficient Chinese-English bilinguals to name the same pictures in Chinese and then in English or in the opposite order. Because the pictures were identical in both languages, they predicted that the language that came second would benefit from this repetition. Using ERPs, they found a facilitatory pattern consistent with repetition priming when the L was named following the L. In contrast, when the L was named following the L, there was greater negativity in the ERP record, a pattern consistent with inhibition. Critically, Misra et al. reported that the inhibitory pattern when the L followed the L persisted despite repeated opportunities to use the L, a context that should have enabled recovery. The results suggest an inhibitory process for the L that is relatively long lasting following L use. A similar pattern was described by Van Assche et al. () using a verbal fluency paradigm. A number of recent neuroimaging studies have contributed evidence on the differential effects of bilingualism for the L that suggests both long lasting structural changes to the L with extended L use (e.g., Parker Jones et al., ) and differential functional consequences for the L that depend on language processing demands (e.g., Guo et al., ). The Guo et al. study demonstrated that there are both local and global contributions to the observed inhibitory patterns in the native language when bilinguals switch from one language to the other. While there is not agreement about the scope or time course associated with different types of inhibition, the evidence converges on the finding that it is the L that is differentially affected by L use. Are the changes we have described in these studies an indication of language attrition? We think not. They are a reflection of the ordinary dynamics and modulation of the native language as bilinguals negotiate their ways through a landscape of language use that
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
sometimes requires that one language be used in isolation and other times requires that both languages be used together. Critically, these dynamics are not restricted to the least proficient L speakers but characterize language processing for highly proficient bilinguals. Proficiency affects the observed patterns, but the changes to the L occur even once individuals acquire a very high level of skill in the L. It is quite possible that the long term disuse of the native language will differentially change the weighting of these inhibitory patterns. This is congruent with the claims of the Adaptive Control Hypothesis (Green & Abutalebi, ), according to which language control processes are adjusted based on the communicative demands of the context of interaction. What is not yet known is which of these changes are short term and easily recovered and which are long term and endure even after the native language is reintroduced. It may be important to note that in the context of this discussion we have not reviewed the research on code switching at the sentence or discourse level, a phenomenon of language mixing that characterizes highly proficient bilingual speakers who may switch from one language to the other in the middle of a sentence but following grammatical constraints (e.g., Myers-Scotton, ). Not all bilinguals code switch, but for those who do, there appears to be a modulation of the relations between comprehension and production to enable prediction of an upcoming switch of language and to achieve the coordination that maintains proficient use of both languages (e.g., Fricke et al., ; Tamargo et al., ). It seems likely that the dynamics of native language modulation and attrition are affected by whether bilingual speakers are habitual code switchers or not. Indeed, one claim of the Adaptive Control Hypothesis is that frequent code switching places unique demands on the language control system and thus requires a different set of control processes than those required by single language or dual language (but non-intermixed) contexts. This is a topic that has only recently been investigated with respect to control processes (e.g., Green & Li, ) and merits further investigation.
. R
.................................................................................................................................. Over the past two decades, research has demonstrated that the retrieval of information from long-term memory leads to reduced accessibility of related but non-retrieved information—a phenomenon described as retrieval induced forgetting (RIF; e.g., Anderson et al., ). In the classic paradigm, participants study multiple exemplars of various categories (e.g., fruits: banana, apple, orange), then are provided a category (e.g., fruits) and asked to produce a category member based on a presented cue (e.g., starts with a particular letter). For example, ‘FRUITS—B_____’ might be presented to cue retrieval of the word ‘banana’. Participants are cued to recall some highly prototypical category members (e.g., banana, apple)—known as the Retrieval Practice plus (RP+) condition— but, importantly, they are purposefully not cued to recall other prototypical category members (e.g., orange)—the Retrieval Practice minus (RP ) condition. Later in the experiment, participants are given a new task that requires retrieval of both the RP+ and RP items, but this time with a novel cue (e.g., semantic judgement, rhyme generation). The key finding is that previous retrieval of the RP+ items leads to a reliable decrease in the
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
accessibility of the RP items during this subsequent recall task. In this example, retrieving ‘banana’ during the initial category cued task makes the participant significantly slower and/or less likely to successfully retrieve ‘orange’ during the second recall task. A primary theoretical model of RIF implicates inhibition as the underlying mechanism driving the reduced accessibility of the RP items (e.g., Levy & Anderson, ). The idea is that retrieval of a category member during the initial retrieval task requires the individual to overcome interference from other highly-related category members. This may be achieved by inhibiting the related but non-target RP representations, as evidenced by the reduced accessibility of those items in subsequent, independent tests requiring recall of the RP items (for a recent meta-analysis, see Murayama et al., ; but see Jonker et al., , for a context-based, inhibition-free account). Dozens of follow-up studies have been conducted to clarify the nature of these RIF effects and to identify boundary conditions, such as relative dominance, integration of practised and to-be-remembered items, and semantic similarity of category exemplars (e.g., Anderson, ; Jonker et al., ). A critical finding is that the memory representations in question must be sufficiently in competition during the retrieval process in order for the inhibitory mechanism to be engaged. One feature that may create or limit this competition is the relative dominance or representativeness of the category members. For example, retrieving a more dominant category member, such as ‘banana’, does not induce RIF of a less dominant FRUIT category member, such as ‘tangelo’. However, the opposite pattern is likely: retrieving ‘tangelo’ would likely induce inhibition of the more highly representative exemplar ‘banana’. In other words, RIF effects require that a non-target representation be in direct conflict or competition with the target representation in order for the inhibitory mechanism to be engaged. Extending the RIF theoretical framework to the bilingual case, it seems plausible that the inhibitory mechanism implicated in RIF might be one mechanism that bilinguals rely on to successfully manage cross-language competition. For example, for a native English speaker who is learning Spanish, retrieval of the word ‘gato’ (English: cat) likely induces competition from the non-target but more accessible L representation ‘cat’ and therefore successful L production may require inhibition of the L lexical entry. This account fits with claims of the IC model of bilingual speech production previously discussed (Green, ). Note also that the relative dominance feature of the IC model parallels the relative dominance boundary condition of RIF effects. This raises the possibility that RIF-based inhibition may be one mechanism for the instantiation of inhibition within the IC model framework. Of course other mechanisms are likely at play, given that the IC model’s claims extend beyond the case of retrieval induced inhibition. But this hypothesis that RIF may be supporting bilingual language control addresses recent calls in the literature to better specify the nature of inhibitory mechanisms within the bilingual language system. The theoretical frameworks of the IC model and the RIF literature are indeed congruent, but can RIF occur across languages? As reviewed above, there is ample evidence that when producing a less-dominant L, the speaker may encounter interference from the dominant L and that L use has consequences for subsequent access to the L. In the context of lexical access, some studies have specifically reported cross-language RIF effects, although the results are somewhat mixed. In the first published RIF study with bilinguals, Levy et al. () asked native English speakers learning Spanish to repeatedly name a series of line drawings of common objects (e.g., cat, snake, book) in either the L or L based on an
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
external cue (i.e., colour of the line drawing). Half of the drawings were named in the L and the other half in the L, with each drawing being consistently named in one language for a given participant. The critical manipulations within the naming task included the number of repetitions (, , , or ) and the language of naming. The L names for all pictures were then cued in a subsequent retrieval task. They found that naming an object in the L ten times rendered the L translation of that object to be less readily accessible on a subsequent, independent retrieval task (rhyme generation). The results suggest that repeatedly producing a lexical item in the L has a measureable impact on the accessibility of the lexical representation for the L translation. RIF has also been found when focusing on the grammatical gender of lexical items. Morales et al. () asked speakers of Italian (L) and Spanish (L) to name pictures in the L. For some pictures, the grammatical gender matched across languages, whereas for others the grammatical gender mismatched between the L and L. In a subsequent task, the same pictures were presented and participants were required to produce the article of the L picture name. Parallelling the results of Levy et al. (), repeatedly producing the gender for a given item in the L led to subsequently slower naming latencies when producing the L article of items with incongruent grammatical gender relative to items with congruent grammatical gender. This suggests that conflict at the level of grammatical gender can induce RIF as a result of successful retrieval during the initial naming phase. The robustness of these effects has been questioned by some researchers. For example, Runnqvist & Costa () failed to replicate the RIF effect using a similar paradigm as that employed by Levy et al. (). In their study, Runnqvist & Costa included three groups of Spanish bilinguals who varied in their L proficiency. Across all three groups, they found evidence of retrieval induced facilitation: repeatedly naming pictures in the L was associated with better subsequent recall of L translations. Of course, bilinguals vary substantially in their language experiences and contexts of use, which has important implications for how language control mechanisms are called upon (e.g., Green, ), as captured by the Adaptive Control Hypothesis (Green & Abutalebi, ) and detailed above. A clearer mapping of language experiences to specific cognitive mechanisms will inform constraints on when and how RIF mechanisms may support language control.
. T
.................................................................................................................................. Research on language immersion suggests the L becomes less readily accessible within the L-dominant context. Linck et al. () tested native English speakers approximately four months into a semester abroad in Spain and compared them to a cohort of classroom-only learners enrolled in language courses back at the home university who were at a similar L proficiency level. Participants completed a battery of tasks measuring lexical access in the L and L. On both comprehension (translation recognition) and production (semantic fluency) tasks, participants in the immersion context demonstrated reduced access to the L, as evidenced by the elimination of L-related distractor interference on the comprehension task and a significant reduction in their L verbal fluency. A subset of the immersed
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
learners were then retested approximately six months after returning to the L-dominant home university; although L verbal fluency rebounded to a level equivalent to the classroom learners, participants continued to show reduced L interference on the comprehension task. The results were interpreted to suggest that the L-rich immersion context encouraged inhibition of the L. The retest results suggested that the immersed learners developed IC skill that they could deploy to mitigate L interference during comprehension. Baus et al. () also studied the effects of immersion on L lexical processing using a longitudinal research design. They tested immersed learners (native German speakers learning Spanish in Spain) upon their arrival to the L context and four months later, prior to their departure. On a picture naming task, the immersion experience was associated with slower naming latencies in the L. That is, the L was less accessible at the conclusion of the immersion period, and this effect was most apparent with low frequency non-cognates. Differences on an L semantic fluency task (similar to the task used by Linck et al., ) were also modulated by cognate status: semantic fluency increased for cognates but decreased for non-cognates (i.e., participants produced more cognates and fewer noncognates in the departure testing session relative to the arrival testing session). These patterns of reduced L access are congruent with an inhibition account, by which one might expect cognates to be relatively immune to inhibition-based reductions in accessibility due to the strong phonological and/or orthographic overlap with the L. That is, L and L forms for cognates are (nearly) identical and therefore would not be expected to compete for selection—a necessary condition for RIF effects to emerge. Or from another perspective, non-cognates and lower frequency words might be more vulnerable to retrieval-based inhibition mechanisms, as there is greater potential for cross-language competition between the word forms. We note that these effects on lower frequency words and non-cognates are also congruent with the claims of the weaker links hypothesis (e.g., Gollan et al., ). As we noted earlier, this account does not require inhibitory mechanisms, but rather assumes that reduced accessibility of the L is driven by a weakening of the links between conceptual and form representations as a consequence of the reduced frequency of using L words in the immersion context. Although the two studies reviewed here cannot adjudicate between these two accounts, taken together these results are congruent with the inhibition hypothesis, with RIF providing one specific mechanistic account of the dynamics of these inhibition effects.
. I
.................................................................................................................................. Given the findings of research on RIF, the implication we consider here is that language attrition phenomena may emerge, in part, from the underlying dynamics of the memory system. Regardless of the specific mechanism behind RIF (be it inhibition or something else), when considered in the bilingual context, these results suggest that repeated retrieval of the non-dominant language may lead to reduced accessibility of the
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
corresponding lexical representations in the dominant (native) language. This is an important consideration, as it underscores the insight, emerging from the literature on bilingual language processing, that developing proficiency in an L has consequences for the L—even for skilled bilinguals who show no obvious signs of L attrition. Indeed, there is growing evidence that proficient bilinguals may not function like monolinguals in any of their known languages (e.g., Ameel et al., ), lending support to Grosjean’s (, p. ) earlier warning that ‘the bilingual is not two monolinguals in one person.’ This directly contradicts the previously held beliefs that the L is unchanged by the learning of an L (e.g., Clahsen & Felser, ) but is compatible with a Dynamic Systems Theory (DST) view of language development, by which the two language systems are assumed to interact and cause non-linear change over time (e.g., de Bot et al., ). L attrition may be seen as representing the extreme end of the spectrum of L changes that co-occur with increasing proficiency in an L. The finding that the process of learning an L has measureable impact on L processing highlights an important opportunity to develop a shared theoretical framework that incorporates features of theories of language attrition (with DST perhaps also as a framework to study attrition, as it views language acquisition and attrition as two sides of the same coin), models of bilingual language development, and models of the dynamics of memory retrieval. In such a framework, factors known to drive more subtle changes in memory representations, as demonstrated by RIF, may also scale up to provide an explanatory mechanism for broader changes in the bilingual’s dynamic language system. This framework is congruent with existing models of bilingual processing. As previously discussed, one claim of the IC model is that inhibitory mechanisms originating from outside of the language system support bilingual language control (Green, ) and certain claims of the IC model parallel the dynamics of RIF (e.g., relative dominance). This offers specific points of overlap that can serve as anchor points when integrating these distinct models of memory and language processes. Similarly, the Adaptive Control Hypothesis (Green & Abutalebi, ) may help us to formulate specific predictions about the impact of specific features of a bilingual’s language use patterns and language context on mechanisms underlying language attrition phenomena. In the context of the current discussion, one might hypothesize that RIF could contribute to the interference control component of the Adaptive Control Hypothesis, with both single language and dual language interactional contexts placing increased demands on this specific aspect of control. Clearly, RIF is not likely to be the only memory retrieval mechanism responsible for language attrition. But existing models of language attrition—particularly those that take a dynamic systems approach—may benefit by incorporating general mechanisms of memory retrieval, such as RIF, as one of multiple mechanisms driving language change. What we hope to have accomplished in this review is to point out these opportunities for future development of such a unified theoretical framework.
. C
.................................................................................................................................. There is a growing research literature on language attrition phenomena, as seen in this volume. The findings from psycholinguistic studies of bilingualism reviewed above suggest
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
that the bilingual language system is dynamic and constantly adapting to changing demands placed on the bilingual. We suggest that L attrition can be viewed as reflecting the dynamic nature of the language system operating within the human cognitive architecture. Within this framework, shifts in language accessibility are predicted to result from the processing mechanisms that allow successful language engagement and disengagement within the bilingual language system—and features of the memory system more broadly.
A The writing of this chapter was supported in part by NSF Grants BCS-, OISE-, and OISE-, and NIH Grant HD to J. F. Kroll.
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
......................................................................................................................
Evidence from eye movements ......................................................................................................................
. , . ´ , , ´
M studies published over the past two decades have used measures of online language processing to show that the languages of a multilingual speaker are engaged in constant, pervasive interaction, and that these interactions are bidirectional: not only does the first language affect the second language, but the second language also influences the first. Remarkably, these bidirectional influences can be seen at every level of language use. In immersion contexts, second language (L) speakers experience reduced access to the first language (Linck et al., ), and extensive contact with the second language can affect first language (L) phonology (e.g., Flege, ; Flege & Eefting, a), syntactic processing (Dussias & Sagarra, ), sentence production (Hartsuiker et al., ), and naming performance of common objects (Malt & Sloman, ). There is now growing evidence demonstrating that the seemingly stable L system is open to influence when individuals learn and become proficient in an L (e.g., Gollan et al., ; Ivanova & Costa, ; Runnqvist et al., ; Kasparian & Steinhauer, a). The insight that there are bidirectional influences between the two languages of a bilingual speaker has implications for the study of language attrition; because the linguistic system is dynamic and interactive at all levels of representation, it is reasonable to expect that variations in language dominance, use, and exposure can lead to changes in the system—some subtle, some significant. In our view, these changes are not necessarily the result of what has increasingly been labelled ‘attrition’ in the literature, but rather are a natural consequence of the interactive nature of the languages at play. The goal of this chapter is twofold. The first is to survey recent contributions to the research on bilingual language processing that serve to demonstrate how exposure to a second language, even for
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
a brief period of time, can impact processing in the native language. Given this evidence, we will argue that claims of language attrition may not be as clear-cut as one may think when online language processing is taken into account. A second goal is to show that eye-tracking is a premier behavioural method by which we can come to understand fine-grained changes in online language processing. In doing so, we hope to illustrate how the study of online language processing via eye-tracking can help to clarify issues in language attrition (e.g., Berends et al., ).
. T
.................................................................................................................................. One finding that remains uncontroversial after almost two decades of psycholinguistic research with bilingual speakers is that the two languages of a bilingual speaker are active, even when the intention is to use only one language. When bilinguals read, when they listen, or when they prepare to speak in one of their two languages, the language not in use is also active (Kroll et al., ; for an extensive review, see Dijkstra, ). For example, when bilinguals are asked to name cognate words (such as ‘piano’ in Spanish and English) in only one language, the language not in use is also active and influences performance, resulting in a cognate facilitation effect (i.e., bilinguals produce and recognize cognates faster than non-cognates). Although this work could be criticized on the grounds that the ‘out-of-context’ nature of word recognition and word naming experiments may itself induce the observed effects, extensive research also demonstrates that it is much more difficult than we might have thought to reduce or eliminate the presence of cross-language activity. First, the same cognate facilitation effects observed in out-of-context word recognition experiments can be observed when cognate words are embedded in sentences (e.g., Schwartz & Kroll, ; Duyck et al., ; Van Hell & De Groot, ; Libben & Titone, ). Second, the cross-language competition is not restricted to the lexicon. Although more research has examined these interactions for words than for sentences, recent studies show that the grammars of the bilingual’s two languages are also open to one another in a manner far more permeable than what might have been predicted. Studies within the syntactic priming literature have shown that monolingual speakers are more likely to produce a sentence using a grammatical structure if they have just produced a sentence using that same structure (Bock, ). In bilingual research, the main question has been whether priming will be observed when switching languages from the prime to the target sentence. Although there are some exceptions (e.g., Bernolet et al., ), the general finding is that cross-language priming can be obtained (Hartsuiker et al., ). In general, the parallel activity of a bilingual’s two languages creates cross-language interactions that influence performance at every level of language, including phonology (e.g., Spivey & Marian, ; Jared & Kroll, ; Marian & Spivey, ; Ju & Luce, ; Blumenfeld & Marian, ), orthography (e.g., Dijkstra & Van Heuven, ; Van Heuven et al., ), syntax (e.g., Hartsuiker et al., ), and meaning (e.g., Sunderman & Kroll, ). The parallel activation of the bilingual’s two languages is not restricted to languages that share structural or functional characteristics. An example of cross-language interaction
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
, ´ , ,
between structurally distinct languages is the case of bimodal bilinguals, who use a spoken and/or written language and a signed language. Morford et al. () found that the speed with which deaf American Sign Language (ASL) English bilinguals made semantic relatedness judgements for English word pairs (movie-paper) was affected by whether the hand formations of the ASL translations were themselves similar or different in form. This result is remarkable because no ASL was present during the experiment. In an earlier study involving a logographic and an alphabetic language, Thierry & Wu () found early neural sensitivity (measured via the recording of EEG (electroencelphalogram)) to a word’s Chinese translation when Chinese-English bilinguals performed a semantic relatedness judgement task in English only. These bilinguals performed the semantic relatedness task without Chinese actually being present during the experimental session, and in a context in which they were immersed in an English-speaking environment. This demonstrates that the languages of a bilingual speaker are engaged in constant, persistent interaction, even in the presence of obvious differences between the bilingual’s two languages (e.g., Emmorey et al., ; Hoshino & Kroll, ). What’s more, the results in Morford et al. and in Thierry & Wu show that conscious awareness of the other language is not required to observe interactive effects between the L and the L, suggesting that parallel activation is a feature of the architecture of the bilingual language system (Morford et al., ). The finding that cross-language activation and interaction is pervasive in bilinguals has ramifications for the study of language attrition. This is especially true in the case of language processing, which can be modulated with relative ease. At one extreme end, some bilingual speakers experience an abrupt shift in language use towards their L, perhaps due to marriage, immigration, or education. After decades of lack of use of the L, these speakers may experience difficulty in lexical retrieval and large-scale interference from the L (Schmid, a; Schmid & Keijzer, ; Schmid et al., ), and in many cases can be labelled as having undergone L attrition. At the other extreme, study abroad students, after mere weeks of immersion, experience subtle but quantifiable differences in lexical access vis-à-vis non-immersed peers in their first language (Linck et al., ). Is the latter case also a clear-cut example of attrition? In part, the field lacks a clear consensus on () what attrition is; () how soon attrition effects can begin; and () how permanent the effects of attrition are. The goal of this chapter is to survey recent contributions to the research on bilingual language processing that serve as illustrations of how exposure to a second language, even for a brief period of time, can impact processing in the native language. Critically, we take the evidence of change to the native language not as an indication of permanent language attrition or loss, but rather as evidence that variation to the native language is a natural consequence of the intrinsic plasticity of the linguistic system (Kroll et al., ). In fact, monolingual speakers demonstrate this plasticity too, adapting to and extending the use of unfamiliar structures from a language variety different from their own variety (e.g., Fraundorf & Jaeger, ), and even showing reduced sensitivity when presented with ungrammatical structures for a short period of time (Hopp, a). These effects, both in bilingual and monolingual speakers, may be attributable to priming, but to the extent that the facilitatory effects of priming during language comprehension are not limited to known structures but also affect entirely unknown structures (Fraundorf & Jaeger, ), it suggests that priming is not necessarily a reflection of transient activation (Pickering & Branigan, ; Traxler & Tooley, ); rather, it can strengthen the mappings between
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
meanings and linguistic structures (Leonard, ), thereby leading to the learning and modification of linguistic representation (e.g., Bock & Griffin, ; Konopka & Bock, ; Savage et al., ; Ferreira et al., ; Hartsuiker et al., ; Fine & Jaeger, ). We begin with a brief overview of eye-tracking and how eye movements are a fine methodological tool to investigate shifting sensitivities to linguistic cues in sentence processing. We then illustrate our approach to language interaction by examining two instances of the reconfiguration of the L, which we argue are natural extensions of a dynamic linguistic system—monolinguals to a lesser extent demonstrate this dynamicity as well. Then, we ask whether eye-tracking can illustrate the reactivation of L processing strategies after re-exposure in a compressed timeframe.
. E- L
.................................................................................................................................. How is eye-movement data informative to studying changes in the processing of the first language? Theories of sentence processing are generally interested in the online or incremental nature of comprehension processes. As soon as each word is encountered, readers and listeners are assumed to make structural decisions about how to integrate each word within the ongoing syntactic structure. The incremental nature underlying sentence processing allows researchers to use eye movements and fixations to make inferences on the ease of integration into prior sentential context. Eye-tracking has been used for both auditory and reading comprehension studies and with a variety of populations. The logic that underlies these studies is quite simple: there is a plausible link between what readers/ hearers look at and for how long as they process linguistic input. Thus, when reading, readers linger longer and regress back to words that are harder to process as compared to easy or predictable words (see Rayner, ; Clifton & Staub, ; Keating, , for comprehensive reviews). Similarly, researchers working within the auditory domain utilize the visual world paradigm (Tanenhaus et al., )—an eye-tracking technique whereby participants listen to auditory stimuli in the presence of a visual scene—to study the timecourse of spoken language processing. Because listeners will look at relevant referents as they are named in the auditory input, researchers can manipulate the presence/absence of linguistic features (e.g., phonetic overlap, grammatical gender, case marking, anaphora, givenness, etc.) to investigate the time course of looks to a target item (see Tanenhaus & Trueswell, ; Altmann, ; Huettig et al., , for reviews). In what follows, we present a necessarily brief overview of the central findings of eye-tracking research in reading and auditory processing. Over three decades of eye-movement research have shown that when eye movements are recorded during reading, there are systematic relations between fixation durations and the characteristics of the fixated words (Rayner, ; Just & Carpenter, ; Ehrlich & Rayner, ). Readers spend more time fixating on harder words (e.g., lower frequency words) and on more important words (e.g., nouns and verbs) than on easier words. Longer words are also more likely to be fixated on than shorter words, and words that are likely to be skipped
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
, ´ , ,
are short, function words. We also know that eye movements are particularly sensitive to textual variables. For example, when the text becomes more complex or contains uncommon or contextually implausible words, eye-fixation duration increases, and saccade length (i.e., small jumps made by the eye to move through text) decreases (Duchowski, ). This variation in fixations can be captured in the gaze duration of readers (i.e., the initial amount of time a reader spends in a region, as defined by the researcher—typically as single words or phrases—from first entering it until the eyes move to another word). Word frequency has the most influential effect on gaze duration: a high-frequency word decreases gaze duration (O’Regan & Lévy-Schoen, ; Rayner & Pollatsek, ) compared to a lower-frequency word, even when length, number of syllables, meaning, and sentence frame are controlled for. Unpredictable words also have immediate effects on fixation duration. Readers tend to look at unpredictable words longer than predictable words and they skip over predictable words more frequently than unpredictable words (Ehrlich & Rayner, ). Moreover, when disambiguating information in a structurally ambiguous sentence is inconsistent with the syntactic interpretation assigned by a reader, there is considerable disruption in eye movement. Thus, participants reading a syntactically ambiguous sentence show long fixation durations at the disambiguating region, launch regressive saccades from the disambiguating point to the syntactically ambiguous constituent, or reread the sentence for a second time. The fact that inconsistencies associated with the structural analysis of a particular word (or collection of words) are noticed by readers as soon as they arise suggests that recordings of eye movements can be very informative when studying the structural decisions that people make during reading. Another major advantage of the eye-movement recording technique is that it allows researchers to obtain evidence about what is happening during the comprehension of a sentence moment by moment, as processing unfolds, without significantly altering the normal characteristics of either the task or the presentation of the stimuli. Eye movements are a normal characteristic of reading, and while eye-movement records are collected, participants are free to move their eyes along the printed line of text. In a similar vein, eye-movement records are highly informative to auditory comprehension (i.e., the visual world paradigm, Tanenhaus et al., ). As in the case of reading studies, lexical (e.g., lexical frequency, phonological cohorts, and neighbourhood density) and grammatical manipulations (e.g., grammatical gender, cloze probability, verb semantics) during auditory processing may induce faster and greater looks to target stimuli in cases of facilitated processing, and slower and reduced looks to target stimuli in cases of delayed processing. The visual world paradigm has helped to elucidate that listeners do not wait until the end of a word before they begin to process its meaning (Allopenna et al., ). Similarly, listeners can anticipate upcoming target items based on the verb’s meaning (Altmann & Kamide, ). Listeners are also adept at using grammatical cues such as grammatical gender, word order, or case marking to anticipate target items in informative contexts (e.g., Lew-Williams & Fernald, ; Hopp, a, for grammatical gender; Kaiser & Trueswell, , for word order; Kamide, Altmann, & Haywood, ; Knoeferle et al., , for case marking). Because participants are freely scanning a visual scene as they listen to auditory input, the visual world paradigm is particularly useful for investigating online sentence processing in populations with reduced literacy, speakers of minority languages and stigmatized forms of speech, as well as language attriters. In the latter case, the visual world paradigm may be a powerful tool for language attrition researchers
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
because it does not require the use of judgement or acceptability tasks, and most participants are not conscious of how linguistic stimuli may impact how they inspect a visual scene. Recent advancements in eye-movement technology also make available eye-tracking equipment that is extremely versatile, and replaces traditional, fixed eye-tracking systems with more flexible head-mounted systems, or remote systems that do not require the use of a headband or head (i.e., chin or forehead) support. In addition, to obtain the dependent measure, participants are not required to perform a secondary task (such as a button press) that might disrupt the normal comprehension process. Furthermore, thanks to several decades of eye-movement research, we have a very good understanding of the amount of visual information processed while our eyes fixate on text or when processing images.
. I
.................................................................................................................................. Eye-movement records have been employed in second language and bilingualism research primarily to ask how proficient speakers of two or more languages manage the presence of two languages in a single mind. If it were the case that these individuals could be characterized as two monolinguals in one head, then this question would not be very interesting. However, as stated earlier, the available evidence from the word recognition and sentence processing literature suggests that when two linguistic systems are housed in a single brain, they interact closely with one another, and these interactions influence the way in which L speakers read and understand spoken words in each of their languages. Below, we review several studies that will serve as illustrations about how eye-tracking methodology can be used to study these interactions.
.. Reconfiguration of the L Valdés Kroff et al. () examined whether intense contact with Spanish–English codeswitched speech had consequences for the processing of grammatical gender in L Spanish. In many bilingual communities, speakers regularly switch from one language to another, often several times in a single utterance. The ability to engage in fluent code-switching is a hallmark of high proficiency in two languages (Miccio et al., ), as successful and fluent code-switching requires a high degree of knowledge of and sensitivity to the grammatical constraints of both languages. Grammatical gender was examined because it is a lexically specified feature of nouns that triggers syntactic agreement with other function and openclass elements, such as determiners and adjectives. Grammatical gender was interesting for yet another reason. In the production literature of Spanish–English code-switches, one widely attested pattern is that when a code-switch occurs within a noun phrase composed of a determiner and a noun, the determiner overwhelmingly surfaces in Spanish and the noun in English, e.g., el building and not the edificio (Liceras, Fernández Fuertes, et al., ; Herring et al., ). Researchers have also documented a production asymmetry in grammatical gender assignment in these mixed noun phrases. The Spanish masculine article el (‘the’) surfaces with English nouns regardless of the grammatical gender of the noun’s translation equivalents—for example, el juice (Spanish ‘jugo’, masculine), el cookie
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
, ´ , ,
(Spanish ‘galleta’, feminine). In contrast, mixed noun phrases involving the Spanish feminine article la (also translated as ‘the’) are rare and occur in restricted environments, such that only English nouns whose Spanish translation equivalents are feminine surface with la in code-switching (e.g., la cookie but not la juice; Jake et al., ; Otheguy & Lapidus, ). These production distributions in Spanish–English code-switching stand in contrast to monolingual Spanish, where the grammatical gender of a noun and its accompanying article must obligatorily match, and where masculine and feminine nouns are evenly distributed (Eddington, ; Otheguy & Lapidus, ). As mentioned above, Valdés Kroff et al. () investigated whether L Spanish–L English speakers who had been immersed in a code-switching environment used grammatical gender information encoded in Spanish articles to anticipate the gender of an upcoming noun (as Lew-Williams & Fernald, , have shown with monolingual speakers of Spanish). Given the asymmetry observed in production data, it seemed plausible that the gender marking of articles would facilitate to a lesser extent the processing of codeswitched speech. To investigate this, the eye movements of Spanish-English bilinguals were recorded. Participants saw two pictures that represented objects whose nouns were either of the same or different grammatical genders, and listened to code-switched sentences (i.e., Hay un niño que está mirando el candy/‘There is a boy looking at the candy’). Words were spoken with a Spanish article that either matched the gender of the word’s Spanish translation equivalent (e.g., el candyMASC , Spanish el caramelo) or did not match (target: el candleFEM, Spanish la vela). A control group of monolingual speakers of Spanish was also recruited. Participants were asked to listen to each sentence and to click on the named object. Where the grammatical gender of the Spanish names for each of the pictures was different between the two pictures, the gender information in the article was informative. Monolingual Spanish speakers showed the expected anticipatory effect on masculine and feminine different-gender trials. Results for the Spanish-English bilingual speakers revealed an anticipatory effect, but only on different-gender trials where the auditory stimulus was feminine. When the auditory stimuli were masculine in different-gender trials, participants did not launch anticipatory looks but rather waited to hear the target noun, meaning that they did not use masculine articles as cues for anticipatory processing. At first glance, these results could potentially be consistent with an attrition account; after all, the bilingual participants experienced what appeared to be a permanent change to their L linguistic system (cf. Bergmann, Meulman, et al., for a discussion that L attrition studies in post-puberty migrants have not found evidence for weakening of gender); however, a follow-up production study with a subgroup of the same participants revealed that these bilinguals were highly accurate at producing the correct gender in Spanish-only noun phrases. The findings suggest that while there may be constraints on the sort of cross-language exchanges in bilinguals, the presence of these influences suggests a dynamic language system that changes in response to contact with other languages and to how the two languages are used among other bilingual speakers. In an eye-tracking-while-reading study, Dussias & Sagarra (; see also Fernández, ) investigated the effect of intense contact with English on the resolution of syntactically ambiguous relative clauses in Spanish. Native Spanish and native English speakers differ in how they interpret temporarily ambiguous relative clauses such as Alguien disparó al hijo de la actriz que estaba en el balcón (‘Someone shot the son of the actress who was on the balcony’). When asked the question ¿Quién estaba en el balcón? (‘Who was on the
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
balcony?’), monolingual Spanish speakers typically respond with ‘the son’ (i.e., high attachment preference) and monolingual English speakers with ‘the actress’ (low attachment preference, e.g., Carreiras & Clifton, ). Dussias & Sagarra found that L SpanishL English bilinguals immersed in an English-speaking environment for a prolonged period of time favoured the low attachment strategy associated with English when reading in Spanish (their native language) in contrast to a non-immersed but proficiency-matched bilingual group who continued to favour a high attachment strategy. Importantly, overall comprehension was not affected (non-immersed group: %, immersed group: %) nor was average reading time. Although not reviewed here, the influence of the first language on the second is a welldocumented and robust finding (see Schwieter, , for an extensive overview). The studies reviewed in this section demonstrate the lesser-known influence of the second language on the first. Taken together, these studies reflect the permeability of the two linguistic systems and not a gradual loss of sensitivity to L processing per se. How do we reconcile these bidirectional influences with an attrition account? Potentially, one criterion is how permanent these influences may be. In the next section, we review whether L speakers can be shifted back into demonstrating monolingual-like processing through reactivation.
.. Reactivation of the L Several recent studies have provided examples of how an L assumed to have undergone attrition can be reactivated despite years of non-use. Park () examined the effects of re-exposure in Korean adoptees living in Sweden. The adoptees had been exposed to Korean before adoption for differing lengths of time, and were all learning Korean (effectively) as an L after years of non-use. To test whether or not re-exposure to Korean was affected by their experience during early childhood, Park administered two phonetic perception tasks to compare these adoptees to L Swedish speakers who were L learners of Korean. While the results were marginally significant, the trends were in the expected direction: the Korean adoptees performed better than their Swedish counterparts in distinguishing both vowels and stops in Korean. Likewise, those who had been adopted at a later age showed the greatest performance overall. Thus, the author concludes that, even after years of non-use and likely attrition, the native language can be ‘reactivated’ with new exposure, resulting in subtle yet detectable differences (see also Flores, Chapter , this volume). More recently, Choi, Cutler, & Broersma () demonstrated that the benefits reported for adoptees’ perception of the sounds of their birth language also transfer to production. In the area of sentence processing, an eye-tracking study conducted by Chamorro, Sorace, & Sturt () examined the effects of language attrition and re-exposure to the L on the use of the differential object marker (DOM) in Spanish. Three different groups were tested: Spanish monolinguals with little to no knowledge of English (who had recently arrived in Edinburgh, UK), and two groups of L Spanish attriters who had been living in the UK for more than five years and were near-native speakers of English. One of the latter groups of language attriters had been re-exposed to Spanish before participating in the experiment. The results were striking: all three groups showed sensitivity to violations involving the DOM in an offline judgement task but only the attrition group who had not
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
, ´ , ,
been re-exposed to Spanish showed reduced sensitivity in online reading task using eye-tracking. Remarkably, the group that had been re-exposed to Spanish exhibited sensitivity patterns similar to the monolingual group after just one week of exposure. But do these findings extend to other language domains? A series of on-going eyetracking experiments seems to suggest so. We mentioned earlier a study by Dussias & Sagarra (), who found that bilingual speakers experienced changes in the processing of their L, and which the authors attributed to exposure to their L. Given this finding, one might wonder whether proficient bilinguals who had adopted processing strategies of the second language when parsing their L can return to using L parsing after a period of reexposure. To examine this question, Carlson et al. (in prep.) first ran a pre-test to identify Spanish-English bilinguals as ‘high attachers’ or ‘low attachers’ while they processed syntactically ambiguous relative clauses (e.g., Alguien disparó al hijo de la actriz que estaba en el balcón ‘Someone shot the son of the actress who was on the balcony’). Next, they exposed participants to a five-day ‘intervention study’ (that ran for one hour each day), during which they read short paragraphs containing relative clauses in which the syntactic ambiguity was resolved opposite to their natural attachment preference. That is, participants who favoured high attachment received a low attachment treatment, and those who favoured low attachment received a high attachment treatment. In addition, within each group half of the participants were randomly selected to receive the intervention in Spanish and the other half in English. Participants returned to the lab after the intervention study to participate in two subsequent eye-tracking studies: one that assessed the immediate effect of the intervention and one that assessed the effect of the intervention a week after it was completed. The findings showed that those participants who originally preferred high attachment switched to a low attachment preference after the intervention, and maintained the low attachment preference by the second post-test. Similarly, participants who originally showed a low attachment strategy switched to high attachment. Critically, whether the intervention was in Spanish or in English did not affect the pattern of results, suggesting that structures that are shared, by virtue of the fact that they overlap between the two languages, can be fairly easily ‘lost’ but also ‘reactivated’. Importantly, the shared nature of the structure seems to confer a particular advantage: exposure to the structure itself rather than exposure to the language can revert an attachment preference, underscoring the highly dynamic nature of the language systems and not prima facie language loss. The studies reviewed here show that when online processing is taken into account, it can be said that even individuals considered to be language attriters still maintain sensitivity to potentially vulnerable elements in their L. This suggests that online measures of language processing can provide a more complete picture of language use in populations of possible attriters, and that even brief periods of re-immersion in the first language can shift processing strategies back to monolingual-like preferences.
. C
.................................................................................................................................. In this chapter, we have reviewed some of the recent developments on the permeability of the native language system, focusing primarily on the sentence comprehension. Our review is far from exhaustive because this is an active area of research that is rapidly expanding,
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
but our hope was to illustrate the view that changes to the native language are better understood as an indication of the openness of the networks that support language knowledge and language use in general, and as a consequence of a fundamental trait of the architecture of the linguistic system. Although the malleability of the native system can be evidenced in both monolingual and bilingual speakers, bilinguals represent a ‘natural experiment’ on the openness of the native language system; changes to the native language in response to the experience and context of use of an L result naturally when individuals come in contact with and use more than one language. At the same time, the use of eyetracking has provided a unique set of insights about emerging questions that we have only had an opportunity to review briefly in the present chapter. If we step back from the specifics of the experiments that we have discussed, the overarching theme in our review is that the bilingual’s two languages are remarkably open to one another, with cross-language interactions that persist from word to sentence processing, and with changes to the native language that provide a model for testing claims about the plasticity of cognitive and neural representations. One unanswered question moving forward is whether there is a threshold when this natural permeability and interaction between the two linguistic systems becomes a more permanent loss. The bilingual’s two languages operate within a rich cognitive network that supports the comprehension processes on which we have focused. It is beyond the scope of our chapter to address this question; however, we hope that in the next phase of research on L attrition, the view that we have presented here represents a counterpoint to constrain what research defines as attrition.
A The authors thank the reviewers for very helpful and insightful comments, and for their close reading of the chapter. The writing of this chapter was supported in part by NSF grants BCS-, NSF grant OISE-, NIH Grant HD, and NIH Grant HD to Paola Dussias.
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
......................................................................................................................
......................................................................................................................
, ,
. I
.................................................................................................................................. C who are raised in migrant families are mostly exposed to their first language (L) from birth. This language is also referred to as their home language (Valdés, ) or heritage language (Valdés, ) (see Part V of this volume). Exposure to the language of the country they reside in typically starts later, making this language their second language (L), and increases in frequency when children grow older and spend more time outside their homes. Previous research has demonstrated that the delayed onset of L exposure can lead to overlap between the language profiles of typically developing (TD) successive bilingual children in the L and monolingual children with Developmental Language Disorder (DLD) learning this language as their mother tongue (e.g., Håkansson & Nettelbladt, ; Grüter, ; Paradis, ).1 In the last decades, a mounting number of studies in the field of DLD have therefore taken a particular interest in bilingualism (Paradis, ; Armon-Lotem, ), mainly focusing on successive bilingualism as a result of migration. Research on the intersection of bilingualism and DLD has been conducted for several reasons. First, the partially overlapping language profiles of successive bilingual children without DLD and monolingual children with DLD can provide more insight into the nature of DLD. For instance, if the errors of children with DLD are similar to those made by bilingual children learning an L, then this could suggest that the impairment affects input In keeping with a recent proposal (Bishop, ), we decided to use the term DLD. Other commonly used terms are Specific Language Impairment (SLI), Primary Language Impairment (PLI), or Language Impairment (LI). 1
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
processing. Within this view, impaired input processing (in children with DLD) and reduced input frequency (in bilinguals) have similar effects on children’s language outcomes (see Paradis, , for a more detailed discussion on this topic). In addition, another more practical, but not less important, reason is related to how the effects of language impairment can be isolated and teased apart from effects of delayed L exposure in diagnostic settings. This is a pressing issue, due to the large numbers of bilingual children on clinical caseloads and lack of appropriate instruments for assessing language proficiency in a bilingual context (Mennen & Stansfield, ; Kohnert, ). Finally, besides studying the overlap between monolingual DLD and bilingual TD described above, investigating the combined effect of bilingualism and DLD in children who are both bilingual and diagnosed with DLD is also important, for theoretical as well as applied reasons. For example, do effects of bilingualism and DLD on language accumulate, and does bilingualism alleviate or aggrevate the symptoms of DLD? The above issues and questions concern the L of successive bilingual children. Less attention has been paid to the L, and therefore to the possibility of L attrition, which is the particular case of language loss at the individual level (Montrul, , p. ). Like a delay in the L, attrition in the L of successive bilinguals might lead to a diagnostic confound: effects of L attrition on children’s language proficiency may resemble effects of DLD on skills in that particular language. Effects of L attrition and effects of DLD could also accumulate and it is possible that DLD influences the quantity and/or quality of L attrition in bilingual children with DLD. In this chapter, we aim to demonstrate that knowing more about attrition is important for accurately diagnosing bilingual children whose language abilities raise concerns. We discuss whether the effects of attrition and DLD are expected to show overlap, whether they would accumulate, and whether DLD would influence attrition. In so doing, we often refer to research that has looked at the L of bilingual immigrant children, as much less is yet known about their L. Section . starts with parallels between the effects of DLD and successive bilingualism on children’s language abilities. Section . continues with the combined effects of DLD and bilingualism on the L and L. Section . highlights factors that may increase the chances of attrition in immigrant children and that may be responsible for the diagnostic confound with DLD. In Section ., issues are discussed that arise when the aim is to isolate the effects of DLD from those caused by attrition. We conclude by summarizing the main issues and suggesting promising avenues for future research in Section ..
. O : D L D
.................................................................................................................................. DLD is an impairment specific to language that affects about % to % of all children (Tomblin et al., ). DLD is commonly defined as an impairment in the comprehension and, more prominently, in the production of language in children; the area that is affected most persistently by the impairment is grammar (Leonard, a). A robust ingredient of
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
, ,
the definition concerns exclusion criteria, specifying the presumed absence of another clinical condition that can adequately explain the language difficulties (Leonard, a; Bishop, ). The inclusion criterion, the language disorder itself, is typically defined in comparison with the language skills of TD. Besides language difficulties, various studies have demonstrated that children with DLD have reduced processing abilities (Leonard et al., ). These limitations surface both in the verbal/auditory and in the visual domain (e.g., Ebert & Kohnert, ; Vugs et al., ; Henry & Botting, ). However, the magnitude of the effect of DLD on verbal processing is often larger than the effect of DLD on visual processing, and, in addition, findings in the visual domain are mixed. For instance, sustained selective attention limitations are found for auditory but not for visual stimuli (Spaulding, Plante, & Vance, ). An important issue, when establishing a benchmark for DLD, is what the nature of the difference between language-impaired children and TD children actually is. Leonard (a) discusses different options. First of all, it could be said that DLD represents a general language delay (‘not only [ . . . ] late emergence of language but also slower than average development’ (Leonard, a, p. )). However, it appears that language development in DLD is not a carbon copy of that in younger TD children. Instead, the difference between DLD and TD may be described as follows: () DLD may be characterized by an abnormal frequency of the same errors also found in younger children. In other words, children with DLD produce more errors than younger TD children, but the errors are the same and emerge in the same language domains. Alternatively, () DLD may represent an uneven profile, in that some domains are more impaired relative to TD development than others, even though the errors are generally similar. Finally, () the difference between children with DLD and TD children may be qualitative in nature. A qualitative difference would imply that children with DLD not only show delayed development, but also deviant development and, crucially, produce errors that are not or hardly ever found in TD children. Rice (, p. ) describes children with DLD as having a ‘delayed departure’, a characterization that refers to a quantitative rather than a qualitative difference. This delayed departure may, however, (come to) affect certain aspects of language, such as tense morphology, more than others, leading to an uneven profile (as described above) or, as Rice (, p. ) puts it, a ‘delay within a delay’. The notion of delay is highly relevant when comparing DLD and bilingualism, as successive bilingual children experience a delay in exposure to the L, potentially leading to (temporary) similarities between the language outcomes of monolingual children with DLD and TD age peers who are learning the language as their L (e.g., Håkansson & Nettelbladt, ; Paradis & Crago, ; Crago & Paradis, ; Grüter, ; Paradis, ). Note that these similarities are expected to be temporary as the delay in L children will be transient, assuming that they are equipped with well-working language learning mechanisms, whereas the language problems of children with DLD are found to be persistent (Rice, ).2 It should be stressed that the additional ways in which monolingual children with and without DLD differ (profile difference, abnormal quantity of errors,
2 For this reason, Reilly et al. () explicitly recommend against the use of the notion ‘delay’ in relation to DLD.
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
deviant structures) are not expected to characterize TD L learners, although in early L development child L learners are likely to go through a phase in which their L shows signs of L transfer leading to non-target like productions (Whong-Barr & Schwartz, ; Zdorenko & Paradis, ). While delayed L development does appear to have a parallel in monolingual DLD, L attrition by definition does not have such a parallel because it is specific to bilingual learning contexts (see definition below). However, incomplete acquisition, another term used in research on bilingualism, and related both to (L) delay and (L) attrition, may be applied to DLD. Anderson () defines the two notions as follows: L attrition occurs when L features are lost as a consequence of the child becoming dominant in the L, while incomplete acquisition is defined by a failure to acquire native-like structures as a result of insufficient exposure during crucial developmental stages. Following Montrul (, p. ), we assume that incomplete acquisition refers to a mature linguistic state and the outcome of language acquisition. In the first edition of his seminal book, Leonard () discussed the possibility that the language abilities of children with DLD ‘plateau’: ‘mastery levels are never reached [ . . . ] a plateau is reached before certain aspects of language are mastered. The basis for such arrested development is presumed to be biological age.’ (Leonard, , p. ). The latter remark echoes the notion of a critical age for language acquisition that is also often invoked to explain lack of success in L acquisition (Hyltenstam & Abrahamsson, ). In a recent article about the language development of children with DLD, Leonard (b) wrote: After a period of late emergence, the pace of subsequent lexical and grammatical development does not appear to be appreciably slower than that seen in TD. However, a plateau effect may start to occur because the late start makes it difficult to reach full mastery at the age when language learning is still efficient. (p. )
Rice (, p. ) adopts a metaphor to describe a process that closely resembles Leonard’s description. With respect to the language development of children with DLD, and the contrast with TD, she writes: ‘Their language train, however, seems to decelerate before reaching the final destination of the adult grammar, not quite reaching the end goal’ (italics added).
. C D L D
.................................................................................................................................. DLD is to a large degree heritable (Bishop, ). By implication, it affects not only monolingual but also bilingual children (Paradis et al., ). In bilingual children with DLD, the joint effects of impairment and bilingualism present researchers and practitioners with a challenge. After all, successive bilingual children with DLD are likely to demonstrate the phenomena that are typical of bilingual acquisition but also the symptoms of their language impairment. If the effects of impairment and bilingualism affect the same
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
, ,
domains of language in a similar way (e.g., a negative effect), the expected net result of both effects would be the sum of both effects. One way to determine if the effects of DLD and bilingualism indeed accumulate is to use a multiple-group design in which bilingual children with DLD are compared with monolingual children with DLD on the one hand and bilingual TD children on the other hand. Adding a fourth group consisting of monolingual TD children enables comparisons of the effect size of the TD–DLD difference in a monolingual and bilingual context. Effect size is, as Paradis () observed, the index that should be used to determine if bilingualism aggravates the effect of DLD. Such an aggravating effect is present if in the bilingual context the difference between TD and DLD is larger (as evidenced by a larger effect size) than in a monolingual context. An alleviating effect is, in contrast, shown by a smaller effect size.3 It is not only relevant to ask if bilingualism aggravates the effects of DLD, but also if DLD aggravates the effects of bilingualism. This latter question is specifically relevant with respect to attrition and will be discussed in Section ... There, we will show that the notions of delay and attrition invoke specific considerations when applied in the context of bilingual DLD.
.. Delay in the L Two Dutch studies by Verhoeven and colleagues (Verhoeven, Steenge, & Balkom, ; Verhoeven, Steenge, Van Weerdenburg, & Van Balkom, ) arrived at the conclusion that bilingual children (L Dutch, various Ls) with DLD have an additional disadvantage in the L because their Dutch language skills showed effects of bilingualism as well as effects of impairment. In this context, these effects of bilingualism, which originate in the input to bilingual children that is distributed over two languages, should not be equated with incomplete acquisition, because it is unknown whether or not children’s language outcomes attested in these studies represent a mature linguistic state. Actually, because the children were rather young ( to years old) and frequently exposed to the L, it is likely that the outcomes do not demonstrate incomplete acquisition but represent delayed acquisition (compared with monolinguals), caused by reduced input. It cannot be ruled out, however, that the delay will ultimately lead to incomplete acquisition. As Leonard (a) notes, the performance gap for bilingual children with DLD (as compared with bilingual TD children) in the studies by Verhoeven et al. was not greater than a similar gap for the monolingual DLD children as compared with their monolingual TD peers. In other words, there was no difference in effect size between monolinguals and bilinguals when comparing TD and DLD, showing that although the effects of bilingualism and DLD may add up, bilingualism does not aggravate the effects of DLD. In a recent study, we compared - to -year-old bilingual and monolingual children with DLD and TD, learning Dutch as their L (bilinguals) or L (monolinguals). The four groups were matched on age, non-verbal intelligence and also socio-economic status to the extent that this was possible. The bilingual TD and DLD groups were furthermore matched on
3 If the effects of impairment and bilingualism are in opposite directions, it is possible that bilingualism reduces the effect of DLD. In this case, the difference between TD and DLD is smaller in a bilingual than in a monolingual context. Such an effect might happen if transfer or cross-linguistic influence in bilinguals leads to accelerated language development.
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
amount of exposure to Dutch. We found that the bilingual DLD group was outperformed by the bilingual TD group as well as by the monolingual DLD group, both on Dutch inflectional morphology and Dutch receptive vocabulary tasks, confirming that effects of bilingualism and impairment add up (Blom & Boerma, ; but see Paradis, ; Gutiérrez-Clellen et al., ; Rothweiler et al., ). For receptive vocabulary, but not for grammatical morphology, the effect of impairment was larger in the bilingual than in the monolingual sample, as shown by comparisons of the effect sizes of DLD in bilingual and monolingual contexts. This led us to the conclusion that bilingualism—or rather reduced input in one language—can aggravate the effects of language impairment in the domain of vocabulary. Because vocabulary is a moving target, as it continues to grow (Brysbaert et al., ) and contrasts in this respect with inflectional morphology, reduced input may have more of an impact on vocabulary development than on the development of grammatical morphology. In line with this idea, Chondrogianni & Marinis () observed that in a sample of Turkish children learning English as their L, vocabulary was affected by environmental factors such as mother’s proficiency and English use in the home, whereas tense morphology was not.
.. L attrition The conditions under which the L of immigrant children is learnt are often less favourable than those for the L, because the L is used by a limited number of people, in a limited number of domains, and proficiency in the L is often valued less by dominant societal norms than L proficiency (see Section ..). The result of such a bilingual learning situation is ‘subtractive bilingualism’ (Wong-Fillmore, ), which refers to the process in which the L is learnt at the expense of the L and gradually replaces the L, resulting in incomplete L acquisition and attrition. This will affect the L development of bilingual TD children and bilingual children with DLD alike, but it may not affect them in equal measure. As discussed earlier, incomplete L acquisition as observed in bilinguals is at first glance not unlike what we see in DLD (‘not quite reaching the end goal’), but the cause of incomplete acquisition in DLD is not primarily a lack of exposure in a crucial developmental phase. Rather, it is a diminished ability to learn language that prevents children from reaching a sufficient language level in a crucial developmental phase. In bilingual DLD, this diminished ability may be coupled with a diminished window of opportunity if input in a particular language is reduced and children stop being abundantly exposed to one of their languages. It is conceivable that these two factors, which are internal and external to the child respectively, add up and lead to a more incomplete L acquisition in bilingual DLD than in bilingual TD (or monolingual DLD). Also, internal factors could enlarge the effects of external factors, leading to aggravation of L attrition in bilingual DLD compared with bilingual TD. To our knowledge, there is a hardly any research on attrition in children with DLD, but, with respect to the influence of DLD on attrition, a comparison of two cases by Restrepo & Kruth () may be insightful, keeping in mind that observations based on cases cannot be generalized to the population. Restrepo & Kruth () followed two -year-old Spanish-English bilingual children in the United States, one with and one without DLD, over a period of one year. Despite similar patterns of exposure, the child with DLD appeared to lose Spanish faster in comparison with the TD child, as indicated by a decrease in sentence length and a faster increase in errors with grammatical gender and verb
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
, ,
inflection. The same errors were also attested in other TD children, but these children seemed to show a more protracted pattern of attrition (cf. Anderson, ), suggesting that DLD affects quantitative aspects of attrition and speed more than qualitative aspects of attrition (such as the types of errors).
. F L
.................................................................................................................................. Attrition does not necessarily affect all bilingual migrant children. One factor that influences attrition, which is part of the definition by Anderson (), is dominance. A dominant bilingual is ‘someone with greater proficiency in one of his or her languages’; who ‘uses it significantly more than the other language(s)’ (Li, , p. ). L attrition happens as the result of becoming dominant in the L (Anderson, ), a dominance shift which, in turn, is caused by more exposure to and use of the L. In addition, there are other factors that might contribute to attrition in the L. If the effects of attrition are exacerbated by a language impairment (Section ..), these factors likely have an even larger effect on the L abilities of bilingual children with DLD. Moreover, it is important to be aware of these factors when deciding on (the level of) attrition and comparison with appropriate norms, which is an issue we turn to in Section .. Below, the role of age of exposure to the L, a child-internal factor, and socio-economic status and L status, two child-external factors, will be discussed. This is by no means an exhaustive list of factors, but they are chosen as an illustration and because these factors have been investigated in recent research on children becoming bilingual in immigration settings.
.. Age of exposure to L and L development Recently, Bedore et al. () explored the extent to which age of exposure to L English impacts the development of L Spanish in Spanish-English bilingual children in Grade , with an average age of months, and Grade , with an average age of months. Their findings suggest that children who start acquiring English the earliest scored the lowest on Spanish, indicating that early L exposure may come at a cost to the acquisition of the L. From the perspective of distributed input this pattern is not surprising. When L exposure starts, input becomes divided across languages and this might prevent children from acquiring new grammatical constructions in their L (Montrul, ). An early age of L exposure may particularly diminish a child’s chances of acquiring those L constructions that tend to be acquired late. The consequence could be that the earlier a child’s attention shifts to the L, the earlier L development may come to a halt (stagnation) and may, subsequently, start to show attrition. At this point we want to emphasize that this is one developmental scenario out of many, as L development in successive bilingual children does not necessarily stagnate, but could also continue to grow at a slower pace due to an increase of L exposure. Stagnation is, moreover, not necessarily followed by attrition, but could also lead to incomplete acquisition, which is a relatively stable state. In addition, growth, incomplete acquisition, and attrition could co-occur in an individual
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
child. A child may show growth on one aspect of language, incomplete acquisition on another and at the same time lose other aspects of the L (Montrul, ).
.. Socio-economic status: parent–child interaction and L development Children from low socio-economic status (SES) homes, whose parents have a background of low education, income and/or occupational prestige, have been found to lag behind in their language outcomes compared with children from higher SES homes (see, e.g., Hoff, , for an overview). Maternal education, specifically, may influence vocabulary development, regardless of culture (Hoff & Tian, ). The seminal study by Hart & Risley () suggested that the smaller vocabularies of children from lower SES families are related to parent–child interaction patterns. The importance of parental input in explaining the effect of SES was further confirmed by Huttenlocher et al. (), who observed that SES effects on lexical, constituent, and clausal diversity in children’s language are, in part, mediated by caregiver speech. Independent of impairment or reduced input in bilingual contexts, SES is a factor that may prevent children from reaching language milestones, even though the effect of SES may not be as persistent as suggested by some studies and could influence certain aspects of language more than others (Letts et al., ). Accumulating effects of low SES and DLD were not found in the study by Roy & Chiat (), who observed similar language profiles across children with DLD from high and low SES families. However, effects of bilingual exposure and SES may add up: in a study by Calvo & Bialystok (), bilingual children growing up in working class families scored lower on receptive vocabulary tests than monolingual children from working class families, but also lower than bilingual children from middle class families and monolingual children from middle class families. If the effects of reduced exposure in bilingual contexts and low SES add up, the prediction could be that incomplete acquisition and attrition are more prominent in children from lower SES backgrounds than in children from higher SES backgrounds. Low SES can be more prevalent among certain groups of immigrant children. For instance, in the Netherlands, many unskilled Turkish and Moroccan migrant workers came to the country because of labour migration and family reunification (Swanenberg, ). In these families, low SES may go hand in hand with limited stimulation in the L, in particular when the L is further suppressed due to an increase of L exposure when the children growing up in these families start to attend school. However, low SES has also been found to be associated more with L than L use at home, at least by Turkish mothers in the Netherlands (Prevoo et al., ; but see Goldberg et al., , for a somewhat different pattern in migrant mothers in Canada), suggesting that low SES can also create a relatively favourable context for L development and may reduce L attrition.
.. Language status and L development In many cases, the L of migrant children, also labelled minority language, lacks a high social status—and is thus low in prestige in the host country, in sharp contrast to the
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
, ,
majority language, which is the language used in educational settings and official media. This applies to, for instance, Spanish or Somali children in the United States or Turkish children in the Netherlands and Germany. As a result, there are fewer opportunities for those children to develop and use the L, and fewer resources are available to support their L development compared with the L. For instance, literacy may be exclusively associated with the language of school, which is the L. Reading and literacy-related activities will therefore often be done in the L, supporting L rather than L development (Prevoo et al., ). Moreover, high proficiency in the L is generally not highly valued, unlike reaching a high language level in the L. This situation is the typical subtractive context in which the L is likely to replace the L, as described earlier (Section ..). This contrasts with additive contexts in which the acquisition of L is supported and can facilitate L learning, as in, for instance, Canada where both French and English are majority languages (Baker, ). The effect of L status varies across minority languages, as illustrated in a study by Scheele et al. (), who investigated vocabulary development in the L and the L of bilingual Turkish-Dutch and Moroccan-Dutch (L: Tarifit-Berber) preschoolers. The Dutch vocabulary scores of the Turkish children were lower than those of the Moroccan children, which was related to the more frequent use of Dutch in the Moroccan homes. This difference between the Moroccan and Turkish migrant families can be explained by a myriad factors (Laghzaoui, ). Tarifit-Berber is a predominantly oral rather than written language in which fewer resources (books, television) are available than in Turkish. Its prestige is moreover low and, in the Moroccan community in the Netherlands, Dutch is often used to communicate among each other because the different Berber varieties are not mutually intelligible. As a consequence, attrition may happen more often and to a larger extent in Moroccan individuals who learn Tarifit-Berber as their L than in Turkish individuals. To our knowledge, there is, however, no research that has systematically compared the levels of attrition across speakers of Tarifit-Berber and Turkish in the Netherlands. Language shift, the loss of the L that concerns an entire speech community (Fishman, ; Lambert & Freed, ), may also be more prominent in the Moroccan than in the Turkish community in the Netherlands. If language shift occurs in an immigrant community, it creates different characteristics in the child’s language input, which is relevant for every child learning a language, but could create a specific risk for children with DLD. Restrepo et al. () approach bilingual DLD from the perspective of Dynamic Systems Theory. Crucially, in that framework, language abilities in multilingual children change depending on the quantity and quality of input needed to induce a change. When input is strong, language changes can be positive on all the languages that the child speaks, but they can also be negative for one language and positive for the other. (p. )
When we consider language shift, the quantity and quality of the input may be at stake. Since children with DLD already have more difficulty learning language, this can be a serious threat to the acquisition of the L and, as a consequence, it might affect children with DLD even more strongly than their TD peers. However, given the sparseness of relevant research specifically addressing the impact of these variables in bilingual DLD, predictions can only be tentative.
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
. I D L D
.................................................................................................................................. Successive bilingual children experience a delay in L exposure. In such cases, evaluating the child’s abilities in the L (in which they have had input from birth) could be more insightful than assessing the L. However, like L assessment, L assessment is not always straightforward. In this section, we discuss the ways in which attrition may complicate L assessment. In addition, possible alternatives are described when reliable assessment cannot be ensured, either in the L or in the L.
.. Assessing the L: comparing an individual child using appropriate norms In order to isolate effects of DLD from effects of attrition or bilingual development in a more general sense, information could be gathered on estimated language dominance, as attrition typically happens in a situation when the L becomes dominant (Anderson, ). Many children who become bilingual in immigration contexts are more dominant in their L at younger ages. They gain experience with the L when they spend more time outside their homes in educational settings and have increased interactions with the broader community, most typically leading to a dominance shift (see Kohnert, , and references cited therein). The timing of this dominance shift, and the associated likelihood that a child may show attrition, will thus depend on the child’s age of L exposure as well as the amount of exposure to and use of both languages. Consequently, it is of paramount importance to collect relevant background information on a child’s linguistic environment as part of a multilingual anamnesis. On top of this, isolating effects of DLD requires a comparison with normative data collected from TD bilingual children with the same language background and level of attrition. However, in particular, degree of attrition is hard to assess and measuring dominance might, in this case, be an alternative.
.. Assessing the L: linguistic markers and valid tools When assessing the L, it is important to be aware of the fact that the effects of DLD and attrition might be easily confounded because they can affect the same areas. For instance, errors with verb inflection are not only linguistic markers of DLD in Spanish, but they are also a characteristic of children who undergo loss of Spanish (Anderson, ). Not all linguistic markers that are indicative of DLD in monolingual contexts can thus be generalized to L assessment in bilingual contexts, unless appropriate normative data are available. Note that the same holds for the assessment of the L of bilingual children, as a delayed L development can also show similarities with an impaired language development (as described in Section .).
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
, ,
Moreover, the L in the host country may be subject to language change as a result of language contact. The implication is that the validity of assessment tools for the L created in the country of origin is restricted in the host country. For example, in Turkish, transitive verbs mark the direct object with accusative case, depending on the position of the object and definiteness. Case marking is vulnerable in DLD across many languages, including morphologically rich languages (Leonard, ), and might be seen as a clinical marker. However, in Turkish multiword expressions as spoken in the Netherlands, accusative case can also be omitted or substituted, presumably under the influence of Dutch, which has no case marking (Doğruöz & Backus, ). Because the input (and thus the target language) properties may thus differ between Turkish speakers in Turkey and the Netherlands, assessment tools developed in Turkey may lack validity in the Dutch context. Note that this process of language change is not necessarily equated with language shift, as the Turkish community has a strong tendency to maintain their L (Backus, ).
.. Possible alternatives: measures that are insensitive to L attrition (or L delay) The ideal situation, in which a child’s level of attrition is estimated correctly and relevant normative data and assessment tools are available, is often out of reach. The question arises as to what professionals could do if reliable assessment in the L or in the L cannot be ensured. In this section, we review two possible alternatives. One option is the use of language processing measures instead of standardized tests of vocabulary and grammar, which are based on previous knowledge and thus are crucially susceptible to effects of input (i.e., experience with the language) (Campbell et al., ; Kohnert et al., ). Knowledge-based measures may underestimate the language learning abilities of children who have less experience with the target language, in particular bilinguals, because knowledge of one specific language is tested (Restrepo & Silverman, ). Processing-based measures, in contrast, tap into the cognitive underpinnings of language and are less biased towards language-specific knowledge (Chiat, ). Factors like attrition will therefore affect performance on processing-based measures to a lesser extent than performance on knowledge-based measures. Taking this idea one step further, Chiat () developed a non-word repetition task—a processing-based measure that taps into verbal short-term memory—that includes features that are common across languages, enabling children to use knowledge that is not specific to the L or the L. In a study by Boerma et al. (), bilingual children did not differ from monolingual peers on this ‘quasi-universal’ task, but performed more poorly on a non-word task that was based on Dutch phonotactics. The quasi-universal non-word repetition task was insensitive to the effects of bilingualism but sensitive to DLD, and its diagnostic validity was excellent in both a monolingual and bilingual context. Another type of task that may support the identification of DLD in bilingual child populations is a narrative task. Narrative tasks can target a story’s microstructure, focusing on a child’s linguistic expression of story events, or macrostructure which refers to the structure of the story, including plot elements such as the goals of the protagonists (Stein & Glenn, ). In contrast to microstrucure, measures of narrative macrostructure are relatively independent of experience in a specific language and do not disadvantage
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
bilinguals (Pearson, ; Hipfner-Boucher et al., ), which may be explained by the finding that, unlike in microstructure, linguistic factors do not predict children’s expression of the macrostructure of a story (Blom & Boerma, ). Instead, cognitive limitations, and specifically sustained attention, have been found to influence performance on narrative macrostructure (Blom & Boerma, ). While sustained attention is compromised in children with DLD (Ebert & Kohnert, ), it is not affected by bilingualism (see Bialystok et al., ). Hence, even when there is attrition, bilingual TD children may still perform well on a narrative task when analysed on the macrolevel. Bilingual peers with DLD, however, might perform more poorly due to limitations in sustained attention. A study by Boerma et al. () confirms these predictions. Large negative effects of DLD were found on the measures of narrative macrostructure, whereas no effects of bilingualism emerged. As a result, the diagnostic validity of the narrative task was adequate in both a monolingual and bilingual context. In general, in this section we showed that tasks which require less language-specific knowledge (or less experience with a specific language) are more valid indices of impairment in a bilingual child than tasks that do require such knowledge in those cases where reliable assessment in either the L, due to attrition, or the L, due to delay, is not reliable. These tasks should, however, not be used as stand-alone tools, and, instead, the use of multiple instruments is recommended.
. C
.................................................................................................................................. In this chapter, we discussed attrition in relation to DLD, asking the question whether the effects of attrition and DLD would show overlap, whether they would accumulate and whether DLD could influence attrition. What emerged from this discussion is a potentially very interesting and urgent field of study, but one in which hypotheses have to be pieced together on the basis of evidence from various subfields that have until now not been directly associated: L development, L attrition, and language impairments such as DLD. Another issue that we discussed and that directly relates to our topic is the possibility that L attrition leads to a diagnostic confound due to the fact that both L attrition and DLD affect children’s language level, albeit for different reasons. Because of a dearth of research on L attrition in the context of DLD, our point of departure was often research on the L of bilingual immigrant children. This research could be informative because the same mechanism—limited exposure and usage—underlies both L attrition and L delay. Moreover, linguistic phenomena that are susceptible to attrition effects might also be prone to delay, following the ‘last in, first out’ principle that follows from the regression hypothesis (Jakobson, ). This hypothesis states that those linguistic features which are learned last are the first to show attrition, and vice versa (Keijzer, ). However, an important difference between L and L development concerns the conditions under which the two languages are learnt, which are often more favourable for the L and more adverse for the L in migrant families. Because of their language impairment, children with DLD may need more input to reach, or maintain, the same language level as their TD peers (e.g., Gray, ; Rice et al., ; Riches et al., ; for a meta-analysis, see Kan & Windsor, ). We hypothesized
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
, ,
that, consequently, the L may erode relatively fast in children with DLD. This effect may be modulated by language-level and child-level factors, both internal and external, such as input-dependency of linguistic phenomena, age of L exposure, socio-economic status, and status of the L. Understanding more about L attrition is furthermore relevant for diagnostic purposes, as overlap between the effects of L attrition and DLD could create a diagnostic confound. Regarding the diagnosis of DLD in bilingual contexts, it is generally recommended that both the L and L are assessed. However, like delays in the L, attrition in the L presents a complicating factor that clinicians need to be aware of, because of the possibility that it renders L normative data unreliable and assessment tools in the L invalid. For this reason, new instruments have been developed that are sensitive to the effects of DLD, but relatively insensitive to the child’s knowledge of a specific language and, therefore, to effects of bilingualism. That these instruments are promising has been shown by the equal performance of monolinguals and successive bilinguals in their L. Future research has to show if the outcomes are indeed replicated for their L, regardless of L attrition. Moreover, to fully understand the interaction of L attrition and DLD, future studies need to longitudinally follow the L and L development of bilingual children with and without DLD, next to systematically keeping track of patterns of L and L exposure and use. Although L attrition may even then be difficult to disentangle from incomplete acquisition, language shift, and language change, longitudinal work would be a good starting point to investigate this understudied reality. It may allow us to identify a plateau in the L, determine if children’s development stabilizes after reaching the plateau or if linguistic features start to show attrition due to L dominance, and keep track of the speed with which attrition takes place. Such longitudinal research might inform us about the nature of attrition in children with DLD and allow us to address attrition in this group in a more direct manner than we were able to do in this chapter, due to the present lack of resources.
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
......................................................................................................................
Lexical retrieval, language use, and cognitive and neural changes ......................................................................................................................
, , , , .
. I
.................................................................................................................................. A- native-language attrition most often occurs following a move to an environment where a non-native language is spoken. To date, the focus of attrition research has been on identifying what aspects of language are lost and which factors are crucial for the retention or attrition of the native language. Attrition is a gradual and fairly subtle process with no clear beginning or end. To best assess the effects of attrition, researchers tend to choose study participants who have lived in a non-native environment for a decade or longer (e.g., de Bot & Clyne, ; Schmid, ). The assumption is that the longer they have been removed from the native-language environment, the greater the degree of language attrition that should be observable. However, this principle regarding length of time and its relationship to language use overlaps with another, largely ignored, phenomenon: language changes associated with ageing. Are language changes due to long-term disuse conflated with age-related language changes in older adults who experience language attrition? This chapter explores changes to the adult lexicon as a result of attrition and ageing since the lexicon is considered a vulnerable part of the language system in both attrition and ageing. We consider neurophysiological changes that may play a role in language attrition and in non-pathological ageing to speculate whether the neurobiological sources of these two processes are similar or different. If attrition and ageing exert independent effects on
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
, , , ,
lexical retrieval decline, we must consider the effects of each of these factors on word retrieval for older adult bilinguals immersed in a non-native-language environment.
. L
.................................................................................................................................. Language attrition frequently occurs when a person’s use of the first language decreases or ceases completely, often as a result of moving to an environment where another language is primarily used. One of the earliest noticeable features of attrition is word retrieval difficulty (Opitz, ; Kasparian & Steinhauer, b). This difficulty may be manifested in slower lexical retrieval, retrieval failures, increased tip-of-the-tongue (TOT) states (when the speaker knows a word but cannot access its full phonological representation), as well as an increase in disfluency markers (e.g., hesitations, filled pauses, self-corrections, and repetitions) (Schmid & Köpke, ; Schmid & Beers Fägersten, ). Word-retrieval problems in attrition are reported to be greater for low-frequency words (such as accordion) than for high-frequency words (such as guitar) (e.g., Hulsen, ). There are several issues to keep in mind when considering word frequency effects, however. First, word frequency counts are likely to differ for bilinguals and monolinguals, whether or not bilinguals are in an attrition scenario (i.e., living in an L-environment). Since bilinguals divide their time between their two languages, they have fewer overall encounters with words in each of their languages than monolinguals. Given that the highest frequency words will be encountered frequently by both bilinguals and monolinguals, differences between the groups are smaller for high-frequency words and become larger for lower frequency words (Gollan et al., ). Gollan et al. () explain this phenomenon in terms of vocabulary size and a lexical search process that uses frequency-based rankings. If lexical access involves a search through a frequency-ordered list, access time would be a function of lexicon size. Bilinguals have on average double the lexicon size as monolinguals, which may account for the slower overall lexical retrieval speed of bilinguals compared to monolinguals and the greater frequency effect observed for bilinguals. Second, cognates may have a special status in the lexicon that can boost word frequency counts. These words have overlapping semantics and phonological forms in two languages (such as the English and Spanish words flower and flor). Production of cognate words seems to benefit from encounters in either language (Costa, Caramazza, & SebastiánGallés, ), and attriters are better able to retrieve cognate words than non-cognate words (Ammerlaan, ), suggesting that the similarity with words in the second language preserves their accessibility in the first language. Thus, patterns of performance seen in attrition may be more extreme versions of patterns that are typical for bilinguals in general. Reduced accessibility of low-frequency, non-cognate words has particular implications for lexical specificity because many highly specific words (e.g., finch) are lower in frequency than their superordinate counterparts (e.g., bird). The result is that attriters’ accessible vocabulary tends to consist of more general words. For example, Olshtain & Barzilay () investigated lexical retrieval accuracy using a structured narrative task in which Americans living in Israel for eight years or more described the story in the picture
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
book Frog, Where Are You? They found that the participants experienced more difficulty when trying to retrieve low-frequency words in the pictures (such as jar and gopher), and they often substituted the specific low-frequency term with a more general one sharing the same semantic features (e.g., producing bottle or container instead of jar). Another strategy used by the participants was paraphrasing and circumlocuting, such as describing the floor as the ground of the living room. Although such studies show that attriters fail to access certain lexical items that they presumably knew at an earlier time, they do not necessarily demonstrate that this group of bilinguals has ‘lost’ these words completely. In fact, when participants are presented with a picture-name recognition task (which does not require production, as opposed to a recall task), they are often able to correctly identify the names for pictures that they previously could not retrieve and express (Ammerlaan, ). Consequently, it appears that it is the accessibility to those items required for production that suffers as a result of the forces of attrition rather than complete loss of the lexical information. Reductions in expressive vocabulary can be investigated by analysing the lexical diversity of attriters’ and non-attriters’ speech or writing. Lexical diversity is often calculated using the Type–Token Ratio, the ratio of the number of unique words (types) to the total number of words produced (tokens). The higher the Type–Token Ratio, the more variety there is in the lexical choice of the speaker. Comparing Type–Token Ratios, Schmid () found that lexical diversity was lower for attriters than non-attriters, even for attriters who had maintained high proficiency in the first language. Laufer () found an inverse relationship between length of residence in the second-language environment and lexical diversity of the first language, as measured by the Type–Token Ratio. An even stronger effect of length of residence was found for lexical richness, which was measured as the proportion of infrequent words in the text (i.e., words not occurring in the , most frequent words). Using a more complex measure of lexical diversity that incorporated not only types and tokens but also distribution of tokens, distance between repetitions of the same type, and rarity of words used, Schmid and Jarvis () were able to discriminate between attriters and non-attriters while simple Type–Token Ratio measures were not able to discriminate between these groups. Thus, discourse analysis measures indicate that attriters’ lexical production is less diverse and that the longer the time spent in the second-language environment, the less diverse the lexical choices become in the first language, with fewer low-frequency words chosen.
.. Causes of lexical attrition Attrition studies have often focused on identifying whether the source of the deviant patterns seen in attriters is due to changes in the memory system or interference from another language. However, this is a bit of a false dichotomy. For example, slower response times on a lexical retrieval task may be due to reduced accessibility to the item in the memory system as well as the need to overcome interference from a strong second-language word. Indeed, as Schmid & Köpke () have pointed out, the source of word-retrieval difficulty is likely to be a combination of both. Even in cases of obvious cross-linguistic transfer, such as the use of the English phrase walk the dog in Spanish (caminar al perro) instead of the standard Spanish phrase sacar al perro
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
, , , ,
(take the dog out), the phrasal borrowing may be accompanied by or facilitated by sources of memory decay. A long tradition of research on memory and ‘forgetting’ provides insights into possible sources of language attrition that might stem from disuse. Ecke () presents seven sources of ‘forgetting’ that, he argues, are common for both language and non-language information. Two of these include the decay of memory representations and a loss of access to memory representations. Memory for a variety of different skills or types of knowledge weakens with periods of disuse or exposure, and language is no exception, though rates of memory decay may vary with different knowledge systems and even different aspects of language (Arthur et al., ; Köpke, ). In connectionist models, the strength of connections is increased with each new occurrence (Ellis, ; McMurray et al., ). So, for example, the connection between the concept of a cat and the phonological form in English (kæt) is strengthened with each use. Lindsay & Gaskell () suggest that in the case of language loss, the connection weights in the network may still contain some vestige of the original knowledge, which may be accessed implicitly but not through explicit tests like assessments of vocabulary or grammatical knowledge. If lexical access challenges in language attrition are due to weakening of the formmeaning relations as a consequence of disuse, then we would expect to find that lower amounts of continued L contact and longer residence in the host country would lead to greater signs of attrition. Typically, the more contact one maintains with the L, the less attrition will occur (e.g., Soesman, ; Cherciov, ; Kargar & Rezai, ), but some studies have failed to find an effect of L contact (e.g., Schmid, b). This may be because the type of contact might matter more than the amount. Schmid & Dusseldorp () found that using the L for professional purposes (i.e., at work) was a significant predictor of L maintenance, while Hulsen () found that the number of different L contacts a person had positively influenced L maintenance. Surprisingly, length of time in the L environment seems to have little predictive power for degree of L attrition. This may be because the relationship is non-linear, with more accelerated rates of attrition in the first decade after immigration compared to later years (Schmid, b) or because length of residence is only a strong factor when L contact is very low (de Bot et al., ; Soesman, ; Schmid, b). The lack of clear correlations between either amount of L use or length of residence in the L context and the degree of L attrition make an explanation of attrition based primarily on weakening of connections more difficult. Another possible source of language attrition is competition from the other language(s). Ecke () describes the loss of access to information due to cross-linguistic processes as stemming from interference from the newly learned language and interactivity between the languages. For example, attriters may learn new collocations in their second language (e.g., Spanish: tomar una decisión, literally: take a decision), and when asked to judge whether or not a collocation exists in their first language, they find the translations of a large number of foreign collocations acceptable (% in Laufer, ). Words in the first language may also undergo semantic extension or reduction due to differences in the semantic scope or features of translation equivalents (Pavlenko & Malt, ). Pavlenko () describes four types of cross-linguistic influence: borrowing words from the second-language when no first-language equivalent exists, restructuring the semantics of L items to accommodate the semantic scope of the translation equivalent, convergence of meanings from both
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
languages, resulting in a unique form, and shifting of the L items toward the L norm. While the first type is an example of lexical enrichment (i.e., additions made to the mental lexicon), the other three types represent deviation from an L ‘norm’, which may be taken as evidence of language attrition.
. N
.................................................................................................................................. Very little research has been done to investigate the neurophysiological changes that underlie the attrition of acquired skills such as language use. Two studies have used electrophysiological measures (Event-Related Potentials (ERPs)) to examine the brain processes of attriters during lexical processing. Datta () investigated how lexical characteristics of the two languages affect processing of words in each language. She contrasted ERP responses to English and Bengali words with high and low familiarity. Low-familiarity words typically elicit a larger N component than high-familiarity words. Using a picture-word matching paradigm, she found that native Bengali speakers who had been residing in the United States and had become English-dominant showed an effect of English familiarity on the amplitude of the N but no effect of Bengali familiarity, demonstrating that these attriters were more sensitive to properties of the English names than they were to properties of the Bengali names. Greater N amplitudes have been proposed to reflect more difficult lexical access, driven by both recent and longterm experience (Kutas & Federmeier, ). Datta’s () study thus revealed that highly familiar English words have a processing advantage for the attriters while highly familiar Bengali words do not have such an advantage as both high- and low-familiarity Bengali words showed similar amplitudes for the N. These findings suggest that attriters’ sensitivity to lexical properties such as frequency/familiarity can decrease with reduced exposure to the language. Another study that used ERPs to investigate lexical processing in L attriters examined whether attriters recognize a violation when ‘confusable’ word pairs in Italian (words that differ only in their final vowel, such as cappello (hat) and cappella (chapel)) are swapped in a sentence context (Kasparian & Steinhauer, b). They found that Italian speakers who had immigrated to Canada in adulthood and had lived there for on average eleven years showed the native-like pattern of an N effect when a word in a sentence was substituted by its highly confusable pair-member but only when their proficiency in Italian remained very high. The researchers also included sentences with more blatant word-swap violations such as The fisherman wears the briefcase of wool. Longer length of residence in Canada was associated with a weaker N effect for these types of sentences, though not for the confusable word-swap types. The authors suggest that language proficiency is the principal driving force of the N effect for word-swap violations, which is bolstered by the fact that Italian second-language speakers with high proficiency in Italian also showed an N effect in their study. This adds to what Datta () found and supports the idea that sensitivity to both lexical characteristics like familiarity and word form-meaning relations is weakened when language use and proficiency decline.
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
, , , ,
The underlying neurophysiological changes associated with language attrition are still unknown. The ERP evidence described above reveals that neural processes underlying lexical comprehension (either in isolation or within sentence context) change in the context of attrition in a way that suggests a loss of sensitivity or differentiation to factors that native-speaking non-attriters are sensitive to. Attriters also appear to exhibit greater effort in attempting to reconstruct a sentence meaning when a semantic violation is detected, as evidenced by a greater P for attriters compared to controls in Kasparian & Steinhauer (b). What these patterns represent at a cellular level, though, is still unknown. A good deal of attention has been applied in recent years to questions of experience-dependent neuroplasticity and the effect of various types of training or expertise on brain morphology (for a review, see Lövdén et al., ). Research using animal models has demonstrated that skill acquisition results in synaptogenesis, alterations in synaptic connectivity, dendritic lengthening and branching, and functional reorganization of neural networks (e.g., Kleim & Jones, ; Lövdén et al., ). Surprisingly little research has examined the neural changes involved in skill loss, likely due in part to the extended time scale of skill loss as compared to skill acquisition. It appears that the ‘use it or lose it’ principle would apply here, although behavioural data has shown us that in many cases a skill that has not been used for a long time may be dormant rather than lost, as evidenced by the speed of re-acquisition, as if the skills simply needed to be ‘awakened’. This rapid re-learning has been demonstrated also in cases of first language attrition (Higby & Obler, ). The strongest evidence for the ‘use it or lose it’ phenomenon comes from studies of sensory deprivation in animals and humans (e.g., Hubel & Wiesel, ; Bedny et al., ). It is not entirely clear, however, whether the same types of neural reorganization apply to disuse as they do to deprivation. In the case of language, there is rarely a complete cessation of native-language use, but rather a reduction of use, which may be abrupt or gradual, and which may or may not fall below the level of use of the second language. Thus, the neural changes involved in language attrition may be both quantitatively and qualitatively different from the types of effects seen in deprivation cases. Competition from the second language may explain the possible neurophysiological mechanisms of attrition effects better than weakening due to disuse. Functional neuroimaging studies show that second-language processing uses many of the same neural circuits as native-language processing (Higby et al., ). At a more fine-grained level, the activation patterns for the first and second languages are likely partially overlapping and partially distinct. Bilinguals have to map two different word-forms to the same concept. Commonly, attriters use the second language more often than they use the first. Thus, the neural populations involved in going from meaning to the second-language word-form during lexical retrieval become more practised and automatic. This creates a situation in which the second-language route is primed by recent and frequent use. When attriters attempt to access the now less-used route from meaning to the native-language word-form, there is competition from the primed route. Current neuroimaging techniques have not been able to demonstrate this type of competition at a fine-grained level, but as technology improves, we should be able to test whether this type of competition is indeed affecting access to the first-language lexicon. In sum, the findings with regard to lexical attrition of the first language demonstrate that attrition occurs even under conditions where proficiency has been maintained. Attrition
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
manifests as retrieval slowing and retrieval failures (particularly for low-frequency, non-cognate words), reduced lexical diversity and richness, as well as semantic restructuring and other modifications to lexical knowledge. The amount of attrition correlates with certain language-use factors, but length of residence in the L environment has little effect on the degree of attrition. The precise neural processes underlying these behavioural changes are still unknown, but recent ERP evidence suggests that neural sensitivity to certain aspects of the attriting language is reduced, particularly as proficiency declines.
. L
.................................................................................................................................. As people age, they experience declines in memory and some aspects of language. Word retrieval problems are one of the most pervasive complaints among older adults (e.g., Ossher et al., ). In laboratory studies, which often use confrontation naming tasks to probe word retrieval abilities, older adults show lower accuracy and longer response times for lexical retrieval tasks such as picture naming compared to younger adults (Au et al., ; MacKay et al., ; Goral et al., ; Verhaegen & Poncelet, ). These age-related declines are gradual, making it difficult to determine when they begin, but researchers have suggested that declines in accuracy become apparent after about to years of age (Nicholas et al., ; Albert et al., ; Kavé et al., ), though retrieval speed may begin declining after age (Verhaegen & Poncelet, ). Word-finding difficulties have also been found in the narrative discourse of older adults. In a picture description task by Schmitter-Edgecombe et al. (), while % of younger adults had behaviours indicative of word-finding difficulty (e.g., substitutions, repetitions, fillers, etc.) in more than % of utterances, the rate of older adults who showed the same level of word-finding difficulty increased to % for older adults in lower age ranges and % for older adults in upper age ranges. Like the attriters described in the previous section, substitutions made by older adults generally consist of semantically related words, vague terms, and circumlocutions (Nicholas et al., ; Obler & Albert, ; Albert et al., ). Schmitter-Edgecombe et al. () reported that the number of substitutions and word reformulations increased with age, indicating a continued increase in word finding difficulties with advancing age. However, unlike the attriters, older adults show greater lexical diversity (as measured by the Type– Token Ratio) than younger adults (Kemper & Sumner, ). This may be due to a critical difference between attriters and older adults: vocabulary size. Older adults continue to add new words to their mental lexicon throughout adulthood (Verhaeghen, ; Goral et al., ). This levels off and/or declines only at much older ages, such as in the s or s (Alwin & McCammon, ; Singer et al., ). Furthermore, vocabulary size predicts word retrieval accuracy, even after one accounts for education (Albert et al., ), suggesting that continued increase in vocabulary in adulthood provides protective effects against age-related word retrieval decline. Unlike decline seen in attrition, the lexical retrieval decline seen in ageing is not modulated by word frequency; older adults show similar declines for both high- and low-frequency words (Newman & German, ). Some research shows that older adults
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
, , , ,
can access more low-frequency words than younger adults (Gollan & Brown, ) and choose lower frequency words in production (Kavé et al., ). However, word familiarity, a related measure to lexical frequency, does affect naming accuracy in older adults, with better accuracy observed for more familiar words (Newman & German, ). While word frequency is calculated as the relative frequency of appearance across a variety of (usually printed) sources across a corpus, word familiarity is a subjective measure based on scalar ratings. Word frequency and familiarity typically correlate, but they do not measure the same thing, as there are many examples of words that are rated as highly familiar (e.g., puddle) but which are relatively uncommon in corpora, and thus are considered lowfrequency items. Newman & German () suggest that highly familiar words may be in some sense protected from age-related declines while the same does not appear to be true for high-frequency words. The source of word retrieval problems in ageing appears to be related to access to the word-form rather than loss of word knowledge, at least until a much older age or the onset of pathology (e.g., Barresi et al., ). This is similar to the proposals put forth for lexical decline in attriters. The majority of the evidence for this in older adults comes from TOT states and cuing. In a TOT state, the speaker recognizes that the word has been partially accessed (that is, semantic and sometimes grammatical and sublexical information like number of syllables has been recovered), but the speaker cannot access the phonological form of the word. Older adults experience more TOTs than younger adults (Burke et al., ; Gollan & Brown, ). Furthermore, providing participants with a phonemic cue (the word’s initial sound) when they experience a word retrieval failure leads older adults to the correct answer in most cases (MacKay et al., ). Neuroimaging studies provide further support that word retrieval impairment in older age is primarily due to poorer access to phonological information rather than to semantics (Diaz et al., ; Rizio et al., ; Shafto et al., ). Together these findings strongly suggest that the source of word retrieval failures among older adults is primarily a failure to retrieve the phonological form of the word. A well-supported proposal to explain the characteristics of word retrieval difficulty in ageing is the Transmission Deficit Hypothesis (Burke et al., ). This theory is based on a connectionist framework in which various levels of representation (e.g., semantic, phonological) are weakened in ageing. When a word’s semantic nodes are activated, the activation primes lexical nodes and consequently phonological nodes. Weakened connections hinder the transmission of priming, particularly from lexical to phonological nodes, for which a single connection exists (by contrast, between the lexical and semantic nodes, multiple connections are available). A similar proposal has been put forward to explain the poorer performance on word retrieval tasks by bilinguals compared to monolinguals. The Frequency-Lag (a.k.a. Weaker Links) hypothesis of Gollan and her colleagues (e.g., ) proposes that bilinguals have weaker connections between nodes than monolinguals due to their less frequent use of each language, which creates particular problems for phonological access. If, as suggested in section II, word retrieval difficulties in attrition may be an extreme version of a typical bilingual characteristic, then the notion of weakened connections that has been used to explain ageing effects and bilingual effects may also extend to first language attrition. Given that age-related word retrieval difficulty appears to be, in essence, an access problem, changes in other cognitive faculties, in particular executive function, may also contribute to word retrieval failures. For example, impaired inhibition affects older adults’ word recognition abilities (Sommers & Danielson, ; Wingfield et al., ; Peelle et al.,
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
), but it does not appear to impact word retrieval in the same way (Higby et al., in press). Inhibition is recruited to suppress competitors for selection, but there is little evidence that age-related word retrieval deficits are due to competitors (Burke, ). However, other executive functions, like set-shifting or working memory updating, may be important for successful word retrieval (Shao et al., ; Higby et al., in press).
.. Neurophysiological changes in cognitive ageing Ageing brings about a variety of neurophysiological changes that affect behavioural performance. Among the most widely observed changes is axonal degradation and loss (Grillo, ). The loss of entire neurons appears to be relatively rare in non-pathological ageing, and since loss of a neuron’s entire axon would lead to its death, it is likely that axonal changes in ageing reflect a loss in the number of axonal branches. Furthermore, demyelination of axons, which refers to damage to the myelin sheath that insulates the axon and permits information to be conveyed quickly, may be responsible for some of the age-related changes in cognitive processing speed and makes the axon more vulnerable to deterioration (Kövari et al., ). Synaptic structures and functions that permit inter-neuron transmission do deteriorate with advancing age in humans (and other living beings, including rodents and drosophila (fruitflies)). Two prominent sets are synapses in the hippocampus (which is involved in explicit memory) and in the pre-frontal cortex (related to executive function) (Morrison & Baxter, ), areas that have been linked to age-related cognitive declines in memory and executive control functions (Peters et al., ). Even in those two regions, synaptic types operate differently for their respective functions, and they degrade in different ways as well, so we may expect that networks and regions more directly involved in language will include numerous types not only of neurons but also of synapses that degrade, more or less, and in different ways. Little research has been done investigating the relationship between age-related grey matter or white matter changes and lexical retrieval decline. Obler et al. () found that greater grey and white matter volume and white matter integrity in left and right periSylvian regions, as well as additional areas in the frontal and temporal lobes, predicted better word retrieval success. Shafto et al. () focused on TOT occurrences as an index of word retrieval difficulty and found that more TOT occurrences were found in individuals with more grey matter atrophy in left insula, an area of the cortex that lies underneath parts of the frontal and temporal lobes. The scale used for these types of MRI studies is far greater than the microscopic scale required to examine changes in axonal length or synaptic density; thus, we can only speculate that the changes in grey and white matter that correlate with word-finding difficulties in older adults reflect some of the cellular changes that appear to be typical of ageing animals.
.. The interaction between age-related declines and L attrition Language attrition is a slow, gradual process, so researchers often study attriters who have been away from the native-language environment for decades in an effort to capture the
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
, , , ,
characteristics of attrition. Moreover, attrition research typically includes only those who immigrated as adults, when the native language is assumed to be fully developed. This combination of factors—late age of immigration and long length of residence—is likely to result in participants who fall within the age range normally considered to be older adults (e.g., age and older). Thus, it is important to consider whether some of the attrition effects reported may be due, or partially due, to ageing effects (Goral et al., ; Schmid, a). As we have demonstrated, there are several similarities in the patterns of lexical attrition seen in the first language of bilinguals and in monolingual older adults. One important way in which they differ, however, is the purported source of weakened phonological connections that result in word-finding difficulty. For bilinguals, the weakened connections impeding lexical access have been attributed to reduced language use and/or to competition from word forms in the second language. The weakened phonological connections in older adults, by contrast, have been attributed to brain changes associated with advancing age (Diaz et al., ), which may weaken neural connections through reduced synapses and axonal branches and either directly affect access to a word’s form or indirectly affect access through impaired executive functions crucial to successful word retrieval (Higby et al., in press). It is currently unclear how first language attrition is associated with brain changes (i.e., the strength of synaptic connections). Since the sources leading to word-retrieval difficulty differ, it is theoretically plausible that bilingual older adults may experience even greater difficulties with lexical access as a result of additive negative effects from multiple sources: language use, cross-language competition, and brain changes associated with ageing.
. L
.................................................................................................................................. Despite the evidence of ageing effects on lexical retrieval in monolinguals, few studies have explored the effect of ageing in late bilinguals, even in recent publications about bilingualism throughout the lifespan. Age-related lexical retrieval decline is similar for bilinguals living in their native-language environment and for monolinguals (Ashaie & Obler, ). Therefore, a key question is: do bilinguals undergoing attrition demonstrate an even greater deficit, suggesting that ageing and attrition effects are independent and additive? Within a connectionist network model, it is reasonable to expect that due to both ageing effects and attrition, age-related weakening of connections in the lexical system may impact an already weakening L system more than an L system that is in more frequent use. This is supported by evidence testing the Transmission Deficit Hypothesis from TOT studies of ageing monolinguals in whom low-frequency words (which are presumed to have weaker semantic-phonological links) are more strongly impacted than high-frequency words (Burke et al., ). Moreover, as inhibitory control abilities tend to decline with advancing age, it may become even harder for older adults to inhibit a more dominant L (Stolberg & Münch, ). In order to understand the interaction between ageing and reduced language-use effects, it is necessary to look at studies in which there is a possibility of taking both factors into account—that is, comparing older bilinguals and younger bilinguals who are living in their L environment. However, to date, there are few studies that have done this. Most studies
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
on attrition in older adults living in their L environment do not mention possible effects of ageing on their language measures (e.g., Altenberg, ; Jaspaert & Kroon, a; Hulsen et al., ; Kargar & Rezai, ). One study that did consider the possibility of ageing affecting L attrition limited their participant sample to those below age years in order to avoid the possibility of ageing effects on the results (Soesman, ). Goral et al. () attempted to disentangle the effects of ageing from attrition in older adults (aged to years) who spoke L Hebrew and L English and were living in their L environment. They reasoned that if the declines in language processing are primarily due to ageing, similar effects should be seen when bilingual older adults are tested in their first and second languages. If deficits are seen only for the first language, then it is likely that the processes of language attrition are primarily responsible for the decline. The authors compared younger and older Hebrew-English bilinguals living in the US (attriters) and younger bilinguals living in Israel (controls) on a lexical decision task. In Hebrew, the older attriters had the longest response times followed by the younger attriters. In English, the control group was the slowest, but the two attriter groups did not differ from each other. Based on the response time patterns, the authors concluded that the slower responses in Hebrew of the older bilingual group are more likely due to effects of attrition than ageing because the older adults were slower than younger adults only in the L. However, as mentioned previously, it is possible that ageing and attrition interact such that ageing processes disproportionately affect an already weaker language system. In fact, the pattern of results reported for the L across the three groups implies that both attrition and ageing contribute to the older attriters’ performance. Not all studies have shown a clear ageing or attrition effect in elderly bilinguals. Jaspaert and Kroon () studied letters written by a native Dutch speaker who had been living in the US for over sixty years and who had had limited contact with Dutch during that time. The participant, called A.L., wrote these letters over a six-year period between his late s and early s. The authors identified lexical choice problems, such as English borrowings, in about % of the verbs used. Given the low level of L contact the participant had had over a period of more than six decades, the rate of lexical borrowings seemed low. Subsequent studies have reported a similar number of lexical problems in attriters (Schmid, ). The fact that A.L. showed a similar number of lexical problems as younger attriters suggests that ageing may not have had much of an impact on A.L.’s L lexical choices. However, written production is likely to show fewer effects of ageing than oral production since writers can take as long as they want to choose a word or phrase, use sources like a dictionary or thesaurus, and can go back and self-correct. Circumstances that change in ageing, such as retirement, may affect language use patterns, complicating the study of attrition in older populations even more. In many older people who immigrated as adults but continue to use the L at home, a shift towards using L more predominantly comes with retirement (Schmid & Keijzer, ). Schmid & Keijzer tested three groups of attriters and two groups of non-attriters (controls) within five age groups of approximately equal numbers of participants: French > English). As for code mixing, the patient almost always responded in a different language from that of the examiner. Similarly, in their study of the language behaviour of two late bilinguals (German‑Swedish and Swedish‑Finnish) with AD, Hyltenstam & Stroud () reported that both patients showed a tendency to use the language that was not shared by their Swedish-speaking addressees. Moreover, the patients’ productions showed numerous intrusions of one language, reflecting the difficulty of keeping their languages separate. Finally, the nature of the elicitation task appeared to have an effect on the patients’ choice of language. For instance, they preferred their L to recall distant memories. Although the patients’ pragmatic difficulties were particularly marked in the severe stage of the disease, they depended on a variety of factors, including the age of acquisition of the languages involved, the premorbid degree of proficiency achieved for each of them, and/or the language(s) spoken in the environment. Similar observations were reported by De Santi, Obler, et al. () and Obler et al. (). Results of the latter study, which included four English‑Yiddish bilingual patients, only one of whom was an early bilingual, indicated that with the exception of the early bilingual, all the patients had difficulty addressing their interlocutor in the appropriate language. This difficulty increased with the severity of the disease, with both languages being impaired to the same extent in the most severely affected patient. Sometimes, the two languages were so jumbled up that it was impossible to tell which language was being spoken (De Santi et al., ). When Hyltenstam & Stroud () and Hyltenstam () studied two patients with mild AD, two with moderate AD, and two with severe AD—all late bilinguals with L Finnish and L Swedish—they found that the majority of participants (four out of six) often mixed languages, despite the strictly monolingual character of the experimental conditions. There was a clear trend for patients to use their L in the situation where L use was expected. In a longitudinal study conducted with three late bilingual (German‑Dutch) patients with AD, Luderus () investigated the directionality
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
-, , ö,
of intrusions (L to L versus L to L) in the context of a conversation experimentally defined as strictly monolingual. Overall, their findings highlighted considerable variability in patients’ conversational profiles and, consequently, the nature and extent of code mixing. The problems of language choice and language separation were shown to be strongly predicted by the balance of bilingualism in the premorbid period. The longitudinal approach made it possible to demonstrate that when bilingualism was balanced, the problems of language selection/separation appeared later on in the course of the disease. For patients with dominant bilingualism, these problems appeared earlier, and also had a predetermined directionality, depending on the premorbid dominance of L or L. Likewise, Mendez et al. () showed that bilingual patients with AD produce significantly more L intrusions in a later learned L. Language choice and separation in bilinguals rely on the ability to take account of the specificity of the communication situation in order to inhibit the non-relevant language. The causes of separation and code mixing problems should therefore be sought primarily in executive functioning deficits. Moreover, the processes involved in bilingual language production are based on a network of connections between the prefrontal cortex, anterior cingulate cortex, inferior parietal region, and basal ganglia (Abutalebi & Green, ). Solving conflicts in language selection normally requires the intervention of specialized structures, namely the dorsolateral prefrontal cortex and anterior cingulate gyrus. Conflict arises in the inferior parietal cortex, but spreads to linguistic areas in the inferior frontal cortex. All these cortical structures are connected through subcortical structures to the basal ganglia, in particular the caudate nucleus, which is responsible for conflict resolution. Given that AD gradually affects (among other things) frontal and cingulate regions, as well as the subcortical nuclei, we can assume that bilingual patients increasingly encounter between-language competition and conflict resolution difficulties (i.e., inhibition of the non-target language). The term inhibition is used to describe a cognitive control mechanism that tunes out the language that is irrelevant to the communicative context at hand (Abutalebi & Green, ; Costa & Sebastián-Gallés, ). In bilingual conversations, it prevents the speaker from producing utterances in the undesired language by keeping the irrelevant language below the selection threshold. The monitoring process that determines whether cognitive control mechanisms need to be applied to keep speaking in one language (i.e., activation) or switch to the other one (i.e., inhibition) is supported by the executive control system. Neuroimaging studies have shown that bilingualism affects the brain structures involved in the executive control system (e.g., greater density of grey matter in bilinguals’ anterior cingulate cortex). The logical outcome of a cortical impairment affecting the neural basis of executive control (i.e., frontal and cingulate regions) is therefore language choice difficulties. The study of how bilingualism affects the neural basis of executive control processes has only recently commenced, and research on language changes in patients with AD is a promising avenue for extending current understanding of language control in bilingual speakers. It remains unclear whether code mixing/switching behaviours in bilingual patients with AD are due to a deficit in inhibitory control or to language loss (i.e., language attrition/regression). For the time being, the variety of experimental paradigms, the heterogeneity of patients’ medical and linguistic profiles, the limited number of observations, and the inter-individual variability mean that it is still difficult to assess the impact of AD per se.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
. C
.................................................................................................................................. As underlined in Stilwell et al.’s recent review (), studies of language change in bilingual patients with AD are relatively inconclusive regarding the differential effects of AD on the two languages. Although they are not very informative regarding the order of L versus L deterioration in the course of the disease, these studies nonetheless raise important theoretical and methodological questions about how to test the multifaceted nature of bilingualism. For example, further research is needed to verify whether richer connections between concepts and words—which are characteristic of the dominant language—should be regarded as a strength or a weakness in the context of neurodegenerative diseases and attrition. Likewise, the issue of how to determine participants’ dominant language, which surprisingly was never experimentally assessed in the above-mentioned studies, should be carefully taken into consideration when testing bilingual patients with AD, as it is a crucial dimension for understanding the direction of language decline. Previous discrepant results may have stemmed from the fact that patients’ first-acquired language was not necessarily their dominant language (often the case in the context of migration). Furthermore, it is still unclear whether pathological ageing reverses language attrition as postulated by the reversion hypothesis put forward by de Bot & Clyne in .
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
......................................................................................................................
......................................................................................................................
T chapter provides an overview of the first few Event-Related brain Potential (ERP) studies on L attrition that have been conducted in recent years and will briefly discuss their results as well as future directions. As we will see, ERP studies are particularly suited to advance attrition research due to their power to track even subtle changes in cognitive processing with a considerable level of detail. In our opinion, the labels ‘attrition’ and ‘attriter’ must be reserved for cases where individuals undergo neuro-cognitive changes in how they process language that are nonpathological and induced by (environmental) changes in their language use (Kasparian & Steinhauer, b). Thus, while migrants immersed for years in a second language environment clearly constitute the prototypical example, neither are L immersion or bilingualism in absence of cognitive changes in L processing sufficient to justify the label ‘attriter’ (but see Schmid, a; Schmid & Köpke, a, for different views), nor is long-term immersion a necessary prerequisite. That is, attrition may also occur at relatively early stages of bilingualism (cf. Linck et al., ), possibly even without migration, or after extended periods of discontinued L exposure with little or no influence of a second language (e.g., Baladzhaeva & Laufer, )1. If the central criterion for attrition is linked to neuro-cognitive changes in language processing, and if some such attrition effects are relatively subtle (and controversial), then a research technique capable of tracking subtle cognitive changes is crucial. Importantly, a number of previous electrophysiological studies on L acquisition have demonstrated that ERPs may reliably reveal certain processing differences even when behavioural data do not (yet) reflect them (Friederici et al., ; McLaughlin et al., ; Morgan-Short et al., ). Since the early s, ERP techniques have enjoyed increasing popularity as an affordable online measure in psycholinguistic research, including in 1 Relevant scenarios of possible L1 attrition in absence of bilingualism may also include monks/nuns who gave a vow of silence, or individuals stranded on a desert island.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
bilingual populations. Given that numerous introductory chapters and review articles on ERP research in L and L are available (Kaan, ; Osterhout et al., ; Steinhauer & Connolly, ; Kotz, ; Van Hell & Tokowicz, ; Steinhauer, ), here we will only summarize a few methodological essentials, central findings, and current controversies that set the stage for the novel field of ERP research on L attrition.
. E-R P -
.................................................................................................................................. With two to five words per second, listeners and readers process language at a very fast rate. In order to investigate how phonological, morphological, semantic, syntactic, and discourse information is integrated in real time, most behavioural approaches and brain imaging techniques (such as PET or fMRI) are too slow. In contrast, ERPs reflect a reader’s or listener’s neural responses to linguistic stimuli with virtually unlimited temporal resolution in the range of milliseconds, without any delay, and continuously across entire sentences (i.e., for each single word). Moreover, as a number of psycholinguistic processes are reflected by distinct brain waves (ERP components) that differ in latency, polarity, and scalp distribution, ERPs are ideal to uncover the nature of processing difficulties and the contribution of different types of information (e.g., prosody, syntax, and semantics). Thus, ERP data permit us to (a) identify which word(s) in a sentence may cause processing difficulties, (b) link the findings to psycholinguistic processes, and (c) investigate if and how L learners and attriters may differ from monolinguals and from each other. To briefly illustrate the general approach, let us assume we are interested in the ERP profile for conceptual-semantic anomalies (e.g., ‘John ate democracy at dinner’), as compared to matched correct control sentences (‘John ate broccoli at dinner’; target words for the ERP analysis are underlined). As a first step, electroencephalographic (EEG) data are collected from an array of cap-mounted scalp electrodes while the participant reads or listens to a randomized list of sentences from the various experimental conditions (Figure .). Each time a target word is presented, a condition-specific ‘trigger code’ is sent to the EEG system marking the word onset in the continuous stream of EEG data. These triggers are then used to identify and extract the relevant epochs of the EEG that reflect the processing of target stimuli (‘time-locking’). Most psycholinguistic processes related to a given target word occur within the first , milliseconds (ms) after word onset and are reflected by characteristic negative and positive deflections in the ERP (socalled ‘ERP components’). Early brain waves during the first ms after word onset (such as the N and the P components) primarily reflect physical stimulus characteristics (such as font size or acoustic intensity), whereas later components are linked to higher cognitive processes. For instance, the lexical-semantic anomaly in our example elicits a larger negativity between and ms for the implausible word ‘democracy’ than the plausible word ‘broccoli’ (see Figure .). As this deflection has a negative polarity and peaks around ms, it is referred to as the N component. The N amplitude increases gradually the more difficult the lexical retrieval and semantic integration of a target word is (Kutas et al., ; Steinhauer et al., ). As illustrated by the voltage map in
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
EEG → Event-related brain potentials (ERPs) John ate broccoli at dinner. John ate democracy at dinner.
–5μV
N400 effect N100 Voltage map of N400 effect
EEG amplifier
+5μV
200 400 STIMULUS P200 ONSET
600
800 Time
1000
ms
300 – 500 ms
.. EEG set-up, lexical-semantic N effect, and a corresponding voltage map EEG data collected from an electrode cap during sentence processing are amplified, averaged, and time-locked to the onset of the target words (black line—control: ‘broccoli’, red line—mismatch: ‘democracy’). The (schematic) ERP plot shows similar brain waves in both conditions during the first ms and an enhanced N for the semantic mismatch condition (negative polarity is plotted upwards). The voltage maps illustrate the centro-parietal scalp distribution of the N effect between and ms (mismatch minus control). See main text for details. Source: K. Steinhauer (), Applied Linguistics, (), –, with permission from Oxford University Press.
Figure . (right panel), the N effect (i.e., the N difference between implausible and plausible words) has a broad distribution over the scalp but is most prominent at central-parietal electrodes near the midline of the scalp. Once we have established such a pattern in native monolinguals, we can see if L learners or attriters display the same ERP profile, or if their N effect is smaller, delayed, prolonged, or accompanied by other ERP effects that may point to deviant online processing. Corresponding ERP data for the online processing of more subtle semantic anomalies in all three groups will be presented below in Section .. In contrast to lexical-semantic anomalies, increased structural processing costs due to complex syntax or grammar violations are most reliably associated with enhanced posterior positivities between and , ms (P components; Osterhout & Holcomb, ). In many group studies on grammar violations, Ps are preceded by Ns or left anterior or left temporal negativities (LANs or LTNs, respectively; see Steinhauer et al., ; Molinaro, Barber, et al., ). Some researchers have argued that early LAN-like negativities may be elicited by the fast and automatic detection of a grammar violation (possibly subserved by language-specific circuits in Broca’s area), while late Ps may reflect subsequent and more controlled attempts to reanalyse or mentally repair the grammatical structure (Friederici, ). More recently, other authors have focused on the substantial inter-individual variability (even among monolinguals) as to which effect (negativity or P) is dominant (Tanner & Van Hell, ), potentially suggesting the ERP components may reflect distinct processing strategies that rarely co-occur within the same trial. In this chapter, we will address possible factors underlying inter-individual variability only in the context of bilinguals.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
A last question concerns the domain specificity of ERP components. While the information provided by ERP data is undoubtedly much richer than response time measures and error rates in behavioural studies, ERP components should not be viewed as distinct markers for language-specific cognitive processes. Research in other cognitive domains has shown that ERP effects similar to those found in language studies can also be elicited by other stimulus materials, especially in music (e.g., Patel et al., ). Moreover, there is some evidence suggesting that at least certain ERP effects (such as the P) may consist of sub-components that reflect different kinds of cognitive operations (e.g., Steinhauer & Connolly, ). However, ‘even if these ERP components are not specific to language, they may still reflect cognitive processes that are both characteristic and tightly linked to natural language processing’ (Steinhauer, , p. ).
. H ? E-R P L
.................................................................................................................................. The first ERP studies on morpho-syntax comparing late L learners to native speakers found significant group differences that increased with increasing age of acquisition (AoA) (e.g., Weber-Fox & Neville, ), thus apparently providing hard brain evidence supporting the ‘critical period hypothesis’ (Lenneberg, ). A common conclusion was that loss of brain plasticity in puberty may prevent late L learners from relying on native-like brain mechanisms (and thus from attaining native-like L proficiency), and that these maturational constraints primarily affect L phonology and morpho-syntax, as compared to vocabulary learning and lexical processing (Newport et al., ). However, confounds between AoA and L proficiency levels in these early studies have been criticized, and ERP studies that unconfounded the two factors reported that L proficiency was a better predictor for nativelike ERPs than AoA, with little evidence for a critical period, arguing instead for a proficiency-dependent convergence on native-like processes even in late L learners (e.g., Steinhauer et al., ). Currently, the data across studies are mixed, and many other factors contributing to the variability in ERP profiles are being discussed, including the length and type of L exposure, the role of classroom instruction versus immersion, transfer effects in different L–L pairings, modality effects, motivation, and the specific structures under investigation (Steinhauer, ; Tanner et al., ; Meulman, Stowe, et al., ). Attriters constitute an important complementary group of interest for various reasons. First, where robust differences between native speakers and L learners have been found, these may be partly due to the difference between ‘being monolingual’ and ‘being bilingual’. As pointed out by Schmid (c), data from native but bilingual attriters may help fill this gap (as might data from simultaneous bilinguals; e.g., Molnar et al., ). Second, an interesting question is whether it is actually possible to ‘unlearn’ a full-fledged L system and whether and to what extent this process of losing one’s L resembles the stages of L acquisition, but in reverse. If structure-specific language proficiency is a strong predictor for neurocognitive language mechanisms as reflected by ERPs (Steinhauer, ), then one might expect similar ERP profiles in L attriters and proficiency-matched L learners of the
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
same language. Recall, however, that ERPs may sometimes reveal differences that are not accurately detected in behavioural tasks (e.g., McLaughlin et al., ). Third, if attriters are more proficient in their L than their L, does this mean their ERP profiles are more ‘nativelike’ in their L than in their L as well? This kind of question requires one to test attriters in both of their languages (Kasparian, ). Fourth, which language domains are most vulnerable to attrition, and do they overlap with those discussed for critical periods in L acquisition (potentially pointing to shared restrictions in brain plasticity across learning and forgetting)? Fifth, what is the relationship between success in L acquisition and the occurrence of L attrition? Are they two sides of the same coin? Are they influenced by the same factors (such as type and intensity of L exposure; cultural identity, etc.)? Before future research can address these specific questions in detail, much current research attempts to determine which linguistic domains are actually subject to L attrition effects. Especially in the area of morpho-syntactic processing, there is little agreement on whether attrition occurs at all once a full-fledged first language has been established. To our knowledge, there are currently only two research groups that have started to systematically investigate the neurocognitive underpinnings of L attrition using ERPs: Monika Schmid’s group (now in Essex) focusing primarily on spoken sentences, and our own group in Montreal (in collaboration with Francesco Vespignani in Italy), testing written sentence materials. Relatedly, Judith Kroll and colleagues (e.g., Linck et al., ) have investigated L-induced changes to L in immersed and classroom L learners, and have thus far discussed their findings in terms of L and L co-activation effects and in terms of reduced L access during L immersion (without explicitly labelling such effects ‘attrition’). Although we will argue that these investigations of ‘early attrition’ are definitely warranted and can serve to address several open empirical questions (see Section .), we focus our current discussion on those ERP investigations that have targeted L attriters who have migrated to a new country for the longer term. Monika Schmid’s group has established a large-scale research program on the morphosyntactic processing of grammatical gender (GG) across various European languages (especially Dutch and German). GG is known to be notoriously difficult in L acquisition, while only few data are available for attriters. Important aspects of this research program include (a) comparisons of ERP data from attriters, L learners, and monolinguals to address effects of AoA and bilingualism, (b) the consideration of proficiency levels in the target language, and (c) cross-linguistic comparisons for different L–L pairings. The research program in Montreal started in and is largely based on the second author’s PhD thesis project (Kasparian, ). It focuses on adult Italian immigrants in Montreal who now use English as their main language and have identified signs of losing their first language. While this project does not include multiple L–L pairings, it is special in that it tests attriters in both their L (Italian) and L (English) and compares them with proficient English learners of Italian L as well as with both English and Italian monolinguals (the latter were tested in Italy). Similar to the European research program, multiple behavioural measures on L and L proficiency, length of residence, and language-use were collected in addition to ERP and behavioural data. Moreover, the linguistic paradigms tested with ERPs are quite diverse and include lexical-semantic processing as well as morphosyntactic processing in a range of different sentence structures (in Italian and English). In the following, we will provide an overview of recent findings from these groups, in the areas of morpho-syntactic processing as well as lexical processing.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
. M- ’ L
.................................................................................................................................. The first three ERP studies in this review set out to determine whether and to what extent attriters show altered online processing for grammatical anomalies at the sentence level. As already discussed in previous chapters, whether attrition effects in this linguistic domain exist at all is highly controversial.
.. Grammatical gender violations in German and Dutch In a collaboration with ERP expert Laurie Stowe, Schmid and colleagues conducted a series of ERP studies on GG processing in monolinguals, L learners and L attriters of various European languages. When studying bilingual sentence processing, focusing on GG has a number of advantages. First, acquiring a new gender system is known to be very difficult for L learners, and there is some debate whether it is more challenging if one’s first language has a different mapping or no gender system at all (Tokowicz & MacWhinney, ; Sabourin et al., ). In many languages such as Dutch, French, and German, each noun has a GG that needs to be learned, and the mapping is largely arbitrary (i.e., independent of conceptual-semantic or phonological cues). During real-time sentence processing, a reader or listener must retrieve the idiosyncratic gender information of a given noun from longterm memory (LTM) and check this feature with those of the determiner and the inflectional morphemes of other elements. Although GG mismatches (e.g., using the wrong determiner) do not usually change the meaning of a given noun phrase (NP), native speakers perceive them as relatively salient morpho-syntactic violations (especially in speech) and such instances therefore reliably elicit P effects, often preceded by either LANs or Ns. In fact, there is substantial ERP literature on gender violations across languages, in both native speakers (e.g., Molinaro, Barber, et al., , for a review) and L learners (see overview in Meulman, Stowe, et al., ), but not yet for attriters. Schmid and colleagues’ spoken sentence materials included subject NPs in which the determiner either agreed or disagreed with the GG of the noun. For example, the German noun Garten (garden) is masculine and requires the masculine determiner ‘der’, while the neuter determiner ‘das’ would result in a gender mismatch on the target noun (underlined in ()): () . . . der Garten / das *Garten . . . ‘themasc gardenmasc / theneut *gardenmasc’ () . . . der kleine Garten / das kleine *Garten . . . ‘themasc small garden / theneut small *gardenmasc’ () . . . hat die Rose geblüht / *blühen . . . ‘the rose has bloomed / *bloom’ To prevent simple chunking strategies based on familiarity, a second experimental condition () added an intervening adjective that separated the determiner and the noun. Because the adjective is marked only for number and case but not gender, its presence does not directly contribute to the GG agreement or disagreement. Lastly, the researchers also included a third morpho-syntactic violation condition not involving GG that was expected
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Controls Monolingual L1 speakers (n = 29)
Attriters Late L1 attriters (n = 27)
Early learners L1 without L1 with gender gender (n = 17) (n = 26)
Late learners L1 without gender (n = 17)
L1 with gender (n = 40)
Verb form
Gender: Det + N
Gender: Det + Adj + N
.. Voltage maps summarizing the (preliminary) main findings for agreement violations in ‘finite verb’ and GG conditions These voltage maps represent top views of the human scalp (i.e., the ‘nose’ in each map points upwards) and illustrate the ERP differences between conditions (violation minus control) in a representative time window. Colours correspond to a ‘heat’ map, that is, dark blue indicates strong negative potentials and dark red indicates strong positive potentials. Schmid and colleagues focus primarily on the presence versus absence of a large P, represented by a large red area over centro-parietal scalp sites. While monolingual controls and Attriters display such a large P component, late L learners show consistent Ps only for the verb condition. In early L learners, having a GG system in L seems to lead to more native-like P patterns. Source: Schmid (c).
to be detectable even by less proficient L learners. As illustrated in example (), this ‘finite verb condition’ replaced a past participle (e.g., geblüht; bloomed) with the infinitive (*blühen; to bloom). A corresponding design was also used in Dutch, where native speakers were compared to L French late L learners of Dutch and each group performed both in an auditory and a visual version of the ERP experiment (Meulman, Stowe, et al., ). Since only some of the final data have been published to date (Bergmann, Meulman, et al., ; Meulman, Stowe, et al., ), the results reported here are partly based on conference presentations (Schmid, c; Schmid et al., ). As illustrated in Figure ., monolingual native speakers reliably elicited the expected P components for all three types of violations (an additional N preceding the P was found for the Dutch verb condition, by Meulman, Stowe, et al., ). The presence of an adjective in () versus () did not seem to have an effect in any group. Second language learners generally elicited native-like Ps for the finite verb condition (as well as the N in Dutch auditory sentences), whereas their ERPs for gender agreement varied primarily as a function of AoA. Early L learners (AoA < years) showed a P similar to monolinguals (especially when their L had a gender system), whereas late learners (AoA > years) displayed no typical P (Schmid, c). In the Dutch study by Meulman, Stowe, et al., the majority of French L learners did not show any ERP violation effect for GG violations. Importantly, and in stark contrast to the L
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
learners, the group of attriters was indistinguishable from monolinguals in that all of them elicited P components for all violation types (Bergmann, Meulman, et al., ; Schmid et al., ). These findings were interpreted as evidence that morpho-syntactic knowledge stabilizes around puberty. For late learners acquiring their L after the age of years, this means that they may be able to rely on native-like processing mechanisms for the finite verb condition (possibly transferring knowledge from their L) but fail in this respect for structures that are different or absent in their L, such as GG (Meulman, Stowe, et al., ). In attriters, on the other hand, the ‘stability of a deeply entrenched L system’ (Bergmann, Meulman, et al., , p. ) apparently served as a protective factor from losing their native morphosyntactic processing mechanisms, such that all ERP profiles of the attriters under investigation remained native-like even after many years of dominant L exposure and use after immigration. This view is in line with the behavioural findings of one of the first cognitive attrition studies by Scherag et al. (), who also found no indication for differences between monolinguals and bilingual migrants, either for lexical-semantic or morphosyntactic processing. With respect to Schmid and colleagues’ research objective of teasing apart the contributions of various factors to differences between L learners and native speakers, the attriters’ data suggest that merely ‘being bilingual’ has little or no impact on the real-time processing of one’s first language. A number of points regarding these studies are noteworthy. With respect to L learners, the article by Meulman, Stowe, et al. () reminds us that data for GG violations are not consistent across studies. While their own analysis convincingly showed that neither proficiency, length of residence, L use, nor AoA (after age ) had any impact on the P for GG violations, several other investigations did observe these effects, as well as native-like ERP profiles including P effects for GG violations—even in late L learners (Tokowicz & MacWhinney, ; Gillon Dowens et al., ). This variability across studies certainly deserves attention in future research, and the authors provide a number of interesting suggestions (e.g., to focus on modality and L/L pairings). Another aspect of Meulman, Stowe, et al.’s () study is the striking observation that the L learners’ accuracy in a grammaticality judgement during the EEG experiment was much lower for the GG condition (.%) than for both the finite verb condition (.%) and an offline task in which participants had to assign GG to a list of nouns used in the EEG experiment (.%). We will return to this point below in Section .. Finally, given that GG online processing involves both the quick retrieval of a noun’s idiosyncratic gender information from LTM and subsequent morpho-syntactic feature checking with the determiner, low accuracy scores in the judgement task and the absence of P effects can be due to problems with either of these sub-processes. According to most psycho- and neurolinguistic models, the two sub-processes involve very different cognitive mechanisms and brain circuits (e.g., declarative and procedural memory systems, respectively, according to Ullman, ). As the finite verb condition (which relied almost exclusively on morpho-syntactic processing/procedural memory) did not pose a major challenge to the L group and elicited native-like ERP effects, it may very well be that the GG data collected by Schmid and colleagues do not point to morpho-syntactic problems in a late acquired L, but rather to problems in quickly and reliably retrieving the correct gender information from LTM—and thus a lexical problem (see also Section .). This interpretation would be in line with Hopp’s
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
‘lexical bottleneck’ hypothesis for a range of apparently morpho-syntactic processing difficulties in L acquisition (e.g., Hopp, , b). With respect to attriters, the available GG data clearly support Schmid and colleagues’ notion of a well-preserved L system for GG processing. Especially since this paradigm critically relies on fast native-like lexical access as well as fast native-like morpho-syntactic integration, an ERP profile that is indistinguishable from that of monolingual speakers provides evidence that both of these sub-processes must be intact in attriters, at least for certain structures. In other words, with a single paradigm the researchers in Groningen seemed to have demonstrated attriters’ preserved native-likeness in two linguistic domains (lexicon and morpho-syntax) and for multiple complex neural circuits in the brain (e.g., declarative and procedural memory), at least for certain linguistic structures. Regarding neural plasticity over the life span, the attriters’ GG data appear to confirm what Meulman, Stowe, et al.’s results in late L learners already suggested, namely that GG processing may not be subject to much neurocognitive plasticity—especially in adulthood: GG seems to be both difficult to acquire (by Lers) and to lose (by attriters). An important question is to what extent these findings in attriters can be generalized. For example, Hopp & Schmid () reported that some L attriters do show signs of nonnative-likeness in L pronunciation (‘foreign accent’), whereas a few late L learners are indistinguishable from native speakers. They concluded that ‘acquiring a language from birth is not sufficient to guarantee nativelike pronunciation, and late acquisition does not necessarily prevent it’ (p. ), suggesting that native-likeness is not a universal phenomenon in attriters across linguistic domains. But as illustrated by the inconsistency of ERP profiles across L studies on GG violations, even within the domain of morphosyntax, a single ERP experiment is unlikely to tell the whole story. The next sections will present complementary ERP data on attrition, collected from Italian immigrants in Montreal.
.. Number agreement violations at various positions in Italian sentences A study by Kasparian et al. (, ) also investigated morpho-syntactic agreement violations. However, here the focus was not on GG but on number agreement (singular vs. plural; sg/pl) between (a) the subject NPs (e.g., the workers) and the adjacent verb and (b) the subject NP and a distant modifier (adjective) in a given sentence, as illustrated in () and (): () I lavoratori tornano dalla fabbrica sporchi di grasso The workers(pl) return(pl) from the factory dirty(pl) with grease () I lavoratori *torna dalla fabbrica *sporco di grasso The workers(pl) *returns(sg) from the factory *dirty(sg) with grease The four conditions of this study crossed agreement and disagreement on the verb and the adjective such that (a) both agreed with the subject NP (as in ()), (b) only the verb disagreed, (c) only the adjective disagreed, or (d) both verb and adjective disagreed with the subject NP (as in ()). The number of the subject NPs (sg/pl) was balanced.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
The general paradigm was adopted from a previous study with native Italian monolinguals by Molinaro, Vespignani, et al. () who also examined the ERP responses on both the verb and the modifier. In this study (as well as all other Kasparian & Steinhauer studies reviewed in this chapter), the behavioural task during EEG recording was to rate the acceptability of each sentence on a scale from (unacceptable) to (perfect). Because various recent ERP studies on agreement violations in L learners had found that long-distance agreement relationships are more likely to reveal differences between bilinguals and monolinguals (e.g., Gillon Dowens et al., ), one important research question was whether differences might be found for attriters as well. A second question had to do with cross-linguistic differences between Italian and English. Disagreement between an NP and a subsequent verb is a reliable indicator for a violation in English, but not in Italian (due to its freer word order). If attriters reading sentences in their L (Italian) were in fact influenced by their dominant L grammar (English), one might expect that their ERP response to agreement violations might be stronger than that of the monolingual control group (for details on the various tasks measuring L and L proficiency in both groups, please refer to Table of Kasparian et al., ). Kasparian et al. () found group differences at both the verb and the modifier position. As shown in Figure ., number disagreement on the verb elicited a weak LTN in the control group (n = ), followed by a frontal Pa component (likely reflecting a shift of attention towards the anomaly) and a subsequent posterior P that lasted from to , ms post target word onset. In contrast, attriters (n = ) showed a significantly larger and more broadly distributed left-lateralized negativity, followed by a Pa and a P that ended around , ms. The significant group differences for the negativity were interpreted as reflecting a more salient violation in attriters, due to the influence from English word-order. An alternative account in terms of component overlap with the subsequent Pa (which may have cancelled out the negativity in the control group; cf. Tanner, ) could be ruled out, as this positivity was actually larger in attriters. However, the shorter duration of the P in attriters is somewhat at odds with the notion of a stronger salience of the agreement violation in this group. Did the stronger initial response (reflected by the larger negativity) facilitate and speed up subsequent reanalyses? Or do the duration differences point to qualitative processing differences? These questions gain additional importance from the finding that the disagreement P elicited by the modifier was again several hundreds of milliseconds shorter in attriters than controls (while the preceding negativity was numerically larger in attriters). Although this issue cannot be completely solved with the data at hand, Kasparian et al. () provide a number of arguments why the shorter P in attriters is likely to reflect less ‘in-depth’ structural repair. First, more efficient processing should have resulted in faster response times, but attriters were slower than controls in this experiment (see Figure b in Kasparian et al., ). Second, the authors checked if this group of attriters just happened to be generally slower in their behavioural responses, or if their P was generally shorter across all experiments. However, data from other sentence conditions included in the same experimental session showed this was not the case. Third, Kasparian et al. argue that the P has been shown to reflect multiple cognitive processes, that is, it is not a monolithic component. Among other accounts, Friederici et al. () have suggested that the first portion of the P in violation paradigms may reflect a diagnosis of the problem while the late portion is associated with actual structural repair. The observation
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
(a) Controls –4
–0.2
Pz
0.2 0.4 0.6 0.8 1.0 1.2
4
Negativity 300–500ms
Frontal pos. 550–650ms
P600 650–1000ms
Late P600 1000–1200ms
Negativity 300–500ms
Frontal pos. 550–650ms
P600 650–1000ms
Late P600 1000–1200ms
(b) Attriters –4
–0.2
Pz
0.2 0.4 0.6 0.8 1.0 1.2
4 I1 lavoratore| *tornano The worker | *return I1 lavoratore| torna The worker | returns
.. ERPs elicited by the verb in response to number agreement violations (red) compared to correct sentences (green) in Controls (a) and Attriters (b), with voltage maps Time ranges (in milliseconds) depicted on the x-axis are relative to the onset of the verb ( ms). Negative values are plotted up. Voltage maps illustrate the scalp distribution of the effects observed for the time-windows of interest. Controls showed a small LTN, followed by a frontal positivity and a posterior P lasting until , ms. Attriters showed a robust N, a numerically larger frontal positivity and a broadly distributed P that was shorter in duration than that of the Controls. (Note that for this contrast we used a non-standard baseline interval from to ms; for details, see Kasparian & Steinhauer, .)
that the scalp distributions of the early and the late portion differed in the present study supports the notion of distinct neural generators, and hence distinct cognitive processes. Moreover, the early P portion (whose topography did not differ between groups) was modulated only by proficiency, whereas the late portion was influenced by the factor of group (present only in monolingual Italians). Interestingly, the more the attriters were still exposed to Italian on a daily basis, the more they resembled the monolinguals in showing the late P effect (as revealed by a highly significant correlation). In sum, contra the findings by Schmid’s group, this study reports some initial ERP evidence for L attrition effects in the domain of morpho-syntactic agreement violations. Influenced by the stricter word order in their co-activated English grammar, attriters seemed to have perceived agreement violations in Italian as more salient than monolinguals, but engaged less in reanalysis processes. The shift towards this new processing strategy was the stronger, the lesser attriters were still exposed to their native language. The hypothesis that some processing changes in these attriters were due to the coactivation of their English grammar was even more directly addressed in another ERP study investigating Italian relative clauses, as will be described in the next section.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
.. Relative clauses in Italian immigrants: can one’s L render the L ungrammatical? Kasparian & Steinhauer (a) compared the online processing of Italian relative clauses in attriters to that in native monolinguals still residing in Italy. Given important crosslinguistic differences between Italian and English relative clause constructions, these structures offer an interesting opportunity to study the processing of complex morphosyntax as well as the influence of the L grammar on the L (cf. Dussias & Sagarra, ). Both languages allow V(erb)-NP-subject2 () and NP-V-object () word-orders. In contrast, V-NP-object () and NP-V-subject () structures are ungrammatical in English, though syntactically permissible in Italian (note: (S) - subject NP; (O) - object NP). () V-NP-subject:
Il poliziotto (S) che arresta i ladri (O) registra i nomi. The policeman (S) that arrests the thieves (O) registers the names.
() V-NP-object*:
I ladri (O) che arresta il poliziotto (S) attendono in macchina. The thieves (O) that arrests the policeman (S) wait in the car.
()
NP-V-subject*: Il poliziotto (S) che i ladri (O) arresta registra i nomi. The policeman (S) that the thieves (O) arrests registers the names.
()
NP-V-object:
I ladri (O) che il poliziotto (S) arresta attendono in macchina. The thieves (O) that the policeman (S) arrests wait in the car.
As English has a less detailed system of morphological markers, English speakers have been shown to rely more heavily on word-order for sentence interpretation. Conversely, Italian has a relatively free word order and rich morphological marking system, thus number agreement and semantic information (e.g., animacy, prominence) are more salient cues than word-order in identifying grammatical and thematic roles (subject/object, agent/ theme) of NPs in a sentence (see ‘Competition Model’, Bates et al., ; BornkesselSchlesewsky et al., ). In this study, sentences were disambiguated at the verb of the relative clause, by virtue of its number (sg/pl) as well as its semantic bias; strong agentpatient–verb combinations (e.g., policeman/arrest/thief ) made it so that only one of the two NPs was a plausible subject of the verb. Unlike monolingual Italian controls who could rely on these cues, attriters were expected to be influenced by their dominant L-English grammar and to process V-NP-object and NP-V-subject sentences as outright grammatical violations, with little or no reliance on semantic cues to revise the interpretation of the sentence. Results showed that attriters differed from native controls in their acceptability judgement ratings and ERP responses when relative clause constructions were ungrammatical in English, though grammatical in Italian. Attriters gave lower end-of-sentence acceptability judgement ratings than Controls for the two conditions that were ungrammatical in English, but did not differ in ratings of sentences that were grammatical in both languages. In response to V-NP-object sentences which are unpreferred compared to subject relative clauses but still syntactically acceptable in Italian, controls showed an N effect between 2 Note that ‘V-NP-subject’ is used as the short hand for ‘NP-V-NP subject-first’ as all structures start with an NP and are either subject-first or object-first.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
(A)
–4.0
Fz 0.5
Controls
Attriters
–4.0
0.5
1.0
300–400 ms
N400
4.0
4.0
650–850 ms
–4.0
Fz 0.5
Fz
0.5
1.0
4.0
N400
No N400
–4.0
Fz 0.5
–4.0
1.0
4.0
4.0
–4.0
Fz
Larger and earlier P600 No N400
850–1050 ms
1.0 Late P600
–2.5
μv
1.0
Fz
0.5 4.0
+2.5
1.0
Larger and earlier P600
V-NP-subject V-NP-object
Controls
(B) –4.0
Attriters –4.0
Fz 0.5
1.0
Fz 0.5
4.0
ns.
4.0
Frontal positivity
–4.0
ns. Cz
–4.0
N400 Cz
0.5
1.0
4.0
–4.0
0.5
1.0
1.0
4.0
ns.
–4.0
Pz 0.5
1.0 Small P600
4.0
N400
Pz
0.5 4.0
1.0 Large P600
NP–V–subject NP–V–object
.. ERPs for Italian relative clauses (A). ERPs elicited by the verb of the relative clause in V-NP-object sentences (pink) compared to V-NP-subject sentences (green). Controls elicited an N effect and a small P. In contrast, Attriters did not show an N effect, and the P effect was earlier, larger and broadly distributed. (B). ERPs elicited by the verb of the relative clause in NP-V-subject sentences (red) compared to NP-V-object sentences (blue). Controls only elicited a small posterior P effect in response to subject sentences, whereas Attriters showed an N-like negativity, an early fronto-central positivity followed by a larger P effect for subject sentences.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
– ms at the disambiguating verb (Figure .A), indicating that they were sensitive to the semantic cues that served as extra disambiguating information to identify the subject of the sentence (Di Domenico & Di Matteo, ; Mecklinger et al., ). The N effect in native controls was larger in participants who gave more favourable ratings and therefore did not index a violation effect. The N in controls was followed by a late P effect ( to , ms) indicative of a mild late effect that reflects re-analysis and repair processes. In contrast, attriters did not show an N effect (irrespective of whether they provided high or low acceptability ratings) and elicited an earlier and more broadly-distributed P ( to ms). In their article, Kasparian & Steinhauer (a) present arguments as to why the P in attriters reflected the perception of a stronger anomaly and higher processing costs related to their reduced use of semantic information and stronger expectations of number agreement (due to English) between the sentence-initial noun and the verb. Larger P amplitudes in attriters were associated with a higher degree of L-English proficiency and a longer length of residence (increased L immersion). In response to NP-V-subject garden path sentences in Italian (Figure .B), controls elicited a small, late, posterior P starting around ms. In contrast, attriters displayed a strong, widespread N-like negativity, followed by a larger frontal positivity and a more robust early P starting around ms. As in V-NP sentences, this pattern fits the view that, due to L English influence, attriters were more sensitive to word-order preferences than to semantic cues, and processed permissible Italian sentences as outright morphosyntactic violations. In sum, attriters with limited L exposure/use and predominant exposure to their L showed distinct processing routines compared to native controls. Within attriters, higher levels of L immersion and proficiency were associated with increased L influence on the L. This study was novel in showing that high levels of L proficiency and exposure may render a grammatical sentence in one’s native language ungrammatical. This finding can be viewed as strong evidence that L attrition does take place in the domain of morpho-syntax, at least for some structures. It also raises the (old) question of when and to what extent L attrition is due to interference with the competing second language (rather than discontinued use of the L). This question can only be answered by testing different L–L pairings that differ in terms of structural overlap, or (at least in principle) by investigating language processing in people who have experienced strongly reduced use of their L for an extended period, but without exposure to another language (e.g., certain monks or nuns following a vow of silence; see Sharwood Smith & Van Buren, ). We will now turn from morphosyntactic to lexical processes.
. L- ’ L L
.................................................................................................................................. Word finding problems are among the most common phenomena reported by attriters, and a number of behavioural studies using production tasks have confirmed that lexical access is slower and L vocabulary smaller or less diverse in attriters compared to
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
monolinguals (e.g., Schmid & Köpke, ; see also Schmitt, Chapter , this volume). This section presents two ERP experiments contrasting attriters’ brain responses to lexical anomalies with those in L learners and monolingual controls. The first study focuses on confusable words in attriters’ L (Italian), whereas the second one addresses the processing of cross-linguistic homographs and cognates in English sentences (i.e., the attriters’ L), with a particular interest in whether attrition affects co-activation and inhibition effects in the bilingual lexicon.
.. A study on confusable words in Italian sentences Kasparian & Steinhauer () investigated the lexical-semantic integration of confusable Italian noun pairs such as cappello/cappella (hat/chapel) in both attriters and L learners as compared to monolingual native speakers of Italian. Phonologically, these semantically unrelated word pairs differ only in their word-final vowel (–o versus –a, also marking their GG) and are extremely difficult for learners of Italian L, as indicated by many cautionary notes in Italian textbooks (e.g. Fogliato & Testa, , p. ) and online language blogs. In a German ERP study, Rüschemeyer et al. () had already shown that, unlike native speakers, L learners tended to confuse similar German word pairs that differed word-initially such as Maus (mouse) and Haus (house), and also displayed enhanced negativities starting in the N time range in a priming paradigm (Maus-Dach; mouseroof) that were interpreted as reflecting problems inhibiting the phonological neighbours (mouse and house). Kasparian & Steinhauer’s Italian ERP study presented three different sentence conditions, in which a given context was either completed with the appropriate target noun from the word pair (Correct), with the confusable other noun from the word pair (Swap), or with a completely unrelated noun (Mismatch): Correct
Per coprire la testa, il pescatore porta il cappello di lana. To cover his head, the fisherman wears the hat of wool.
Swap
Per coprire la testa, il pescatore porta la *cappella di lana. To cover his head, the fisherman wears the *chapel of wool.
Mismatch
Per coprire la testa, il pescatore porta la *cartella di lana. To cover his head, the fisherman wears the *briefcase of wool.
One aim was to determine whether sentences with ‘confusable nouns’ (Swap condition) would be processed as semantically acceptable by late L learners and L attriters, especially by those individuals with lower Italian proficiency scores. As hypothesized, monolingual Italians in Italy showed large N effects for both Mismatch and Swap conditions (Figure .). This pattern was also found for those attriters and L learners who had a high proficiency (HP) level in Italian (as assessed by various written, oral, and comprehension tasks, see Tables and in Kasparian & Steinhauer, ). In contrast, participants with lower Italian proficiency scores elicited reliable Ns only for the Mismatch but not for the Swap condition, irrespective of whether Italian was their L (attriters) or their L (learners). This finding suggests that online confusability of phonologically similar words (as reflected by N amplitudes) is primarily driven by
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Controls –4.0
HP Attriters
LP Attriters
HP L2 learners
LP L2 learners
–4.0
–4.0
–2.0
0.2 0.4 0.6 0.8 1.0 1.2 –2.0
4.0
4.0
4.0
4.0
4.0
–4.0
–4.0
–4.0
–4.0
–4.0
–2.0
0.2 0.4 0.6 0.8 1.0 1.2 –2.0
0.2 0.4 0.6 0.8 1.0 1.2 –0.2
4.0
4.0
4.0
–2.0
–2.0 4.0
0.2 0.4 0.6 0.8 1.0 1.2
0.2 0.4 0.6 0.8 1.0 1.2
–2.0 4.0
0.2 0.4 0.6 0.8 1.0 1.2
–4.0
0.2 0.4 0.6 0.8 1.0 1.2 –2.0
–4.0
0.2 0.4 0.6 0.8 1.0 1.2 –2.0
0.2 0.4 0.6 0.8 1.0 1.2
Mismatch Swap Correct
0.2 0.4 0.6 0.8 1.0 1.2
.. ERPs for confusable words, time-locked to target noun onset (at ms) Responses (at electrode Pz) to Swap (blue) relative to Correct (green) sentences (upper row) as well as Mismatch (red) relative to Correct (green) sentences are compared between native controls and high/low proficiency subgroups of Attriters and L learners. Proficiency modulates the amplitude of the N for the Swap condition across Attriters and L learners. The N in the Mismatch condition appears is less modulated by Proficiency level than Swap violations. The P in both violation conditions is largest in Attriters. Time ranges (in milliseconds) depicted on the x-axis are relative to the noun onset ( ms). Negative values are plotted up.
proficiency, not by AoA. However, a second ERP effect suggested that the group of attriters as a whole employed a processing strategy that was different from both monolingual native speakers and L learners. That is, for both Mismatch and Swap trials, attriters also elicited a large, broadly distributed P component between and , ms, whereas monolingual controls only showed very small posterior Ps, and L learners did not display any P effects. This late positivity in attriters was reminiscent of a similar effect reported by Lazlo & Federmeier () in response to sentence-final nouns that were semantically incongruent but orthographic neighbours of the expected target word. This suggests that attriters, unlike the other two groups, may have ‘doublechecked’ the similarity between the actually presented word and a likely target word, resulting in ‘second thoughts’ in the Swap condition, compatible with the ‘monitoring’ interpretation of the P (e.g., Van de Meerendonk et al., ). Thus, while lowproficiency attriters may have initially missed the inappropriateness of a noun in Swap trials (like low-proficiency Lers), they seemed to have noticed it with a slight delay. This would also explain why attriters’ offline judgements were more similar to monolinguals than to L learners. Overall, these data provide first ERP evidence for lexical-semantic attrition effects and suggest that, despite similarities between attriters and L learners, the former may be able to rely on additional knowledge not accessible to the latter group.
.. Lexical processing in attriters’ L Attriters are typically assessed in their L and compared either to monolingual speakers or, less often, to learners who are acquiring the same language as an L. To date, only two ERP studies have examined attriters’ lexical-semantic (Kasparian et al., under revision) and morpho-syntactic processing (Kasparian & Steinhauer, unpublished) in their predominantly used L, while comparing them to native monolingual and bilingual speakers processing the same language as their L. The question of interest
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
(A)
N400 difference wave: Italian homograph-English homograph
(B)
4.0
–0.2
N400 difference wave: English correct-Cognate correct –4.0
0.2 0.4 0.6 0.8 1.0 1.2
–4.0
English-Italian learners Italian-English attriters English controls
N400 reduced for IH in Attriters
Controls
Attriters IT Learners
–0.2 4.0
0.2 0.4 0.6 0.8 1.0 1.2
N400 reduced for CC in Attriters
Controls
Attriters
IT Learners
.. ERP responses to homographs and cognates (A). A comparison of the difference waves for the N effect elicited by IH violations relative to correct EH sentences, for all three groups. Italian-English Attriters (red) showed a reduced N in response to Italian homograph (IH) sentences that are semantically anomalous in English but not in Italian. The two native-English groups showed a robust N response for IH sentences. (B). A comparison of the difference waves for the N effect elicited by English correct sentences relative to Cognate correct sentences (CC). In line with cognate facilitation effects, the N for CC sentences is reduced in Attriters. Due to this reduction, the English correct (EC) condition elicits a larger N than the CC condition (only in Attriters), as illustrated by the EC-CC difference wave.
is twofold: () from an L processing perspective, do highly proficient and immersed adult L learners resemble native-speakers despite their late age-of-acquisition; and () is L processing in attrition characterized by reduced competition from the L (i.e., greater L inhibition)? As the L ‘takes the lead’ in circumstances of L attrition, it is only logical to examine attriters’ L in addition to their L, in order to study the ERP correlates of this change. Using ERPs, Kasparian et al. (under revision) examined the processing of sentences containing Italian–English ‘false-friend’ homographs (e.g., estate (English: ‘property’ versus Italian: ‘summer’) and matched cognates (music/musica) in Italian-English attriters, English monolinguals and English-Italian late L learners. English sentence contexts either fit with the English meaning of the homograph, the Italian meaning of the homograph, the shared meaning of the cognate, or were semantically implausible in either language. The aim was to assess whether the Italian lexicon of attriters and L learners was automatically co-activated while reading in English, and whether factors such as proficiency, language use and/or length of residence modulated this co-activation. In line with language non-selective models of bilingual lexical access (Kroll & Stewart, ; Dijkstra & Van Heuven, ), both bilingual groups showed co-activation of Italian during English-processing not only in the N time-window but also on the P further downstream. Sentences that fit the Italian homograph meaning were perceived as less of a lexical-semantic violation (reduced N), particularly for Italian-English attriters (Figure . shows difference waves for simplicity). A cognate effect was found for both bilingual groups, though in different processing-stages; while Italian attriters showed reduced N and P effects (i.e., facilitation) in English for correct cognate conditions, L-English Italian learners only showed this facilitation effect in the P time-window. Instead, they showed a tendency
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
toward cognate interference effects in the N window. In both bilingual groups, homograph-targets elicited larger posterior Ps than cognate target words. This effect was positively correlated with bilinguals’ performance on a translation (production) task containing different false-friend items, suggesting that the P in ‘better inhibitors’ reflected a conflict-monitoring process (Van de Meerendonk et al., ). These realtime ERP effects were not directly mirrored in end-of-sentence acceptability judgements. Importantly, L co-activation was weaker (and L processing more native-like) with higher English proficiency, longer length of residence, and reduced Italian exposure. These novel findings suggest that L attrition may be partly conceptualized as a reduction of L activation during real-time L processing as a result of strengthening L formmeaning links in contexts where attriters must continuously inhibit the L in the L environment (Linck et al., ; Whitford & Titone, ).
. S , ,
.................................................................................................................................. We believe the first ERP studies on L attrition have already generated very exciting data. In the domain of lexical-semantic processing, we saw that Italian immigrants in Montreal showed ERP profiles that differed from both L learners and monolingual controls in their L (Section ..). Whereas the N effect to confusable words was primarily driven by proficiency (across all groups), the finding of an additional P in attriters pointed to a different processing patterns in this group only, most likely linked to subsequent checks of knowledge not available to L learners. The less automatic activation of L lexical representations in attriters also seemed to affect their L processing of English words, where less exposure to Italian resulted in less competition (Section ..). Perhaps the most interesting results come from morpho-syntactic studies. For agreement violations, the native-like GG data from Monika Schmid’s group suggested no ERP attrition effects at all, whereas the Montreal data on number agreement indicated consistent group differences between attriters and monolingual controls for various ERP components at both verb and modifier positions. In our opinion, these apparently inconsistent findings should be taken as a warning not to overgeneralize any data from individual studies, but to be prepared for a more complex pattern influenced by various experimental and psychometric factors. That is, the research question of ‘whether there exists L attrition for morpho-syntax’ may not lead to a simple ‘yes’ or ‘no’ but rather require more differentiation regarding structures, language modality, L–L pairings, and so on. Starting this research endeavour with apparent inconsistencies in ERP data may be beneficial in preventing us from adopting oversimplified answers, as was the case for ERP research on late L learners in the early s, when all data seemed to confirm the critical period hypothesis. Recent research on both L acquisition and L attrition suggests that there is at least some degree of post-puberty brain plasticity that can affect morpho-syntactic real-time processing. For L attriters, perhaps the best example is the data for (grammatical) Italian relative clauses that Italian immigrants in Montreal (but not monolingual Italians)
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
perceived as ungrammatical (Section ..). However, only data from different L–L pairings can tell to what extent the results were caused by English grammar influence, and this is only one of many open questions. Future research needs to question initial interpretations and systematically test alternative accounts. All ERP studies presented in this chapter have their shortcomings; a common one across experiments was that the groups of L learners and attriters were not perfectly matched in terms of proficiency. Attriters were typically in the proficiency range of the monolingual controls, whereas L learners had lower scores for at least certain proficiency measures (which in itself is sometimes overinterpreted as evidence for critical periods). The main reason for this asymmetry is that recruiting large and homogenous groups of ‘strong’ attriters and native-like L learners is a non-trivial (and very time-consuming) task that, moreover, usually requires the involvement of multiple labs in two or more countries. Whether some group differences (such as the finding of a P for confusable words in attriters only) were partly due to (undesirable) matching problems of this kind needs attention in future studies. Both research teams in Groningen/Essex and Montreal/Italy have tried to address some of these issues by running additional correlations and regressions that use measures of proficiency, length of L exposure, etc., as continuous predictor variables for ERP patterns. The use of generalized additive mixed models (or GAMMs; Meulman, Wieling, et al., ; Nickels & Steinhauer, , in preparation) that can model non-linear relationships between continuous predictor and predicted variables promises to offer even more sophisticated statistical approaches to deal with these challenges. A related question is whether differences or limitations in experimental design may explain some of the apparent inconsistencies across studies on agreement violations (i.e., an attrition effect for number disagreement but not for GG violations). Although in both Bergmann, Meulman, et al.’s () GG study and Kasparian et al.’s () number agreement study, attriters were quite well matched with the control group, the different outcomes may be influenced by the fact that Kasparian et al. only recruited attriters who explicitly reported problems with their L. That is, only a subset of immigrants whose daily exposure to their L is very limited and who reported experiencing problems in their L may display clear attrition effects in ERPs. Another factor contributing to different findings may have to do with the slightly more complex (and longer) sentences in the number agreement study (which were presented intermixed with many other structures). Finally, Kasparian and colleagues employed a somewhat more fine-grained ERP analysis in multiple P time windows, whereas Bergmann, Meulman, et al. focused on one large P time window ( to , ms). Future research will have to clarify which of these factors are the most relevant ones. For the absence of any GG violation ERP effects in late L learners (Meulman, Stowe, et al., ), however, their low performance during the online judgement task may indeed have played a role, because only a relatively small number of trials may have contributed to actual ERP differences between the conditions.3 3
The L2 learners’ accuracy for GG conditions was 62.8% (compared to 92.2% in the control group), with a chance level of 50%. In a ‘worst-case scenario’, only about 26% of trials in each condition may have reflected L2 learners’ correct judgements, while up to 74% of trials did not contribute to ERP effects between grammatical sentences and GG violations. The much higher scores in the offline gender assignment task did not turn out as a good predictor for online processing during the ERP study, which is a common issue with proficiency measures that tap different types of knowledge or cognitive
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Other open questions have to do with determining to what extent L attrition is due to (a) the reduced exposure to one’s L, or (b) the influence of L on L in terms of coactivation, competition, transfer, and L inhibition. As reduced L exposure almost always takes place in the context of predominant L exposure and active L use, teasing apart these different aspects will be challenging. The exact L-L pairing likely also plays a role. For example, Bergmann et al.’s () attrition study used English as the L, which however does not have a grammatical gender system that could compete with the German one. The resulting lack of competition between L (English) and L (German) may be a main reason for the the absence of attrition effects in their study. Similar to comparing ERP profiles of L and L during different stages of language acquisition, ERP research contrasting L and L attrition may help us better understand shared and distinct trajectories of losing a language, including the role of changes in brain plasticity over the life span. For the time being, we believe the present ERP studies show that anecdotal reports and ‘feelings’ of change in L can be substantiated with real-time processing patterns. Will ERP attrition effects be a rare exception, or rather the rule? We expect that the truth will be somewhere in the middle, but it will be the specific details that define this ‘middle’ that count and will advance the field.
processes (for discussion, see Steinhauer et al., ; White et al., ). In these situations, responsecontingent data analyses may reveal significant ERP effects for a ‘subjective’ (rather than objective) online categorization of grammar violations, as has been prominently demonstrated in a recent ERP study by Lemhöfer et al. () who initially also did not find any significant ERP effects in L learners, interestingly also for GG violations in Dutch. Note, however, that the logic of relying on more detailed behavioural data is not universally applicable. In some studies, reliable ERP effects may occur long before behavioural data show corresponding effects (e.g., McLaughlin et al., ).
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
......................................................................................................................
......................................................................................................................
, , .
. L :
.................................................................................................................................. F language (L) attrition is the temporary or permanent non-pathological loss of language ability in a native language, such that language performance is not what would be considered to be typical performance in native/monolingual speakers of a similar age and proficiency. Language attrition is a field of research at the interface of linguistics, psycholinguistics, and neurolinguistics. Historically, neurolinguists have investigated language loss in the face of pathology, such as aphasia, in monolingual, bilingual, and multilingual individuals. In the last two decades, research on non-pathological language loss has been increasing, with the realization that L language attrition is a highly complex, multifaceted phenomenon, influenced by a number of linguistic, socio-linguistic, cognitive, and neural variables, thus sparking a number of studies that have capitalized on different methodologies (see Köpke, ; Köpke & Schmid, a; Köpke & Keijzer, Chapter , this volume). The overwhelming majority of L attrition studies have focused on behavioural paradigms that have tackled language change and loss in adult speakers who fully acquired their L before emigrating to a country where they were consistently exposed to, and used a second language (L) (Weltens et al., ; de Bot, ; for a review, see Köpke, ) with the goal of advancing the discussion on the nature of language representation and processing (see Schmid & Köpke, , for a detailed description of the various theoretical proposals). It is only very recently that non-pathological L attrition investigations have been informed by neural investigations, with a relatively small number of published articles. The scarcity of neural studies on L attrition may arise from the idea that L attrition has been considered to be an isolated phenomenon rather than a typical facet of
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
bilingualism (cf. Schmid & Köpke, ), and perhaps also because of the relative difficulty of recruiting a large enough population of L attriters (but see Keijzer, , for a large behavioural dataset on L Dutch attriters in Canada). To date, there are only a handful of studies that have looked at the neurophysiological signature of L attrition using EventRelated Potentials (ERPs; see Steinhauer & Kasparian, Chapter , this volume), and even fewer studies have investigated the functional and/or structural neural underpinnings of attrition using neuroimaging methodologies such as Magnetic Resonance Imaging (MRI). The goal of this contribution is to review the extant MRI literature on non-pathological L attrition. We will first introduce the neuroimaging methods that will be discussed throughout the chapter, focusing especially on fMRI methodology, and we will then review the most relevant neuroimaging literature on L attrition. Finally, we will illustrate how recent neuroimaging data (both from fMRI and EEG) on changes to the L performed within the context of bilingual language processing more generally could be informative in our understanding of L attrition.
. I M R I
.................................................................................................................................. MRI is a non-invasive imaging technique that can provide measures of brain function (i.e., fMRI) and brain structure (e.g., anatomy, measures of white matter integrity) with incredibly high spatial resolution (~mm; Brown et al., ; Huettel et al., ). fMRI measures the changes in the ratio of oxygenated to deoxygenated haemoglobin in the blood and thereby can provide an indicator of which brain regions are more engaged during a myriad motor, cognitive, and linguistic tasks. fMRI is considered to be a relative measure. It allows us to see regions that are more engaged during one experimental condition than another, or more engaged during one condition relative to some baseline measure (Huettel et al., ). Because of these issues, task design in fMRI is of critical importance in being able to interpret a study’s results. There are two basic study designs that have been used in functional studies: event-related and blocked designs. Event-related designs allow one stimulus to be paired with one haemodynamic response. This typically involves a random order of presentation of stimuli (across conditions) and a varied inter-stimulus-interval for stimuli to be less predictable, which leads to less pre-planning of responses, and also allows for variability in the haemodynamic response itself, both of which strengthen the experimental design. The limitation of event-related designs is that they often require many trials per condition (at least thirty-five to forty), with jittered intervals, which can lead to a lengthy experimental session (Huettel, ). In contrast, blocked designs involve presenting stimuli from one condition in rapid succession, leading to to ‘blocks’ of stimuli, which are alternated with rest periods or blocks of other conditions. This prolonged presentation of stimuli allows for a single, often larger, haemodynamic response, which can mean fewer blocks are needed to reach a meaningful average. A key limitation of blocked designs is that it is not possible to tease apart the fMRI signal for any one stimulus as the signal is linked to the entire block (Maus et al., ).
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
, ,
In addition to functional imaging, MRI can be used to investigate structural changes in grey and white matter. Standard anatomical images (e.g., T- and/or T-weighted imaging) can provide volumetric measures of grey matter, white matter, sub-cortical structures, and fluids, such as cerebrospinal fluid (CSF). Recent advances have allowed for more sophisticated structural assessments such as measures of cortical thickness and gyrification (Koelkebeck et al., ; Gautam et al., ). These methods can also be used to quantify cortical decline, such as the volume of white matter hyperintensities, which make them invaluable tools with which to assess change, as for example in structural changes mirroring language change after immersion, or language change through attrition. To date, there is considerable MRI evidence based on bilingual language use. Structural differences in grey matter density were reported in a seminal study by Mechelli et al. (), who demonstrated that bilinguals had greater grey matter density in the left inferior parietal cortex relative to monolinguals. For more comprehensive reviews on the effects of bilingualism on grey matter volume, see Stein et al. () and Li et al. (). Finally, Diffusion Tensor Imaging (DTI) is a methodology that examines the anisotropy (i.e., directionality) of water flow in white matter fibres, and has been used as an indicator of white matter integrity (Madden et al., ; Tournier et al., ). Although there have been no structural MRI (DTI specifically) studies investigating L attrition, research suggests that bilingualism is associated with higher white matter integrity (e.g., Luk et al., ; Mohades, Struys, et al., ; Garcia-Penton et al., ; but see also Gold et al., ) even for those who acquired their L relatively late in life (Pliatsikas et al., ), and even at intermediate levels of L learning (e.g., Mohades, Struys, et al., ; Mohades, Van Schuerbeek, et al., ; Rossi et al., ). Overall, studies on structural changes in grey and white matter connectivity due to second language learning or bilingualism have demonstrated a very high level of neural plasticity. Even though no studies to date have investigated changes in white matter connectivity as a consequence of L attrition, it would be possible to hypothesize that decrease in grey and white matter density, and changes in measures of white matter connectivity should be observed as a consequence of language loss. This will be an empirical question to be answered by future research in L attrition, but is also one that can greatly aid a deeper understanding of what attrition is exactly.
. M R I L
.................................................................................................................................. The number of neuroimaging studies that have specifically investigated L attrition is very low. However, in a seminal series of studies designed to ask the question of whether longlasting neural traces of the L can be found after years of non-use, Pallier and colleagues (Pallier et al., ; Ventureyra et al., ; Pallier, ) have investigated L language processing in a number of attriters with the main goal to answer the question of whether those speakers had any detectable neural activation related to their L. Participants were Korean children who had been adopted by Francophone families at a relatively early age and who were exposed to French only after adoption (details of these studies are reported
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
by Pierce et al., Chapter , this volume). Importantly, after being adopted, Pallier’s () participants were completely isolated from Korean. This is similar to what has been reported in another L attrition study that described the displacement of German Jewish children during World War II to English-speaking foster families at an early age ( to years, Schmid, ). In other words, there was complete L cessation at a relatively early age. Crucially, these studies probe the Critical Period Hypothesis for language learning (Flege et al., ), which predicts that long-lasting traces of the L should be detected even after years of non-use. To measure possible residual knowledge of Korean, Pallier et al.’s () participants completed two tasks: a word recognition task in which they listened to a French word and decided which of two possible Korean words was the correct translation, and a language identification task in which they listened to sentences presented in five languages (including Korean) and rated the likelihood that a specific sentence was spoken in Korean. Additionally, to investigate the residual neural sensitivity to Korean, an event-related fMRI study was conducted while participants listened to French, Korean, Japanese, and Polish sentences and performed a fragment detection task. Results from the adoptees were compared to a French monolingual control group. Consistent with the personal reports from the adoptees group who reported no conscious recollection of Korean, behavioural results revealed that Korean adoptees did not show any sensitivity to Korean in any of the behavioural tasks. Overall, no difference in activation was detected between the two groups during the Korean listening condition. Results further showed that listening to French sentences yielded greater activation in a network of areas encompassing left Superior Temporal Sulcus (STS), left Superior and Middle Temporal gyri, left Inferior Frontal gyrus, and the contralateral areas in the right hemisphere in both the adoptees and French control group. Crucially, the network of activation did not differ between the Korean adoptees and the French monolingual group. The only difference between the two groups was observed in the extent of the network while listening to French. The Korean adoptees activated a significantly smaller network than the monolingual French controls, which the authors interpreted as either a residual effect of their L or, alternatively, a result of language variability in their French due to being recruited from different parts of France, while the French control group was recruited from the Parisian metropolitan area only. Importantly, all participants (native French speakers and Korean adoptees) were equally fluent in French. Pallier et al. () concluded that their data do not support a strong version of the Critical Period Hypothesis for L learning (e.g., Flege et al., ), but rather provide evidence in favour of the plasticity associated with language acquisition in the first few years of life from the native to an L. Crucially, the authors interpreted their results as evidence that sensitivity (even at the neural level) to the native language can be diminished if not erased by ceasing to speak or listen to the native language, and by acquiring an L. Contrary to these findings, effects of the native (forgotten) language on the newly acquired L were reported by Rajagopal et al. (), who found differences in language lateralization between monolingual English speakers, and a group of children who were adopted from China or Eastern Europe during a verb generation verbal fluency task in English. Similar to Pallier et al.’s () study, Pierce et al. () used fMRI to study the behavioural and neural processing of Chinese in twenty-three adoptees from China, who were exposed exclusively to French after being adopted (mean age of adoption, year and months), and reported having no conscious recollection of Chinese. Results were
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
, ,
compared to a group of native Chinese speakers (n = ) who learned French as a second language at a very early age (mean age onset learning French: months), and were therefore classified as early Chinese-French bilinguals, and also to a group of French monolingual children who were never exposed to Chinese (n = ). During the study, participants listened to pairs of three-syllable phrases in Mandarin that differed only in the last syllable’s tone, and to pairs of non-speech hummed sentences that were modified to only maintain their tonal information, while performing a similarity judgement task. Results revealed that the Chinese adoptees and the Chinese-French bilinguals recruited similar neural regions within the left temporal cortex (specifically, left planum temporale (PT)) that had been previously found to be active in the processing of tones. Additionally, among the adoptees, a conjunction analysis revealed that the amount of activation in left PT was predicted by Age of Acquisition (AoA), with greater activation for those individuals who had more exposure to Chinese. In contrast with Pallier et al.’s data, these results revealed maintenance of neural sensitivity for the lost L even after no exposure or use of that language for many years. This difference in results may arise from the specific features that were studied, in that tone processing may have a similarity to other kinds of processing, such as music, whereas the voiceless stop consonants that were investigated by Pallier et al. are less likely to have overlap with other kinds of auditory processes. Moreover, the Pallier et al. () and Pierce at al. () studies differ along two crucial variables that might account for the differences in the results. First, Pierce et al. examined lexical tone processing, while Pallier and colleagues’ study included a word matching task and a language identification task, thus encompassing phonological, lexical, and higher language level processing. It is therefore possible that the two studies tapped into partly different linguistic processes that pose challenges for a direct comparison. More importantly, the mean age of adoption, thus the age at which the individuals ceased to be exposed to their native language, was also quite different between the two studies (Pallier et al.: . years, range to years; Pierce et al.: . years, range to months). On one hand, the adoptees tested by Pallier and colleagues had a relatively longer exposure to the L, which in turn could predict greater sensitivity to the L. On the other hand, phonological contrasts are among the earliest linguistic abilities to be learned by infants (Kuhl et al., ), and they also tend to be the features that are more entrenched (Flege et al., ). It is therefore possible that the sensitivity for lexical tone discrimination observed by Pierce and colleagues is characteristic of a more ingrained aspect of a native language, or possibly also an aspect of language that can be transposed to the processing of more general knowledge (e.g., as music). The observed pattern found by Pierce et al. supports the regression hypothesis (for the original formulation see, Jakobson, ; Keijzer, , a; for extensions of this proposal to aphasia, see Bastiaanse & Bol, ; Kolk, ), which posits that linguistic features that are acquired early in childhood tend to be lost last. It is also important to note that the adoptees’ cohort in Pallier et al. () was tested at an older age than the adoptees in Pierce et al. () (mean age at testing: Pallier et al. = .; Pierce et al. = .), making the time lag from the last L exposure and time of testing longer for Pallier et al.’s adoptees (mean time lag = years) than for Pierce et al.’s (mean time lag = years). It is possible that overall differences in lag time between the two studies make the results difficult to compare. A possible approach would be to conduct a meta-analysis of both datasets, which could control and account for variability in age since L exposure, and length of immersion in the L.
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
Overall, the very few fMRI studies that have investigated L attrition (e.g., Pallier et al., ; Rajagopal et al., ; Pierce et al., ) have described neural patterns of individuals who ceased to be in contact with their native language very early during childhood, and were tested as young adults, possibly conflating patterns of L attrition with incomplete L acquisition. To date, there are no fMRI studies on populations of L attriters that fit the more stringent definition of ‘L attrition’ in which the L changes only after complete acquisition. Understanding the neural changes in language processing in speakers who ceased to use their L after complete acquisition is definitely one of the next major steps for the scientific community in this field (see Keijzer, , for a recent discussion on future steps that the literature on attrition could take on neuroimaging). However, an increasing number of studies show L changes in response to L exposure (e.g., Dussias & Sagarra, ) even after relatively short periods of immersion in the L environment (Linck et al., ), or at very early stages of L learning (Bice & Kroll, ), suggesting that the behavioural and neural underpinnings of the L are much more malleable than previously thought.
. N L L
.................................................................................................................................. The relatively few fMRI studies that have investigated L attrition have yielded mixed results. As discussed, variability in the age at which the L exposure stopped, differences in the relative length of immersion in the L, differences in the developmental stage of the individuals tested, and differences in the aspects of languages that were assessed all may contribute to the observed differences across the studies. Undoubtedly, however, the existing studies represent strong evidence for changes that occur to the L as a consequence of a sudden interruption of contact with the L, prolonged lack of use of the L, and/or prolonged immersion in an L environment. These cases, however, represent one extreme tail of a continuum that spans from no L attrition to extreme L attrition. In other words, the current literature on L language attrition represents one particular facet of the phenomenon that could be assumed to be the tail-end of a distribution of change that may occur in the L. We argue that L attrition should be understood as a flexible, continuous adaptation of the language system that is characterized by modulations in the linguistic, cognitive, and neural functioning of the L. Importantly, we propose that the first stages of this adaptive process might start very early during L acquisition, L use, and immersion in the L environment, partly based on the existing evidence that has accumulated so far. Moreover, those early changes might share similar behavioural and neural signatures to more general bilingual language processing. Crucially, neural data (i.e., MRI, fMRI, ERP) will play an increasingly important role in detecting early functional and neural changes in L processing that might not be observable in behavioural measures. The past few years have seen a number of experimental paradigms that have tested the linguistic and cognitive underpinnings of bilingual language processing. Many of those
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
, ,
paradigms have capitalized on tasks that require participants to switch between their two languages in a number of ways. Under the hypothesis that L attrition is one facet of a more general process encompassing linguistic and cognitive changes to the L, these experimental paradigms could then be re-interpreted as an essential test-bed for measuring early stages of linguistic and cognitive changes that might represent, or be similar to, the earliest stages of L attrition mechanisms. Recent blocked language switching fMRI studies have investigated how speaking the L for a relatively short amount of time dramatically impacts L processing, revealing that the costs related to switching from the L to the L are greater than when switching from the L to the L, supporting the hypothesis that the strongest language needs to be temporarily inhibited to allow production in the L (Green, ). For example, Guo et al. () used a blocked language-switching picture naming paradigm during which participants named blocks of lexical items (using picture cues) alternating between their two languages, or switched languages item by item to determine whether the neural basis for the blocked switching effect was the same as the one engaged by mixed item language naming. Results showed that the dorsal anterior cingulate cortex and supramarginal gyrus were activated to a greater extent during mixed language naming and the dorsolateral prefrontal cortex and parietal cortex were more activated when the L followed the L in blocked naming. Crucially, the regions that were activated during blocked naming are associated with boosting relevant information and switching mental sets, suggesting that even short periods of naming in the L engage a global L inhibitory mechanism (Branzi et al., ; Rossi et al., ). Rossi et al. (in preparation) used fMRI during a blocked language-switching naming paradigm in English-Spanish bilinguals. However, unlike Guo et al. (), the task was designed to test whether inhibition of specific items in the L would spread beyond the item level, to the larger semantic category, and to the whole language level. To that goal, participants named items from different semantic categories in their L first. In a second block, they named pictures only in Spanish (L) that were drawn from three other categories, (e.g., clothing, furniture, kitchen items). Finally, in six subsequent blocks they named pictures in their L (English) again that were mixtures of items drawn from the old items and categories presented in the first two blocks, new items from previously named categories, and new items from categories not previously named in either language. Results provided evidence for significant inhibition in naming in L after an intervening block of L, as revealed by decreased accuracy and longer naming latencies. Moreover, there was activation in the brain areas involved in general cognitive control (Anterior Cingulate Cortex, pre-supplementary motor area, and the basal ganglia) while processing pictures that appeared during the initial blocks, but also while processing novel pictures drawn from the same categories previously seen, and even while processing pictures from novel categories not seen earlier in the experiment. Overall, this suggests that the scope of inhibition during bilingual speech extends more globally beyond the word level and that its time course is relatively long. Overall, blocked naming paradigms have highlighted how the native language is modulated and controlled even in the context of short periods of naming in the L, and have started providing data in support of the idea that temporary language suppression of the L, as a consequence of speaking the L can pervade the whole language system (Rossi et al., in preparation). It is therefore tempting to think that the very same mechanisms might be at play during L attrition, or at least in its earliest stages, especially since language attrition has frequently been referred to as an unbalanced form of bilingualism (Schmid, a).
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
Apart from experimental paradigms that have been conducted in laboratory settings, fast and dynamic changes to the native language have been demonstrated to be triggered by immersion in the L environment, a situation that is in itself much closer to the environmental scenarios that have been described as leading to L attrition. Importantly, in line with recent behavioural findings that have demonstrated changes in L processing as a function of a relatively short immersion in an L environment (e.g., see Linck et al., , for verbal fluency data; Dussias & Sagarra, , for parsing preferences), recent fMRI data have revealed that neural activation in bilingual speakers diverges from the neural activation observed in monolingual speakers, even when the task is conducted in the L and in a single language context, suggesting that speaking an L has a strong impact on the neural organization of the L. For example, a recent study by Parker-Jones et al. () tested a group of English monolingual speakers, and a group of highly proficient L speakers of English (who were immersed in the UK). Parker-Jones et al.’s data revealed that bilinguals had increased activation in frontal (Pars Opercularis (POp), Pars Triangularis (PTr), Precentral Gyrus (PrG), and temporal regions (PT and Superior Temporal Gyrus (STG)) relative to monolingual speakers while naming objects or reading words in a single language context. Increased activation in these areas was observed when bilinguals performed the task in the L, and also in the L. Importantly, the areas that were more active in bilinguals than in monolinguals encompassed both language-specific areas, as well as areas that have been shown to be active during interference tasks (POp and PTr), suggesting that processing more than one language does not only have consequence for the L but critically also for the L. Converging results have been found in an fMRI study by Rossi et al. (in preparation) in which English-Spanish bilinguals elicited greater activation in the right caudate and putamen than the English monolingual control group while naming pictures in English, their L, suggesting that bilingual speakers engage areas dedicated to control and language selection, even when they are speaking in their L. Overall, the few fMRI studies on language attrition and the extant research looking at the neural substrates of bilingual language processing and control show that neuroimaging data represent an increasingly important aspect in bilingual language research, including research on L attrition, understood as a specific facet of bilingual language processing. In what follows, we discuss two specific research directions and designs within the neuroimaging domain (i.e., longitudinal designs and neural oscillations) that might be particularly fruitful to advance the research on L attrition.
. F L
..................................................................................................................................
.. Longitudinal designs One fundamental aspect of L attrition is the language change and loss due to prolonged immersion, and lack of L use. As such, longitudinal neuroimaging designs that track neural changes over time, while changes in behaviour might not be yet detectable, could become paramount in the next stages of research on L attrition. Previous longitudinal
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
, ,
neuroimaging studies have demonstrated grey matter changes in language regions such as left inferior frontal gyrus and left anterior temporal lobe (Stein et al., ) and left parietal cortex (Osterhout et al., ) during L learning. These changes have been shown to occur after both naturalistic immersion experiences (Stein et al., ) as well as classroom experiences (Osterhout et al., ; Martensson et al., ). Moreover, these structural changes in grey matter in left frontal (Stein et al., ), left temporal, and right hippocampal regions (Martensson et al., ) were positively correlated with proficiency, suggesting that the observed structural changes may reflect meaningful changes in language. Similar structural changes have been observed for white matter tracts. Schlegel et al. () examined English-speaking learners of Chinese over months and found increased connectivity in left hemisphere language tracts, as well as their right hemisphere homologues. Interestingly, Hosoda and colleagues () observed changes among Japanese university students learning English in white matter regions connecting right inferior frontal regions and the caudate and temporal regions that were positively correlated with increases in second language proficiency. Their findings suggest that even non-language regions may be sensitive to language-experience-induced changes. Similarly, future studies could capitalize on longitudinal neural designs to track language changes in an L attrition context over time, thus providing fundamental information about the neural and behavioural changes that occur during the long-term process of L attrition.
.. Neural oscillations The use of neuroimaging methods, including EEG/magnetoencephalography (MEG) and fMRI are paramount in examining the rapid neural changes that accompany changes in language use or language processing. While fMRI has great spatial resolution, its temporal resolution is slower than electrophysiological measures. Investigating neural oscillations, recorded with MEG or high-density EEG, allows to combine both fine-grained temporal and spatial resolution. Such measures could help advance the study of language attrition by addressing the question of whether there are traces of the forgotten L even in the absence of behavioural sensitivity. We will first briefly introduce the techniques and then provide an example of how they can be used to tackle the question of maintenance in L language attrition. Electrophysiological activity of the brain (measured with EEG) and the magnetic fields induced by it (measured with MEG) provide a rich source of data that can be analysed in a number of ways. Traditionally, psycholinguistic research has relied on ERPs—the result of time-locking neural activation (fluctuations in voltage generated by large populations of neurons) recorded by scalp electrodes to the stimulus of interest (e.g., single written/spoken words, words in a sentence, images, sounds, etc.) and averaging across many trials and participants (for a review of this technique and its use in psycholinguistics, see Kutas & Van Petten, ). ERPs are thus a time-series of voltage measurements associated with the event of interest which are usually visualized as a waveform, with time plotted on the horizontal axis, and amplitude plotted on the vertical axis. In this way, the time course of activation evoked by the stimulus can be tracked with high temporal precision. Differences in the amplitude and latency of waveform peaks associated with experimental manipulations indicate differences in the underlying neural mechanisms employed to process stimuli
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
in different conditions (see Steinhauer & Kasparian, Chapter , this volume, for a more elaborate discussion of ERP methodology). However, information about induced activity associated with the critical manipulation, but not necessarily time-locked to it, is lost after averaging. Decomposing the recorded signal into different frequencies, investigating local synchrony (the amount of power in each frequency band), and long-range synchrony (oscillatory phase relation measures between different sources/channels) adds a frequency dimension to the electrophysiological signal (in addition to time and amplitude). Thus, frequency analyses might help capture the dynamics of purely linguistic processing (syntactic and semantic analysis) as well as domain-general cognitive processes (memory encoding and retrieval, attentional mechanisms, spatial and temporal neural binding) associated with language comprehension and production. Crucially, the study of oscillatory brain responses allows one to tackle different subcomponents of complex cognitive processes that can be simultaneously implemented by means of synchronization/desynchronization of neuronal activity at certain frequencies, times, or locations. The same brain structures can be recruited simultaneously for different cognitive processes due to their ability to operate at multiple temporal scales. Neuroimaging measures such as ERPs and the BOLD (blood-oxygen-level dependent) signal used in fMRI studies are not specifically geared towards disentangling different synchronic cognitive processes, while this is possible when looking at neural oscillations. Traditionally, brain oscillations are subdivided into five frequency bands: delta ( to Hz), theta ( to Hz), alpha ( to Hz), beta ( to Hz), and gamma (> Hz). Here, we will focus on memory and language related processes. Memory research has related synchronization in theta and gamma bands as well as desynchronization in alpha and beta bands to episodic memory encoding and retrieval (Hanslmayr et al., ). Language research has related theta activity to lexical retrieval, alpha and beta activity to semantic and syntactic unification, and gamma activity to prediction during sentence processing (Lam et al., ). Since its discovery, alpha activity has been repeatedly found to correlate with attentional demands (for a review, see Klimesch, ). Gamma oscillations are thought to be a binding mechanism that integrates information from distributed groups of neurons (Colgin et al., ). For a more detailed review of how brain oscillations relate to dynamic cognitive processes, see Ward (). High-density EEG allows for decent spatial resolution as long as ‘anatomically precise forward models of subject-specific brain and skull anatomy’ (Cohen, , p. ) are used and spatial filtering techniques are applied. Spatial resolution of fMRI still remains the best; however, because of the slower haemodynamics, these measures result in ‘a rather static picture, emphasizing mainly the structural aspects of the Brain and Language network’ (Bastiaansen & Hagoort, , p. ). Linguistic processing, like any other cognitive operation, is very rapid and characterized by the dynamic formation of ‘transient functional language centers’ (Weiss & Mueller, ), thus methods with precise temporal and spatial resolution are required to track these centres throughout space and time as linguistic input unfolds. In what follows, we will illustrate how the use of neural oscillations could be applied to the study of language attrition and effects of re-exposure, by using examples from work on word learning. Previous work on word learning has shown that the analysis of brain oscillations is sensitive to the degree of lexicalization of novel and existing words. For example, Bakker et al. () compared neural representations of words at different stages of lexicalization:
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
, ,
untrained existing words, untrained novel words, words learned immediately before testing, and words learned twenty-four hours before testing. Words that have wellgrounded representations have been shown to elicit theta and gamma power increases and beta and alpha power decreases. Bakker et al. () found that novel words seen by participants for the first time were characterized by a smaller theta sychronization compared to existing words. Words trained before the experiment elicited a theta response marginally lower than the one observed for the existing words. Words learned twenty-four hours before the test elicited a response identical to that of existing words, suggesting memory consolidation of the learned information. Beta desynchronization was characterized by a similar pattern: existing and consolidated words elicited identical neural responses; however, unfamiliar and recently learned words elicited a significantly smaller power decrease in beta band than existing words. Converging evidence has been reported in the fMRI literature suggesting that the brain differentially represents words learned at different points across the lifespan. For example, Fiebach et al. () have shown that the processing of words learned earlier in life recruits the precuneus (a brain structure important for episodic memory retrieval), while late learned words activate lateral inferior frontal areas (associated with lexical access and linguistic processing in general). Overall, recent work on brain oscillations and fMRI illustrates the effects of AoA on lexical processing suggesting that the brain responds to linguistic structures differentially depending on when they were acquired during development. The analysis of lexical items that were either forgotten or reinforced later in life (after being forgotten) could be a fruitful further direction for L attrition research. For example, by looking at theta and beta power induced by words from the forgotten language that were supposedly acquired by the attriters during childhood one might be able to see whether—based on what has been discussed above—they would elicit more theta increase and more beta decrease than novel unfamiliar pseudowords learned immediately before testing. Employing MEG and highdensity EEG would also provide information about the sources of these effects.
. S
.................................................................................................................................. The goal of this contribution was to review the current neuroimaging (fMRI) literature on first language attrition, and to discuss how current fMRI research informs theories of bilingual language representation and control. Overall, the few existing fMRI studies of L language attrition (i.e., Pallier et al., ; Rajagopal et al., ; Pierce et al., ) have only looked at neural patterns of sensitivity to the native language in individuals who ceased to be in contact with their native language (very) early during childhood. The results of those few studies are mixed, with some suggesting that the early adoptees do not maintain any neural or behavioural sensitivity to the native language (Pallier et al., ), while others reveal traces of neural sensitivity to the forgotten language despite the lack of behavioural sensitivity (Pierce at al., ). It was also noted that, in contrast to the few studies on early adoptees, there are no fMRI studies for speakers for whom attrition processes started only after complete acquisition of the L. Despite the still high financial costs related to utilizing MRI, this methodology has become increasingly available in labs around the world. It would therefore be important to embark on studying how the neural representation
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
and function of the native language changes for ‘late’ attriters. This research direction seems particularly important, because MRI-related techniques could highlight changes in language representation and processing even when there are no appreciable changes in behavioural measures. We also proposed that two fruitful approaches for future studies on L attrition, specifically in the neuroimaging domain, would be to utilize longitudinal designs and to study brain oscillatory signals. Longitudinal designs allow studying functional and structural changes in response to language change while attrition takes place, and even in the absence of behavioural changes. The study of oscillatory brain signals across different frequencies will help disentangle the dynamics of linguistic changes due to L attrition, as well as domain-general cognitive processes such as memory, and attentional mechanisms, which are associated with language comprehension and production. Moreover, oscillatory studies are feasible to incorporate with behavioural, longitudinal, or cross-sectional designs. Most importantly, we discussed recent fMRI literature that has highlighted how learning, speaking, and processing an L, even at very early stages of exposure and immersion, have important functional and structural consequences for the representation and control of the L (e.g., Guo et al., ; Parker-Jones et al., ), suggesting that changes in neural representation and control of the native language are a normal facet of bilingualism, and not limited to attrition processes. We therefore argue that the phenomenon of L attrition should be understood as a flexible, continuous adaptation of the language system that is characterized by modulations in the linguistic, cognitive, and neural functioning of the L in the face of bilingualism. Importantly, the first stages of the observed adaptive neural processes that start very early during L acquisition, and immersion in the L environment might be the very same ones that are at play in the earliest stages of native language attrition.
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
........................................................................................................................
LINGUISTIC FACTORS IN LANGUAGE ATTRITION ........................................................................................................................ Edited by
.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
......................................................................................................................
......................................................................................................................
.
. I
.................................................................................................................................. T chapters in the present section provide an overview of research concerning the linguistic properties of attrited languages—that is, of those features which typically make up the dependent variables in investigations of attrition. As such, they represent what is probably the oldest interest of this research field: from the earliest stages of attrition research, studies have attempted to provide not only descriptive accounts of how linguistic properties may change under specific circumstances of attrition (e.g., based on their similarity or difference across the languages in question) but also to integrate these findings into theoretical models of language contact and bilingual development. It has often been pointed out that the recognition that a mature speaker’s language can change within this person’s lifetime is a fairly recent one. While language change across generations and across centuries has been a key preoccupation of linguistic study for centuries (most notably represented by the nineteenth century Neogrammarian approach) and is thus at the root of virtually all modern day linguistic knowledge and theory, the general assumption has always been that, once a native speaker reaches maturity, their first language (L) becomes stable and, hence, uninteresting—and this view prevails to this day (e.g., Gregg, ). The first linguistic researcher to challenge this assumption was the great Einar Haugen who, in his essay in ‘Language and Immigration’ (), not only laid the foundations for an entire research field but also coined the term attrition to describe it (p. ). Berating ‘both lay and learned’ for regarding the phenomenon of ‘the dialect of the immigrant’ with ‘scorn or amusement’, and in particular the linguists for considering this phenomenon ‘as beneath their dignity’ (p. ), Haugen points out to what extent our understanding of the host of immigrant languages spoken in America falls short of capturing the overall linguistic experience and reality: ‘No general perspective has been reached, and few general conclusions drawn’ (p. ). He sets out to catalogue his own observations from extensive
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
investigations of American Norwegian, in the hope ‘that students of immigration may find here some intimations of what can be derived from continued researches into the problems touched upon’ (p. ). Focusing mainly on the domain of vocabulary, Haugen identifies phenomena such as a decrease in accessibility of low-frequency L lexical items, which are being replaced by their L equivalents, the loss of lexical distinction made in the L but not the second language (L) (e.g., separate terms for maternal and paternal relatives), and the cross-linguistic blend of forms and meanings. Haugen also acknowledges that English influence on Norwegian is more readily seen on lexical than on grammatical/ functional items. Haugen describes with insight and detail different processes that combine to bring about the phenomenon of attrition, such as dialect levelling among the migrant communities, the social dynamics within these communities, the decrease of Norwegian literacy across generations, and difficulties associated with L acquisition. His analyses of how and why lexical mergers have occurred include a wide range of background phenomena, from the social makeup of the immigrant communities to the influence of characteristics of the new environment, such as its technology, its geography, and its climate. Crucially, he points out that the aggregate influence of English on Norwegian, which he calls a ‘psychic compulsion’, is at work ‘from the day of immigration’ (p. ). Despite Haugen’s early work, and the efforts of others to draw attention to the fact that cross-linguistic influence in bilingualism is not a unidirectional phenomenon (e.g., Weinreich, ), it took several more decades for linguistic research to develop an interest in the phenomenon that went beyond the qualitative and the anecdotal. One of the earliest attempts to provide a set of predictions for language attrition as a process of linguistic change is Roger Andersen’s () ‘blueprint for language attrition research’. Andersen proceeds from the assumption that, under conditions of inadequate input and interaction in a given language, not all lexical, phonological, morphological, and syntactic distinctions can be maintained, but that deterioration will be selective (p. ). Drawing on evidence from research on language contact and language acquisition, the author sets out a number of specific and testable predictions for potential changes through which language attrition may affect different linguistic levels. He invokes factors such as frequency, cross-linguistic similarity, and the functional nature/contribution to meaning of certain categories or features as important drivers of the attritional process. While many of these predictions can still be considered relevant today, thirty-five years of research into the linguistic properties of language attrition have refined our understanding of the contribution of these factors but also established further factors and attempted to account for their findings based on a variety of theoretical frameworks (see also the contributions in Part I of this volume). Following on from this and other early attempts to kick-start language attrition research, empirical findings initially were quite slow to emerge, as much of the early work done on language attrition fell into one of three categories: (a) theoretical/hypothetical work; (b) qualitative work or small-scale pilot studies; and (c) initial outlines of larger projects (which, unfortunately, rarely saw eventual completion) (Köpke & Schmid, ). It was not until around the start of the new millennium that a solid evidence base began to emerge. Köpke & Schmid () review the evidence available at this stage, with a view to moving the field forward, and suggest a number of theoretical and methodological considerations, arguing for a stronger integration of purely linguistic observations of language attrition with, on the one hand, extralinguistic factors such as language exposure and attitudes (see Part IV of this
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
volume), and theoretical approaches to the development of the representation and use of linguistic knowledge (see Parts I and II of this volume) on the other. Probably the most important and most controversial issue in this context is the question of what kinds of changes to linguistic features actually constitute language attrition. Investigations of effects of the second language on the first often attempt to distinguish ‘superficial’ from ‘structural’ changes—in earlier treatments, this is framed in terms of the competence versus performance distinction (e.g., Seliger & Vago, ), while later investigations often apply different but related concepts, such as representation versus processing. In early work, this distinction was assumed to be crucial, with representational changes privileged over processing-related changes in terms of their importance to attrition research. For example, for Seliger & Vago ‘it is erosion that reaches the level of competence that allows for interesting claims about and meaningful insights into the attrition process’ (, p. ). In this vein, as Schmid & Köpke () point out, attrition was originally thought of as a distinct developmental category, experienced by a small subset of bilinguals who had experienced very long-term radical deprivation of input from and exposure to their L, and who had—presumably as a consequence of this deprivation—experienced changes to their L that could clearly be shown to be structural/representational. On this view, the more ‘superficial’ differences in processing the native language, experienced by all bilinguals as a consequence of the co-activation of two language systems, should not properly be considered attrition (e.g., Meisel, ). Schmid & Köpke argue, however, that the reality of attrition research suggests that instances where a representational change can be demonstrated are rare, if not unheard of, among late bilinguals; that the absence of a clear picture of the impact of length of exposure and frequency of use (see Schmid, Chapter , this volume) goes against the notion of a small subpopulation of ‘true’ attriters among the larger population of bilinguals, and that, in contrast to Meisel’s view, all instances of L-to-L transfer should therefore be considered attrition phenomena. To illustrate this point with an example, consider phonetic versus phonological changes in the L of immersed bilinguals: phonetic changes might be considered to be ‘superficial’ and not affect the underlying structure, while a phonological change such as the loss of a phonemic contrast (rather than simply change within a phoneme, as in the case of phonetic attrition), would indeed affect the underlying grammar. According to Meisel’s assessment, presumably, only phonological attrition would qualify as ‘true’ attrition. Interestingly, however, similar changes can yield phonetic (i.e., supposedly superficial) versus phonological (i.e., supposedly structural) outcomes in different populations. This is the case in two investigations of the lateral approximant in German-English (de Leeuw, Mennen, & Scobbie, ) and Albanian-English (de Leeuw et al., ) attriters. Both studies found a darkening of the light lateral approximant in coda position in an adaptation to the English pronunciation. However, while this change in German only affects the pronunciation, not the meaning, in Albanian the contrast between light and dark laterals is phonemic, and so the difference in pronunciation actually changed the meaning of the words. In other words, the same observation of attrition would be interpreted as a superficial one for the German-English bilinguals but as a structural and representational one for the Albanian-English bilinguals. More likely, however, as this example illustrates, is that the effects of L-to-L transfer exist on a continuum where development is gradual though not necessarily linear and
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
where the strength of any measured effect is strongly related to both the task, the linguistic level under observation, and the properties of the relevant feature in both languages. Using the example above, this would mean that any neutralization of a phonological contrast in Albanian laterals would have been preceded by an initial phonetic change (as observed in many of the participants), and that, therefore, the phonetic change in the German lateral may very well also have been a phonological change in the phoneme inventory of the individual German native speakers. It is just that in German, this change was not able to constitute a phonemic neutralization effect—because there is only one lateral in German. As this example shows, it is often very hard to establish at what point—if at all—a change that initially affected online processing—that is, the surface level becomes structural/ representational (nor, indeed whether representational changes must necessarily be preceeded by online ones). In addition, as Schmid & Köpke () point out, and as the chapters in this section illustrate, the vast majority of studies on language attrition deal with phenomena which are unlikely to be indications of a ‘competence change’. They therefore consider all such phenomena part and parcel of one single developmental process— whether one chooses to subsume this process under the label of ‘language attrition’ (as Schmid & Köpke propose), or prefers a more neutral term, such as ‘effects of the second language on the first’ (Cook, ). The point here is that we believe that attrition constitutes both online and representational changes, if a distinction between these two types of change is to be drawn at all. It may be that online changes are necessary before representational changes take place, but we would nevertheless consider these online changes to be a necessary component of the attrition process. Moreover, as in so many cases it is not unequivocally clear whether a change represents a surface level or representational change, this line of debate could be seen as the ‘black box’ of attrition research. The nature of the contents of this black box at present eludes our full understanding and analysis, but that there is something inside it is beyond doubt. The contributions in the present section examine what is inside the black box—the linguistic manifestations of attrition—and we believe that, whether surface or representational, they reveal changes in how bilingual individuals use, process, and potentially represent their native language, and therefore constitute attrition.
. T
.................................................................................................................................. Like most work in the area of multilingual development, language attrition investigations typically adhere to the classic division of the study of linguistics into different domains, or linguistic levels. The domain of phonetics deals with the production and perception of sounds, while the language specific phoneme inventory is related to the domain of phonology. The domains of morphology and the lexicon deal with the structure and property of words, whilst syntax deals with how words can be combined into sentences and phrases, and the domain of semantics deals with meaning. There are overlaps between these domains and it is possible to break them down into further sub-domains. Nonetheless, traditionally, when examining attrition, research has generally been conducted in relation
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
to these general domains, showing that the native language of bilinguals can indeed become affected by another language at all of these levels. These phenomena of language transfer, interference, or attrition can be observed most straightforwardly in the lexicon. From the earliest phases of language attrition research, it has commonly been suggested that this domain may become restructured and that lexical items may become less accessible, leading to the use of underspecified or highly frequent items (Olshtain & Barzilay, ), delays in naming latencies (Ammerlaan, ) and other phenomena of lexical convergence (Pavlenko, ). These phenomena have since been described in a range of quantitative, empirical, and psycholinguistic studies, which have also shown how lexical productivity and sophistication in free speech can be affected by attrition, with a concomitant increase in disfluencies (e.g., Bergmann, Sprenger, & Schmid, ; Schmid & Jarvis, ; Yılmaz & Schmid, ). While these studies have typically focused on changes in immersed populations years or decades into the attrition process, many recent studies also find subtle changes related to lexical access at much earlier stages of L acquisition, including in instructed populations with no experience of L immersion. For example, Gollan et al. () have shown that L learners are not only outperformed by monolinguals on tasks relating to lexical access and retrieval in their L, but that this effect is modulated by word frequency (the ‘weaker links hypothesis’). Similar findings have shown that L word naming in early L learners is affected differently by phonological neighbourhood density and spelling-sound regularities than in monolinguals (Bramer et al. ). Taken together, these findings imply that the pathways of lexical access are affected by the mere presence of a second language. Lexical attrition is therefore not, or at least not only, a process whereby items become unavailable due to underuse (as has often been hypothesized), but the paths and mechanisms which facilitate word retrieval may become less effective due to the co-activation of a second language. The second linguistic domain which became the focus of investigations of language attrition at a relatively early stage is that of grammar. Research in this area was initially mainly situated within the framework of generative grammar, focusing on issues such as the potential reparametrization of a native grammar under the influence of an L and issues of markedness (e.g., Sharwood Smith & Van Buren, ; see also Gürel, Chapter , this volume). With the advent of the Minimalist framework, attention turned to the distinction between core grammar on the one hand (which has often assumed to be invulnerable to attrition processes; but see Iverson, ) and interface phenomena, on the other (see Chamorro & Sorace, Chapter , this volume). Later approaches to the attrition of grammar and morpho-syntax adopted more processing-based models, such as the -M model proposed by Myers-Scotton & Jake (e.g., ) and based on Levelt’s () Speaking Model (see Schmitt, Chapter , this volume) or models based on the parallel activation of languages (see Dussias et al., Chapter , this volume). While thus adopting a variety of explanatory approaches, based on their underlying assumptions about the architecture of the human capacity for language (see also the contributions in Part I of this volume), these studies underscore the original assumption that morpho-syntactic attrition is selective, with different parts of the repertoire being differentially vulnerable (see Schmid & Köpke, ). In addition, two recent studies of morpho-syntactic attrition, using the EEG paradigm (see Steinhauer & Kasparian, Chapter , this volume), seem to suggest that, while attriters may not differ markedly on those EEG components which they share with monolingual controls for violations of certain structures, they may develop additional responses, over and above the ones exhibited by monolinguals,
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
suggesting different overall processing strategies which may be indicative of somewhat more effortful processing among attriters (e.g., Kasparian & Steinhauer, a, find a P among attriters, but not among controls, in their study of confusable words; while Bergmann, Meulman, Stowe, Sprenger, & Schmid, , report that their attriters exhibit an N not shown by the controls, in addition to a native-like P, in response to verb finiteness violations). Investigations of phonetics and, in particular, phonology represent a relatively late arrival within the otherwise well-established field of language attrition studies. With some early exceptions of studies investigating the cross-linguistic adaptation of VOT in immersed bilinguals (Flege, ; Major, ), research in these areas did not become firmly established until the second half of the first decade of the s (see de Leeuw, Chapter , this volume). Even then, studies initially tended to rely on broad and global measurements, such as foreign accent ratings (e.g., de Leeuw et al., ; Hopp & Schmid, ) and only later on turned to more fine-grained segmental and prosodic measurements (e.g., de Leeuw, Mennen, & Scobbie, , ; Mayr et al., ). Recent developments in this field, however, are rapidly closing the knowledge gap between phonetics/phonology and other aspects of language attrition, as is evident in a recent Special Issue of the Journal of Phonetics on ‘Plasticity of native language phonetic and phonological domains in the context of bilingualism’. The findings presented in this collection lend support to the idea that bilinguals, through the very act of being bilingual, undergo bidirectional interaction in their languages which is not delimited by a critical period (de Leeuw & Celata, ). Moreover, it appears that the effects of L acquisition on the native speech of first generation immigrants is magnified in subsequent generations with regard to attrition of both segments (Nodari et al., ) and lexical tones (Kan & Schmid, ). Early exposure might be a necessary condition for the acquisition of certain language-specific speech patterns, but continued extensive exposure to the language is needed in order to maintain speech patterns (Mayr, López-Bueno, & Tomé Lourido, ). Although sustained plasticity of native language phonetic and phonological domains is apparent, variation in the extent of plasticity is also observed. Phonetic and phonological changes in the native language, potentially induced by an L, are suggested to be dependent on factors such as L and L input as well as L proficiency levels, and to proceed in tandem with biological changes across the lifespan, as observed in a longitudinal case study of Stefanie Graf over four decades (de Leeuw, ). The contributions to the special issue also provide evidence confirming previous findings on the rapid cross-linguistic adaptation of some phonemes (Chang, ), and open up a new area of linguistic investigation for attrition studies, namely phonotactic restructuring: a preliminary cue-weighting perception study into Russian expatriates in the United States (Russian-English bilinguals) revealed that these participants relied on vowel duration as a cue to consonant voicing in Russian significantly more than Russian monolinguals and also produced vowel duration distinctions not found in monolingual Russian (Dmitrieva, in review). While vowel duration is an important cue to voicing in English, in monolingual Russian it is virtually absent (given that, in this language, all word-final consonants are devoiced). A further perception study examines phonotactic restructuring in L Brazilian Portuguese (BP) speakers who had been living in the US for almost a year (Cabrelli Amaro et al., ).
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Participants completed a vowel classification task and an ABX task with nonce stimuli as BP speakers tend to perceive and produce an illusory /i/ between illicit consonant clusters. The study finds an increase in accuracy on the perception task (that is, a reduction of illusory vowel perception) in both languages among the L learners of English, suggesting phonotactic restructuring of their native Brazilian Portuguese as a result of exposure to English.
. K
.................................................................................................................................. The findings summarized above demonstrate that language attrition research has made great strides and gained a large amount of knowledge in the thirty-five years since Lambert and Freed’s seminal publication, The Loss of Language Skills (). However, the insights that have become available remain limited in that, with few exceptions, investigations tend to be confined to: • a single linguistic level • a single language pairing • a single language (i.e., the L—this is mirrored in studies of L acquisition by the neglect of potential changes to the native language). While there are thus solid insights into domain-internal attritional processes, we know very little about how different linguistic domains may be affected by attrition within the same individual, that is, how attrition may unfold as a holistic process across different domains. There is, for example, the possibility that attrition at one linguistic level may have ramifications for others (see de Leeuw, Chapter , this volume), that different language combinations may provide strikingly different outcomes, or that some of the phenomena found in attrition may be relatable to concomitant changes in the L, or vice versa (see Schmid & Yılmaz, ). In addition, there is currently a substantial research gap where linguistic properties outside of these traditional domains of lexicon, phonology, and morpho-syntax are concerned—for example, to date there is little to no research that we are aware of investigating language attrition with respect to pragmatic conventions of politeness and formality (but see, e.g., Montoya-Abat, ; and Passoni et al., ). We thus propose that the future research agenda for language attrition studies should include the following questions: . Are some linguistic domains more stable than others? For example, is pronunciation (phonetics) more susceptible to attrition than grammar (phonology, morpho-syntax), or is it true that the lexicon is the ‘most vulnerable’ linguistic domain, as has often been claimed (see, e.g., Schmid & Jarvis, ; see also de Bot, Concluding remarks, this volume, on the difficulty of making meaningful comparisons between attrition effects at different linguistic levels)? . Is there a ‘domino’ effect in attrition? For instance, does phonetic attrition always precede phonological attrition—that is, would it only be possible to observe
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
phonological attrition when phonetic attrition has already occurred, or only morphosyntactic attrition when lexical attrition has already occurred? . Do predictor variables affect different linguistic domains to various extents? It could be the case, for example, that language use strongly affects the phonetic domain whereas age of L acquisition affects morphology and syntax more strongly. . To what extent are more general processes of cognition interconnected with attritional processes? Recent research has drawn attention to executive function in bilinguals. Are, for example, general cognitive processes of inhibition, activation, and task selection intertwined with the processes of attrition? We believe that these questions will move forward research in the field of attrition more specifically, and also help us to understand the human faculty of language more generally.
. C
.................................................................................................................................. In Chapter , Chang discusses phonetic drift, which he defines as short-term L-influenced phonetic change in an individual’s L system, attributable to recent L experience. Accordingly, Chang differentiates drift from attrition and proposes that drift may be viewed as the beginning of a trajectory of L-influenced change in the L that, if not reversed, leads either to attrition or to incomplete acquisition. As an example of drift, he points to the studies by Chang () and Guion () in which overall vowel raising, that is a decrease of F frequency, was the main finding, whereas in Mayr et al. () the main finding was a lowering—that is, an increase in F frequency. Chang refers to such vowel space changes as systematic phonetic drift, thereby highlighting the idea that phonetic drift does not occur sporadically. In Chapter , de Leeuw makes a similar distinction between short-term and longer-term phenomena. She considers phonetic attrition to encompass long-term changes in the native language pronunciation of an individual as a result of the acquisition of a new dialect or language acquired post adolescence. This definition focuses on examining plasticity in adult speech—that is, the implication of phonetic drift in child bilinguals would be less than that of phonetic attrition in adults, as similarly argued in de Leeuw, . De Leeuw draws primarily on the Speech Learning Model (Flege, ) and discusses studies which reveal dissimilation and assimilation in segmental and prosodic aspects of speech, as well as in overall changes to an individual’s native accent. She emphasizes findings which reveal interpersonal differences between adult participants (de Leeuw, Schmid, & Mennen, ; de Leeuw, Mennen, & Scobbie , ; de Leeuw, Tusha, & Schmid, ), suggesting that phonetic attrition is not a uniform process and that careful formal speech with a high information content appears to be less likely to exhibit signs of phonetic attrition than more casual speech styles (Shockey, ; Major, ). Such intrapersonal differences according to style and setting might indicate that drift and attrition are interconnected processes which influence each other throughout the lifespan—that is, short-term as well as long-term changes.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
In Chapter , this volume, Celata builds on the previous chapters, postulating that phonological knowledge is dependent on the way speakers experience and use particular phonetic patterns. She defines phonological knowledge as ‘the way humans group different pieces of phonetic knowledge together in a coherent system of relationships that is practical or useful for communication (in the sense of both speech planning and speech decoding)’ and suggests that phonological changes in the form of attrition may include contrast blurring as well as boosting the distinctiveness among system elements as a reaction to contact with a different phonological system. Her chapter delves into both speech production and perception and discusses how attrition in the first generation may impact subsequent generations. She also suggests that socio-stylistic variation should be further investigated in research into phonological attrition, as well as cognate effects in the speech production of bilinguals, heritage language speakers and attriters, to enlarge the scope of available models of cross-language phonological interference. Together, these three chapters thus suggest that, in the phonetic and phonological domain, drift, and attrition are connected, and that changes potentially unfold in the order of phonetic drift > phonetic attrition > phonological attrition. Moving forward, in Chapter , Schmitt defines morphological attrition as changes in inflectional and derivational morphemes and their combinatory possibilities in language production and perception. Interestingly, in her first example she refers to the omission of the genitive feminine suffix in a Russian native speaker, indicating morphological attrition. However, in the same utterance, the speaker implements other morphemes in line with standard Russian grammar, prompting Schmitt to ask why some morphological markers remain resilient to attrition while others are lost more rapidly and easily. Potentially, insight into this question may be gained through examining morphological markers in relation to other linguistic domains, such as their phonological constituents. Schmitt discusses the Markedness Theory, described by Sharwood-Smith () in relation to attrition; the Regression Hypothesis (RH), first formulated by Jakobson (); the Activation Threshold Hypothesis, proposed by Paradis (, ); the Interface Hypothesis (IH; Tsimpli, ; Montrul, ; Sorace, ; see also Chamorro & Sorace, Chapter , this volume); and the -M Model (Myers-Scotton & Jake, ). Through this discussion she demonstrates that there is no single theory that accounts for the complexity and variety of morphological operations and morphemes that undergo restructuring, modification, and ultimately language attrition, whilst pointing to new research which is forging paths (Alexiadou, Anagnostopoulou, & Schäfer, ). In Chapter , Jarvis discusses lexical attrition which he defines as a process that leads to a person’s loss of the ability to retrieve words from memory, adding that lexical attrition is an observable phenomenon which is now widely understood as a collection of undesirable changes to a person’s lexical skills. Jarvis examines attrition in children and in adults in various circumstances, expanding the more restricted definition used by de Leeuw ( and Chapter , this volume). Interestingly, in one particular example, Jarvis mentions that nonce words produced by a Russian adoptee were phonologically compatible with Russian, but that their forms had apparently become distorted in memory (Isurin & Seidel, ). Here again, we see a link between different linguistic domains as the process of attrition unfolds. Moreover, he points out that many studies have confirmed that individuals who have acquired a new language after the acquisition of the first tend to recognize and retrieve L words more slowly than do comparable speakers of the same L whose life circumstances
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
make them less susceptible to attrition (Goral et al., ; Schmid & Jarvis, ; Isurin & Seidel, ). Finally, in Chapter , Gürel discusses attrition research into null and overt pronouns, which span the domains of syntax and pragmatics. In contrast to Chang, Celata, and Jarvis, but in line with de Leeuw (see also de Leeuw, ), she confines her chapter to attrition studies involving late bilinguals, typically residing in an L country as first generation immigrants, who become dominant L users after puberty. Thus, as L grammatical competence is believed to have stabilized in adolescence (see, e.g., Montrul, , a), qualitative changes in the L grammar of post-puberty bilinguals may have far-reaching implications for the alterability of L linguistic competence (Gürel, , a). Gürel emphasizes that although in some null-subject languages there are certain null pronounlicensing morpho-syntactic mechanisms, the use of null and overt pronouns is not random but subject to discourse-pragmatic constraints. For example, a null pronoun in Italian is sometimes preferred if this sentence is uttered as a response to a question, whilst topic shift requires overt pronouns. She draws on the IH (Sorace, ; Chamorro & Sorace, Chapter , this volume) which stipulates that features that are at the interface of syntax and other cognitive-linguistic domains will be susceptible to attrition whilst purely syntactic features would pose no considerable problems. As such, the IH predicts how domains interact with one another in the process of attrition.
. C
.................................................................................................................................. As both the considerations presented in this introductory section and the chapters contained within the section demonstrate, language attrition research has come a long way from Haugen’s mainly qualitative observations made over eighty years ago. In particular, it is becoming ever clearer that changes to the native language in the context of bilingualism are a phenomenon not limited to extreme circumstances of non-use and longterm immersion in an L setting, but that they, to some extent, affect all bilinguals and all linguistic levels, often in ways which are unexpected and surprising. We hope that this summary of the state of the art will help the field progress and make an even larger impact on the mainstream of bilingual research in future.
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
......................................................................................................................
......................................................................................................................
.
. I
.................................................................................................................................. T chapter addresses the phonetic changes that occur in one’s native language (L) due to recent experience in another language (L), a phenomenon referred to as . Before proceeding, it is important to define precisely what I mean by ‘phonetic drift’ in light of various other uses of this term. The earliest published instance of the term ‘phonetic drift’ in the linguistic literature is from Sapir (/), who uses it to discuss contactinduced diachronic sound change: We may suppose that individual variations arising at linguistic borderlands—whether by the unconscious suggestive influence of foreign speech habits or by the actual transfer of foreign sounds into the speech of bilingual individuals—have gradually been incorporated into the phonetic drift of a language. (p. )
Other researchers have also used this term in a diachronic sense, but to refer to languageinternal phonetic shifts in a category distribution (e.g., Solé & Recasens, ; Garrett & Johnson, ). Thus, the term ‘phonetic drift’ is often used to refer to diachronic change in ‘macro’ language (i.e., language at the level of the speech community), which may or may not be due to contact with another language. In this chapter, too, the focus is on change over time; however, I am concerned specifically with those L phonetic changes that follow from language contact (i.e., L experience) within a speaker’s lifetime. Consequently, I will discuss these developments primarily at the level of the individual (i.e., ‘micro’ language, or an idiolect), with a view toward elucidating the individual cognitive mechanisms behind such developments and their relationship to diachronic sound change in a speech community. Therefore, I will use the term ‘phonetic drift’ (or ‘drift’ for short) mainly to refer to L-influenced phonetic change in an individual’s L system. Since most of the research on phonetic drift examines the L production of L users, that is the focus of this chapter; however, it should be noted that phonetic drift is also reported in perception (Tice & Woodley, ; Namjoshi et al., ; cf. Cutler et al., ) and in L non-users (i.e., L speakers with only ambient L exposure).
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
In the context of the following chapters in this section as well as others in this volume, an additional terminological question that arises is whether there is a difference between phonetic drift and phonetic or phonological attrition (see also Chapters and , this volume, on Phonetic Attrition and Phonological Attrition by de Leeuw and Celata, respectively). In short, I will be discussing as cases of drift ostensibly short-term L changes, both in early- and late-onset L learners, which are attributable to recent L experience (e.g., because the change or divergence from L norms coincides with concomitant L exposure), whereas I will reserve the term ‘attrition’ to refer to long-term L changes in late-onset L learners which are unlikely to be due to recent L experience only (e.g., because the change perseveres long after a decline in L exposure). Naturally, this leads to the question of how drift relates to attrition, a question to which I will return in Section .. Studies of phonetic drift have been concerned with a number of different questions, some related to the dynamics of language contact and bilingualism (e.g., Yeni-Komshian et al., ; de Leeuw et al., ; Yao & Chang, ) and others related to the nature of phonetic processing and linguistic representation (e.g., Chang, ; Namjoshi et al., ). In this chapter, I will review findings that relate to three main questions: () What features of the L are subject to phonetic drift? () What is the cognitive mechanism (or mechanisms) behind phonetic drift? () What factors increase or decrease the likelihood of phonetic drift? To preview some of the current answers to these questions, virtually all aspects of L speech are subject to drift, but different aspects do not drift in the same manner, possibly due to multiple routes of L influence coexisting at different levels of L phonological structure. However, because the literature has generally examined L change in terms of sound categories rather than routes of influence, the rest of this chapter is organized by category type: Section . covers drift in consonants (e.g., in voice onset time (VOT)), Section . covers drift in vowels (e.g., in formant frequencies), and Section . covers drift in suprasegmentals (e.g., in fundamental frequency).
. P
.................................................................................................................................. An abundance of acoustic studies have reported phonetic drift in consonants. The majority of this research has focused on oral stops (i.e., plosives)—in particular, the feature of VOT, the primary acoustic correlate of stop voicing contrast. In addition, some studies have reported drift in fundamental frequency (f0), a secondary acoustic correlate of voicing contrast; these are discussed further in Section . since f0 also serves as a primary cue to suprasegmental contrasts (e.g., tone, intonation). Finally, there are a few studies reporting drift in consonants other than stops (namely, fricatives and approximants). The recurring theme throughout these studies is that it is common for L consonants to drift in one or more ways following the onset of L experience, even in adult L learners for whom L categories might be considered to be relatively stable (i.e., ‘fully acquired’). As for the mechanism behind phonetic drift of consonants (as well as vowels), Flege (, , ) articulated an influential proposal based on
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
in a theory called the Speech Learning Model (SLM). According to the SLM, speech sound categories continue to develop over the lifespan; L and L sounds coexist in a shared mental phonetic space; and an L sound that is, at a position-specific allophonic level, ‘similar’ to an L sound tends to undergo a perceptually-based, automatic ‘equivalence classification’ with that L sound. Equivalence classification results in similar L and L sounds becoming linked under the same category and, consequently, converging towards each other phonetically. Since L–L equivalence classification becomes more likely as L categories widen over the course of normal L development, the probability of cross-linguistic perceptual linkage increases with a later onset of L learning. Late L learners are thus particularly subject to L influence on the L, whereas early L learners are more likely to notice the systematic phonetic differences between close L and L sounds and, therefore, to establish a distinct category for the L sound. When L and L sounds are separated categorically in this way, they may develop independently, resulting in native-like realization of both; however, they may also dissimilate from each other to maximize crosslinguistic distance within the shared L–L phonetic space, leading to divergence from native phonetic norms (e.g., Mack, ; Yusa et al., ). Several studies of late L learners have reported results consistent with the SLM view of the cross-linguistic interactions leading to phonetic drift. In one study of L French to L English and L English to L French speakers (Flege, ), the VOT of the voiceless coronal stop /t/, which is short-lag in French but long-lag in American English, showed drift in both L groups for speakers immersed in their L: L English speakers living in France produced English /t/ with shorter-than-native VOT, while L French speakers living in the US produced French /t/ with longer-than-native VOT. Similar results were reported by Major (, ) for L English to L Portuguese speakers living in Brazil. In addition, L Korean to L English speakers living in the US were shown to produce Korean fortis stops, which have slightly shorter VOT norms than English voiced stops (Chang, ), with longer-than-native VOTs that were virtually identical to the VOTs these speakers produced for English voiced stops (Kang & Guion, ). Although the above studies give suggestive evidence of phonetic drift in late L learners, they were not concerned with the extent to which the L phonetic changes persist and, consequently, were based only on cross-sectional comparisons, which have the inherent limitation of assuming that any cross-group differences providing a picture of ‘apparent time’ are due to the independent variable of interest (in this case, L experience) rather than some other factor that has not been brought under experimental control. This limitation makes it crucial to confirm such findings in longitudinal research, and indeed there are a few studies that have documented drift longitudinally within individuals. One example is a case study that followed an L Portuguese late learner of English as she travelled between Brazil and the US (Sancier & Fowler, ). This study showed that the VOT of L (as well as L) voiceless stops drifted toward the VOT norms of the most recently experienced ambient language; thus, the speaker’s VOT in L stops was observed to be significantly longer following a few months of L immersion in the US, an effect perceptible to native listeners. This drift was attributed to three factors: similarity between the L and L sounds (cf. Flege’s SLM), humans’ tendency toward imitation, and the recency effect on memory. The effect of recent (as opposed to temporally distant or cumulative) L experience on L production was further clarified in a series of longitudinal studies on L English late
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
learners of Korean who were recent arrivals to Korea (Chang, , ). This work contributed three novel findings to the literature in this area: () early drift, () modulation of the recency effect, and () multi-level cross-linguistic linkages. First, in contrast to previous studies, which had focused on experienced L learners, Chang () examined inexperienced (i.e., novice) learners, finding that phonetic drift was not only evident, but relatively pronounced, during the first weeks of elementary L instruction. As in Sancier & Fowler (), the VOT of learners’ L voiceless stops drifted toward the longer VOT norms of a similar L stop series (in this case, the Korean aspirated stops); however, the amount of drift was larger than in Sancier & Fowler (). This disparity led to a follow-up study comparing inexperienced to experienced learners (i.e., those with prior exposure to the target L) enrolled in the same language programme (Chang, ). Contrary to the assumption that phonetic drift would be positively correlated with L experience/proficiency (e.g., Major, ), this study revealed that experienced learners showed less drift during the same time period of L learning, as shown in Figure . (which illustrates this effect for VOT as well as f0, discussed further in Section .). This result supports the view that the potential for recent L experience to lead to drift decreases over the course of L acquisition, which may be due to an L novelty effect for novice learners (Chang, ) and/or developmental changes in the relative strength of the L versus L systems ( Jacobs et al., ). As for the third finding of multi-level cross-linguistic linkages, this relates to the theoretical basis for phonetic drift assumed in Chang ()—namely, a cross-linguistic acoustic distance that is perceptible. For an L sound to be ‘similar’ (in SLM terms) to an L sound such that the two become perceptually linked, the L sound should be not so different that it is perceived as ‘new’, yet different enough that it is not perceived as ‘identical’. Thus, for any given acoustic dimension, L sounds are predicted to drift only when the distance from a ‘similar’ L sound is at least as large as the just-noticeable
(b)
VOT (ms)
110
inexperienced experienced
100 90 80 70 1 2 3 4 5 Time(weeks into language program)
F0 onset, standardized (z-scores)
(a)
0.5
inexperienced experienced
0.0
–0.5
1 2 3 4 5 Time(weeks into language program)
.. Phonetic drift of English voiceless plosives in learners of Korean (A) drift in VOT for all learners; (B) drift in onset f0 for female learners. The two learner groups are inexperienced (circles) and experienced (triangles). Error bars show % confidence intervals. Source: Chang (: Figures & ).
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
difference (JND) in that dimension. For VOT of the L (English) voiceless stops in Chang (), it was therefore predicted that /p/ and /k/ would drift, as the respective VOT norms for the similar Korean /ph/ and /kh/ are each longer by an amount that exceeds the JND for VOT in this range; in contrast, /t/ was predicted not to drift, as the VOT norm for Korean /th/ is longer by an amount that does not reach the JND. Results separated by a stop indicated that /t/ did not fail to drift, but rather drifted along with the other voiceless stops; at the same time, /t/ showed less drift than both /p/ and /k/. These findings thus suggest that production of L /t/ was influenced by linkages to the L at two levels: the individual sound and the natural class of sounds. The class-level linkage of L voiceless stops to L aspirated stops precipitated the drift of all voiceless stops (including /t/) due to the perceptibly longer overall VOT norm for the L class, whereas the segment-level linkage of L /t/ to L /th/ moderated the amount of drift for /t/ due to the relatively close VOT norm for L /th/. Given that the examples of drift discussed above were all found in L immersion contexts, a question that arises is whether drift also occurs in non-immersion contexts. Findings from a few studies examining L English late learners of Spanish in the US suggest that this is the case. In one study of advanced learners who used the L regularly in their professions (Lord, ), drift was evident in the VOTs of L voiceless stops, which were shorter-than-native for all stops but significantly so only for /k/. Further evidence along these lines was reported in a larger cross-sectional study of learners exemplifying a wide range of L proficiency levels from elementary to master’s level (Herd et al., ); the learners in this study, the majority of whom had never been immersed in the L for more than a week, produced VOTs for L voiced stops that were more negative (thus, more L-like) the higher their level of study, with master’s students consistently showing long negative VOTs. The cross-sectional results in these two studies converge with the results of a longitudinal study that tracked, and phonetically trained, five learners in an elementary Spanish course (Schuhmann & Huffman, ). Some (but not all) of these learners showed downward drift in VOT (i.e., VOT shortening) of L voiceless stops and/or a tendency toward downward drift in VOT of L voiced stops, suggesting again that phonetic drift is not dependent on an L immersion environment. To close this section, it should be mentioned that, although most findings of drift in consonants come from studies of stops, other types of consonants are clearly subject to change as well. For example, in a cross-sectional study of fricatives in L Taiwanese Amoy learners of Mandarin Chinese in the US (Peng, ), proficient, but not unproficient, Mandarin speakers showed movement of their L fricative /h/ toward the similar L fricative /x/ with respect to frequency range and spectral energy. Recent cross-sectional studies of L German late learners of English living in Canada (de Leeuw, Mennen, & Scobbie, ) and the UK (Ulbrich & Ordin, ) provided further evidence of L change in the formant properties (i.e., F1, F2, F3) of lateral and rhotic approximants, due to L–L differences in velarization and rhoticity. Note that the L changes in these studies were examined only after a period of extensive L experience, so it is unclear whether they reflect drift per se, as opposed to attrition (see Section .). Nevertheless, findings of L-influenced change in fricatives and approximants as well as stops suggest that phonetic drift is a wide-ranging phenomenon that may affect a variety of L consonants.
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
. P
.................................................................................................................................. Like consonants, vowels also drift due to L experience, and in accordance with the SLM, this drift has often been explained in terms of cross-linguistic influence at an allophonic (i.e., segmental) level. However, the literature on vowels is similar to the literature on consonants in showing L influence at a higher (in particular, global or systemic) level as well. Segment-level and system-level effects of an L are not necessarily mutually exclusive, although there is some indication that the systemic influence of an L inventory may override the segmental influence of one L vowel. In this section, we consider both sources of vocalic drift as well as their interaction. A few cross-sectional studies have reported drift of selected L vowels due to the influence of a similar L vowel. In one of the earliest studies of this type, Flege’s () study of L French to L English and L English to L French speakers, late L learners were tested on their production of the high back rounded vowel /u/, which shows a significantly higher norm for F2 in American English than in French. Since drift was expected to involve movement toward the phonetic norms of a similar L vowel, L French to L English speakers were predicted to produce French /u/ with higher-than-native F2 values, and L English to L French speakers to produce English /u/ with lower-than-native F2 values. Indeed, the L French speakers showed the predicted drift pattern; however, the L English speakers did not, producing their L vowel close to native norms. These results were replicated in a later study of L Mandarin to L English and L English to L Mandarin speakers (Chang et al., ). Although the reason for this asymmetry in drift remains unclear (see Chang, , pp. –, for one possible explanation based in properties of human audition), what the asymmetry suggests is that drift cannot be predicted solely on the basis of cross-linguistic differences between similar sounds; rather, there are other intervening factors (which we discuss further in Section .). In addition to the data from these cross-sectional studies, further evidence of vocalic drift has come from a number of recent longitudinal studies. In one study of L Japanese learners of English taking up residence in an L environment, a year of L immersion was found to result in drift of some L vowels (specifically, /i a/) in terms of increased F2 values, but only for child learners and not for adult learners (Oh et al., ).1 On the other hand, research on adult L French speakers receiving phonetic training has shown clear effects of concentrated L exposure on L vowels (Kartushina, ; Kartushina et al., ). In this work, monolingual French speakers underwent an hour of production training on each of two target vowels, Danish /ɔ/ (which is perceived as similar to, as well as a good exemplar of, the L /o/) and Russian /ɨ/ (which is not perceptually similar to any L vowel, but acoustically closest to /ø y i/, in that order). Examination of pre- and post-training production of these L vowels revealed no drift of /o/ or /i/, but significant drift of /ø/ and /y/; furthermore, the L training effect was stronger for /ø/ than for /y/. These results thus suggest, in line with the SLM and Chang (), that drift is dependent on the disparity between L and L 1
Note that the inclusion of child learners in the participant sample crucially distinguishes this study from studies of attrition, which typically focus on adult learners whose L acquisition starts clearly after a critical/sensitive period.
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
3
F1 (Bark)
4
late bilingual early bilingual
i i
ʊ ʊ
5
a
6
a
7 14
13
12
11
10
9
F2 (Bark)
.. L Quichua vowels (/ɪ ʊ a/) according to timing of L onset Note: The late and early L learner groups are plotted in grey and black, respectively. Source: adapted from Guion (: Figure ).
sounds falling within a ‘sweet spot’ of cross-linguistic distance (i.e., not so small that there is no room to drift, but not so large that the L sound is no longer in the phonetic neighbourhood of the L sound); however, drift is not dependent on perceptual similarity specifically, as even the Russian /ɨ/, which would qualify as a ‘new’ L sound under the SLM, appeared to precipitate drift in this case. Whereas the above studies examined a subset of learners’ L vowels, other studies have looked at the whole L vowel inventory and found evidence for vocalic drift at a systemic level. In one such study of L Quichua to L Spanish speakers (Guion, ), early L learners produced all L vowels with lower F1 values than late L learners; that is, more L experience was linked to a global raising of the L vowel space, as shown in Figure .. Crucially, this pattern of drift could not be explained in terms of the attracting influence of the closest individual L vowel, because the L vowels did not consistently converge toward this vowel. Instead, Guion () accounted for this pattern in terms of maximizing dispersion with respect to the L vowel system, which contains two mid vowels /e o/ that could put pressure on the high vowels of Quichua to go higher. However, a different mechanism for this type of systemic drift was proposed in the longitudinal work of Chang (, , ) on L English late learners of Korean. Besides showing drift in VOT (as discussed in Section .), these studies showed, for female learners, a consistent raising of the L vowels, much as in Guion (). Unlike the case of L Quichua to L Spanish, however, the systemic drift observed in Chang () could not be explained in terms of cross-linguistic dissimilation, as the drift resulted in a lower degree of cross-linguistic dispersion. It could also not be explained in terms of separate individual vowel shifts, as the L vowels did not uniformly assimilate toward or dissimilate from the closest L vowel. Instead, the drift was attributed to convergence toward the L via a cross-linguistic linkage between the global formant levels associated with the L and L vowel systems. Since the L (Korean) inventory contains fewer low and mid vowels (i.e., those with relatively high F1 values) than the L (English) inventory, the average F1
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
of the L vowel system is lower than that of the L system, resulting in phonetic drift taking the form of a decrease in F1 (i.e., vowel raising). This systemic raising of L vowels might also be explained in articulatory terms, as the result of cross-linguistic differences in articulatory setting (see, e.g., Gick et al., ). To be specific, the default (i.e., rest) position for the tongue in a given language might be relatively higher when the language’s phonological inventory leads to more production of high vowels on average; such cross-linguistic differences in default tongue height could explain the systemic raising observed in Chang () in terms of a lower default tongue position for the L being pulled toward the higher default tongue position for the L. Crucially, however, the articulatory account predicts the same direction of vocalic drift for female and male learners (because the L–L inventory differences are the same regardless of talker sex), yet this is not borne out by the male learners examined in Chang (, ). These male learners were, instead, found to drift in the opposite direction of female learners, which was attributed to the fact that, for male learners, the average F1 of the L vowel system (modelled by L instructors who were female) was higher, rather than lower, than that of their L system. Although the global features of L vowels that learners acquire, and drift toward in their L, remain to be fully specified, what these results suggest is that these features are probably not articulatory, but acoustic in nature. Despite involving quite different sets of L vowels, a point of overlap between Guion () and Chang () is the fact that drift took the form of vowel raising. This leads to the question of whether the raising in these studies was based on L–L comparisons at all or was simply a default pattern of drift. One study that helps to address this question is a cross-sectional investigation of two L Dutch to L English speakers, identical twin sisters who differed in terms of their L experience (Mayr et al., ). While one sister (‘TZ’) remained in the Netherlands, where both sisters grew up, the other sister (‘MZ’) had moved to the UK as a young adult and had thus been immersed in the L for many years. Acoustic analyses showed that MZ produced most (seven out of nine) of the L monophthongal vowels with higher F1 values (i.e., lower in the vowel space) than TZ did. This vowel lowering pattern was consistent with L–L differences in average F1, which, in this case, was higher for the L system than the L system (Mayr et al., , p. ). Additionally, it supports the view that systemic drift of vowels does not necessarily involve raising, but instead depends on the nature of acoustic differences between the L and L. The findings of recent research on L Polish to L English and L Polish to L German speakers provide additional evidence that systemic drift may be realized in a variety of ways (Sypiańska, ). The role of system-level differences in influencing vocalic drift was further highlighted in a longitudinal study of L English late learners of French (Lang & Davidson, ). This study tracked novice learners during a study-abroad programme in Paris and also compared these novice learners to advanced learners who were long-term Paris residents. Whereas the advanced learners showed evidence of vocalic drift, particularly in F1, the novice learners did not, in contrast to the results of Chang (, ). This apparently delayed onset of L effects from French vis-à-vis Korean suggests that the progression of vocalic drift may be influenced by specific properties (e.g., size, crowdedness) of the L and L vowel systems that come into contact in L learners. The last study to be discussed in this section relates to the effect of phonological context and the lexical (i.e., whole word) level. The relevance of these factors for vocalic drift
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
was shown in a study of L Shanghainese to L Mandarin speakers in Shanghai, which examined how production of Shanghainese /ɛ/ has been influenced by the similar Mandarin vowel /ej/ (Yao & Chang, ). This study tested two groups of speakers (younger and older), two language modes (mono- and bilingual), and three types of L lexical items cognate with L lexical items (one type whose phonological form was maximally similar to that of the L cognate, and two types that were each less similar). The crucial result has to do with item type: although all three types of L item contained /ɛ/ and were cognate with an L item containing /ej/, more drift of /ɛ/ was observed in the first type than in the other two, suggesting that L influence from /ej/ was modulated by the degree of similarity between the lexical contexts in which the vowels were embedded. That is, vocalic drift was influenced by L–L linkage not only at the level of the segment (i.e., /ɛ/ and /ej/), but also at the level of the lexical item. This case of drift is thus similar to the case of English /t/ in Chang () in evincing multiple layers of cross-linguistic comparison affecting phonetic drift.
. P
.................................................................................................................................. Although the literature on drift in suprasegmentals is not as extensive as the literature on drift in segmentals, there are now a number of studies demonstrating that suprasegmental features of the L, much like segmental features, are influenced by L experience. This literature contains both cross-sectional and longitudinal data and mostly addresses drift related to pitch (namely, f0 level and alignment). The wider literature on contact varieties suggests that rhythm, too, may be influenced by bilingualism (e.g., Low et al., ; Carter, ), but there are as yet no systematic investigations of drift in L rhythm during lateonset L learning. Drift in f0 level has been documented in a few studies on Korean and English in contact. For example, a cross-sectional study of L Korean to L English speakers found that L experience was correlated with increased onset f0 values after L lenis stops (Yoon, ). In the longitudinal research on L English late learners of Korean discussed in Section . (Chang, , ), drift was found both in the VOT of voiceless stops and in the onset f0 of the vowel following these stops. There was an effect of gender on drift in f0, which arose only for female learners. Further, there was an effect of prior L experience, as drift was greater for inexperienced than experienced learners (Figure .B); nevertheless, drift in f0 was evident in both groups, as well as for both voiced and voiceless stops. In fact, upward drift in f0 was expected after both L stop types due to the pitch accent associated with the perceptually most similar L stop types (i.e., fortis, aspirated). Unexpectedly, however, f0 also drifted upward in the absence of any stops—namely, in vowel-initial items (see Figure .). This result was explained in terms of a control mechanism for f0 that is, to some extent, shared across languages (and, thus, influenced by global f0 properties of an L including overall f0 level). At the same time, f0 increased more in stop-initial than vowelinitial items, suggesting that the f0 properties associated with the fortis and aspirated stops specifically were also at play. Thus, similar to the case of VOT, drift in f0 was influenced by
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
240
D voiced stops
230
A vowel-initials
F0 onset (Hz)
T voiceless stops
T
220 210 200
T
D A
T D A
T D A
T D A
A D
190 1 2 3 4 5 Time (weeks into language program)
.. Phonetic drift of onset f0 in English words produced by female learners of Korean, by initial segment type. Note: Error bars show % confidence intervals. Source: Chang (: Figure ).
a multifaceted relationship between the L and L, here involving overall f0 level (i.e., a global property) as well as the natural class-level properties of specific stop types. In addition to f0 level, change may also occur in f0 alignment. This type of change was found in a cross-sectional study by Mennen (), which examined the temporal alignment of the f0 peak in an intonation contour as realized by L Dutch late learners of Greek and native speakers of Dutch and of Greek who did not know the other language (i.e., control groups). The Dutch learners and controls were living in the Netherlands or in Scotland, while the Greek controls were living in Scotland. They were compared on their production of a prenuclear rise (LH*), which is present in both languages but peaks earlier and is subject to different phonological conditioning (by vowel length) in Dutch than in Greek. Most Dutch learners of Greek produced the rise in Dutch differently than the Dutch controls, in addition to producing the rise in Greek differently than the Greek controls. On the other hand, one learner produced the rise in both Dutch and Greek like the respective control groups. Taken together, the findings of this study, as well as of a later study of L German to L English speakers (de Leeuw et al., ), converge with findings on segmental properties (Sections . to .) and overall accent (Yeni-Komshian et al., ; de Leeuw et al., ) in showing that L experience often, but not always, leads to L changes that diverge from native phonetic norms. Importantly, however, the intonational changes in Mennen () cannot be explained in terms of the same type of L–L linkage (and phonetic convergence) posited in the SLM for segments. This is because positing convergence between L and L tonal phones would predict that the relatively late peak alignment pattern of the L should result in later L peak alignment, whereas this is not the pattern observed. Instead, the Dutch learners of Greek tended to produce earlier L peak alignment than the Dutch controls in short vowels (where the peak typically occurs late, although not as late as in Greek), while producing similar L peak alignment as the Dutch controls in long vowels (where the peak typically occurs early). Thus, in this case L influence did not take the form of assimilation of L-specific phonetic implementation; rather, it appeared in a smaller differentiation of L long and
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
short vowels in terms of peak alignment (i.e., attenuation of an L-specific contrast). These findings are consistent with the view that drift of certain suprasegmental features may operate differently than drift of segmental features, although much more research on suprasegmental changes (including targeted studies of drift per se as opposed to divergences that may reflect attrition) is needed to shore up any claim about segmentalsuprasegmental disparities in L change.
. D
.................................................................................................................................. To return to Questions () to () from Section ., recent experience with an L can affect many features of the L, but the specific effect varies across features as well as other dimensions such as the developmental profile of the L. The literature on L learners contains findings of L change in the VOT of stops, spectral characteristics of fricatives, approximants, and vowels (e.g., F1, F2, F3), f0 level, and f0 alignment. One route of L influence on the L is through equivalence classification of an L sound with an L sound and their resulting perceptual linkage, as proposed in the SLM. However, this type of L–L linkage only predicts phonetic drift at a segmental level, whereas drift has also been observed at other levels: the natural class of sounds, the vowel system, and overall f0 level. This multifaceted nature of drift implies that L knowledge may be linked back to the L in multiple ways, which are often seen to jointly influence L production. As for the factors affecting the likelihood of phonetic drift, (i.e., Question ()), these include linguistic, cognitive, and socio-demographic variables, some of which are related to the individual differences in drift observed in the literature. First, the specific L feature at issue is relevant, as some features may be inherently more grounded in somatosensory feedback (see, e.g., Chang, ), decreasing the likelihood of drift, or include languagegeneral aspects of representation or control, increasing the likelihood of drift. This crossfeature variation is seen, for example, in differences between drift in vowel quality (Section .) and drift in f0 (Section .). Furthermore, the acoustic distance between an L structure and a close L structure is crucial, as it needs to be ‘just right’: large enough to be perceptible and/or provide room for the L structure to drift, but small enough for the L structure to qualify as an attractor (Flege, ; Chang, ; Kartushina et al., ). Prior L experience is also relevant, as it diminishes the potential influence of recent L exposure, and L categories that are inherently less variable (i.e., more stable in their realization) may generally be more resistant to drift (Chang, ; see also Tobin, ). Apart from variation in prior L experience, factors contributing to individual differences in L acquisition are also likely to be correlated with at least some of the individual differences that have been observed in phonetic drift (Huffman & Schuhmann, a,b; Schuhmann & Huffman, ; Schwartz et al., ). These include cognitive, affective, and attitudinal variables such as executive control, affinity for the culture and/or people associated with the L, and motivation (for an overview, see Dörnyei & Skehan, ). Although individual differences in drift remain largely unaccounted for, recent work relating variation in drift to variation in inhibitory abilities (Lev-Ari & Peperkamp, ) presents a promising first step toward a better understanding of the source of these individual differences in L malleability.
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
Given the surface similarity between drift and attrition, one may ask how a particular instance of L phonetic change can be classified as one versus the other. In Section ., we distinguished drift from attrition on the basis of perseverance as well as underlying cause, with attrition lasting in the absence of a proximal L stimulus (i.e., recent L experience) and drift not. Few studies, however, have actually tracked L learners through alternations in language environment, with the result that it is often unclear whether observed L changes are short-term, long-term, or medium-term (i.e., reversible, but not quickly or easily) and, more problematically, impossible to tease apart the effect of L exposure from the role of L acquisition, L knowledge, or L use. Thus, the exact source of phonetic drift, within the broad construct of recent L experience, remains largely unaddressed, raising a number of interesting questions. Can phonetic drift arise, for instance, solely through ambient L exposure, without L knowledge or use? Very little research has directly addressed this question, but one study of monolingual French speakers (Caramazza & Yeni-Komshian, ) suggests that the answer may be yes. In this study, French speakers in Canada produced less prevoicing (i.e., less negative VOTs) for voiced stops and longer positive VOTs for voiceless stops than French speakers in France, indicating convergence toward the norms of Canadian English despite not actually speaking English. Note, however, that these results were not replicated in a later study (Fowler et al., ), which may be due in part to a change in the relative sociolinguistic status of French vis-à-vis English in Canada during the time between the two studies. Complementing Caramazza & Yeni-Komshian (), results from a follow-up study to Chang (, ) further suggest that some aspects of drift are due primarily to L exposure (and, therefore, persist as long as L learners are in an L environment) whereas others are tied to L use (Chang, , ). In this study, a subset of the L learners examined in Chang (, ) were examined again at the end of a year in the L environment. An early drop-off in L use was found to lead to a reversal of drift in VOT and F1, but not in f0 or F2, suggesting that once an L has been acquired to a certain level, it is difficult for phonetic drift to subside completely in an L environment. When L learners return to an L environment, it may be the case that many of the L changes observed in an L environment are eventually reversed, and certain cross-sectional comparisons are consistent with this hypothesis. For example, the inexperienced and experienced learner groups in Chang (), both arriving directly from an L environment, were similar in their L production at the start of the study, suggesting that the L (re)immersion that followed the experienced learners’ previous L learning largely erased any drift that may have occurred during the previous L experience. It bears repeating, however, that no longitudinal research has provided direct evidence of what happens to the L phonetically when L learners disengage from the L and return to an L environment. Despite the relative lack of evidence on the effects of L reimmersion, some researchers (e.g., Stolberg & Münch, ) have argued that L-influenced changes in an L should not be defined in terms of their perseverance, because it may be the case that all behavioural changes resembling attrition are reversible once robust contact with the target language is renewed. Under this view, there is no principled reason (such as a distinction between ‘competence’, or knowledge, and ‘performance’, including factors such as language activation and accessibility) for distinguishing between long-term and short-term changes to the L system. In contrast to this view, I consider the null hypothesis, at this point in our
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
understanding of L development over the lifespan, to be that some aspects of L change do not reverse themselves even after renewed L contact. In other words, regardless of language environment, L learners may continue patterning differently from monolingual L users in certain ways, a view supported by the findings of bilingual-monolingual differences throughout the psycholinguistic literature (for a review, see Grosjean & Li, ). In closing, the conceptual proximity of drift and attrition invites the essential question of how phonetic drift relates to phonetic and phonological attrition. I consider phonological attrition to comprise long-term changes that cross a phonemic boundary; this includes, for example, change in the application of L phonological rules that add, delete, or switch segments from one category to another (e.g., Dmitrieva et al., ; Joh et al., ; Cho & Lee, ). Phonetic attrition, by contrast, comprises long-term changes that are subphonemic, while phonetic drift comprises short-term changes of this type. Recall that the crux of the distinction between drift and attrition is in the timeline and, by implication, the source of change: whereas drift is a short-term effect attributable to recent L experience, attrition is a long-term effect arising from extensive, and not necessarily recent, L contact. Drift may thus be viewed as the beginning of a trajectory of L-influenced change in the L that, if not reversed, leads ultimately to attrition (for late L learners) or to ‘incomplete acquisition’ (for early L learners; see Montrul, ). As suggested in Chang (), one factor that seems to reduce the magnitude of drift is alternation between periods of L and L immersion, which gives the L system the opportunity to retune to L norms; this suggests that the scenario of continuous L immersion is particularly conducive to progressing from drift to attrition. In this regard, it should be noted that the literature on L change is, in fact, dominated by cases of ongoing L immersion, which is why many of the findings reviewed in this chapter are inherently ambiguous between exemplifying drift or attrition. Looking forward, there are many open questions regarding phonetic drift, but two overarching questions are particularly worth highlighting here. First, how exactly does L exposure, as opposed to L use, lead to phonetic drift as well as contact-induced diachronic sound change? In the case of Caramazza & Yeni-Komshian (), for example, it is not clear whether the divergence from L norms in monolinguals with ambient L exposure is a direct L effect (i.e., phonetic drift arising from ambient L exposure) or an indirect L effect (i.e., faithful acquisition of different L norms from the drifted or attrited production of L-speaking parents). Second, how does drift of the L in L learners resemble and/or differ from drift of the L and/or L in third-language (L) learners? The burgeoning field of L/Ln acquisition (see, e.g., Cabrelli Amaro, ) is beginning to address this type of question and represents an exciting area of future research into the dynamics of language development and interaction in an increasingly multilingual world.
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
......................................................................................................................
......................................................................................................................
. I
.................................................................................................................................. A, it is not uncommon for people to know friends or family members who moved abroad, and return home with speech differently accented in comparison to before their departure. Or, perhaps, one has experienced such changes in pronunciation oneself. Indeed, there is consistency among the research discussed in this chapter that the native phonetic domain is malleable upon competition from a new language or dialect in adulthood, although the extent of such effects differs between individuals (Shockey, ; Flege, ; Major, ; Munro et al., ; Mennen, ; Evans & Iverson, ; de Leeuw, , ; de Leeuw, Schmid, & Mennen, ; de Leeuw, Mennen, & Scobbie, , ; Mayr et al., ; Hopp & Schmid, ; Ulbrich & Ordin, ; Bergmann et al., ; de Leeuw, Tusha, et al., ). Research into phonetic attrition (here, attrition is used solely with reference to the L) is important because native language phones are the building blocks from which all other language representations are built (i.e., in L acquisition, children must firstly perceive sounds without an underlying category, before distinctive phonemes are acquired, words built, and sentences constructed, e.g. Maye, Werker, & Gerken, ; Polka & Werker, ; Stager & Werker, ; Werker & Lalonde, , ). Phonetic attrition is therefore indicative of an unbalancing of the native language system, with potential trickle down effects evidenced in phonological attrition, see Chapter , this volume, on phonological attrition by Celata. Moreover, findings suggest that with training the effects of a new competing linguistic system on the native language phonetic system can occur rapidly (Chang, ), and that short-term modifications to the phonetic system occur dependent upon the ambient linguistic system (Sancier & Fowler, ), see Chapter , this volume, on phonetic drift by Chang. For these reasons, research into phonetic attrition is of great consequence to our understanding of the plasticity of native speech, how languages and dialects interact and are represented in the mind across the lifespan, as well as to how individuals who possess multiple languages and dialects are perceived by other native speakers.
.. Definition of attrition Research into phonetic attrition usually aims to describe changes in pronunciation which occur within an individual’s L when an L is acquired in adulthood, as well as to explain
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
why such changes occur. Within the context of bilingualism, attrition is most often investigated within the scope of long-term immigration (Schmid & Köpke, ). In this chapter, the characterization of attrition is broadened to include studies that have examined pronunciation changes in the native language upon acquisition of a new dialect in adulthood. The inclusion of dialects into a definition of attrition is relevant when one considers that the terms language and dialect represent a continuum and that their ‘edges are extremely ragged and uncertain’ (Haugen, , p. ). Attrition is not the only way in which an individual’s native speech can change post adolescence. Dialects themselves change over time which affects the individual’s speech production (Harrington, ; Harrington et al., a,b). Individuals can style-shift according to variable social contexts (Bell, ; Coupland, ; Eckert & Rickford, ), and, as previously mentioned, short-term rapid changes to the L phonetic system have also been reported upon L training and exposure (Chang, ; Sancier & Fowler, ). However, studies into phonetic attrition are considered to reveal long-term changes in the native language pronunciation of an individual as a result of the acquisition of a new dialect or language acquired post adolescence. Emerging longitudinal research may reveal how such long-term phonetic changes in the form of attrition are affected, or perhaps governed, by such short-term changes, or whether there is a cut-off between the two at all. Together, research which reports phonetic attrition, as described in this chapter, combined with changes to the L phonetic system due to the aforementioned factors, lend support to the understanding that there is an inherent degree of plasticity to the native phonetic system both during childhood (Pallier et al., ; Ventureyra, Pallier, & Yoo, ), as well as post adolescence as it develops and adapts throughout the lifespan (see, e.g., de Leeuw, Opitz, & Lubinska, ).
. A
.................................................................................................................................. This contribution aims to summarize research into phonetic attrition by examining the question of whether the pronunciation of a linguistic system learned in childhood and adolescence can change in adulthood, potentially, in an extreme case, even becoming ‘foreign accented’ (de Leeuw et al., ; Hopp & Schmid, ), that is, being perceived by native speaker listeners to be non-native speech. In doing so, firstly, the term phonetics will be defined as applied in this chapter, and thereby differentiated from phonology (see Celata, Chapter , this volume). Thereafter, the Speech Learning Model (SLM) and its relevance for studies examining phonetic attrition will be discussed, with a view towards assimilation and dissimilation: both processes can occur in phonetic attrition. Subsequently, findings from research into both segmental and suprasegmental phonetic attrition will be summarized, as related to both bilingualism and monolingualism. Research that has examined native speech which is globally regionally or foreign accented will be discussed thereafter. Questions of psycholinguistic, socio-linguistic, and formal relevance are as follows. . Why do some individuals appear to undergo more phonetic attrition than others? . What can the processes of assimilation and dissimilation tell us about the nature of changes to the phonetic system? . Is there a difference between phonetic attrition in a bilingual versus bidialectal context?
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
. W ?
.................................................................................................................................. Phonetics can be defined as the study of the realization of sounds in spoken language. For example, in the word love, phoneticians might examine the actual realization of the lateral sound at the beginning of love—that is, [l]. Notably, the phonetician might investigate possible variability in the production of the lateral sound, for example whether it was realized with a higher pitch in the case of a child, or a lower pitch in the case of an adult. Indeed, there are infinite ways in which individual lateral phones, [l], and indeed all sounds, can be realized. In contrast, phonology can be defined as the study of the organization of the sounds in spoken language. In the example of the word love, a phonologist might be interested in the fact that the lateral phoneme can be replaced by /d/ and that the new word arising is dove. One way of conceptualizing phones and phonemes is through a snowflake metaphor. Phones are like individual snowflakes: each one is different; phonemes are like the general category of snowflake in our mind. Although each individual snowflake is different, we are nonetheless able to perceive all snowflakes to belong to the category of snowflake; and we have a separate category for snowflake than we do for hail stone, than we do for rain drop—the infinite variability is constrained. In terms of attrition, the difference between phonetics and phonology is crucial. Studies which reveal phonetic attrition (Shockey, ; Flege, ; Major, ; Munro et al., ; Mennen, ; Evans & Iverson, ; de Leeuw, Schmid, & Mennen, ; Mayr et al., ; de Leeuw, Mennen, & Scobbie, , ; Hopp & Schmid, ; de Leeuw, ; Ulbrich & Ordin, ; de Leeuw, Tusha, et al., ), indicate changes with regard to how speakers of a particular language realize particular phones in their native language. In contrast, studies which reveal phonological attrition, of which there are far fewer (see, e.g.; Dmitrieva et al., ; and de Leeuw et al., ), reveal representational changes to the phonemic organization of the native language. Socially, both phonetic and phonological changes to the native language or dialect may vastly influence how an individual is perceived by others, and indeed potentially their own self-perception.
. S L M (SLM)
.................................................................................................................................. Although most widely applied in studies of L acquisition, the SLM was identified early on as relevant to studies into phonetic attrition (see de Leeuw, ), as, in addition to predicting how the L will influence the L, the SLM makes claims about how the acquisition of an L impacts the L. Crucial for phonetic attrition, the SLM posits that the phonic elements making up the L sound system and the phonic elements comprising the L system (either newly established categories, or adaptations of L categories) exist in a ‘common phonological space’, and so will mutually influence one another. (Flege, , p. )
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
either through the process of assimilation or dissimilation (Flege, ). With regard to phonetic category assimilation, the SLM posits that assimilation will occur when ‘the L learner continues indefinitely to judge the instances of an L category to be instances of an L category’ (Flege, , p. ). Alternatively, category dissimilation occurs because ‘bilinguals strive to maintain phonetic contrast between all of the elements in their L/L phonetic space, just as monolinguals strive to maintain phonetic contrast among the elements making up their L phonetic space’ (Flege et al., , p. ). However, factors which would explain whether two sounds from the given L and L undergo assimilation or dissimilation, are not given. Moreover, the SLM explicitly constrains itself to segmental elements of speech, rather than prosody, which may also undergo attrition, and is hence also discussed in this chapter. Nevertheless, the bidirectional application of the SLM is extremely relevant to research into phonetic attrition and its simplicity provides a fruitful foundation for studies.
. S
.................................................................................................................................. Evidence of assimilation in phonetic attrition has been revealed in both segmental and prosodic analyses (note again that although the SLM does not directly make predictions with regard to prosody, assimilation of prosodic categories will be discussed hereafter). For example, although the term ‘attrition’ was never used, a seminal study by Flege () showed that in native American-English speakers immersed in a French environment as adults, and in native French speakers immersed in an American-English environment as adults, voice onset time (VOT) of /t/ productions in English and French were ‘merged’ (Flege, , p. ), becoming intermediate to the target language norms (longer VOT values in English than French). Additionally, within the same study it was found that native French speakers, ‘for whom English was clearly their principal language’ (p. ), produced French /u/ with a mean frequency of the second formant (F) that was higher ( Hz) than the value obtained for French monolingual subjects ( Hz), approximating English monolingual values (p. ). Although the results from the vowel analysis were not significant, they suggested that the prolonged acquisition and use of an L phonetic system affected the L phonetic segments. With regard to the contrast between phonetics and phonology, note that this does not constitute the loss of a phonemic contrast in the L categories, but rather a change in the production of the phonetic realization. Interpersonal variation was not specifically examined in this study; however, the standard deviation of the VOT in the French L speakers in Chicago, in their French, overlapped with the French monolinguals’ standard deviation of VOT. Similarly, the standard deviation of the English L speakers in Paris, in their English, approached the standard deviation of the English monolinguals. Such interpersonal variation suggests that not all subjects evidenced assimilation effects in their /t/ and /u/ productions, others may have undergone less segmental attrition or, alternatively, they may have displayed dissimilation effects. Another study (Flege & Hillenbrand, ) investigated a similar group of French native speakers, who had all learned English as adults and were living in an English-speaking
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
environment for an average of just over twelve years. These late sequential bilinguals were all married to native English speakers and, similarly, the VOT of their /t/ in both French and English was examined. Here too, the averaged results of the subjects indicated that in general the VOT of /t/ in their French was substantially longer (and more English-like) than that of monolingual French speakers (Flege & Hillenbrand, ). Moreover, in the same study it was reported that ‘the bilingual French native speakers produced French /u/ with a substantially higher mean F value ( Hz) than previously reported for monolingual French speakers ( Hz)’ (p. )—that is, long-term exposure to English influenced their production of French /u/, causing it to be more English-like, just as it influenced their production of French /t/. The focus of this investigation was again on group differences, rather than on inter- or intrapersonal differences within the speakers who comprised the groups, with the take-away finding that phonetic attrition at the segmental level is exhibited through assimilation effects in sequential bilinguals resident in a prolonged L environment. Building on these group analyses, an influential study of five native American-English speakers who had acquired Brazilian-Portuguese in adulthood, examined such inter- and intrapersonal variation. Major () was the first to explicitly use the term ‘attrition’ in a phonetic context. Even though the subjects in this study reported personal and professional reasons to maintain English in Brazil, all sequential bilinguals, who had been living in Brazil for between twelve and thirty-five years, exhibited some degree of phonetic attrition in their L realization of VOT (it became shorter and therefore more Portuguese-like, i.e., assimilation effects). In general, there was also a correlation between proficiency in the L, measured according to the realization of Portuguese-like VOT, and rate of attrition in English. On average, the lower the VOT (less native-like) in the English casual speech of his subjects, the lower the Portuguese VOT (more native-like). However, this correlation was not displayed in formal English speech (elicited through word and sentence lists, whereas casual speech was taken from informal spontaneous conversations). Moreover, one subject performed outside of the monolingual norms in both English and Portuguese, whilst another performed within the monolingual norms of both languages, at least in formal speech. Therefore, it is possible to summarize that socio-linguistic factors—that is, degree of formality, also contributed to the extent of phonetic attrition in this segmental analysis, suggesting that although phonetic attrition of segments may be inevitable in a prolonged L environment, more naturalistic forms of speech elicitation may be more likely to promote phonetic attrition—that is, if speech is elicited formally, the effects of phonetic attrition might not be evidenced at all. To a certain extent, these studies within a bilingual context built on an early impressionistic study of monolingual phonetic attrition (Shockey, ). In this study, four American English-speaking residents of the United Kingdom, who had been living in England for between eight and twenty-seven years, were examined. Her study, within a context of dialect acquisition, involved counting the number of flaps in American speech production, where in Standard Southern British English (SSBE) one would expect a voiceless or voiced alveolar plosive—that is, /t/ or /d/ in words like, respectively, latter and ladder. Shockey found differences between subjects, with some individuals exhibiting a greater degree of attrition of segments, and in different phonetic contexts. The voiceless alveolar plosive /t/ was much less likely to flap if it was in a relatively slow phrase, or if it had a high information content; and ‘both /t/ and /d/ showed a strong tendency to flap in fast,
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
unstressed, highly redundant speech, with the tendency being much stronger for /d/ than for /t/’ (pp. –). Although never drawing on the term ‘attrition’ specifically, Shockey interpreted that the changes in the American English speech were the result of a ‘slow, ongoing process which is not completed for a number of years’ (p. ). In line with Major (), her findings suggest that not only amount of exposure to the new linguistic system might account for interspeaker differences in phonetic attrition, but that speech style also affects intrapersonal variation, with careful speech and high information content eliciting less phonetic attrition. Aside from these early studies into segmental assimilation effects in plosives, formant frequencies in laterals, rhotics, and vowels have also revealed assimilation effects. For example, within the context of dialect acquisition in monolinguals, a more recent study (Evans & Iverson, ) investigated changes in vowel production among university students from Ashby de la Zouch, Leicestershire, UK, a small town in the Midlands where the local accent is classified as a variety of northern English (Wells, ; Evans, ). This study enhanced the Shockey () research because it involved both an acoustic analysis, as well as a regional accent rating by fellow British English speakers. Moreover, the subjects in this study had only been resident in their new environment for three years as they had moved to London to study, where they were exposed to SSBE more frequently than in their home town. In the group analysis, the results showed that their accents became more like SSBE, again revealing plasticity in the L phonetic system through assimilation effects. Regarding the acoustic formant analysis, the subjects changed their production of bud, cud, and could so that they produced them with a more fronted (higher F frequency) and lower (higher F frequency) vowel after only two years at university, closer to how southerners produce the vowel in words like bud and cud. However, there were differences in the amount and direction of accent change: although the majority of subjects were rated as sounding more ‘southern’ after two years than when they were first recorded; in a minority, there was only a very slight change in accent; three subjects were rated not to have changed their accent at all; and one subject was judged to have a more ‘northern’ accent at the final testing session than when she first moved to London. This study indicates not only a high degree of interpersonal variation, but also that the process of phonetic attrition, here in a monolingual context, can occur relatively quickly, and that it therefore might not be a slow process, as suggested by Shockey (). Moreover, although only revealed in one subject, it also hints at dissimilation effects, to be discussed subsequently. Formant frequencies have been the subject of further research into phonetic attrition within the context of bilingualism (de Leeuw, Mennen, & Scobbie, ; Ulbrich & Ordin, ; de Leeuw, Tusha, et al., ). An investigation into coda lateral production in the German speech of ten German-English late bilinguals who had been living in Vancouver, Canada for between eighteen and fifty-five years, indicated that the F frequency of the lateral phoneme in German words like viel and Kiel was significantly higher in the German of both the female and male bilinguals than in a respective German control group, becoming more English-like, and therefore indicating segmental attrition through assimilation effects. For the males, F frequency was also significantly lower in the German of the bilinguals than in the German monolinguals, as is characteristic of the ‘darker’ North American English lateral, again, revealing phonetic attrition. Moreover, some of the F frequencies were characteristic of the dark values, whilst the F values were characteristic of
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
the light values, suggesting a truly assimilated production of the lateral. Interpersonal variation across subjects was also revealed: one male subject (FS) produced the German lateral completely within the English monolingual norm whilst two female subjects produced the German lateral entirely within the German monolingual norm. In line with the previously discussed studies, it was again revealed that phonetic attrition in the form of assimilation effects is an outcome of late bilingualism, although the degree of plasticity of the phonetic L system varies across individuals. The results from this study prompted further investigation into one individual, FS, who stood out, and revealed ‘extreme’ phonetic attrition in his speech (de Leeuw et al., ). Subject FS was selected for this case study because he displayed phonetic attrition in all of the group phonetic analyses he underwent, whilst the others nine bilinguals who had moved from Germany to Canada did so as well, but less consistently. In addition to the lateral analysis (de Leeuw, Mennen, & Scobbie, ), in which it was found that his realization of German /l/ in coda position was entirely within the ‘dark’ English norm, in a prosodic analysis it was revealed that his realization of the prenuclear rise in his native German was significantly earlier than is characteristic of German; instead, falling within the English monolingual norm (de Leeuw et al., ), to be discussed shortly. Moreover, FS was salient because he was consistently rated to be a non-native speaker of his native German by German monolinguals in Germany in a global foreign accent rating (de Leeuw et al., ), as will also be discussed. The particular case study investigated phonetic attrition in the rhotic realizations of FS, whose biographical data indicated that, out of all the participants, he was among one of those who had been with their partner the longest, who was a native English speaker and with whom he only spoke English. They had two adult daughters and three grandchildren, with whom he only spoke English, and he had moved to Canada when he was , and thereafter had lived in Canada for fifty-three years. Crucially, when he spoke English, which he did almost always, he did so with next to no code-mixing into German, as none of his English interlocutors spoke German. He therefore differed from the other late bilinguals, who, in one way or another, were all considered to use German more than FS. The impressionistic analysis of his rhotic production revealed that % of his German rhotic productions were perceived to be English-like, much higher than that of any other participants (ranging from % to %). The average F frequency of his German rhotic (, Hz) approximated values typically given for American-English, such as , Hz (Dalston, ), and the range of between , and , Hz for English rhotics (Thomas, ), and substantially differed from values typically given for Standard German of between , and , Hz (Ladefoged & Maddieson, ). Although all participants in this study into rhotic realization revealed some degree of phonetic attrition, FS did so more extremely than any of the other participants, and realized the English retroflex rhotic in his German. The authors tentatively suggest that, essentially, complete immersion within an entirely English language setting is what made FS unique, and triggered his extreme attrition, in the domain of phonetics. As such, the findings shed new light on the SLM in that his German rhotic was not merged intermediately between German and English monolingual norms, but rather, in this extreme case of phonetic attrition, his rhotic production entirely assimilated within the monolingual norm of the English retroflex. Another related study examining formant frequencies in rhotics reported findings from an analysis into vocalization and rhotic realization of post-vocalic /r/ in the German native
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
speech of sequential bilinguals who had acquired either a rhotic variety of English as an L (Belfast English), or a non-rhotic variety of English (Oxford English) (Ulbrich & Ordin, ). In Belfast, the earliest age of arrival was years of age, whilst the oldest age of arrival was ; in Oxford, the earliest age of arrival was years of age, whilst the oldest age of arrival was . The results of the auditory and acoustic analysis of post-vocalic /r/ in the speakers’ German and English suggested that attrition occurred in the form of transfer from the L—that is, the German native speakers in Belfast and Oxford produced acoustic correlates of rhoticity in their native German speech. However, as noted by the authors, those exposed to Oxford English would not be expected to produce post-vocalic /r/, yet the German native speakers in Oxford did produce the acoustic correlates of rhoticity in their German, just like those with Belfast English as their ambient language. The authors therefore suggest that direct transfer from the L to the L could not have been the sole factor determining phonetic attrition. Instead, they suggest that between subject variation should take into account ‘actual creative language use and communicative events’ (p. ). What particularly stood out in this study is that dialectal variation in an L did not necessarily modulate the effects of attrition, and that, although sequential bilinguals may be exposed to a particular variety, other factors, dissimilating from the target L group in L speech, potentially towards a perceived more standard-like variety, for example American English, may also play a role. In line with these studies, a case study examining the speech of a monozygotic twins compared two sisters through an investigation into numerous phonetic variables (Mayr et al., ). MZ had moved from the Netherlands to the United Kingdom in early adulthood and where she had been living for thirty years at the time of data collection, when she was years of age. Her speech was compared to that of her twin sister, who had remained in their country of birth, the Netherlands, and phonetic attrition was therefore measured in a perfectly controlled environment. The results revealed that MZ produced Dutch voiceless plosives with VOT values longer than the native Dutch norm of her twin sister, but not as long as the aspirated plosives of English, and that, overall, there was evidence for ‘crosslinguistic assimilation patterns for all plosives, with voiced ones realized with a voicing lead in both languages, and voiceless ones with VOT values intermediate between short-lag and long-lag categories’ (p. ). In addition, MZ’s monophthongs and diphthongs followed a general trend towards more open (higher F) realizations compared with her twin sister’s Dutch norm, which is in line with the expected more open pronunciation of English vowels (Mayr et al., ), with no dissimilation effects reported. Moreover, ‘not only did L categories assimilated to L ones shift towards a more open position, but also L categories with no counterpart’ (p. ). Therefore, the authors suggest that cross-linguistic interactions operate at a system-wide level, rather than at the level of individual sounds, which is consistent with related research into early bilinguals (Guion, ) short-term phonetic drift effects (Guion, ; Chang, ), as well as with the notion of language specific phonetic settings (Mennen et al., )—that is, the individual undergoing attrition does not only acquire individual sounds in the L, but rather a particular articulatory setting, and this configuration impacts the L phonetic space generally, rather than token-by-token. However, dissimilation effects, which have also been reported in phonetic attrition studies (Flege & Eefting, a; Evans & Iverson, ; de Leeuw et al., ), do not feed into this interpretation cleanly. If there is a system wide shift, there must also be scope
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
for phonetic attrition to occur at a level of individual sounds, as some bilinguals evidence dissimilation whilst others evidence assimilation for the same sounds.
. P
.................................................................................................................................. In addition to such segmental changes in the speech production of the L in sequential bilinguals undergoing attrition, it has also been shown that prosodic realizations are susceptible to attrition and can undergo assimilation effects. Mennen () investigated native Dutch speakers who were at near-native level in their acquisition of L Greek. Although her sequential bilinguals did not live in Greece, they had all learned Greek as an L in adulthood; all had extensive experience with the L (between twelve and thirty-five years); all held a university degree in Modern Greek language and literature; were teaching Greek at university level in the Netherlands, and all used Greek regularly in their daily lives. As such, they were similar to subjects investigated in attrition studies, and the findings from her study are therefore of consequence to the questions of this chapter. Four out of five of her subjects were not only unable to realize Greek tonal alignment according to monolingual norms, they also showed a change in their native Dutch tonal alignment patterns. Specifically, the differentiation in the alignment of pitch peaks across Dutch long and short vowels was greatly reduced in their L speech. However, similar to the aforementioned interpersonal variation revealed (Major, ; de Leeuw, Schmid, & Mennen, ; de Leeuw, Mennen, & Scobbie, , ) one of the bilinguals did produce tonal alignment with native-like values in both the L and L, again suggesting that the extent of plasticity of native speech differs between individuals.
. S
.................................................................................................................................. In addition to assimilation effects, research into the speech of late bilinguals has suggested that the late acquisition of an L can have dissimilation effects on the phonetic system of the L. One study which explicitly reported dissimilation effects is that of Dutch native speakers who were highly proficient in English as an L, which they had begun learning at years of age in the Netherlands. These sequential bilinguals, who were between and years of age at the time of the experiment, and highly proficient in English, produced their Dutch /t/ with shorter VOT values than a group of Dutch L speakers who were less proficient in English (Flege & Eefting, a). Specifically, in the highly proficient subjects, the Dutch /t/ moved away from both the typical English value and the typical Dutch value (becoming shorter). Flege & Eefting (a) suggest that this may have been a result of ensuring sufficient discrimination between the L and the L segments, but it is difficult to reconcile that these results are in contrast to the previously reported assimilation effects in
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
VOT for the same language combination (Mayr et al., ). Perhaps, the earlier age of English acquisition in the Flege & Eefting study plays a role, although one would expect that in Mayr et al. (), MZ was similarly exposed to English at an early age in the Netherlands; alternatively, it may be that MZ was even more proficient in English than the Dutch-English highly proficient bilinguals in the Netherlands, with even higher proficiency evidenced through assimilation effects, and lower proficiency through dissimilation effects. Nonetheless, it seems plausible that an entirely system-wide explanation cannot account for these discrepant findings, as the competing systems are the same, but yield different results.
. P
.................................................................................................................................. Dissimilation effects have also been reported in an analysis of interpersonal variation of prosodic attrition in ten late German–English bilinguals (de Leeuw et al., ). In German, both the start and end of the prenuclear rise occur later than in English (Atterer & Ladd, ), and it was therefore examined whether German native speakers who had moved to Canada in early adulthood would produce the prenuclear rise with an earlier alignment than the German monolingual matched control group. Although most of the bilinguals evidenced assimilation in their tonal alignment, it was also reported that two females out of ten native German speakers ‘overshot’ the monolingual German norm with respect to the tonal alignment of the pre-nuclear rise. In ‘overshooting’ the German monolingual norm, the alignment at the end of the rise occurred even later in their German productions than the already late German alignment, and was thus more dissimilar from both the German norm and the English norm (similar to the segmental dissimilation effects reported in Flege & Eefting, a). This finding was not highlighted in the research, potentially because dissimilation effects are so counter-intuitive when considering the consequences of two competing linguistic systems, and may even be interpreted as outlier effects in other studies (see, e.g. high standard deviations in Flege & Hillenbrand, ; and Flege, ). However, dissimilation effects in some speakers, coupled with assimilation effects in others in this study (de Leeuw et al., ), again suggest that, at the very least, a system wide shift has a wide range of freedom, or, potentially, that competition from two language systems is also governed by both interspeaker variation and token-by-token competition.
. G
.................................................................................................................................. In the present section, studies which have examined global regional or foreign accent in native speech are discussed. In a way, such research goes beyond the analyses of individual sounds and prosodic variables by investigating the perception of an individual’s accent, and
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
whether it is possible for someone to no longer be perceived to be a native speaker of his or her native language, as in the previous acoustic investigations, it was not examined whether such fine phonetic differences from monolingual native speaker norms are actually perceived to be accented by listeners. An early study into global regional accent perceptions in monolinguals reported that on average a group of Canadians from the province of Alberta, who had moved to Birmingham, Alabama, USA after years of age, sounded more American-like than a comparable group of Canadians resident in Alberta (Munro et al., ). Some of the Canadians in Birmingham were judged to be just as American-sounding as actual American speakers, whilst others fell within the Canadian range. Although there was some indication that the amount of time spent in the US may have predicted interspeaker variation in the bidialectal speakers, with Canadians living longer in Alabama having more American-accented speech, some speakers did not follow this pattern. Other predictor variables were not investigated in this study into global regional accent, but what is clear from the findings is that the native linguistic system was permeable within the domain of phonetics at a global level, which was perceivable by L listeners. Accordingly, this research largely corroborated the findings from Evans & Iverson (), as previously discussed; however, in the latter study, it is worth emphasizing once more that one speaker dissimilated from the ambient variety, sounding more northern-like after two years in London than upon her arrival. Building on this study in the context of bilingualism, de Leeuw et al. () examined whether it was possible for native speech to become foreign accented upon acquisition of a new language in adulthood. Specifically, the objective was to determine whether native speakers of German living in either Canada or the Netherlands could be perceived to have a foreign accent in their native German speech. German monolingual listeners (n=) assessed global foreign accent of thirty-four L German speakers in Anglophone Canada, twenty-three L German speakers in the Dutch Netherlands, and five German monolinguals in Germany. The bilinguals had moved to either Canada or the Netherlands at an average age of and had resided in their country of choice for an average of thirty-seven years. The findings revealed that the German listeners were more likely to perceive a global foreign accent in the German speech of the late bilinguals than in the monolinguals, and that, although twenty bilinguals were rated clearly to be native speakers of German, fourteen bilinguals were rated clearly to be non-native speakers of German (although they were native speakers of German). Further analysis revealed that language use had a significant effect on predicting the extent of foreign accent in the L, rather than, for example, age of arrival in their country of choice or length of residence. Those bilinguals who were expected to code-switch more often were more likely to undergo phonetic attrition—that is, be perceived as non-native speakers of their native German—than those who largely conversed in monolingual settings, where less code-switching was predicted, suggesting that monolingual modes (Grosjean, ) are likely to maintain phonetic properties of the L (de Leeuw et al., ). A similar study, which investigated foreign accent in a group of ‘L attriters’, in comparison to a group of ‘L acquirers’, largely confirmed these findings (Hopp & Schmid, ). A main goal of this study was to determine whether a group of German native speakers who had moved from Germany to Canada or the Netherlands after age , and had
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
used their L most frequently in daily life after having lived abroad for more than fifteen years, were rated to be less native-like in a global foreign accent rating (as in de Leeuw et al., ), than a group of late L English and L Dutch L acquirers of German. At the group level, the findings revealed that the ‘L attriters’ did not differ from the monolingual German speakers. However, of the forty ‘L attriters’, twenty-nine scored within the range of the native controls, whereas eleven scored outside this native range. The authors interpreted their findings to show ‘that acquiring a language from birth is not sufficient to guarantee nativelike pronunciation, and late acquisition does not necessarily prevent it’ (p. ). However, in exploring why some individuals were more likely to undergo phonetic attrition than others, the results were inconclusive: no independent variable significantly predicted the extent of phonetic attrition, which contrasted with the findings from de Leeuw et al. (), which suggested that language use plays. Finally, in a more recent study, it was investigated whether foreign accent ratings would be associated with acoustic measurements of vowels and consonants—that is, whether those late bilinguals who were categorized as phonetic ‘attriters’ in a foreign accent rating task would likewise exhibit phonetic measurements which deviate from the monolingual group (Bergmann et al., ). The study examined thirty-three L speakers of German, residing in the USA or Canada, where they made use of their L. Results from the ratings showed that native-likeness was negatively associated with length of residence abroad and positively associated with L use. Second, formant analyses of four speech sounds in German—/aː/, /ɛ/, /ɔ/, and /l/—showed that most acoustic differences were observed in /a:/ and /l/ whilst the other formant frequency values were not significantly different from the monolingual group. However, it was not the case that a stronger foreign accent in the late German-English bilinguals was associated with more deviant formant frequency values in these sounds, therefore suggesting that other segmental, or more likely prosodic features, were responsible for the foreign accent ratings.
. S
.................................................................................................................................. The studies discussed in this chapter revealed a common finding that the phonetic system of the L is malleable upon competition from a new language or dialect in adulthood. Although intersubject variability with regard to the extent of phonetic attrition was evidenced, in their entirety, the findings indicate continued plasticity in native speech throughout the entire lifespan. Examining the studies together, the following ten key findings, in no order of importance, can be presented. . Research which reports phonetic attrition, as described in this chapter, combined with changes to the L phonetic system due to dialects changing over time (Harrington et al., a,b; Harrington, ), style-shifting (Bell, ; Eckert & Rickford, ; Coupland, ), and short-term rapid changes to the L phonetic system upon L training and exposure (Sancier & Fowler, ; Chang, ), lend support to the understanding that there is plasticity in native speech beyond adolescence.
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
. Although phonetic attrition is consistently observed in analyses of speakers who have an increased dominance in their new dialect or language, there is interpersonal variation in the extent of phonetic attrition, and some research suggests that increased L use in a monolingual mode maintains native speech pronunciation (de Leeuw, Schmid, & Mennen, ; Bergmann et al., ; de Leeuw, Tusha, et al., ). . Assimilation effects between L and L sounds have been evidenced in the VOT of plosives in phonetic attrition (Flege & Hillenbrand, ; Flege, ; Mayr et al., ), although dissimilation effects in the VOT of plosives in the L have also been reported in advanced L learners (Flege & Eefting, a). This discrepancy suggests that phonetic attrition incorporates both systematic merging (Guion, ; Chang, ; Mayr et al., ), as well as token-by-token level changes (Flege, , , ; de Leeuw, Mennen, & Scobbie, ; Ulbrich & Ordin, ). . Assimilation effects between L and L sounds have been reported through formant frequency analyses of vowels and liquids (Flege & Hillenbrand, ; Mayr et al., ; de Leeuw, Mennen, & Scobbie, ; Ulbrich & Ordin, ; de Leeuw, Tusha, et al., ), although further research is necessary to examine the precise form which intermediate tokens might take (e.g., whether the ‘merging’ becomes entirely Llike, or whether one acoustic variable—that is, F frequency—becomes more L-like, whilst another acoustic variable—that is, F frequency—remains L-like (de Leeuw, Mennen, & Scobbie, )). . Assimilation effects have been reported in prosodic analyses of L speech (Mennen, ; de Leeuw et al., ) and are suggested in global foreign accent effects (de Leeuw et al., ; Hopp & Schmid, ), but note that dissimilation effects have also been reported in prosody (de Leeuw et al., ) and in global regional accent ratings (Evans & Iverson, ). . Assimilation processes can involve intermediate ‘merging’ between the L and L categories (Flege, ), and in the case of ‘extreme phonetic attrition’ the realization of the L phoneme can occur entirely in the space of the L phoneme (de Leeuw, Tusha, et al., ). . Informal speech is more likely to evidence phonetic attrition (Shockey, ; Major, )—that is, if speech is elicited formally, the effects of phonetic attrition might not be evidenced at all. . Careful speech with a high information content appears to be less likely to elicit phonetic attrition (Shockey, ). . In the case of more than one standard variety of the L, the ambient language might not drive phonetic attrition, but rather ‘actual creative language use and communicative events’ could influence processes of phonetic attrition (Ulbrich & Ordin, , p. ), potentially suggesting that a non-ambient target variety may induce phonetic attrition. . Phonetic attrition, as it occurs in the context of new dialect (Shockey, ; Munro et al., ; Evans & Iverson, ) or new language (Flege & Hillenbrand, ; Flege, ; Major, ; Mennen, ; de Leeuw, Schmid, & Mennen, ; de Leeuw, Mennen, & Scobbie, , ; Mayr et al., ; Hopp & Schmid, ; Ulbrich & Ordin, ; Bergmann et al., ; de Leeuw, Tusha, et al., ) acquisition, has been reported at both the segmental and prosodic level of native speech, as well as in overall changes to an individual’s native accent.
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
. D
.................................................................................................................................. If, as this chapter suggests, the phonetic system is more vulnerable than other linguistic domains, such as that of morpho-syntax (see Köpke & Schmid, , who suggest that linguistic domains are differently affected by attrition), it is plausible that that from which all other language representations are built is destabilized as well, once phonetic attrition occurs. Phonetic attrition would thus reflect the start of an unbalancing of the native language system, which arises from the perception of phones, before distinctive phonemes, lexemes, and syntactic structure (e.g., Maye et al., ; Polka & Werker, ; Stager & Werker, ; Werker & Lalonde, ). Therefore, relevant future research questions might relate to, for example, how phonetic changes in the L potentially trickle down into other linguistic domains. Further questions of relevance relate to how regulation of the L and L phonetic systems impacts cognition more generally (de Leeuw & Bogulski, ), as well as to how the regulation of the L and L phonetic systems change over time, in different contexts in longitudinal analyses (de Leeuw, Tusha, et al., ). Questions might also focus on comparing different groups (e.g., those with brief exposure to the new language, versus those with long-term exposure), to determine whether interactional effects differ, and the extent to which accommodation effects resemble phonetic attrition. A final interpretation emerging from this chapter, but certainly not unique to it, is that the languages of a bilingual, and indeed the dialects of a monolingual, are separate linguistic systems, but not fully autonomous from one another (Caramazza et al., ; Grosjean, ; Paradis, ; Baker & Trofimovich, ; Sundara et al., ; Fowler et al., ; Hopp & Schmid, ; de Leeuw, ; Mayr et al., ). As such, phonetic attrition, as a symptom of increased cognitive load in bilinguals over monolinguals (de Leeuw, ), is simply one case in which the languages of a bilingual, or indeed the dialects of a monolingual, interact. This interaction between the competing linguistic systems differentiates those with competencies in more than one linguistic system from those who possess only one linguistic system, should such individuals exist at all (Cook, ; Grosjean, ; Hopp & Schmid, ; de Leeuw, ; Rothman & Treffers-Daller, ). Future research into phonetic attrition will surely tell us more about the plasticity of native speech and cognition more generally.
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
......................................................................................................................
......................................................................................................................
. I
.................................................................................................................................. T pronunciation patterns of a second language (L) can influence the pronunciation patterns of a native language (L). Analysing the speech of adults who have experienced social changes in their lives, such as migration to a non-native language setting, international adoption, or even less extreme forms of contact with an L (such as full immersion in an L course), shows that the use of one’s L can be challenged by the competing demands of an L acquired later in life. Phonological knowledge is referred to in this chapter as the way humans group different pieces of phonetic knowledge together in a coherent system of relationships that is practical or useful for communication (in the sense of both speech planning and speech decoding). The relationships among entities of the phonetic domain can be of different natures, involving either mutual exclusion (e.g., when the replacement of one phoneme with another or of an intonational contour with another produces a change in the meaning of a word or a sentence) or alternations on the basis of contextual restrictions (such as when the phonetic form of a phoneme varies in pre-tonic versus post-tonic position in a predictable way). The grouping of phonetic patterns into coherent systems of relationships results in higher-level phonological entities in such a way that, for example, a group of vocalic patterns used within a given linguistic community can be referred to as the vowel system of that language or the set of consonant clusters allowed in word-internal position can be seen as one of the basic ingredients of the language’s lexical phonotactics. Understanding how changes in the phonological knowledge of attriters occur thus means understanding how the L grouping systems change as a consequence of contact with a different language. Attrition deals with transient states of linguistic knowledge. Much like in acquisition, a transient state does not automatically imply an unsystematic behaviour; in the context of L acquisition, scholars intentionally developed the notion of interlanguage to account for both variability and systematicity in learners’ productions (e.g., Selinker, ). Note, however, that the temporal dimension is a more complex issue in language attrition than in language acquisition. In fact, according to a one-dimensional
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
view of attrition as a phenomenon occurring across the lifespan, prototypical attriters begin as monolinguals (in their L), become bilinguals (in an additional L) at a certain point of their life (usually as adults or adolescents), and eventually become monolinguals again (in what was initially their L). Those individuals thus clearly differ from early, stable, fluent bilinguals. Other non-prototypical cases of speakers whose L is influenced by an L acquired in adulthood, such as participants in full-immersion language courses or second-generation speakers of heritage languages, also experience changing degrees of contact with the L at different epochs of their lives. A phonological account of attrition phenomena should therefore include transition (in the sense of variability over time) as a core concept in the grammar of a language. The issue is probably of particular interest for those theoretical approaches that postulate that phonological knowledge is strongly dependent on, and possibly emerging from, the way the speakers experience particular phonetic patterns and use them. These approaches are based on the idea that memorization and learning occur in terms of clusters of remembered episodes of individual experiences and that the ‘best exemplar’ or prototype of the cloud is positioned where the highest number of episodes is densest (e.g., Goldinger, ). Changes in category location and number are therefore expressions of essentially temporal developments in the process of updating this abstract knowledge from continuously varying input episodes. However, including time in phonology essentially makes phonology continuous, much like phonetics, rather than categorical, in contrast with much thought in phonological research (challenging substance-based approaches as well; Scobbie, ). Phonological attrition, as a manifestation of transient bilingualism, therefore offers great opportunities to test concurrent hypotheses on the nature of phonological representations and the relationship between phonetics and phonology. In this chapter, a selection of studies dealing with phonological change in language attrition is reviewed. First, we look at studies that enquire whether native phonological oppositions are maintained or lost when there is increased use of an L with different phonological oppositions in either the vowel system (Section ..) or the consonant system (Section ..). Then, we review studies that analyse changes in native phonotactic patterns after contact with an L (Section .). In the concluding section (Section .), possible ways of further developing the investigation of phonological L attrition are sketched.
. P
..................................................................................................................................
.. Contrasts in the vowel systems Some attrition studies have focused on L changes in the vowel systems of speakers of declining varieties subject to the pressure of a dominant L (e.g., Guion, ; Bullock & Gerfen, ; Bullock et al., ). Bullock & Gerfen () analyse the variety of French in Frenchville, a village in Pennsylvania where the descendants of nineteenth century French immigrants still live. The study focuses on two brothers aged and who were born in Frenchville and are
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
illiterate in French. Frenchville French is their native language: they spoke this language exclusively at home until their early twenties. Subsequently, as they married Englishspeaking women, their use of French was dramatically reduced over time. The study documents the loss of the quasi-phonemic contrast between mid-high and mid-low front rounded vowels /ø/ and /œ/ in Frenchville French. The contrast has a very low functional load in European French and tends to be neutralized in favour of the open-mid phoneme /œ/. In contrast to speakers of continental French, the two Frenchville speakers analysed in the study neutralize the opposition by replacing both phonemes with a rhoticized schwa (/ɚ/), which is evidently the result of contact with L English. The authors show that the reason for such replacement is not to be found in the low functional load of the opposition because other non-contrastive differences between Frenchville French and English are maintained in the speech of the informants (e.g., uvular trill as opposed to retroflex rhotics). Unlike uvular trills, the /ø/–/œ/ distinction is characterized by low psychoacoustic salience and low frequency of occurrence in the lexicon. The authors thus argue that contrastive oppositions have no special status in language attrition inasmuch as other properties, such as segmental salience and frequency, can be better predictors of maintenance or loss. The fact that frequency can override contrastivity is also evident from the tendency of attriters to enlarge the scope of application of contrasts by overgeneralizing one phoneme (usually the less marked one) to contexts in which it would not be expected, as reported, for instance, by another study on Frenchville French focusing on the /a/ and /ɑ/ contrast (Bullock et al., ). In terms of perception, the interplay between contrastivity and acoustic salience is further evidenced by Flores & Rauber (), who show that Portuguese adults who abandoned Germany and the German language during childhood are still able to discriminate pairs of vowels distinguished along two properties that are contrastive in German: duration and roundedness. The role of perceptual distinctiveness in phonological attrition is thoroughly explored by Guion () with respect to the production of the vowel systems of Quichua and Spanish by Ecuadorean bilinguals. The study compares simultaneous Spanish-Quichua bilinguals with early bilinguals who acquired Spanish between and years of age, mid bilinguals (between and ) and late bilinguals (between and ). While simultaneous bilinguals produce vowels of both languages consistently like monolingual speakers do, late bilinguals do not differentiate Spanish and Quichua vowels and seem to use the Quichua vowel system when they speak Spanish as well. The early and mid bilinguals show the most interesting pattern: most of them are like monolingual Spanish speakers in their L but differ from monolingual Quichua in their L. In fact, the early and mid bilinguals produce their native Quichua /ɪ/, /ʊ/ and /a/ higher than bilinguals who have not acquired distinct Spanish vowels. Based on fine-grained formant analysis of individual vowel changes, it is argued that a systemic raising of all Quichua vowels relative to the Spanish vowels has occurred and that this raising is the most economic and efficient way to provide the two phonological systems with a maximum of reciprocal distinctiveness. According to this interpretation, attrition is guided by maximization of contrasts in the perceptual space within one common phonological system and not by progressive assimilation to the segmental properties of the L. Similar tendencies are found by Chang () in the domain of production for American English speakers enrolled in a language course in Korea. The context of this study is different from Guion’s () study: instead of focusing on a condition of asymmetric
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
language contact and societal bilingualism, Chang () studies the students’ pronunciation of L vowels and stops one to five weeks after the beginning of an L Korean course in Korea. Another difference between the two approaches is that the American English subjects in Chang’s study are all late L learners. Despite these differences, both studies show that the changes in the L do not involve local assimilations of individual L segments to the closest L segments. Rather, the changes tend to encompass global shifts in one dimension (in the case of vowels, F values) for almost all segments in order to approximate the system’s spatial distribution of the L. In some specific cases, as a consequence of this global movement, an L segment may end up being farther away from, rather than closer to, the non-native (L) counterpart. Similar results, if taken in isolation, would appear counterintuitive or at least more difficult to interpret if we did not consider the big picture and look at the system’s movement altogether. This suggestion is also put forth by Mayr et al. () based on the analysis of the vowel systems of two monozygotic twin sisters who were interviewed when they were years old. One of them (MZ) had migrated to the United Kingdom thirty-four years before the time of the study (thus adopting English as her dominant language in all private and social dimensions), whilst the other (TZ) had spent all her life in the Netherlands (though being familiar with English as an L). When speaking Dutch, MZ produces all but two monophthongs with higher F values than TZ; this effect is also attested for L vowel categories that have no counterpart in English. The direction of the change (increased openness of L vowels) can be interpreted as induced by the phonetic properties of British English (F values in female speech are overall higher in British English than in Dutch). However, if this effect was to be explained by a mechanism of token-by-token assimilation, its generalization to Dutch vowel categories that have no English counterpart would remain unexplained.
.. Contrasts among obstruents Another topic frequently investigated in attrition and bilingualism studies is the voice onset time (VOT) of obstruents in obstruent-vowel sequences. Languages differ in the number and quality of VOT duration oppositions used to differentiate among consonants of their phonological inventory (e.g., Cho & Ladefoged, ). Therefore, documenting a change in VOT realization for (pairs of) L contrastive stops can either reveal an ongoing adjustment of the parameters, according to which the contrast is implemented, or document the suppression or creation of VOT oppositions. Only the latter would in principle impact phonological grammar in the sense of increasing or decreasing the number of contrasts contained in it. However, most attrition studies have focused on the former aspect (adjustment of phonetic parameters). That is, VOT production has been mostly investigated for transient bilinguals whose L and L both possess a two-way VOT distinction (typically in stressed position). This is the case for bilinguals speaking English and French (Flege, ), English and Portuguese (Sancier & Fowler, ; Major, ), English and Japanese (Harada, ), English and Spanish (Flege & Eefting, b; Lord, ), and English and Dutch (Flege & Eefting, a; Mayr et al., ), with English being often (but not always) the L of migrants or late language learners. Whether the attriters assimilate the phonetic properties of the L to those of the L, producing intermediate VOT values (as in
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
Flege, ; Major, ; Sancier & Fowler, ; Harada, ; Lord, ), or they exaggerate the L phonetic properties to differentiate them from the properties of the L (what is generally referred to as dissimilation or polarization in attrition studies; cf. Flege & Eefting, a,b), all these studies basically demonstrate that in adult attriters, the phonetic properties of the L can reshape those of the L. In this sense, these studies expand on the observation that the sound systems of two languages interact in a dynamic and ‘creative’ way in the brain of early bilinguals, a finding empirically demonstrated by Caramazza et al. (), again with reference to VOT production by Canadian fluent bilinguals. There is, however, at least one case in which the two contact languages differ for the number of phonological categories distinguished along the VOT continuum: English and Korean, the former featuring a two-way phonological opposition between voiced (shortlag VOT) and voiceless (long-lag VOT) stops and the latter featuring a three-way opposition between fortis (short-lag VOT), lenis (long-lag VOT), and aspirated (‘very long’-lag VOT) stops. Other phonetic parameters also contributing to the distinctions are the f contour and the voice quality of the onset of the following vowel (e.g., Cho et al., ). The early Korean-English bilinguals who migrated to the USA before the age of in the study by Kang & Guion () were able to correctly produce five different types of stops (for VOT, voice quality and f of the vocalic onset) when speaking in both languages. This finding suggests that they kept the phonological contrasts intact in the two languages. However, those late bilinguals who learned English as adults (average age of arrival: . years) differed from Korean monolinguals in the production of both fortis and aspirated stops for at least two parameters, VOT and f of the vocalic onset; their values assimilated to those of monolingual English voiced and voiceless stops, respectively. The study by Kang & Guion () clearly shows the difference between early, fluent bilinguals and late, less proficient bilinguals. Only the latter show an effect of the L on the L, basically leading them to a merger of L and L categories and, consequently, to contrast blurring. The results of Kang & Guion () thus indirectly support the view that the degree of L attrition is correlated with the degree of L knowledge. As anticipated above in the section on vowel contrasts, Chang () deals with L English elementary learners of Korean. The study shows that these learners also modify their L stop pronunciation after short exposure to the L (one to five weeks). As an effect of the increased familiarity with Korean aspirated stops, the VOT of English voiceless stops lengthened at every place of articulation and for both male and female speakers, even though the VOT of English alveolars should not have changed for females on the basis of their distance from Korean aspirated alveolar stops. As previously mentioned for the vowel study, changes affecting natural classes of segments and conveyed by multiple phonetic cues at the same time (VOT, f, vowel formants) are to be seen as systemic rather than segmental. Whether they should also be seen as the earliest stages of a process of L attrition is to be ascertained. The study certainly suggests that for an L system to show a drift towards an L, extensive contact with the L is not needed. This finding integrates the view that L-induced changes in the L are experienced over long stretches of time by long-term immigrants or international adoptees.
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
Further examples of dissimilation or maximization of acoustic distinctiveness are provided in the study by Dmitrieva et al. (), who investigate how Russian word-final obstruent devoicing is produced by Russian speakers with and without extensive knowledge of American English. L Russian speakers with no knowledge of English are found to maintain the distinction between underlying voiceless and voiced word-final stops by varying the duration of the closure phase and of the release, consistent with previous literature reporting incomplete neutralization in Russian. By contrast, L Russian speakers with extensive knowledge of English additionally use vowel duration and voicing into closure to distinguish between underlying voiced and voiceless stops. English lacks wordfinal devoicing but uses vowel length (and voicing into closure) to signal the voicing of the following obstruent. Thus, it appears that knowledge of L English has modified how an L phonological rule is implemented, enhancing a distinction that tends to be neutralized in the L of monolinguals. Contrast neutralizations are investigated in the domain of perception by Celata & Cancila (). Experience with allophonic variation and neutralization rules in the L affects the perception not only of the L (discrimination of allophones is poorer than of phonemes) but also of similar sounds in the L (Hume & Johnson, ; Celata, ). In the study by Celata & Cancila (), first- and second-generation Italian immigrants from the city of Lucca who were living in San Francisco were tested for their ability to maintain the singleton-geminate contrast in a word identification task (words included in frame sentences) and a non-word discrimination task (segmentation of three acoustic continua). In Lucca Italian, neutralization of the contrast selectively occurs for the rhotics (and variably for some other consonants of the inventory). The study shows that the speakers’ performance varies across tasks. A group of homeland Lucca Italian speakers still residing in Italy is tested for comparison. In word identification, the homeland Lucca Italian speakers are selectively impaired on the discrimination of short and long /r/ but not /s/ and /t/ (as predicted by their native experience with the partial neutralization rule); by contrast, first- and second-generation immigrants make the same percentage of errors in all three consonants (although the former score is on average better than the latter). In nonword discrimination, the homeland Lucca Italian speakers produce categorical classifications of the singleton versus the geminate for all three continua, with high correlation values between their performance and that of native Italian speakers from a region where the singleton-geminate contrast is not neutralized for any of the consonants. The same happens as for the classification produced by first-generation immigrants (although the correlation values are somewhat lower); by contrast, second-generation speakers produce noncategorical patterns of discrimination. The study thus shows signs of attrition in the maintenance of a partial neutralization rule, with the L neutralization of the /r/–/r:/ contrast overgeneralizing to other consonant contrasts in word identification by first-generation immigrants. However, the ability to discriminate is maintained in non-word discrimination (segmentation of acoustic continua). The finding is consistent with perceptual studies on adoptees (e.g., Hyltenstam et al., ) or returnees (e.g., Flores & Rauber, , already mentioned in the section on vowels), which show that even in adults with strongly reduced or totally interrupted contact with the native language, the ability to discriminate native phonological contrasts can be maintained in categorical discrimination tasks.
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
. P
.................................................................................................................................. Phonotactic knowledge refers to the set of categorical rules and probabilistic constraints that regulate the form of speech, allowing or disallowing sequences of sounds or features to appear in different positions of a structure (at word edges or word-medially, within or across morphemes, etc.). Recent research has shown that in bilinguals, phonotactic constraints of both languages are actively accessed during the processing of one language (e.g., Carlson et al., ; Freeman et al., ), thus reinforcing the view that multiple sound systems are integrated and mutually influence each other in fluent bilinguals’ speech. Studies on phonological attrition and transient bilingualism apparently confirm the same view but additionally provide information on the potential changes occurring in such integration as a function of time development and the degree of bilingualism. Perceptual repair is one of the best-documented strategies exploited by speakers when they are faced with segmental sequences that do not match the phonotactic rules of their L. For instance, as a consequence of perceptual repair, speakers can hear an ‘illusory’ epenthetic vowel within consonant clusters that are illegal in their L (e.g., Dupoux et al., ; Berent et al., ). Parlato-Oliveira et al. () factor the subjects’ degree of bilingualism into the study of cross-linguistic perceptual illusion by analysing four different populations of Japanese-Brazilian Portuguese bilinguals living in São Paulo: first- and second-generation immigrants, simultaneous bilinguals, and L Portuguese learners of L Japanese. Based on the phonotactic principles of Japanese and Brazilian Portuguese, monolinguals in these two languages should perceive an illusory vowel in stimuli containing a consonant cluster, but the quality of the epenthetic vowel should differ in the two languages, most often as /u/ for Japanese listeners and /i/ for Brazilian Portuguese listeners (e.g., Dupoux et al., ). First-generation Japanese immigrants are indeed found to perceive like Japanese monolinguals (with slightly higher percentages of non-/u/ responses), whereas second-generation immigrants and simultaneous bilinguals are found to perceive like Brazilian Portuguese monolinguals (with higher percentages of non-/i/ responses). The learners of L Japanese behave differently from any other group, showing an intermediate behaviour with a relatively higher percentage of both /i/ and /u/ responses, suggesting that the learning of Japanese as an L has an impact on how speakers process the phonotactic structures of their L. In sum, first-generation Japanese immigrants show evidence of L perceptual attrition compared to Japanese monolinguals, whereas secondgeneration immigrants and Brazilians learners of L Japanese show an influence of the L on their perceptual processing of illusory vowels. Interestingly, these effects are found in an explicit identification task with word-sized stimuli but disappear in an implicit forcedchoice categorization task, suggesting that explicit perceptual tasks involving linguistic or meta-linguistic awareness are better at highlighting L attrition and L–L interference phenomena than implicit acoustic tasks are (see Celata & Cancila, , reviewed above, for a similar finding). This finding is also consistent with reported data on attrition in the detection of foreign accent (e.g., Major, ), as foreign accent detection involves holistic perception of multiple cues in naturally produced words or sentences. The study by de Leeuw et al. () on the production of /l/ by ten Albanian speakers who acquired English as an L in adolescence or adulthood also provides a window on
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
phonotactic competence in L attrition. The phonotactic structure of English requires a ‘clear’ or apical variant of the lateral approximant /l/ in onset position and a ‘dark’ or velarized allophone in coda position ([ɫ]). By contrast, /l/ and /ɫ/ are two distinct phonemes in Albanian, and they can occur in both onset and coda positions. Thus, the general prediction is that if the L of Albanian-English bilinguals undergoes attrition, a native phonological contrast is restructured in the form of an allophonic rule involving the alternation of two phonetic variants in a contextually predictable way. The results of the study (based on a word list reading task) confirm that such a change in the L Albanian phonology of the subjects can actually occur. In fact, in contrast to three subjects who perform like monolinguals and show no evidence of phonological attrition, five subjects show variable traces of the suppression of the native phonological contrast and its replacement by a contextual restriction predicting dark [ɫ] in the coda and clear [l] in the onset according to the requirements of the L. Finally, two other speakers show evidence of an unexpected realization of clear /l/ in coda position, even clearer than onset /l/, probably as an effect of the tendency to dissimilate or maximize phonetic distinctiveness as referred to above. From a cross-generational perspective, the study by Nodari et al. () on voiceless stop aspiration by Calabrian-English immigrants similarly suggests that when two languages share the same phonetic feature but differ in terms of the feature’s phonotactic distribution, L phonological change across generations may occur through a progressive restructuring of the phonotactic norms, potentially leaving phonetic substance unaffected. The phonetic feature under investigation here is long-lag VOT, and the cross-linguistic phonotactic difference refers to the preference for aspiration to occur in unstressed syllables in Calabrian Italian and in stressed syllables in English. Another fundamental difference between the two languages is that voiceless stop aspiration is a socio-phonetic variable in Calabrian Italian, as it indexes socio-cultural values about the speaker’s ethnic and geographical origin. In contrast to the Albanian speakers analysed in de Leeuw et al. (), first-generation Calabrian immigrants appear to maintain socio-indexical aspiration in their Calabrian speech; the subsequent generations incrementally depart from such a model, with some second-generation speakers showing that the phonotactic norms of both languages can be consistently present in their near-native Calabrian speech.
. F
.................................................................................................................................. Some areas of phonological investigation are being extensively explored with respect to fluent bilingualism and language acquisition in general, but less emphasis is given to transient bilingualism and language attrition. For instance, a large body of research on monolinguals suggests that speech production is sensitive to the lexical and morpho-syntactic characteristics of the specific items produced, starting with their lexical frequency (e.g., Jurafsky et al., ) and moving to their neighbourhood density, phonotactic probability, morpho-syntactic status, etc. (e.g., Wright, ; Plag et al., ). In the domain of bilingualism, it is not surprising that
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
lexical properties such as frequency, concreteness, and cognate status have a facilitative effect in L lexical and morpho-syntactic acquisition. For instance, cognates and concrete words are easier to learn and harder to forget than non-cognates and abstract words for experienced foreign language learners (e.g., de Groot & Keijzer, ); lexical access and word production are faster and more accurate (e.g., de Groot et al., ; Costa et al., ). Production of segmental and sub-segmental properties is affected by cognate status as well, as shown by Flege & Munro (), Amengual (), Mora & Nadeu (), among others. For instance, in the production of Catalan words by Catalan-Spanish bilinguals, /ɔ/ is realized as higher in cognate lexical items (such as in bosc /bɔsk/ ‘forest’, Spanish bosque) than in non-cognates (Amengual, ). Since the /ɔ/–/o/ contrast is present in Catalan and absent in Spanish, these results suggest that there may be more acoustic interference in cognates than in non-cognates. It would therefore be interesting to know how such interactive lexicon-phonetics mechanisms occur in the phonological competence of attriters, who may be exposed to (and be users of) an impoverished L lexicon and a limited variety of morpho-syntactic structures and pragmatic-communicative situations. Preliminary investigations suggest that L attrition is stronger for cognate than non-cognate words (Engstler & Goldrick, ), but future studies should further characterize cognate effects in attriters’ speech and clarify whether attriters differ from fluent stable bilinguals regarding the extent and direction of these effects. Current models of cross-language speech production and perception, such as the Perceptual Assimilation Model (e.g., Best, ; Best & Tyler, ) and the Speech Learning Model (e.g., Flege, ), assume that the degree of interference between sounds of the L and those of the L depends on the phonetic distance between them but cannot predict the relation between the acoustic characteristics of L sounds and the lexical status of the items in which the sounds are embedded. A more thorough understanding of cognate effects in the speech production of bilinguals, heritage language speakers, and attriters is therefore needed to enlarge the scope of available models of cross-language phonological interference. Another aspect that should be included in future studies on phonological attrition is socio-stylistic variation and the way in which attriters cope with this dimension of speech variation (as already suggested in Major, ). Early studies of the phonology of contracting varieties and moribund languages have shown, for instance, that contracting languages undergo phenomena of stylistic shrinkage and reduction of sociolinguistic variation, probably as a result of a decrease of the social contexts of use (e.g., Dressler, ). One consequence of such relaxation of socio-linguistic norms is the proliferation of unconstrained phonetic variation (unmotivated allophones), not attributable to either the L or the L. Recent investigations on endangered languages have confirmed this scenario both at the community level (e.g., Montoya-Abat, , for a Catalan variety) and at the individual level. For instance, heritage Mandarin speakers have been shown to produce more variable phonological targets than both native speakers and L learners (Chang & Yao, ). Moreover, when rated by native speakers of Mandarin Chinese, their production elicits uncertain demographic and sociolinguistic classifications (i.e., their speech ambiguously sounds ‘native’ or ‘non-native’ to native hearers in an apparently non-deterministic way). The extent to which attriters preserve and potentially transmit to subsequent generations features that play a socioindexical role in their L should also be ascertained in future studies. Undoubtedly,
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
a challenging aspect of studying stylistic variation in the context of attrition is the need for applying methods and techniques traditionally used in variationist socio-linguistics (such as social network analyses) and focusing on dialogic rather than monologic speech (to establish the role that concrete input speech plays in the development of stylistic variation in individual grammars). Similar methodological advances will in turn shift the focus from laboratory speech to large and multi-layered corpora based on naturalistic settings for data collection.
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
......................................................................................................................
......................................................................................................................
. I
.................................................................................................................................. () S: Skol’ ko u t’ebya How many with you-sg/gen ‘How many medals do you have?’ S: U menya With me-sg/gen SR: U menya With me-sg/gen ‘I have two medals.’
dv-e two-fem. dve two-fem.
medal-ej? medals-pl/gen/fem.
medal’. medal-sg/nom/fem. medal-i. medal-sg/gen/fem.
This conversation is an example of morphological language attrition, where Speaker (S) produces a non-target-like utterance with an omitted genitive feminine suffix on the word ‘medal’. Importantly, the speaker uses the genitive case correctly in the first noun phrase u menya (‘I have’) and marks the numeral dve ‘two’ for feminine gender. It may be concluded from this exchange that S still has a robust knowledge of Russian morphological system, but some of the case markers in this system have become less stable. The analysis of example () provides a glimpse at the types of questions for which the study of morphological attrition attempts to find solutions: • Which morphological markers remain resilient to attrition while others are lost quickly and why? • Why do some speakers produce all morphological markers in a target-like fashion while others do not? In other words, is morphological attrition highly individual or can it be generalized? • How would lack of stability of one morpheme affect other morphemes? • Can vulnerability of morphemes in one language be generalized to another language in the same typological group and/or other languages in different groups? • What linguistic and cognitive processes are involved in the attrition of morphology?
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
Answers to these and other questions are not easily reached in the study of first language (L) attrition of morphology as individual variation along with morphological idiosyncrasy of different languages and language groups makes it difficult to create a common theoretical paradigm that would predict and explain loss of morphological structures in L. The following overview of the state of research in this area of attritional studies reflects this difficulty.
. B
.................................................................................................................................. Studies of morphological attrition have been less numerous than studies of other aspects of language development, partially due to the fact that not all languages have inflectional morphology and those that do have a limited morphological repertoire. However, the existing studies have been successful in gaining insights into the processes of morphological attrition of the L (e.g., Polinsky, ; Bolonyai, ; Schmid, ). The ultimate purpose of explaining what types of changes morphological markers and systems undergo in bilingualism and language loss is to uncover the more general properties of morphology in linguistic knowledge. On the surface, morphological markers are fairly easy to study as speakers provide plentiful evidence of their target- and non-target-like use in spontaneous production as well as in a variety of elicitation tasks. Moreover, the omission of morphological marking is just as telling as the use of an incorrect form of a morpheme. However, morphology proves to be exceptionally difficult to analyse and generalize across languages due to its internal complexity and idiosyncratic nature (Baerman et al., ). Internal complexity of morphological markers is particularly evident in fusional languages where one affix may encode several meanings. Example () illustrates the intricate nature of the Spanish suffix -o that expresses Present tense, singular number, and first person: () Habl-o Speak-pres/sg/st person ‘I speak Spanish’
Español. Spanish
Oftentimes it is difficult if not impossible to tease out which meanings expressed by one affix may be attriting and which remain stable. The idea of idiosyncrasy of morphology refers to language specific ways of encoding various grammatical meanings. Some languages rely on more or less extensive case marking paradigms to show relationships between elements of the utterance (German, Dutch, Russian, Czech, and Finnish among many others) while others have no overt case system in place (Chinese, Arabic, Japanese); some languages have a system of grammatical gender (French, Italian, German, Polish), but others do not (English, Korean). These intricacies and inconsistences across languages complicate the study of morphology. Morphological attrition explores changes in inflectional and derivational morphemes and their combinatory possibilities (Anderson, ) in language production and perception. Researchers who investigate morphological attrition often catalogue attriting morphological features found in their data and attempt to account for the observed
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
variations in word formation via linguistic and cognitive theoretical perspectives. Processes commonly described in morphological attrition are simplification, paradigm levelling, overgeneralization, and reduction of the number of available morphological markers. These processes that result in morphological errors are often accounted for by ‘mapping problems’ between syntax and morphology (Lardiere, ); by cross-linguistic interference where the new dominant language affects the morphological patterns of the attriting language (Gross, ; Bolonyai, ); and by retrieval problems (Paradis, ) among others. The focus of this chapter is on providing a connection between these morphological processes and the theoretical frameworks that attempt to account for them. I will first briefly review studies of attrition of derivational morphology and then discuss research on inflectional morphological loss from the perspective of a number of theoretical frameworks, including the Markedness Theory, the Regression Hypothesis (RH), the Activation Threshold Hypothesis (ATH), the Interface Hypothesis (IH), and the -M Model.
. A
.................................................................................................................................. Multiple studies on morphology have demonstrated the importance of morphological structures in the organization of mental lexicon (e.g., Marslen-Wilson et al., ; Longtin & Meunier, ; Ghazel, ). Ullman (b) proposed a model of word processing that distinguishes between whole-word and decompositional processing routes that affect processing of some derivational affixes differently than inflectional ones. However, few studies research attrition of derivational morphemes. Dressler’s () work indicates that lack of derivational productivity in one’s L is symptomatic of language attrition. Dressler identifies two main functions of derivational morphology: lexical expansion—that is, creation of neologisms; and cognitive economy— that is, speakers’ ability to create complex word forms following morphological rules, rather than storing complex lexical items in memory. Dressler points out that while the production of neologisms may be lost, attriters are nevertheless able to process neologic creations by applying morphosemantic rules of their L. In their study of the attritional development of Michal, a fluent Hebrew speaker who was immersed in an English-speaking environment at a young age, Kaufman & Aronoff () note that once the initial ‘disintegration’ of L morphology sets in, it affects both inflectional and derivational processes. They claim that the child misuses the derivational templates of Hebrew under the influence of English after twelve to twenty-four months in the L environment, which points to her ‘loss of morphosemantic distinctions’ (p. ). They further specify that the child’s derivational system is reduced to a single form that remains productive and attests to her knowledge of the Semitic verb system. This study only focuses on the production of derivational morphology, not its processing. Intriguing distinctions between relative stability of derivational prefixes and relative volatility of inflectional suffixes used as markers of aspect in Russian have been pointed
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
out for heritage users of Russian (e.g., Pereltsvaig, ; Polinsky, b; Mikhaylova, ), but research in the population that experiences attrition is lacking.1 From the limited study of attrition of L derivational morphology it appears that derivational processes diminish in productivity with the progression of language loss, but remain available for recognition and processing, at least in adults. Clearly, there is a need for more studies in this area from a variety of languages and participants of different ages.
. A
.................................................................................................................................. The vast majority of studies attempt to account for difficulties that attriters experience with inflectional morphemes and variable susceptibility of these morphemes to erosion. Many of the studies on morphological attrition focus on the idea of simplification (e.g., Seliger & Vago, a) and its forms, including reduction and levelling of morphological paradigms. Simplification predicts that a once complex system will become less so in the course of language change due to attrition. While this conceptualization reflects the observed surface changes in morphological use, it does not explain either the original complexity of the items or why there was a need to simplify them. Therefore, simplification studies often rely on the ideas of theoretical frameworks to bring about the explanatory power to attrition data. The most productive of these frameworks for morphological attrition are the Markedness Theory (Sharwood Smith, ), the RH (Jakobson, ), the ATH (Paradis, ), the IH (Sorace, ), and the -M model (Myers-Scotton, ).
. M T
.................................................................................................................................. The notion of markedness that originally stood for a binary opposition of basic versus nonbasic elements in the language (Jakobson, ) has been used widely across linguistic fields, including morphological analysis. Sharwood Smith () aptly synthesizes notions of markedness borne out of different theoretical approaches. He proposes that a form is marked if it ‘has more structure than its unmarked counterpart’ (p. ); if it ‘requires more rules or possesses more informational content than an unmarked form directly associated with it’ (p. ); if it ‘occurs less frequently in the world’s languages’ (p. ); if it ‘required evidence in the input for its adoption by the learner’ (p. ).
1 The debate about generalizability of linguistic consequences in the populations of L attriters, incomplete acquirers, and heritage speakers has been complicated and is discussed in detail in Part VI of this volume. This chapter focuses on L attrition studies as much as possible, but relies on examples from heritage language research when attrition results are not available.
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
Greenberg’s views of markedness () in terms of frequency of use and his suggestion that unmarked categories could be determined by ‘the frequency of association of things in the real world’ (p. ) are of a particular interest for attritionists. The idea of frequency of use has appealed to many researchers as the primary feature that predicts the degree of stability of a morphological marker in the attriting L. For example, Caruso () in her investigation of verbal morphology in Italian adults in Australia reports that speakers produce less marked verb tenses with fewer errors than more marked forms. Importantly, the remaining morphological markers of the unmarked tenses—Presente and Imperfetto, which are most frequent in the participants’ production—are used in a target-like fashion. Albirini & Benmamoun () find that the ‘crosslinguistically marked and infrequent category’ (p. ) of dual nouns presents a problem for attriting speakers of Egyptian and Palestinian Arabic in an English dominant environment while the formation of plural nouns is less problematic due to their less marked status. Similarly, in her study of Greek-English bilinguals, Pelc () found that marked nominal morphological features undergo regularization to the unmarked forms. Importantly this finding affects not only production tasks, but also grammaticality judgement tasks. For example, Greek-American attriters judged non-target-like gender on nouns and their corresponding articles as correct at significantly higher rates (%) than the Greek monolingual participants. The same is true for judgements of ungrammatical substitution of accusative for nominative case in Greek. In a similar vein, El Aissati () reports instances of paradigmatic levelling in marked plural formation in Moroccan Arabic. His participants regularized plural assignment and reduced the number of possible paradigms in their repertoire. Dressler () examines attrition of forms produced via suppletion versus suffixation. He finds that suffixation as a less marked process continues to be used more productively by attriters than umlaut in Breton-French bilinguals and explains it by applying the diagrammaticity parameter, which ascribes more naturalness to suffixation. As a result, less regular suppletion is replaced by unmarked suffixation. Schmid () demonstrated that German Present tense markers remain most stable among German tenses in attriters of German as L. Moreover, her study showed that the Present tense is often used with the past reference—that is, it replaces the Past tense markings. Schmid accounts for such substitutions and the stability of the Present tense in attrition by the unmarked status of the Present as it is derived from the verb stem and is not marked with any additional affixes (p. ). From these studies it is evident that morphological elements are inherently marked or unmarked in a language. Marked items are characterized by more internal complexity and less frequency of occurrence in input and production, which leads to their relative instability in L attrition that results in their simplification, regularization, and/or levelling of paradigms. Therefore, the Markedness Theory accounts for morphological attrition from the perspective of inherent features of morphemes.2 Contrary to these findings within morphological attrition, markedness studies in the domain of syntax point to more stability of a marked value in L. According to Platzack (), ‘the native speaker does not forget the marked values of his mother tongue’ (p. ). Similarly, Sharwood Smith & Van Buren () maintain that the marked values of L persist due to lack of evidence for resetting of parameters (p. ). 2
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
. R H
.................................................................................................................................. The RH, first formulated by Jakobson (), accounts for the order of L attrition as the inverse process of language acquisition. If supported by morphological attrition studies, the RH would specify that morphemes have a differential rate of attrition that can be predicted by the order of their acquisition in L. This, in turn, would provide additional explanatory insights into the nature of morphemes and their complexity for acquisition, maintenance, and attritional selectivity. The RH relies on the following two rules: (a) what is acquired first is lost last (Ribot’s rule; Ribot, ); and (b) what is most frequently used (regardless of the time of acquisition) will be least vulnerable to attrition (Pitres’ rule; Pitres, ) (Berko-Gleason, ). Many researchers find this hypothesis intuitively attractive as it captures morphological processes observed in L attrition (Montrul, ). However, only a few studies have tested the theory within the domain of morphological attrition in part due to methodological difficulties. As Montrul () points out, a minimum of two conditions must be met in order to test the RH: (a) the participants must be complete acquirers of their L; and (b) testing must involve at least two structures, one of which was acquired prior to the other (p. ). The results of the research that has examined morphological attrition from the regression perspective are inconclusive. Several studies that relied on the data obtained from young children (and therefore did not meet the condition (a) of complete L acquisition) found limited support to the RH (e.g. Bahrick, a; Cohen, ; Olshtain, ; Kuhberg, ). Most of these early studies maintain that morphological features that exhibit optionality during the acquisition period are more likely to be lost. For example, English irregular plural forms that are acquired late are lost early on by Hebrew-English children (Olshtain, ); in German the dative case marker remained less stable than accusative that precedes dative in the acquisition process (Kuhberg, ). On the other hand, Jordens et al. () and Schmitt () found that generally case markers remain rather robust and resistant to attrition despite their late acquisition compared to other agreement features like gender, for example. In the same vein, Håkansson’s () research on morphological attrition of L Swedish in expatriates showed no evidence that Swedish was attriting in the sequence of its acquisition. More recent studies in L attrition (e.g. Schmid, ; Keijzer, , , ), adhere to both conditions of testing the RH and provide some evidence for regression. Keijzer (, ) examined the attrition of noun and verb phrase morphology in Dutch immigrants in Canada. She determined that the vast majority of morphological features of Dutch that she investigated, including plural, diminutive, tense, and other markers, ‘revealed mirror symmetries between the attriters and acquirers at the time of testing’ (p. ). Schmid () in her study of German attriters in Canada found that markers of plural morphology that are acquired at a relatively late stage in German exhibit more vulnerability to attrition than those that are acquired early on, like gender (p. ). However, the same participants continue to use all German cases (case markings are acquired fairly late) in all contexts (with the exception of the genitive case). Schmid concludes, ‘there is no overall reduction of the case system where the dative is lost first and a binary system emerges, as the regression hypothesis predicts’ (p. ).
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
To summarize, the RH in its application to morphological attrition mainly focuses on the order of morpheme acquisition as an explanatory tool. The results of morphological analysis in language attrition from the RH perspective are mixed, which points to a less straight-forward correlation between acquisitional and attritional processes at least in the domain of morphology and calls for further investigation of the differential stability of morphemes. Thus far, as Keijzer (a) points out, ‘the fact that evidence for the regression hypothesis was found is not enough for it to serve as an explanatory framework’ (p. ) for morphological attrition.
. A T H (ATH)
.................................................................................................................................. In the theoretical accounts of morphological L attrition discussed thus far the focus has been on inherent markedness of morphemes and the order of their acquisition in determining selectivity of loss. The ATH proposed by Paradis (, , see also Köpke & Keijzer, Chapter , this volume) shifts the focus to frequency, and recency of use, and the related inhibition of the competing items from the involved languages discussed in conjunction with the processes of memory and forgetting. The ATH approaches attrition from a neurolinguistic perspective as it predicts the item vulnerability to attrition based on the amount of neural energy necessary to retrieve it from memory. Paradis () maintains that the level of excitation—i.e., the activation threshold of an item—is regulated by the recency and frequency of its use. A higher activation threshold leads to more susceptibility to erosion. The ATH also tackles the processing mechanism involved in attrition. Specifically, it identifies inhibition as the process of ‘blocking’ competitors of the item that is to be activated (Paradis, ). Green () hypothesizes that speakers ‘ . . . exert control by activating and inhibiting tags at the lemma level’ (p. ). In other words, when the speaker chooses to speak L, the competing items from L are inhibited. Each instance of inhibition raises the activation threshold for future retrieval ‘since it takes time for the effects of prior inhibition to be overcome’ (Walters, , p. ). Conversely, when items are not inhibited their activation threshold is not affected (at least by this factor). To summarize, the ATH predicts that the less frequent of two competing forms will be more susceptible to attrition through inhibition and higher activation threshold. Attrition is the result of long-term lack of stimulation. Intensive use/exposure to one of the languages in a bilingual environment leads to a lower activation threshold for that language (i.e. it requires fewer resources), even in early, fluent, behaviourally balanced bilinguals. (Paradis, , p. )
Relatively sparse research in L attrition of morphology from the position of the ATH attempts to identify specific language forms that are more susceptible to attrition based on frequency of activation and inhibition. The results of the existing studies are inconclusive. For example, the findings of Gürel (a, ) and Schmitt () support the ATH, whereas Köpke () and Schmid (, ) report results that are contrary to the ATH predictions. Gürel’s studies (a, ) rely on the ATH to examine the vulnerability of Turkish pronominals o ‘s/he’, kendisi ‘self ’, and pro to attrition determined by their activation threshold relative to the corresponding L properties that they compete with (Gürel, a,
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
p. ). The only pronoun that has a competing form in English is o ‘s/he’, whereas the reflexive and pro do not compete with English. From Gürel’s analysis it appears that Turkish o is most affected by attrition since the frequent use of English s/he which overlaps, and is congruent with, Turkish o necessitates its inhibition. Consequently, the Turkish pronoun becomes less accessible and is more likely to undergo erosion. On the other hand, since there is no congruency between Turkish kendisi and pro and English pronouns, the competition between them is low, Turkish lemmas are not inhibited and their activation threshold is not affected by inhibition (but continues to be affected through non-use). This results in their relative stability in the speaker’s attriting language. Schmitt () also found supportive evidence to the ATH in her study where she observed the loss of Russian content morphemes and inflectional morphology. The ATH is helpful in explaining decreased vulnerability of Russian content morphemes related to everyday domestic topics on the basis of high frequency of their use, lack of competition with English (since all the participants communicate about those topics only in Russian) and consequently, their low activation threshold. For inflectional case markers, it was determined that the Nominative case which is most frequently used in Russian is least affected by attrition. The Instrumental case shows both the smallest number of the required contexts—that is, it is used least frequently among the six Russian cases—and the highest degree of attrition due to elevated activation threshold. The study concludes that ‘the number of the required contexts and the number of target-like items in production exhibit an almost one-to-one correlation’ (p. ) thus supporting the premise of the ATH. Additionally, Iverson’s data () indicate that his participant Pablo experienced attrition only in the areas where Spanish and Brazilian Portuguese did not overlap. This conclusion further supports the ATH, though there is no specific reference to morphological markers in the study. However, the ATH does not find ubiquitous support in all morphological attrition research. Schmid () suggests that frequency and recency of use do not have as salient an impact on activation thresholds and item accessibility as does inhibition. The results of her investigation of attriters of German in two different contexts indicate that there is little or no correlation between the level of attrition and frequency of L use. Schmid proposes that the quantity of L use is an important predictor of L maintenance only until a ‘saturation point of rehearsal’ (p. ) is achieved and knowledge is stabilized. After this point, frequency may not play as important a role in language accessibility as previously assumed. It is the process of inhibition that will affect the activation threshold the most. In the same line of research, Altenberg () studied the attrition of German gender and determined that word frequency played an important role in the maintenance of gender distinctions (her study, however, was not based on the ATH per se, but did invoke the key component of the ATH—frequency of use). Schmid’s () data of German gender attrition did not confirm the predictions of gender vulnerability based on frequency of use and reinforcement (p. ). Overall, morphological studies grounded in the ATH provide further insights into the connection between frequency of use and inhibition of morphemes in the attriting language and attempt to explain cognitive processes that underlie morphological decline. However, many questions remain unanswered. For instance, what is the level of frequency of use that is needed to avoid an increase in the activation threshold of an item? What features of a morpheme in fusional languages are most likely to compete across languages?
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
At what point does inhibition trigger the onset of morphological optionality? Why attriters continue to comprehend their L morphology, when they are no longer able to produce it due to high activation threshold? New studies may help find solutions to these issues and shed more light on cross-linguistic mechanisms of morphological accessibility.
. I H (IH)
.................................................................................................................................. The IH originally developed to account for non-target-like usage in adult second language acquisition (Sorace, , ) and fruitfully applied in heritage language and L attrition studies (e.g., Tsimpli, ; Montrul, ) moves away from the earlier discussed concepts of markedness, order of acquisition, and frequency and inhibition (see also Chamorro & Sorace, Chapter , this volume). Rather, the IH accounts for the differential susceptibility of items to erosion by the types of linguistic interfaced involved in their production. It predicts that linguistic properties that are dependent on interfaces between language and external domains of cognition are more vulnerable to loss than those that are internal to the linguistic system. The strong version of the IH (Sorace & Filiaci, ; Sorace, ) identifies external interfaces as language modules that operate ‘between syntax and cognitive systems outside of formal grammar’ (Van Osch et al., , p. ). External interfaces are claimed to put higher processing demands on bilinguals due to the need to pull information from two or more systems, for example discourse and syntax, or syntax and pragmatics. This puts language items that operate in the external interfaces at a greater risk of attrition (Montrul, a; Sorace, ). On the other hand, internal interfaces are language modules that operate within linguistic domains such as syntax and semantics, syntax and morphology, and morphology and semantics (Montrul, a; Sorace, ; Iverson, ). Since linguistic items do not have to ‘travel’ outside of their own ‘territory’, they bear fewer processing costs and are predicted to be more stable in bilingual production (Pires & Rothman, ; Sorace & Serratrice, ). To date, only several studies focus on attrition of L morphology from the perspective of the IH. Van Osch et al. () examine gender agreement in Spanish-Dutch bilinguals comparing the difference in the level of weakening of gender stability at the internal interface (adjectival predication) and the external interface (pronominal reference). The results show that both Spanish heritage speakers and attriters in the Netherlands make significantly more gender agreement errors in pronominal reference than in the adjectival predication. The authors account for this difference from the perspective of the IH: adjectival predication—that is, subject–predicate agreement—takes place within the internal interface of syntax and morphology, and remains more stable than the pronominal referencing that involves discourse considerations and takes place at the external interface. This study shows that it is not an individual item from L that becomes vulnerable in attrition, rather it is the context and the amount of processing determined by the interface that account for optionality of linguistic items in an attriting language. A recent study by Chamorro, Sturt, & Sorace () explores the effects of L attrition on Spanish personal preposition a, the use of which does not depend on context, but is determined by the animacy and/or specificity of the direct object. Importantly, the use of the preposition is governed by the interaction of syntactic and semantic properties—that is, internal interface. The results of the online processing and offline interpretation
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
experiments clearly show that this structure has not undergone L attrition in either condition. This result yields strong support to the IH. Other research in L attrition proper deals with syntactic, not morphological changes and supports the predictions of the IH (e.g., Chamorro, Sturt, & Sorace, ). Several heritage language studies also focus on the explanatory power of the IH for morpho-syntax. Their results are not as uniformly supportive of the IH as the attrition research (see, e.g., Montrul, a, ). Clearly, more studies of the attrition of morphology in L that are grounded in the IH would help establish the prognostic and explanatory value of the IH. Current research in this area does not account for the widely attested significant decreases in morphological stability of the attriting L (e.g., Gross, ) in various populations despite the expectation that internal interfaces are less vulnerable to attrition. Thus research that can help explain morphological selectivity in attrition from the IH perspective would look at the following dimensions of morpheme behaviour: morpho-syntactic or internal interface (structural case markers); morpho-semantic internal interface (gender markers, lexical case markers); morphemic-discourse external interface (gender-dependent references, aspect); and morphosyntactic–discourse interface (fusional markers that include case, gender, number, mood).
. T -M M
.................................................................................................................................. Unlike previously discussed theoretical frameworks that can and have been applied to various domains of linguistic knowledge, the -M model was developed with the purpose of capturing morphology in bilingual behaviour. The -M model focuses on the differential distribution and contribution of four types of morphemes—content and early system morphemes that carry the speaker’s intentions, and two types of late system morphemes—bridge and outsiders which encode grammatical relations (Myers-Scotton, ; Myers-Scotton & Jake, ) as illustrated in example (): ()
She plays with dolls of her friend.
Content morphemes in () are conceptually activated and are responsible for conveying semantic and/or pragmatic meaning: she, play, her, friend, and doll. Early system morphemes are selected and conceptually activated by their content morpheme heads to make the meaning of the utterance more specific and to ‘contribute to the realization of semantic and pragmatic intentions of the utterance’ (Myers-Scotton & Jake, , p. ). In () the plural marker -s is an example of an early system morpheme. Late system morphemes (bridge and outsiders) are structurally assigned and are used to build larger units. Bridge morphemes connect phrases or clauses to indicate that they are related. In () of is a bridge morpheme. Finally, ‘outsiders’ look outside of their immediate projection to get assigned. They determine structural relations between constituents and include markers of agreement, case, etc. In () the third person singular marker on the verb -s is an outsider morpheme. Importantly, the distinction between conceptual and structural activation of morphemes (Myers-Scotton & Jake, ) provides an insight as to why they behave differently in attrition and further explains why language items may belong to different types of morphemes in different languages even if they come from the same lexical
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
category. This is exemplified by personal and dummy pronouns in English (e.g., she vs. it/ there). Content personal pronouns are conceptually activated by the speaker in the production process. They are acquired early on in first and second languages and are used in a target-like fashion more often than the structurally assigned morphemes, such as dummy pronouns that are structurally activated (Li, ). The framework created by the -M model has been used in several studies of L attrition. Gross () finds that in German-English attrition ‘content morphemes are most vulnerable to attrition while late system morphemes are the least likely category to undergo attrition’ (p. ). Schmitt (, ) studies the loss of morphological case in RussianEnglish attriters and reports that case markers ‘do not erode, rather it is the contexts where these markers are required become less clear to the informants’ (p. ). Comparing late system morphemes to early system morphemes, Bolonyai () finds that in L attrition of Hungarian late system morphemes such as syntactic case markers like accusative are less vulnerable to loss than early system morphemes like local semantically assigned cases. Fuller () reports that early system morphemes (e.g., English plural markers) appear before late system morphemes (e.g., English third person singular marker) in convergence from the attriting to the dominant language. Thus, there seems to be a growing amount of evidence that structurally assigned morphemes demonstrate less vulnerability to attrition than lexically activated morphemes. In addition, the stage and complexity of activation of morphemes at the level of abstract lexical structure (based on Levelt, ) can help explain why certain pieces of fusional morphology may be lost before others. Bolonyai & Dutkova-Cope () examined differential susceptibility of agreement features, such as person, number, and gender all marked by a single fusional inflection in Hungarian-English and Czech-English attriters. They found that person and number—the late system outsider morphemes in the -M Model classification—remain particularly robust in both languages. Gender, identified as an early system morpheme appears to be significantly more vulnerable to attrition in their participants. The researchers account for this distinction through both cross-linguistic incongruence and the complexity of computational procedures that are involved in the production of these morphemes. It appears that person and number ‘mark the most elementary and universal predicate-argument relations in all three languages that involve the least complex computational procedures and whose access does not interact with semantics-pragmatics’ (p. ) during activation. In other words, as late system morphemes, number and person involve structural activation (Myers-Scotton, ), or in terms of the IH discussed in the previous section, they function at the internal interface. On the other hand, the activation of L-specific gender features evokes more complex processes. Specifically, as an early system morpheme on the noun, the gender feature is first conceptually activated. In order to carry out the agreement function between the subject and the verb, it has to be also structurally assigned as an outsider morpheme on the verb. Thus, gender agreement involves a complex interaction between the two types of activation—conceptual and structural. This complexity along with the uniqueness of gender features and lack of gender agreement congruity between the attriting languages and English explains the instability of gender in Hungarian and Czech as compared to person and number. Just like with the other models discussed here, studies that employ the -M model in the investigation of language loss are sparse. However, the results of the existing research seem to have an explanatory promise. It seems that there is some consistency in identifying
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
structurally activated outsiders as more stable in language attrition. It would be of interest to determine if there is a difference in the order of acquisition of structurally and conceptually assigned morphemes. If such a difference were found, it would be compelling to compare the order of acquisition to the order of attrition based on the type of activation and compare those results to the predictions of the RH. Moreover, the distinction between conceptual and structural activation of the morphemes that seems to play a role in their vulnerability to attrition could be correlated to the types of interfaces involved in their production as discussed in the IH. In addition, since the -M model is the only framework that focuses specifically on bilingual morphology, it is essential to continue investigating how different types of morphemes fair in language loss of a variety of languages in different populations.
. C
.................................................................................................................................. This review of studies on morphological aspects of language attrition indicates two main directions in accounting for differential loss of L morphemes: a. Research focused on inherent properties of morphemes: • More marked morphemes are more vulnerable to attrition, whereas less marked morphemes remain more stable in the attriting L1 (the Markedness Theory) • Structurally activated morphemes are less susceptible to attrition (the 4-M model) b. Research on mechanisms that underlie the processing of morphemes: • Frequency of use and inhibition play a significant role in morphological loss (the ATH hypothesis) • Complexity of morpheme access determines their susceptibility to loss (the IH and the 4-M model) Inherent properties of morphemes may vary from language to language. They have a predictive, but not explanatory power. Mechanisms of morpheme processing, while still minimally understood, have the potential to explain why certain morphological properties lead to more or less vulnerability in attrition and consequently to a better understanding of how linguistic systems interact in both acquisitional and attritional processes. Challenges that morphological studies face include: a. Inconsistency in the obtained results within and across language data sets in all theoretical frameworks discussed above. Examination of the same data sets from the perspectives of different theories as well as collection of more morphologically focused attrition data are ways to enhance current limited understanding of morphology in bilingualism. b. Lack of uniform data collection techniques and methodologies, which makes the comparison of the results exceedingly difficult. Much of the data are collected through free speech production, some data are gathered using online grammaticality judgement tasks, others rely on offline tasks, with or without electroencephalogram (EEG). Schmid (a) provides an exhaustive list of all available and tested data collection
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
methods. For comparative and replication purposes it would be ideal for researchers to rely on these methodologies and use more than one in order to triangulate their results and design replicable studies. c. Lack of a principled agreement on the types of speakers that should be investigated. Data are sometimes collected from different categories of speakers—immigrants and heritage speakers of different ages. This is a complex issue that has been addressed in this volume. What is particularly relevant to issues of morphology is that morphological incomplete acquisition should not be discarded as aspects of morphology that are not acquired by heritage speakers may overlap with the areas of attrition. In this case morphological vulnerability could be understood and explained from a multidimensional perspective. Morphological attrition seems to have different rates among children and adults. Findings in this area could help clarify the connection between the rate of morphological acquisition and erosion. d. Limited number of languages that are researched in the context of attrition. The majority of data come from Romance, Germanic, and Slavic languages. There are few studies on Arabic, Turkish, Hungarian, and Hebrew morphological attrition and virtually no research that involves languages of other typological groups. e. Only one of the five theoretical frameworks is specifically designed for the investigation of morphology. Additional morphology-centred models, including the Distributed Morphology models have a potential for more explanatory and predictive views of attritional processes. The Distributed Morphology model (Halle & Marantz, ) provides an account of lexical primitives combining into complex structures and the operations that apply during the post-syntactic computation to the phonetic form. A particularly useful feature of the Distributed Morphology theory for attrition studies is its hierarchical view of morphological operations that follow syntactic processes. This hierarchical view of morpheme selection and its dependency on syntax rather than lexicon may promote innovative explanations to differential vulnerability of gender, plurality, and case markers, for example. Thus far I have not been able to find any work in L attrition from the perspective of Distributed Morphology model. However, pioneering work in childhood bilingualism (Austin, ) and in code-mixing (Alexiadou, Anagnostopoulou, & Schäfer, ) have laid some foundation for future research in L attrition. This review of research on morphological L attrition has demonstrated that there is no one theory that accounts for the complexity and variety of morphological operations and morphemes that get restructured, modified, and ultimately lost exhibiting varying vulnerability to language attrition. Therefore, in addition to streamlining data collection processes, it is essential to apply multiple frameworks to the observable morphological changes in order to come closer to an understanding and explanation of morphological operations.
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
......................................................................................................................
......................................................................................................................
. W ?
.................................................................................................................................. L attrition has often been described as a process that leads to a person’s loss of the ability to retrieve words from memory (Stolberg & Münch, ; Isurin, ), if not the complete erasure of those words from the person’s mental lexicon (Pallier et al., ). Although attrition is often equated with loss, there is nevertheless a growing tendency in the literature to identify loss as only one of many possible manifestations of attrition. According to Meara (), ‘vocabulary loss is an observable change in the number of activated words in a vocabulary, and will always be measurable’ (p. ), whereas attrition involves small structural changes that weaken a person’s mental lexicon—and can eventually result in loss—but are not always observable in the person’s language performance. First language (L) lexical attrition as an observable phenomenon is now widely understood as a collection of undesirable changes to a person’s L lexical skills that tend to accompany cases of inter alia international adoption, emigration to a new language environment, reduced use of or isolation from the L, second language (L) acquisition, bilingualism, multilingualism, and various pathological conditions that affect word use (Jaespart & Kroon, , p. ; Goral et al., ; Bardovi-Harlig & Stringer, ; Schmid & Jarvis, ). Vocabulary loss is of course the most extreme manifestation of lexical attrition, and some of the clearest examples of this in the realm of L attrition pertain to cases where a person begins acquiring one language in infancy but is adopted during early childhood into a home and community where a different language is spoken, and where he or she is completely isolated from the L. Isurin & Seidel () examined the case of a woman referred to as TJ, who was born in the United States to a mother who is believed to have communicated with TJ in Russian until she was years old. At that point, TJ was taken from her mother and eventually adopted by a family who raised her and spoke to her exclusively in American English. After being taken from her mother, TJ quickly lost her L and soon even became unaware that she had spoken a different language prior to her adoption. This changed when she began treatment for post-traumatic stress disorder at the age of . During her therapy, she was able to remember previously forgotten details about her childhood,
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
including a handful of foreign-sounding words whose meanings she could not remember. Some of the words were identified by the researchers as being Russian (e.g., angliiski ‘English,’ boisja ‘afraid’, podruzhka ‘friend’), and the rest were also phonologically compatible with Russian, but their forms had apparently become distorted in TJ’s memory. Evidence that TJ had known Russian in her early childhood was found in the results of a study where she was retaught Russian words that she would most likely have known at the age of (e.g., korobka ‘box’, ložka ‘spoon’, igraj ‘play’, kušaj ‘eat’). TJ showed a clear advantage in learning these words versus other Russian words that she probably would not have known at such a young age (e.g., kletka ‘cage’, korabl’ ‘ship’, čihaj ‘sneeze’, klej ‘paste’); a comparison group of native English speakers did not show any advantage in learning the former versus the latter set of words. Other studies have also documented cases of what at the level of observation have appeared to represent total or near-total loss (including the loss of comprehension) of a language acquired in childhood; such studies have usually involved international adoptees—some as young as months (Nicoladis & Grabois, ), and some as old as years (Isurin, ) (for an overview of such studies, see BardoviHarlig & Stringer, ; Schmid & Jarvis, ; Isurin & Seidel, ). Besides a complete loss of the ability to produce and, in extreme cases, even to be able to comprehend words that the person previously knew, lexical attrition can in milder cases result in word-retrieval difficulties, where attriters usually do eventually recall the words they need for successful communication and task completion, but many of these words are retrieved from memory only after an effortful and relatively time-consuming mental search. According to Ecke (), ‘retrieval slowdown and failure (mostly reported for lexical production, but also observed in reception) are among the first and strongest indicators of language attrition’ (p. ). Such problems are clearly a matter of access rather than of the loss of memory representations from the person’s mental lexicon, and in language production they result in various forms of disfluency, including the frequent use of either filled or empty pauses (e.g., Das Mädchen wird ah freigelassen ‘The girl is uh released’; und # dann hat er # bisschen (he)rum gekuckt ‘and # then he # looked around a bit’; Schmid & Beers Fägersten, , p. ), repetitions (e.g., als de als de vrachtwagenchauffeur weer naar buiten komt ‘when the when the truck driver comes back out’; Schmid & Beers Fägersten, , p. ), self-corrections (e.g., hat den Brief gekriegt hat den Brief bekommen ‘got the letter received the letter’; Schmid & Beers Fägersten, , p. ), lexical substitutions (e.g., referring to a cliff as a precipice, hill, or mountain; Olshtain & Barzilay, , p. ), circumlocution (e.g., saying The dog goes into this little river, body of water, whatever it is, but it’s very shallow, and they’re safe when referring to a pond; Olshtain & Barzilay, , p. ), or even borrowing from or switching into another language (e.g., ich hab hier eine Bekannte die war sehr sehr ## involved ‘I have a friend here who was very very ## involved’; Schmid, a, p. ). Importantly, as this last example shows, multiple forms of disfluency associated with lexical attrition sometimes occur in tandem. All examples of disfluency in the preceding paragraph come from the freestyle L oral production of attriters who grew up in an L environment and then later emigrated to another country, where they acquired the dominant language of that country and used it in their daily lives for many years. Even though disfluencies such as hesitations, pauses, self-corrections, circumlocutions, and so forth, are common even among L speakers who are not experiencing lexical attrition, they are significantly and sometimes considerably
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
more prevalent in the speech of emigrants than they are in the speech of people who have remained in their home countries (Schmid & Beers Fägersten, ). I will discuss factors related to emigration and the reduced use of the L in Section .. In the meantime, it is important to note that disfluency-related effects of L lexical attrition have also been examined and confirmed in attriters’ reaction times in experimental tasks involving both lexical production (e.g., picture naming, verbal fluency) and reception (e.g., word recognition, picture–word matching, lexical decision). These methods will be described in Section .. For now, it is relevant to point out that many studies that have used these methods have indeed confirmed that attriters tend to recognize and retrieve L words more slowly than do comparable speakers of the same L whose life circumstances make them less susceptible to L attrition (see, e.g., Goral et al., ; Schmid & Jarvis, ; Isurin & Seidel, ). In addition to its effects on fluency, lexical attrition also affects lexical accuracy. Lexical accuracy can be measured in relation to the accurate oral or written production of word forms, the use or comprehension of words in a manner that indicates that the person has an accurate understanding of the meanings (including connotations and usage constraints) of those words, and the accurate production and comprehension of the syntagmatic relationships that can be found in multi-word lexical items. Examples of L lexical attrition affecting form accuracy include the earlier mentioned example of TJ’s oral production of Russian words in a way that made them unrecognizable to the researchers (Isurin & Seidel, , p. ) and cases where an L word becomes partially or fully integrated into an attriter’s use of the L, sometimes without the attriter even being aware of the cross-linguistic influence (e.g., a German-English bilingual saying also hier der Benjamin der hat die Marilyn *angegrowlt ‘imagine Benjamin [a dog] growled at Marilyn’; Schmid, a, p. ). Problems with form accuracy appear to be relatively rare in L lexical attrition, whereas semantic and syntagmatic problems are more common. Examples of lexicosemantic errors include both semantic extensions (e.g., a Finnish-English bilingual saying kattoo ‘look, watch’ instead of vahtia ‘watch, keep an eye on’ when referring to watching/baby-sitting children; Jarvis, , p. ) and loan translations (e.g., a German-English bilingual saying Vergiß es instead of laß es bleiben, lit. ‘let it stay’ when expressing the meaning of ‘forget it’; Seliger & Vago, a, p. ). Loan translations involve not only semantic but also syntagmatic relationships; syntagmatic problems among attriters include problems with compounds, collocations, idioms, and clichés (e.g., a Finnish-English bilingual saying *ottaa suihku ‘take a shower’ instead of the conventional käydä suihkussa ‘visit the shower’; Jarvis, , p. ). Bardovi-Harlig & Stringer () have pointed out that it is important not to neglect multi-word units or conventional expressions when investigating lexical attrition because there is strong evidence that they are stored—like words—in the mental lexicon, and because they reflect the complex networks that link lexical items in memory (pp. –). Beyond fluency and accuracy, lexical attrition can also be observed in relation to lexical complexity. Bulté & Housen () have defined complexity as a property or quality of a phenomenon or entity in terms of () the number and the nature of the discrete components that the entity consists of, and () the number and the nature of the relationships between the constituent components. (p. )
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
Although lexical complexity can be understood in terms of the properties of individual words—or what Bulté & Housen () call ‘the compositionality of lexical elements, that is, the number of formal and semantic components of lexical items’ (p. )—the tendency is to view lexical complexity in relation to the complexity of a person’s mental lexicon or lexical skills, as demonstrated by the nature and variety of words the person produces in samples of speech or writing. According to Bulté & Housen, lexical complexity at this more abstract level can be operationalized with measures that tap into the lexical diversity, lexical density, and lexical sophistication of texts. Lexical diversity refers to the variety of words found in a text (or sample of speech), lexical density refers to the proportion of lexical (or content) words in the text, and lexical sophistication refers to the proportion of rare or less frequent words in the text. Of the three abstract types of lexical complexity, lexical diversity has received by far the most attention in studies of L attrition, whereas lexical density appears not to have been investigated so far. Although some studies of L lexical attrition purport to assess lexical density (Negrisanu, ; Taura & Nakanan, ), the measures they use are indices associated with diversity (e.g., the type–token ratio) rather than measures of the prevalence of content words in a text. The findings of these and several other studies have shown, as expected, that L attrition does indeed lead to a decrease in the levels of lexical diversity that attriters produce in their speech or writing (for a summary of these studies, see Schmid & Jarvis, ). However, lexical diversity has nearly always been investigated using unidimensional measures of lexical repetition rather than multi-dimensional measures of redundancy and variety (cf. Jarvis, a). L attrition research that has investigated lexical diversity has relied almost exclusively on measures that also vary as a function of text length and evenness (i.e., how evenly the lexical tokens in a text are spread across lexical types) (see Schmid & Jarvis, ). The study by Schmid & Jarvis () was designed to overcome these shortcomings by using a combination of diversity indices, such as a measure of lexical repetition that does not vary with text length (i.e., the Measure of Textual Lexical Diversity, or MTLD), a measure of evenness, multiple measures of lexical rarity, and a measure of lexical dispersion (i.e., how evenly the tokens of a given type are dispersed throughout the text instead of clustering near each other). The results of a MANOVA showed that the German-Dutch and German-English attriters in the study did indeed have significantly lower levels of lexical diversity than the German control group. The results of an additional Linear Discriminant Analysis showed that the three groups of participants could be distinguished from each other with % accuracy based on a statistical model made up of eleven variables—ten of which were lexical diversity measures. (The remaining variable in the model was speech rate, measured as words per minute.) Some of the lexical diversity variables in the Schmid & Jarvis () study were measures of lexical rarity—or the relative frequencies (in German) of the words used by the different groups of participants. Jarvis (, a,b) has argued that lexical rarity is one of the components of lexical diversity and should be included in any complete, multi-dimensional model of this construct. Other researchers have treated lexical rarity as a separate construct, and have referred to it with terms such as lexical sophistication (e.g., Linnarud, ) or a text’s lexical frequency profile (Laufer & Nation, ). The amount of research available on the L attrition of lexical sophistication is quite sparse, but the findings of Schmid (a, pp. –) and Schmid & Jarvis () show that L attrition does indeed appear to lead to
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
an increased reliance on highly frequent words in the L (e.g., poor vs. destitute), which in turn tends to coincide with decreased levels of lexical diversity.
. E
.................................................................................................................................. Comprehensive overviews of the hypotheses developed to account for attrition can be found, among other places, in Schmid (), Köpke & Schmid (), Bardovi-Harlig & Stringer (), Schmid & Köpke (a), and in the first several chapters of this volume. I will not attempt to duplicate any of these excellent overviews here, but will instead briefly discuss which of these hypotheses have gained currency in relation to L lexical attrition, and which seem compatible with the patterns of lexical attrition described in the preceding section. Bardovi-Harlig & Stringer () outlined six hypotheses that have been proposed to account for L attrition. (These are the same hypotheses discussed by Schmid, , and Köpke & Schmid, , although they are labelled and described somewhat differently.) Of these six hypotheses, only three appear to be directly relevant to the patterns of L lexical attrition found in the literature and illustrated in the preceding section. These include—in the order I will discuss them—the dormant language hypothesis, the threshold hypothesis, and the interference hypothesis. Importantly, these hypotheses complement rather than compete against one another. The dormant language hypothesis (also called the savings paradigm or the residual memory approach) holds that attriters never completely lose their languages nor any part of them; rather, the parts that seem to have become lost are only dormant in the attriters’ minds and have the potential to be reactivated through interventions such as hypnosis or relearning (Isurin & Seidel, ). The example given earlier of TJ’s relearning of Russian words that she is likely to have known in early childhood does indeed seem to confirm that TJ retained traces of Russian words in memory even after thirty years of isolation from her L and even after living for nearly as long being unaware that she had ever spoken Russian. To date, Isurin & Seidel appear to be the only researchers who have tested the dormant language hypothesis in relation to L lexical attrition. Most of the evidence for this hypothesis comes from studies of L phonological (e.g., Park, ) or morpho-syntactic attrition (Au et al., ), or from studies of various types of L attrition, including L lexical attrition (e.g., Van der Hoeven & de Bot, ). Empirical evidence that is consistent with the dormant language hypothesis is also largely consistent with the threshold hypothesis—the most well-known version of which is the Activation Threshold Hypothesis (ATH). This hypothesis predicts that the mental representation of a word or other item of language requires a certain amount of neural impulses in order to reach activation (its activation threshold). Each time a language item . . . is used, its activation threshold is lowered, making it easier to activate again, but it slowly rises when inactive. (Paradis, , p. )
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
Like the dormant language hypothesis, the threshold hypothesis holds that memory representations for words and other features of language are not completely lost during attrition; rather, they simply become more difficult to access and retrieve after long periods of disuse, especially when they compete for activation with frequently and recently used L words. It is clear from the recent literature that the ATH is the most popular and the most empirically validated explanation of both L and L lexical attrition (see Bardovi-Harlig & Stringer, ). It also nicely accounts for a wide range of lexical attrition phenomena. For example, it accounts for both retrieval slowdown and failure (cf. Ecke, ), which in turn account for several of the types of disfluency described in the preceding section: (a) retrieval slowdown ➔ filled pauses, empty pauses, repetitions, self-corrections; (b) retrieval failure ➔ lexical substitutions, circumlocution, language borrowing, and switching. As pointed out by Schmid & Jarvis (), activation problems associated with lexical attrition can also result in reduced levels of lexical diversity and lexical sophistication. Both findings are predictable on the basis of the ATH: lexical diversity will decrease as more and more words become difficult or impossible for the attriter to retrieve from memory, and lexical sophistication will also decrease because the activation thresholds for individual words will be determined by how frequently (and recently) they have been encountered (see Paradis, and also the discussion in Köpke & Keijzer, Chapter , this volume). The ATH overlaps not only with the dormant language hypothesis, but also with the interference hypothesis, which ‘holds that attrition is directly due to the increasing influence of the newly dominant, competing language’ (Bardovi-Harlig & Stringer, , p. ). Different versions of this hypothesis highlight different ways in which the attriter’s developing L knowledge can affect his or her use of the L; the ATH includes one of its own versions of the effects of the L: ‘the most frequently used elements of L will tend to replace their (less used) L counterparts’ (Paradis, , p. ). Such effects of the L on L attrition are sometimes completely transparent, such as when an attriter unintentionally embeds an L lexical root into an L construction, or when the attriter uses L words during L communication without realizing that those L words do not exist in the L (e.g., Schmid, a, p. ). At other times, the effects of the L on L lexical attrition are so subtle that their detection requires experiments designed specifically for this purpose. For example, Segalowitz () used a primed lexical-decision task to show that bilinguals with higher levels of L proficiency access (or recognize) L words more slowly than comparable bilinguals who are less proficient in the L. Consistent with the ATH, the explanation for these results appears to be that bilinguals with higher levels of L proficiency, over time, have exerted higher levels of inhibition on the L lexicon, and this has led to higher activation thresholds for individual L words. According to Schmid & Jarvis (), ‘Each time a speaker selects a target item from the lexicon, its competitors . . . must be inhibited. This inhibition process raises the activation threshold, making the inhibited items harder to access subsequently’ (p. ) (see also Köpke & Keijzer, Chapter , this volume). Most of the types of L lexical attrition discussed so far in this section involve attrition in the domains of fluency and complexity, but it is important to note that the three overlapping hypotheses are also relevant to the domain of lexical accuracy. As described in the previous section, many of the lexical errors (including errors involving multi-word units) found in attriters’ use of the L are clearly the result of L influence (e.g., Jarvis, ). Others, such as TJ’s inaccurate recall of the pronunciation of some Russian words, however,
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
seem to reflect distortions to memory representations. Alternatively, they may reflect reduced—or partial—access to the phonological, orthographic, semantic, syntactic, or lexical network-related mental representations for words. This last possibility might best be explained by a combination of the dormant language hypothesis and the ATH. The effects of attrition on L lexical accuracy are certainly an area that requires further investigation, although it is important to reiterate that lexical errors not caused by L influence appear to be relatively rare in L attrition.
. F
.................................................................................................................................. Some of the factors that have been found to affect L lexical attrition are either explicitly stated or are implicit in the assumptions underlying the hypotheses described in the preceding section. Some of these factors are linguistic, some are social, and some are psycholinguistic. The linguistic factors include, first and foremost, the frequency and recency with which individual L words have been accessed—and the frequency and recency of the L counterparts of those words (Paradis, ). The frequency and recency with which people produce, encounter, or otherwise access either L or L words are extremely difficult to gauge, but several studies have indicated that attriters overuse (Schmid & Jarvis, ) or show a processing advantage for highly frequent versus less frequent L words (see Schmid and Köpke, ). Another important linguistic factor is how similar L words are to their L counterparts. Cognates with similar forms and meanings across languages are, expectedly, less likely to exhibit attrition effects, whereas words with similar forms across languages but different meanings are often detrimental to both fluency and accuracy (see Schmid & Köpke, , for a review of relevant studies). The most important social factors appear to be age and environment. Age is investigated in relation to the age at which the attrition-inducing condition begins, such as a drastic reduction in the person’s use of the L, adoption into a family that speaks a different language, or emigration to a new country where a different language is spoken. The results of past studies have shown that L attrition (including lexical attrition) is often dramatic in cases where attrition begins before puberty, and is much less so in cases where the onset of attrition occurs after puberty (cf. Köpke & Schmid, ; Isurin & Seidel, ). Recent research has also shown that word relearning improves with age (Van der Hoeven & de Bot, ). The nature of the attrition-inducing condition is of course also of primary importance in that it affects the frequency with which the L is used. Several studies have explored the effects of the amounts and types of exposure attriters have to the L, but, interestingly, the only measure of L use that has reliably predicted the degree of L attrition is ‘the use of the L for professional purposes: the more migrants use the L at work, the less L attrition they exhibit’ (Schmid & Jarvis, , p. ). Education is another social factor that one might expect to affect the maintenance of L lexical skills. However, although education does appear to mitigate L lexical attrition (see Bardovi-Harlig & Stringer, , p. ), the same has not been found for L attrition (see Schmid & Köpke, , p. ).
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
The relevant psycholinguistic factors are inseparably linked to both linguistic and social factors. Of particular relevance is the level of L proficiency—or the exact nature of the L lexical skills—the attriter had just prior to the onset of attrition. As predicted by the activation threshold hypothesis and the interference hypothesis, individuals with stronger and broader skills and experience in the L are less likely to have high activation thresholds for L words and are less likely to experience lexical interference from the L. However, this also depends on their levels of proficiency and experience in the L (cf. Schmid & Jarvis, ). Importantly, the effects of language proficiency on L attrition are largely confounded by the age factor: up to a certain age (perhaps around puberty), age and L skills co-vary. The fact that prepubescent emigrants and international adoptees are often found to show drastic levels of L attrition (see Isurin & Seidel, ) is perhaps often due to incomplete learning of the L combined with the early acquisition of greater levels of proficiency, experience, and dominance in the L than in the L (cf. Hyltenstam et al., ).
. M
.................................................................................................................................. Excellent overviews of the methods relevant for investigating L lexical attrition can be found, among other places, in Schmid & Jarvis (), Isurin & Seidel (), and especially Schmid & Köpke (). Here, I will not try to provide a comprehensive treatment of these methods, but will instead summarize some of the main methods and methodological considerations pertinent to the investigation of L lexical attrition in the domains of accuracy, fluency, and complexity. Traditional freestyle language production tasks, such as spoken or written film descriptions and oral interviews, are representative of natural language use and will always be relevant to the investigation of language attrition and other natural language processes. As I will describe in the following paragraphs, data collected through freestyle production methods are useful for exploring all three areas of proficiency—most especially accuracy and complexity. However, in all three domains— especially fluency—it is valuable to use controlled tests or experiments in addition to (e.g., Schmid & Jarvis, ) or instead of free production tasks. Free production tasks are of course irreplaceable for investigating lexical production accuracy, and it is important to choose tasks that are truly representative of the type(s) of language the attriter has experience with and is likely to produce again in the future outside of a research context. Analysis of the data should focus not just on the accuracy of individual word forms, but also on the makeup of multi-word units (Bardovi-Harlig & Stringer, ), as well as on word choice appropriateness. Lexical accuracy can also be gauged in receptive tasks, such as lexical decision tasks (Segalowitz, ) and picture-word matching tasks (see Schmid & Köpke, ), which can measure, respectively, word recognition accuracy and the accuracy of attriters’ form-meaning mappings. Fluency and fluency-related phenomena (e.g., retrieval slowdown and failure) have received more attention than any other area of lexical attrition—perhaps especially because this is the area that is most directly connected to the activation threshold hypothesis—and the methods for investigating lexical fluency are correspondingly broad and varied.
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
In analyses of free production data, lexical fluency is sometimes measured in relation to speech rate or the total number of words produced (e.g., Schmid & Jarvis, ). However, because lexical fluency is so sensitive to contextual and environmental factors, it is often best to measure it—additionally or instead—with a more controlled task of productive lexical fluency. This might involve controlled tasks of either semantic fluency or phonetic fluency, or both. Controlled tasks of productive semantic fluency include picture-naming tasks with reaction time measures, and especially verbal fluency tasks where participants are given semantic categories (e.g., animals, food, clothing) and asked to produce as many words as possible representing that category within a specified time period (e.g., seconds) (see, e.g., Schmid & Köpke, ). Controlled tasks of productive phonetic fluency, on the other hand, are difficult to find in the attrition literature (but see Guion et al., for an example of how they are used to investigate L acquisition), and it will be important for future research on lexical attrition to develop and make ample use of this potentially promising but severely underused method for measuring productive lexical fluency. Regarding receptive lexical fluency, this is often assessed with picture-word matching tasks and word recognition tasks by measuring how long it takes the attriter to respond. According to Schmid & Köpke (), however, the most useful method for measuring lexical access (and, by extension, receptive lexical fluency) is the lexical decision task, where participants are timed to see how long it takes them to decide whether a string of letters is a real word or not—often following a priming condition. This important method has been neglected in L attrition research (but see Segalowitz, ; and Goral et al., ). Finally, concerning lexical complexity, this can be measured in various ways (Bulté & Housen, ), but it seems most fruitful to do so with freely produced data that are as natural as possible, such as narratives of what the attriter has witnessed, or recorded conversations about the person’s own life experiences (e.g., Schmid & Jarvis, ). The data can be either written or spoken, but the latter has the advantage of being amendable to a broader analysis of attrition-related phenomena (e.g., speech rate, filled and empty pauses, repetitions). Lexical complexity can be investigated from the perspective of either lexical diversity or lexical sophistication, or both, and there is good reason to view both phenomena as closely related, if not part of the same construct (Schmid & Jarvis, ). It is also important to recognize the multi-dimensional nature of lexical diversity, and to use multiple measures that reflect each of these dimensions in the most valid way possible. It is crucial to take into consideration not just the number of types (different words) and tokens (total words) in the sample, but also which specific words occur in the sample, the ways in which they are related to one another, and how evenly they are dispersed throughout the sample (see Jarvis, , a,b). Above all, it is important not to rely on a single, unidimensional measure of lexical diversity—such as the type; token ratio—that measures only the overall proportion of repeated words in the sample and varies as a function of sample size (see Schmid & Jarvis, ).
. F
.................................................................................................................................. To conclude, L lexical attrition phenomena can be found not only in attriters’ loss of the ability to produce and sometimes even comprehend L words and multi-word units, but
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
also in the fluency, accuracy, and complexity of their demonstrated L lexical skills. The patterns of L lexical attrition that have been documented in past research are largely consistent with the predictions of the dormant language hypothesis, and especially the threshold hypothesis and interference hypothesis. L lexical attrition has also been shown to be affected by linguistic, social, and psycholinguistic factors. Linguistic factors include inter alia the frequency and recency with which L words and multi-word sequences have been encountered, and how similar they are in form, meaning, and syntagmatic structure to corresponding L lexical items. Social factors include the age at which attrition begins, as well as the contexts in which a person continues to use the L. Psycholinguistic factors include the level of proficiency the attriter had in the L before the reduced exposure began, and the attriter’s level of proficiency and range of skills in the L. There is a good deal of room for future fruitful research on L lexical attrition. It would be especially useful to see more longitudinal studies that track the beginning and progression of changes in a person’s L lexical skills involving attrition-inducing conditions such as international adoption and emigration to an L environment (e.g., Isurin, ; but see Schmid, , p. , for problems this poses for L attrition research). There is also a need for additional research properly designed to test the dormant language hypothesis (Isurin, ) and various versions of the interference hypothesis (see Bardovi-Harlig & Stringer, , p. ). Future L lexical attrition studies should also give increasing attention to multiword units and the lexical networks and syntagmatic relationships they entail (BardoviHarlig & Stringer, ). In experimental studies, it will be important to make use of valuable methods, such as lexical decision tasks, that have been neglected in past L attrition research but have the potential to shed a great deal of light on how attrition affects lexical accessibility (Schmid & Köpke, , p. ). Finally, future research on L lexical attrition will benefit greatly by giving more attention to the levels of lexical complexity, lexical diversity, and lexical sophistication that attriters demonstrate in freely produced samples of speech and writing. This will be especially useful if the tools used to measure these abstract lexical skills reflect their multi-dimensional character and the ways they are perceived by the human mind (Jarvis, a; Schmid & Jarvis, ).
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
......................................................................................................................
......................................................................................................................
̧ ̈
. I
.................................................................................................................................. A variety of linguistic phenomena related to pronominals has attracted much interest in the field of L acquisition. In this section, I will briefly discuss how this line of L research has developed over the years, subsequently paving the way for L attrition studies in s. One of the earlier theoretical motivations for the study of pronoun acquisition lies in a generative view, which analysed null and overt pronouns as parameterized Universal Grammar (UG) options. For decades, while theoretical linguists have investigated linguistic mechanisms licensing null and overt pronouns across languages (e.g., Chomsky, ; Rizzi, ; Jaeggli & Safir, ; Huang, ), the question of how they are acquired has occupied the agenda of UG-based generative L research (e.g., White, , ; Hilles, ; Phinney, ; Liceras, , ; see also Hyams, ; Hyams and Wexler, ; Rizzi, , for L acquisition). In early adult L acquisition research, the focus was initially on the acquisition of morpho-syntactic mechanisms licensing null pronouns (e.g., subject–verb agreement markers) in reference to cross-linguistic transfer and its directionality in cases where the L and L display the opposite setting of the null subject parameter (e.g., White, , ; Phinney, ; Liceras, , ). In this context, the notion of markednesss was discussed extensively as a factor determining the directionality of transfer. With respect to pronouns, the theory of markedness invoked in learnability-based models (Berwick, ) posits that overt (subject) pronouns are unmarked, because assuming that a language has only overt pronouns is the most conservative hypothesis (i.e., subset grammar) that the
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
̈
learner can initially entertain.1 Nevertheless, in adult L acquisition, L transfer could override the Subset Principle, as might be evinced by learners’ use of null pronouns in a non-null-subject L due to their L. More recent evidence supporting the default status view of overt pronouns has emerged on the basis of child and adult L data documenting the influence of non-null-subject Ls on null-subject Ls in the form of overuse and/or misinterpretation of overt pronouns (despite intact null pronouns in the L). Thus, overt subjects are considered the default option (e.g., Sorace & Filiaci, ; Belletti et al., ; Sorace & Serratrice, ). Moreover, the finding that overt pronoun overuse in null pronoun contexts is observed even in bilingual children with two null-subject languages (Sorace et al., ) also supports this point. As will be discussed below, overt pronoun overuse is also widely considered a primary symptom of L attrition. Subsequent L research on null and overt pronouns has emphasized that the L acquisition amounts to more than just establishing these forms are available in the L. The task also involves learning how pragmatic-discourse constraints govern the distribution of null and overt pronouns (e.g., Liceras, ; Pérez-Leroux & Glass, ; Gürel, ; Rothman & Iverson, ). As will be discussed in the next section, this issue has become relevant in L attrition research that examined the link between the use of null/ overt pronouns and word order variations in null-subject languages (e.g., Tsimpli et al., ; Domínguez, ). An additional theoretical motivation to study pronominals in L acquisition and subsequently in L attrition can be attributed to the fact that binding properties of pronominals (pronouns, reflexives, and reciprocals) had a central place in Chomskyan theorizing, particularly between early s and late s. The subtle differences in referential properties of pronominals have been analysed under Binding Principles A, B, and C, which are assumed to be part of UG. L research on the acquisition of Binding Principles has been particularly fruitful for determining L transfer and UG access in adult learners (e.g., Hirakawa, ; Thomas, ; Wakabayashi, ; White, ; Yuan, ). These theoretical advancements at the time and the ensuing empirical work on L acquisition of pronominals have sparked interest in generative linguistics-inspired research into the maintenance/loss of UG-based properties in the L grammars of bilinguals. The need for linguistic theory-driven L attrition research as part of a more general investigation of bilingualism has been expressed repeatedly in the literature (e.g., Köpke & Schmid, ; Schmid & Köpke, ; Schmid, b, a, Chapter ; see also Schmid & Köpke, a), and the attrition studies discussed in this Chapter constitute an important step in this direction. In the rest of the chapter, after a brief note on key linguistic properties associated with pronominals, relevant adult L attrition studies are presented and their implications are discussed. The following section should not read as a comprehensive overview of all theoretical models proposed so far regarding the linguistic properties (i.e., licensing/ identification, binding) of pronominals. Rather the aim here is to sketch out linguistic characteristics of pronominals as relevant for the L attrition studies discussed in the subsequent sections.
1 If the language also allows the null pronoun, the learner can subsequently acquire it on the basis of positive evidence, without any need for negative evidence (White, ).
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
. T
.................................................................................................................................. In some null-subject languages, there are certain null pronoun-licensing morpho-syntactic mechanisms (e.g., rich/strong morphological features/markers). Yet the use of null and overt pronouns is not random but subject to certain discourse-pragmatic constraints. Overt subjects in null-subject languages perform a variety of functions such as topic shift, contrast, and focus (e.g., Montalbetti, ; Enç, ), as illustrated in the following examples from Turkish, a null-subject language (Enç, ; Gürel, ): ()
pro
alış-veriş-e shopping-
gid-iyor-um go--
() Ben alış-veriş-e gid-iyor-um I shopping- go-- ‘I am going shopping’ While () would be an appropriate answer to a question such as Nereye gidiyorsun? (‘Where are you going?’) in line with the topic continuity function of null pronouns, the answer with an overt pronoun () would not. Sentences such as () would normally be used to signal a contrast such as Ali eve gidiyor ama ben alış-verişe gidiyorum (‘Ali is going home but I am going shopping’) or to establish a topic, for example if the speaker decides to leave the house to go shopping and wants to inform her/his partner about it (Enç, , p. ). The following sentences illustrate similar constraints in Italian (Tsimpli et al., , p. ): ()
pro
E’ is-SG ‘He left’
partito gone-
A null pronoun in () is preferred in Italian if this sentence is uttered as a response to a question such as ‘What did Gianni do’ to maintain topic continuity. Topic shift, however, requires overt pronouns. The presence of post-verbal subjects is also discussed in relation to the null-subject property of Italian. For example, in response to an all-focus question such as Che cosa è successo? (‘What happened?’), an answer with a post-verbal subject () is required (see Sorace, , and Belleti & Guasti, , for other linguistic factors determining post-verbal subjects in Italian). () pro
È is-SG ‘John left’
partito Gianni gone- John
Pre- and post-verbal subjects are also allowed in Greek. Pre-verbal indefinite subjects are interpreted as ‘old’ information (topic), whereas post-verbal indefinite subjects are ambiguous between the ‘old’ and ‘new’ information (Tsimpli et al., , pp. –):
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
()
̈ Context: I gitonisa mu ston trito orofo apektise dhidhima. ‘My neighbour on the third floor had twins.’
() Xtes vradhi ena last night one ()
moro baby
ekleje. was-crying (=one of the twins)
Xtes vradhi ekleje ena moro. last night was-crying one baby (=one of the twins OR some other baby) ‘My neighbour on the third floor had twins. Last night one baby was crying/wascrying one baby’
Another linguistic feature of pronominals pertains to their referential properties. As shown in the Greek examples below, null and overt pronouns involve different coreferentiality (Tsimpli et al., , p. ): () O Janisi prosvale ton Petrok otan proi /aftosk the Janis insulted the Petro when pro / he ‘Janisi insulted Petrok when hei /k approached him.’ () O Janisi idhe ton Petrok otan AFTOSk ‘Janisi saw Petrok when HEk approached him.’
ton
ton him
plisiase. approached
plisiase.
In (), the use of the null pronoun in the subordinate clause subject position leads to a nonshifted interpretation (hence its coreferentiality with the matrix subject, Janis), whereas the overt subject (aftos) is coreferential with the matrix object only. In (), the pronoun is additionally licensed by discourse focus, so the antecedent for the overt pronoun is the matrix object (Tsimpli et al., , p. ). The same referential tendencies have been noted earlier in Italian intrasentential anaphora by Carminati (, p. ) under the Position of Antecedent Strategy, which postulates that the null pronoun has a strong bias towards the most prominent antecedent in the SPEC, IP position (typically the sentential subject), while the overt pronoun, albeit flexible in coreference, tends to select a non-subject antecedent in the non-SPEC, IP positions.2 Another phenomenon related with pronominal binding is discussed under the overt pronoun constraint (OPC) (Montalbetti, ), which stipulates that in null-subject languages such as Spanish and Japanese, when the antecedent is a referential NP, overt pronouns and pro can both be coreferential with the antecedent. However, when the antecedent is quantified, overt pronouns, unlike pro, can have only disjoint reference (see examples () and ()).3,4 2
Not all null-subject languages show similar anaphoric preferences. Even typologically close and morpho-syntactically similar languages, such as Italian and Spanish, display different patterns of overt pronoun resolution (Filliaci et al., ; Romano, ). 3 According to the standard theory of Binding (e.g., Chomsky, ), in a sentence such as Johni believes that [hei/j is intelligent], ‘John’ and ‘he’ can be coreferential as ‘John’ is not in the local domain/ governing category of the pronoun (i.e., embedded finite clause). The pronoun can potentially have disjoint reference (e.g., a third party in the discourse), as shown by the index ‘j’. In contrast, the reflexive anaphor, ‘himself ’ as in Briani says [Georgej likes himself*i/j/*k] has to be governed by an antecedent in its local domain, hence the co-indexation of ‘himself ’ with George only. 4 The distinction between bound and disjoint interpretations can formally be explained as follows (Gürel, a, p. ): (i) Bound variable interpretation: (No x: x is a person) x believes that x is intelligent;
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
() Juani cree que [éli/j / proi/j es inteligente] ‘John believes that hei/j / proi/j is intelligent’ ()
Nadiei cree que [él*i/j / proi/j es inteligente] ‘Nobodyi believes that he*i/j / proi/j is intelligent’
Gürel (, p. ) argues, on the basis of Turkish, that the OPC may not be a property of all null-subject languages. She suggests that the disjoint reading requirement for the Turkish overt pronoun o is syntactically driven (i.e., it derives from Binding Principle B, but not by the OPC) (cf. Montalbetti, ). Since the finite matrix clause (not the nominalized embedded clause) is the governing domain in Turkish, the overt pronoun o cannot be coreferential or bound by the matrix subject (unlike the null pronoun and other overt pronominal, kendisi). Crucially, contrary to what Montalbetti suggests, these binding constraints hold in both referential and quantified contexts alike (indices in the translations show the readings available in English): ()
Zeynepi [o-nun*i/k / proi/k/kendi-si-nini/k dahi Zeynep s/he- pro self genius düşün-üyor think-PRG ‘Zeynepi thinks (that) s/hei/k is genius’
()
Herkesi [o-nun*i/k / proi/k / kendi-si-nini/k dahi Everyone s/he- pro self genius düşün-üyor think- ‘Everyonei thinks (that) s/hei/k is genius’
ol-duğ-u]-nu be--.-
ol-duğ-u]-nu be--.-
The theoretical conceptualization of null and overt pronouns (i.e., the licensing/ identification and binding conditions) has changed over the years particularly after the advent of the Minimalist Program (MP) and its subsequent reformulations (e.g., Chomsky, , , ). The sources of parametric variation across languages are reduced to differences in the formal features of particular items (e.g., the functional heads) in the lexicon, and also to Extended Projections Principle (EPP) features (see Holmberg, ; Biberauer et al., , for further discussions). These grammatical features involving number (singular, plural), gender (masculine, feminine), person, and case are assumed to play a role in morpho-syntactic processes. Some of these features have semantic content hence interpretable at logical form (LF), whereas others are uninterpretable at LF and therefore must be eliminated (checked off) before LF. For example, phi-features (person, gender, number) on an NP (or DP) are considered [+interpretable] and persist at the LF interface to assure interpretability. Categorical features that are strong are also [+interpretable]. However, the phi-features on a verb, auxiliary, adjective, as well as case features are considered [ interpretable] (Chomksy, ). As for null and overt subjects, it is assumed, under one version of the MP, that null pronoun licensing is regulated by the interpretability of agreement features (e.g., Alexiadou & (ii) Disjoint interpretation: (No x: x is a person) x believes that y is intelligent (e.g., Nobody believes that John is intelligent).
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
̈
Anagnostopoulou, ) and obligatory overt subject is explained via feature movement. In other words, in languages like English, merging of the subject in the specifier of Agreement position is required in order to check the EPP feature (but see Holmberg, ; and Roberts, , for alternative views). Despite subtle differences in different versions of Minimalist assumptions, in general, in this reconceptualization, while (second) language acquisition reduces to the learning of formal features of a target language, language attrition is about losing the specifications of these features (e.g., Tsimpli, ; Sorace, ). These theoretical analyses have not only shaped L acquisition research (e.g., Park, ; see also Lardiere, ) but also L attrition work (e.g., Sorace, ). Even though the linguistic analyses adopted and the theoretical models tested have changed over the years, the main objective of L attrition researchers has been to examine how much of the subtle L knowledge of pronominal properties is maintained under extensive L exposure. As intricate characteristics of the L pronominal system constitute an integral part of implicit grammatical knowledge, any attrition effects in this domain would suggest the alterability of mature native competence (Sorace, a; Gürel, ; cf. Montrul, ). In that respect, the studies reviewed in the following section contribute to our understanding of the extent and the nature of the L attrition phenomenon that may emerge either as a competence or a processing problem in the pronominal domain as an (un)avoidable consequence of bilingualism.
. A L
.................................................................................................................................. Sorace (a; and Chamorro & Sorace, Chapter , this volume) is the first among generative linguists who has made predictions for L features vulnerable to attrition in adult grammars. Following the Minimalist assumptions as introduced by Chomksy (), Sorace takes [±interpretable] to be the relevant feature for selective attrition. She argues, based on data from L Italian and L Greek speakers of L English, that the licensing of null subjects is due to an [ interpretable] feature (e.g., an agreement suffix on the verb) while the distribution of null and overt subjects depends on [+interpretable] features such as [+Topic Shift]. Accordingly, because null subjects are the result of the specification of [ interpretable] features, they will not be affected by attrition and will continue to occur in contexts in which they occur in the speech of monolinguals (i.e., in [ Topic Shift] contexts). However, the distribution of overt subjects will be affected. That is, exposure to L English, a language whose pronouns are not obligatorily specified for [+Topic Shift], would cause overt pronouns to become optionally unspecified for [+Topic Shift] in Italian-English bilinguals and lead to the infelicitous occurrence of overt pronouns in [ Topic Shift] contexts. Sorace’s predictions have been tested in several studies. For example, Tsimpli et al. () examined the production and interpretation of null and overt subjects as well as pre-verbal and post-verbal subjects by UK-resident L Greek and L Italian participants who are near-native L English speakers. The researchers examined via elicited production and interpretation tasks whether L Greek attriters (i) overproduce preverbal
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
subjects; (ii) demonstrate non-native-like interpretation of overt indefinite subjects in preverbal and post-verbal positions; and (iii) whether L Italian attriters interpret null and overt pronouns in a non-native-like fashion in bidirectional anaphora resolution. The results of the production task revealed that the Greek attriters produced significantly more preverbal subjects than the Greek controls. In the interpretation task, attriters allowed both new and old information in pre-verbal and post-verbal positions, hence their indeterminate pronoun interpretations. As for L Italian, Tsimpli et al. (, p. ) discussed only the results of a picture verification task, which showed that in backward anaphora sentences with a null subject in the subordinate clause such as Quando pro/lei attraversa la strada, l’anziana signora saluta la ragazza (‘When pro/she is crossing the street, the old woman greets the girl’), both groups preferred the subject (i.e., the old woman) as the referent of pro. In the same sentences with an overt pronoun, the Italian controls strongly favoured an external referent (other than ‘the old woman’ or ‘the girl’), while the Italian attriters accepted, almost equally, all three referents. In sentences with forward anaphora such as L’anziana signora saluta la ragazza quando pro/lei attraversa la strada (‘The old woman greets the girl when she/pro is crossing the street’), with null pronouns, the Italian controls allowed for either the matrix subject or the complement as the referent, whereas the attriters preferred the subject interpretation. When the embedded subject is an overt pronoun, both groups preferred the matrix complement as its referent. Nevertheless, the attriters’ matrix subject preference with the overt pronoun was higher than that of controls. The authors interpreted this as a sign of attrition, as the attrition group showed a significantly greater tendency to allow an overt pronoun to be interpreted as a continued topic. Sorace’s (a) original ideas were later refined under the Interface Hypothesis (IH) (Sorace , ; Sorace & Filiaci, ; Sorace & Serratrice, ; Chamorro & Sorace, Chapter , this volume), which suggests that purely syntactic features (narrow syntax) and features that are at the interface of syntax and other cognitive-linguistic domains (e.g., syntax–discourse/pragmatics, syntax–semantics) undergo differential L attrition (and L acquisition) (Sorace, ). While purely syntactic features pose no considerable problems, interface properties are subject to optional use due to difficulty integrating different levels of information, particularly in real-time language processing. Sorace’s proposal was investigated recently by Kaltsa et al. (), who examined pronominal resolution in Greek by comparing L attriters (adult immigrants in Sweden) with monolingual controls.5 As in Tsimpli et al.’s () study, the focus was on null and overt subject pronominal interpretation in adverbial clauses in forward anaphora contexts I γiaγia xeretise tin kopela otan afti pernuse to dromo (‘The old lady greeted the girl when she crossed the street’). Adult L attriters were predicted to coreference subject antecedents to overt subjects far more than the controls due to L influence from Swedish—a non-null-subject language. The null pronoun interpretation, in contrast, was expected to remain native-like. The study consisted of two self-paced listening experiments (Experiments and ) involving sentence-matching, and testing null and overt subject pronoun resolution, respectively. Participants were asked whether the sentences they heard matched the presented pictures, which depicted the subject (‘the old lady’), the
5 The study also included heritage speakers (early Greek-Swedish bilinguals) and younger Greek controls.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
̈
object (‘the girl’), or another person (disjoint reference) as the agent of the action provided in the adverbial clause (e.g., crossing the street). The results of Experiment revealed that although the attriters generally preferred the object rather than the subject as the antecedent for the overt subject pronoun, their preference for the subject antecedent for the overt pronoun was still higher than that of the controls. In terms of reaction times (RTs), the attriters were, overall, slower than adult monolinguals but both groups showed longer RTs in the ‘subject referent’ and ‘other referent’ conditions than the ‘object referent’ condition, in line with their offline preference in picture matching. With respect to the listening times in the critical region (i.e., subject pronouns of the subordinate clause), the attriters were found to be faster than the controls in all three conditions. In Experiment , which tested null pronoun anaphora resolution, no significant difference was found between the attriters and controls in their subject antecedent preference. The two groups did not differ on the other two conditions, either, which suggests that, as predicted, co-indexation between the null pronoun and the subject antecedent is intact in attriters. The RT data, however, showed slower RTs in the subject referent condition compared to the object referent condition in the attriter group. In terms of listening time on the critical segment (the verb in the subordinate clause), the attriters showed no difference among the three referent conditions. The authors suggested that, in L Greek attriters, the overt pronoun is more vulnerable to change (i.e., it can be associated with subject and object antecedents) than the null pronoun. Given that even the monolinguals’ matching decisions on the overt pronoun were affected, the authors could not confidently attribute these results to L Swedish influence or to the restrictions in bilinguals’ processing resources (cf. Sorace & Serratrice, ). In another study testing the IH, Chamorro, Sorace, & Sturt () examined in three participant groups whether it is knowledge representation (i.e., competence) or ability to process interface structures that is affected in L Spanish attrition. The first group included late bilinguals with near-native English proficiency, residing in the UK for a minimum of five years. The second group was a re-exposed group who were also near-native English speakers with a minimum of five years of UK residency. This group was tested after they returned to Spain and re-exposed exclusively to Spanish for a minimum of one week. The third group included L Spanish-speaking controls who had recently arrived in Edinburgh and had very little knowledge of English. Additionally, the study explored whether attrition effects decrease after increased L input via re-exposure, an issue that is linked to the Activation Threshold Hypothesis (ATH), which suggests that the more frequently a linguistic item is activated, the lower its activation threshold is. The researchers examined intrasentential semantically-neutral anaphora constructions by manipulating the grammatical number of antecedents, so that the pronoun could refer to either the subject or object antecedent (e.g., Las madres saludaron a la chica cuando ella cruzaba una calle con mucho tráfico, ‘The mothers greeted-PL the girl when she crossed-SG a street with a lot of traffic’). In the other item types, the sentential subject of the adverbial clause had a null pronoun. Data was collected via eye-tracking-while-reading and untimed naturalness judgement tasks, which are believed to tap online processing and knowledge, respectively. The prediction was that the attriters would show sensitivity to the pronoun mismatch only in the offline task. The re-exposed group, however, was expected to perform similarly to the native controls in both tasks. In eye-tracking, the critical region was ‘ella/pro cruzaba’, and first-pass time, go-past time, and total time were analysed. The judgement task results revealed that all groups converged on their preferences for subject pronouns.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
In eye-tracking, while the control and re-exposed groups showed early sensitivity to the mismatch conditions, the attriters showed no evidence for a pronoun–antecedent interaction in any of the measures. These results support the IH, which predicts non-native performance only on real-time processing of pronominal properties at the syntax–discourse interface (see Genevska-Hanke, , for similar arguments). In addition, the findings are consistent with the ATH, as the attrition effects disappear after re-exposure to the L. If difficulties with the processing of interface properties are recoverable after only one week of L (re-)exposure, this clearly suggests that processing problems are extremely short-lived.6 Indeed, the transient/restricted nature of the attrition phenomenon was noted in a study by Miličević & Kraš () that examined L English-induced attrition effects in the interpretation of L Italian pronominal subjects in English–Italian translators in comparison to non-translator bilinguals. In a picture selection task, the participants were asked to identify the antecedents of null and overt subject pronouns in intrasentential forward and backward anaphora sentences similar to the ones discussed earlier. The results showed no over-acceptance of subject referents in the overt pronoun condition by the experimental group. The authors suggest that the overuse of overt subject pronouns previously reported for translated texts (e.g., Cardinaletti, ) might simply be a product of the translation process itself. The interpretable features of the null and overt pronouns were also investigated in relation to Binding Principles. For example, Gürel () examined L English influence on pronoun binding in L Turkish following an earlier formulation of the theory (e.g., Chomsky, ).7 The complex sentences tested had overt or null subject pronouns in embedded clauses, which are nominalized DPs in Turkish (e.g., Buraki onun*i/k / proi/k / kendisinini/k zeki olduğunu düşünüyor, ‘Buraki thinks that hei/k is smart’).8 Besides the overt pronoun o, Turkish also has a special type of anaphoric pronominal, kendisi, for which English has no corresponding form. Gürel’s study also included items with quantified antecedents as in (). The attrition group consisted of native Turkish speakers who were late L bilinguals (end-state L English users), who had immigrated to North America (Canada or the US) as adults (mean age of immigration: .), and had been living in the L country for at least ten years at the time of testing. The native controls were residents of Turkey since birth and had basic knowledge of English. The tasks included written interpretation, story-based truth-value judgement, and picture identification/matching listening tasks. Results across the three tasks revealed that while the overt pronoun o was most often assigned the disjoint reading, the null pronoun and kendisi received mostly bound or ambiguous readings in both groups. Nevertheless, the tendency to assign the correct disjoint reading for o was weaker in the attrition group. When participants were given only two options (bound or disjoint readings) (Tasks and ), the difference 6 It is important to note that the re-exposed group differed from the attrition group only by the fact that, by the time of testing, the former had gone to Spain for the Christmas break and re-exposed to Spanish exclusively for a minimum of one week (the mean number of days of L re-exposure was .). The re-exposed group had not been tested before returning to Spain. Therefore, we cannot say with confidence that their native-like performance is due to re-exposure to Spanish. 7 This is actually a comparative study of L attrition and L acquisition of Turkish, exploring English influence on Turkish. 8 As noted earlier, unlike in English, the governing domain in Turkish is the matrix clause, not the nominalized embedded clause.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
̈
between groups became more salient; attriters allowed significantly more (incorrect) bound interpretations for o than controls. For the null pronoun and kendisi, the groups were similar in interpreting them mostly as bound pronouns. These findings were interpreted readopting a subset-superset transfer model, originally proposed for L acquisition (Manzini & Wexler, ), in the sense that when the L constitutes a superset of the L with respect to a particular property, L-induced L attrition emerges. In this particular case, since English allows both DPs and finite IPs as binding domains, it constitutes a superset of Turkish, which does not allow DPs as binding domains. Therefore, Turkish native speakers tended to expand L options based on the L model to eliminate restrictions in the L grammar. In contrast, in configurations with the L offering a more restrictive subset grammar than the L, no L restructuring is expected. A study replicating Gürel () with the aim of establishing similar effects of two typologically-related Ls (English and Dutch) on L Turkish tested Turkish attriters living in the Netherlands under the influence of L Dutch, a non-null-subject language with English-like binding properties and found similar results, confirming the set theoretic assumptions for L attrition (Gürel & Yılmaz, ). Gürel’s () data was also interpreted within the ATH in Gürel (a), proposing that, due to L English, attriters treat o as identical to the corresponding English overt pronoun as a result of activation of L binding properties in lieu of those in the L. When an L linguistic element/rule has no equivalent form in the L (as in the case of pro and kendisi), the L does not compete with the L, hence no interference or attrition (Köpke, ). The set-theoretic predictions proposed in Gürel () were also tested in L Turkishinduced L English attrition in Gürel (). The study compared native speakers of English with advanced Turkish proficiency who had been living in Turkey for an average of . years with native English speakers who had no knowledge of Turkish. The tasks were similar to the first two tasks used in Gürel (). The constructions included possessive DP pronouns (‘The soldiers carried their boxes’), monoclausal sentences with pronouns/ reflexives (‘James sent Henry a picture of him/himself ’), biclausal finite (‘Chris knows that Harry hates him/himself ’) and non-finite sentences with pronouns/reflexives (‘Nancy asked Emily not to blame her/herself ’), and biclausal finite sentences with subject pronouns (‘John said that he lied to the police’). Based on different binding properties in Turkish and English, it was predicted that more attrition effects would occur in configurations where the L Turkish offers broader grammatical options than English. For example, in monoclausal sentences with two potential antecedents, while the English pronoun can allow only disjoint reference, the Turkish object pronoun o can both be coreferential with the indirect object antecedent and have disjoint reference. Thus, under attrition, the L English grammar was expected to move towards L Turkish grammar, which offers less restriction. Similarly, it was predicted that the binding properties of kendisi could be mapped onto the L English reflexive ‘himself/herself ’, leading to attriters’ acceptance of non-local antecedents for it. Nevertheless, the results revealed no evidence of L attrition in this domain. Although the findings contradict the set-theoretic linguistic model of L attrition, frequency-based psycholinguistic models such as ATH can explain them. That is, the absence of attrition effects in English might be linked to the frequency of use and the de facto status of English as a lingua franca (Gürel, ). In another study examining attrition effects in the domain of reflexives, Kim et al. () compared US-resident Korean heritage speakers (early bilinguals), first-generation Korean
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
immigrants (late bilinguals), and adult L Korean learners to monolingual Koreans. The attrition group had been in the US for more than ten years. The task involved a truth-value grammaticality judgement task with stories followed by a sentence to be judged as ‘true’ or ‘false’. On the basis of differences in the binding properties of the Korean reflexive caki, which can function as a long-distance anaphor, and the English himself/herself, which can only be locally bound, it was predicted that in Korean constructions such as Wendynun Tedeykey cakika Charles pota ttokttokha ta ko malhayssta (‘Wendy told Ted that self (she) is smarter than Charles’) and Christineun Tom elopwuthe caki tongsayngi cheypotanghan iywulul tulessta (‘Christine learned from Tom the reason for self ’s (her) brother’s arrest’), bilingual groups would show a lower degree of acceptability compared to the monolingual controls, due to dominant English influence. Results showed that the Korean adult attrition group behaved like controls, while differing significantly from the other two experimental groups. Thus, unlike Gürel’s () first-generation L Turkish immigrants, late KoreanEnglish bilinguals maintained L knowledge of the governing category. The authors link this to the close-knit communities where Korean immigrants live, drawing attention to the role of sustained and frequent L contact. This section provided a selective overview of adult L attrition studies investigating different aspects of null and overt pronouns (i.e., licensing and interpretative/binding properties) as motivated by different theoretical assumptions. All of the studies, except for Gürel () involved late bilinguals with a null subject L (i.e., Greek, Italian, Spanish, Turkish, Korean) and non-null subject L (English, Swedish, Dutch) and revealed no conclusive results concerning the magnitude of L attrition either at the competence or performance level, crucially failing to establish an across-the-board kind of L effects from a non-null-subject language on the L in this domain. Overall inferences that can be drawn from these studies are discussed in the following section.
. S
.................................................................................................................................. All in all, early generative studies on adult L attrition involving pronominals focused on the question of whether L knowledge of syntactic binding/governing domains could change under the influence of a dominant L. Results are inconclusive; while some studies reveal alteration in the L governing domain due to extensive L exposure, along with less accessible L input (Gürel, ; Gürel & Yılmaz, ), others document no evidence of attrition in this area of grammar (Gürel, ; Kim et al., ). Relatively more recent work on adult L attrition of pronominals is generally designed in line with the IH predicting optionality in structures at the interface between syntax and other cognitive domains (Sorace, , p. ). Indeed, as an interface phenomenon, the interpretation/use of null and overt subjects in the context of anaphoric dependencies constitutes a good testing ground for this hypothesis. Several studies confirm Sorace’s (, ) predictions in the sense that, unlike native controls, adult L attriters demonstrate optionality in the mapping of pronouns with a relevant interpretable feature (i.e., [±Topic Shift]). The overt pronoun appears to be particularly vulnerable as attriters fail to associate it consistently with the [+Topic Shift] feature. The null pronoun, however, remains relatively intact, as it is associated mostly with the [ Topic Shift] feature (Tsimpli et al., ; Kaltsa et al., ).
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
̈
Differential vulnerability of null and overt pronouns is indeed predicted in views that analyse overt pronoun overuse as a default ‘fall-back’ option that has the advantage of being redundant rather than ambiguous (Sorace, , p. ) and in other views that essentially predict attrition based on a competition between a null-subject L and a non-null-subject L (Gürel, , a; see also Köpke, , ; Paradis, ). There are also studies that report no major change/restructuring in attriters’ L knowledge of anaphoric reference, as measured by offline tasks (Gürel, ; Chamorro, Sorace, & Sturt, ; Miličević & Kraš, ). This points to the processing versus competence dichotomy proposed by the IH and confirmed by studies revealing attrition effects only in online processing of multiple levels of information (Kaltsa et al., ; Chamorro, Sorace, & Sturt, ; see also Genevska-Hanke, ). Nevertheless, given the scarcity of studies, this is yet an unresolved issue. More data from different language pairs is necessary to clearly confirm that processing problems appear only in adult attriters but not in monolinguals, and that attriters’ core/narrow syntactic knowledge, as measured by offline tasks, always remains converged on native norms (cf. Gürel, ; Kim et al., ; Gürel & Yılmaz, ). With respect to L effects on L attrition in the domain of anaphoric dependencies, while some studies document L influence (Gürel, ; Tsimpli et al., ; Gürel & Yılmaz, ), others fail to show this (Kaltsa et al., ; Chamorro, Sorace, & Sturt, ; Miličević & Kraš, ). It is important to note that L attrition studies typically include only one experimental (attrition) group, whose L differs in certain ways from the dominant L. It is not possible to clearly establish L effects on L grammar in the absence of another attrition group whose L works similarly to the L (cf. Gürel & Yılmaz, ). It is also important to note that in many studies, control groups’ online and offline performance is not unequivocally better than that of adult attriters, suggesting that cross-linguistic influence may not be the major cause of L vulnerability (Kaltsa et al., , p. ). In conclusion, generative research on adult L attrition in the pronominal domain has not, until now, revealed a robust change/restructuring in abstract L grammatical competence as measured through offline acceptability tasks; it has only displayed certain (albeit not completely non-native-like) indeterminacies in online use/interpretation of referential properties of pronominals. In other words, while abstract knowledge representation regarding L pronominal properties remains intact, online processing of relevant linguistic features may be problematic. Furthermore, recent data suggests that processing inefficiencies in the L are largely recoverable after brief re-exposure (Chamorro, Sorace, & Sturt, ). This is a noteworthy finding when we consider the theoretical psycholinguistic literature revealing ‘good enough’ language processing even in monolingual native speakers. More specifically, as research suggests, even native speakers may not always be able to generate representations of the linguistic input that are complete, detailed, and accurate (Ferreira et al., , p. ). Thus in the real world, interpretations are likely to be ‘just good enough’ even in native comprehenders (Ferreira et al., , p. ), possibly because they tend to use heuristics (more than syntactic algorithms) as more readily available interpretations (Ferreira et al., ; Ferreira & Patson, , p. ). Given this, incorrect or incomplete comprehension/interpretation in the pronominal domain may not be a unique condition for attriters or bilinguals in general. Therefore, it is crucial to have more carefully designed attrition studies testing a wider range of phenomena via multiple modality tasks in a variety of languages to identify the underlying mechanism of L attrition and to address
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
the issue of grammatical representation versus processing dichotomy in this line of research involving late bilinguals. On this final note, to be able to establish confidently that attrition affects real-time integration (i.e., processing) of L knowledge, it is first necessary to determine whether the apparently divergent linguistic performance (e.g., comprehension) in attriters in online tasks may be ascribed to a set of human parsing strategies that lead to a cognitively less demanding decoding option. Such simple and parsimonious heuristics for building L interpretation may simply be used more readily by bilinguals than monolinguals. This would indicate that some of the phenomena which have been investigated in the area of first language attrition are less due to loss or erosion but more to general principles of bilingual processing.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
.............................................................................................................
EXTRALINGUISTIC FACTORS IN LANGUAGE ATTRITION ............................................................................................................. Edited by
.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
......................................................................................................................
......................................................................................................................
.
. I
.................................................................................................................................. T interest of first language attrition as a research topic within the wider framework of developmental linguistics and bilingualism is grounded in the fact that bilingual speakers differ substantially from monolinguals in the way in which they use both their languages. The most profound difference is the fact that, in many respects, bilingual populations exhibit more variability than monolingual populations: linguistic features which are used consistently by monolinguals, so that tasks relating to these features generally find a ceiling effect, tend to exhibit a larger degree of optionality and thus a larger amount of stratification within bilingual populations. On the other hand, bilinguals tend to underuse features associated with a high degree of complexity, such as infrequent grammatical structures or lexical items. Group comparisons between monolinguals and bilinguals therefore tend to find between-population differences for such features. The recognition that this is true not only in second language (L) acquisition but also in first language (L) attrition was at the root of the emergence of L attrition studies in the early s (see Köpke & Schmid, , for a historical review). In the early phases of attrition research, interest was focused predominantly on (micro) linguistic phenomena associated with language attrition as such, with little concern for their degree of variability. This approach reflects the early view of language attrition as ‘a special case of variation’ (Andersen, , p. )—that is, as a process which is rare and only occurs in speakers who have had extremely long periods of immersion in an L setting (decades) with few occasions to use their L. On such a view, studying variability in background factors such as length of residence and amount of exposure makes little
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
sense, since the small subpopulations of attriters within the wider population of bilinguals at large would, by definition, be relatively homogenous with respect to these factors. Such studies thus rest on the comparison of two extreme ‘slices’ of the bilingual continuum: attriters occupying one end and monolinguals residing in the country of origin the other. The assumed dichotomy was such that most early investigations only actually looked at data from the attrition end of the continuum without establishing a baseline for comparison: an annotated bibliography of attrition studies up to (Schmid, a) lists a total of seventy-three empirical investigations, of which only sixteen (%) use an unattrited control group. This approach not only reflects the view of language attrition as a relatively rare and extreme form of language development, affecting only long-term immigrants with low levels of language contact, but at the same time embodies an idealized view of monolingual performance: if a study investigates only attriters, taking any and all deviances from the ‘correct’ standard as evidence for attrition, this implies the presumption of monolingual speech as error- and disturbance-free (see Schmid, c). It was largely the contribution of Dutch scholars, in the latter half of the s, which began to move attrition studies away from this purely linguistic focus and idealization of the ‘native speaker’, establishing the topic as one that needed to encompass a socio- and psycholinguistic dimension (see Köpke & Schmid, )—much in the same vein as what has been called the ‘social turn’ did for studies of second language acquisition (SLA) (e.g., Block, ). At this point, several large and ambitious projects were started, aiming to identify the contributions of extralinguistic, personal, and individual predictors, such as the investigation described in Jaspaert & Kroon (). These studies were probably ahead of their time in terms of their ambition and scope and, sadly, most were never completed. However, they did lay the groundwork for many of the investigations that were carried out later, which did take into account considerations of the environment, the circumstances, and the identity of the speaker groups who were experiencing attrition.
. T E G N L A
.................................................................................................................................. The year represents a turning point with respect to the methodology of language attrition studies: during the First International Conference on First Language Attrition (ICFLA), held at the Vrije Universiteit Amsterdam, a network of young researchers was formed, all of whom were at that point involved in early stages of individual projects investigating different aspects of language attrition. Between and this group of some fifteen researchers met annually with a view to developing the common research methodology that had been outlined as a necessary step to move attrition research forward by Köpke & Schmid () and Schmid (b). Arguably the most important outcome of these meetings was the development of the Language Attrition Test Battery, a collection of tools and questionnaires whose purpose it was to render the findings from attrition studies—in particular those related to external and personal background factors— comparable. These tools have been made available in a range of languages on http:// languageattrition.org as well as on the IRIS platform (http://www.iris-database.org) and
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
are described in detail in Schmid (), and they subsequently were used in a number of investigations of attrition in languages as varied as Dutch (Keijzer, c), English (Dostert, ), German (Badstübner, ; Schmid & Dusseldorp, ), Romanian (Cherciov, ), Hungarian (Varga, ), Persian (Gharibi, ), Polish (Lubińska, ), and Turkish (Yılmaz, ) (among others). Collectively these studies revealed a number of methodological challenges for language attrition research which made it necessary to develop, discuss, and refine strategies and analytic designs suitable for the investigation of such data. It has often been pointed out that language attrition is a complex developmental process, driven by the (often nonlinear) interaction of a plethora of factors (e.g., Schmid et al., ; Opitz, Chapter , this volume), and that these factors themselves tend to be multifaceted. This is particularly true for the factor that is most frequently invoked in attrition studies, namely frequency of use and exposure (see also Schmid, Chapter , this volume): studies of bilingual development and language dominance have long since recognized the complexities of quantifying this factor in such a way as would allow it to serve as a predictor in statistical investigations. Approaches that acknowledge this are, for example, Grosjean’s (, ) Complementarity Principle, which rests on the assumption that ‘[d]ifferent aspects of life require different languages’ (Grosjean, , p. ) and therefore assign one or several languages to different domains of use, such as the home, politics, specific leisure activities, etc. A somewhat similar approach was proposed by Birdsong () and modelled on recent approaches to index handedness. Similar problems apply to other extralinguistic factors, such as attitudes (see Yılmaz, Chapter , this volume), and imply that any investigation attempting to probe into the relationship between predictors and outcome variables in the attritional process will have to adopt a formidably multifactorial design, encompassing a wide variety of contexts and domains of language use and an equally varied set of variables relating to attitudes and identification. Such approaches are, however, often hampered by a number of practical and methodological challenges, including (but not limited to) • The problem of power and population size: in order to adequately describe actual linguistic processes of attrition, measures have to be collected that are extremely time-consuming to elicit, transcribe, and code (e.g., Schmid, a). In practice this means that attrition studies with more than thirty to fifty participants per population are not feasible, severely limiting the number of possible predictors (see Schmid & Dusseldorp, ). • The fact that most of these variables can only be elicited by overt self-assessments and introspection, raising concerns about their validity and reliability (see Cherciov, ). • The problem posed by the fact that such self-assessments are typically elicited as ordinal variables, for example by means of Likert Scales. • The problem posed by the fact that data relating to such factors is usually not normally distributed. This is a natural outcome of the Complementarity Principle: people tend to prefer one language over the other in different domains, implying that different factors will either be skewed towards one or the other language or bimodally distributed. Taken together, the problems listed above pose serious concerns for the application of parametric statistics. Non-parametric procedures, on the other hand, are extremely limited
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
when it comes to multifactorial analyses (e.g., Norman, ). One focus of the work of the Graduate Network was therefore to propose a standardized set of steps and analytic procedures that would help alleviate these problems. These are set out in Schmid & Dusseldorp () and Schmid & Yılmaz (), and tools for their application were also made available (on https://languageattrition.org).
. A
.................................................................................................................................. The quantitative and empirical evidence base on language attrition established over the past two decades, partly by members of the Graduate Network and partly by researchers working within other paradigms, has revealed that the early view of attrition being a rare and extreme phenomenon was unwarranted: cross-linguistic transfer from L to L has clearly been shown not to be confined to situations of very long-term exposure and/or minimal L use (see Schmid, Chapter , this volume). Many attrition phenomena appear to occur early on in an immersed setting—for example, phonetic transfer has been observed within weeks (Chang, ) while naming tasks become slower within six months (Baus et al., ). These and other findings have prompted Schmid & Köpke (a) to argue that all phenomena of L-to-L-transfer, at all stages of contact and all levels of L proficiency, should be taken to be indications of the same, larger process; and that the distinction between attrition on the one hand and other transfer phenomena, on the other, is artificial and impossible to establish (see also Schmid & de Leeuw, Chapter , this volume). This view is grounded in the observation that, while most investigations of L attrition do indeed find higher levels of variability, more deviations from the native norm, and reduced complexity and fluency within attriting populations, in virtually all cases there are individuals within the attriting population that perform at the same level as, or even outperform, the monolingual controls. Such observations, however, are often taskdependent and/or related to the linguistic level under observation: the fact that, for example, an individual speaker’s accent is perceived to be entirely native-like does not say anything about the breadth and depth of her productive vocabulary, her fluency or her accuracy on a particular grammatical feature (see Schmid & de Leeuw, Chapter , this volume). One of the main tasks for the field of language attrition research is therefore to establish the conditions under which a particular speaker is more or less likely to experience attrition with respect to any one linguistic feature or task—in other words, to develop models capable of explaining the variability found in these speaker populations. It has been pointed out (Schmid, a) that the predictors which are relevant for the attritional process fall into three larger categories: . Personal background factors a. age at immigration b. length of residence
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
c. education, occupation and socio-economic status d. aptitude . Measures relating to the use of and exposure to the L and the L . Measures relating to attitude, identity, integration, and affiliation
. O
.................................................................................................................................. The chapters in the present section of this volume attempt to provide a summary of the state of knowledge with respect to these predictors. In the opening chapter of the section, ‘Age effects in language attrition’, Bylund (Chapter ) discusses the impact of age of bilingualism on the outcome of L attrition. In the chapter ‘The impact of frequency of use and length of residence on L attrition’, Schmid (Chapter ) explores L attrition and its development under the lens of length of residence in the L country and patterns of L use. The intricate socio-linguistic and psycholinguistic facets of the seemingly inevitable tradeoffs between L and L proficiency levels are thoroughly discussed by Yılmaz in Chapter , ‘L attrition, L development, and integration’. Language contact and its relation to L attrition is analysed by Riehl in Chapter , ‘Language contact and language attrition’. The questions in Bylund’s Chapter , ‘Age effects in language attrition’ focus on three interrelated aspects: the impact of age on the retention of L skills, the interaction between the effect of age and L input, and the underlying mechanisms for the effect of age on language decline. Concerning the impact of age on L attrition, Bylund draws comparisons between studies on children, adoptees, and adult attriters and shows that whereas children’s language skills may undergo extensive attrition, adult linguistic knowledge is much more stable and less susceptible to loss. In cases where input from the L ceases entirely and abruptly, age is the only factor that can predict attrition. In this respect, Bylund cites several studies, among which Pallier et al. (), that report complete loss of the L in internationally adopted children. In order to explain the impact of reduced L contact on attrition, Bylund reviews a number of studies indicating that the main characteristic of L performance is variability, ranging from unreciprocal communication strategies (Gal, ), lexical retrieval difficulties and loss (Ammerlaan, ; Hulsen, ), syntactic violations and morphological disintegration (e.g., Seliger, ; Turian & Altenberg, ; Vago, ), L convergence (Montrul, a) to behaviours that are indistinguishable from non-attrited native speech (Bylund, ). The extant evidence shows that susceptibility to L decline decreases considerably after the age of –, which indicates that this age is a turning point with respect to L attrition. Bylund concludes this section by drawing attention to the question of whether L attrition in children below years is the result of the distance from L norms or the non-acquisition of certain L structures and the explanatory power of these realities with respect to variability in L performance. To understand the underlying mechanisms of age and the age-related outcomes behind language attrition, Bylund discusses three theoretical frameworks. The impediment account draws parallels between the role of age in acquisition and attrition. In SLA, puberty has often (albeit highly controversially) been associated with a turning point after which nativelike proficiency in a second language can no longer be attained. In L attrition, puberty is
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
the age after which susceptibility to loss of the L decreases considerably. This has led to accounts suggesting a trade-off between the two languages, where the L cannot develop to native-like levels due to L entrenchment. However, attrition research reviewed by Bylund suggests that the proficiency in both the L and the L can also be positively correlated (Cherciov, ; Bylund et al., ; Schmid & Yılmaz, ) and remnants of the L continue to be present in spite of native-like proficiency in the L (Norrman & Bylund, ). Next, Bylund examines the maturational account, according to which the success of L acquisition or the degree of L erosion are the result of changes in the neural plasticity in the cortical areas subserving language. Before puberty, the brain is not only more readily able to acquire new linguistic information, but this information can also more easily ‘overwrite’ existing knowledge. Bylund discusses how the ‘sensitive period’ accounts differ from the more traditional ‘critical period’ view in that they assume that the optimal window for L acquisition—which coincides with the period of high susceptibility to L erosion— has a gradual and dragged-out offset, instead of being clearly and narrowly defined (Knudsen, ; Kral, ). Bylund’s review of the psychosocial account looks at how particularly traumatic experiences lived by the bilingual prior to the age of immigration may exacerbate language loss. Several studies cited by Bylund point to the possibility of ‘active forgetting’ (Footnick, ) if the L carries strong negative associations, regardless of the age at emigration. Compelling findings in this respect stem from Schmid’s () investigation of language attrition and maintenance in German-Jewish Holocaust survivors, which shows that the extent and severity of traumatic events individual speakers had suffered at the hand of persecutors who shared the same native language was strongly associated with the degree of language loss in these participants. Bylund’s account of the role of age in language attrition concludes by outlining that: () age is a strong, but not a deterministic variable (heightened attrition susceptibility in children and reduced attrition susceptibility in adults does not imply obligatory attrition for the former, nor invulnerability to L attrition for the latter); and () age is a gradual rather than a dichotomous factor and its effects reside in the degrees of attrition exhibited by bilinguals along the age spectrum. For future research, Bylund argues for the importance of more conclusive research with respect to the mechanisms underlying age effects in attrition, as well as research on selectivity in age effects—that is, how different subsystems and structures are differently affected by age. The second chapter, ‘The impact of frequency of use and length of residence on L attrition’, (Schmid, Chapter ), presents a summary and overview of the research concerning the two factors which were originally believed to be the strongest drivers of the attritional process, namely the intensity of L use and the length of residence in the L environment. Schmid provides the theoretical background for this belief by means of a selective review of theories of memory, forgetting, and access to information stored in the brain. This review demonstrates that the common-sense view of ‘forgetting’ as a single and holistic process is overly simplistic, as information can become available due to a variety of distinct processes: one can indeed ‘forget’ information because the memory traces it has left in the brain gradually decay until it is no longer available for retrieval, but it is also possible that the information itself is perfectly intact but that access to it becomes compromised due to processes of repression, suppression, or interference. Drawing on the seminal study by
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Ecke (), Schmid discusses these different types of forgetting mechanisms, linking them to the Activation Threshold Hypothesis proposed by Paradis (, ) to account for language attrition findings. In a similar vein, Schmid points out that one of the problems from which language attrition studies suffered for a long time, and which prevented solid conclusions on the impact of exposure on the forgetting or attrition of the L, was an overly simplistic notion of the concept of ‘language use’: until the early s, those few studies which attempted to investigate this predictor at all tended to collapse it into a single (and most often dichotomous) variable. She provides a brief introduction to the Language Attrition Test Battery— the collection of research tools and materials developed within the European Graduate Network mentioned above—and demonstrates how it attempts to assess language use across different contexts and domains, making it possible to adopt a multi-faceted and fine-grained approach to the research question. She also discusses the impact of length of residence in this context, particularly stressing the impact of the early years of the immersion experience—a period more often than not completely neglected by attrition studies. Schmid then proceeds to a stocktaking exercise, presenting an overview of the findings of twenty-three investigations of language attrition which quantify either L use or length of residence (or both) as a predictor. She discusses the findings from these studies, taking into account the type of task and the linguistic level under investigation, and attempts to extract general patterns of how they link up to the predictors. The chapter ends with a cautionary note, pointing out not only the limited amount of evidence available to date but also the rapid global changes which—due to the availability of modern communication technology and cheap travel—make the immigration experience and the opportunity for contact with the L a completely different one today than what it was only a few decades ago. She argues that comparisons across studies will need to be aware of these changes in circumstance. The next chapter, ‘L attrition, L development, and integration’ (Yılmaz, Chapter ), addresses the evolution of L attrition in relation to L development and integration into the L environment. Yılmaz uses a triangular approach to review the existing research and brings together related studies from the fields of psycholinguistics, socio-linguistics, and language policy. First, the probability of a linguistic trade-off assumption between L attrition and L development is explained from a psycholinguistic perspective. Yılmaz starts with a review of psycholinguistic investigations that reveal that both languages remain active in the bilingual brain regardless of the language intended for speech production (Kroll & Sunderman, ). Yılmaz notes that the constant interaction between languages not only exerts a toll on bilingual language processing and production, but it also makes language maintenance more effortful due to the sharing of finite resources for memory and processing in the brain. This in turn can lead to competing language systems and proficiency trade-offs, rather than reinforcement of the languages (Costa & Santesteban, ). Yılmaz cites findings from the attrition research indicating that L and L trade-offs were primarily found in those participants most affected by L attrition (Cherciov, ; Opitz, a). However, other patterns, such as bilinguals who retained native-like proficiency levels in both languages, and which are at odds with the trade-off account, also emerged from these same studies. Moreover, the development of attrition can result in the reorganization and expansion of the language system as a result of converging L and L categories (Tsimpli, ; Sorace, ). Yılmaz points to research indicating that divergence from the L does not necessarily
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
resemble the L system, thus expanding our understanding of language interaction outcomes beyond the limits of a two-dimensional perspective (de Leeuw et al., ). Yılmaz proceeds with a review of the acculturation theory (Berry, , ) and its four possible outcomes—integration, assimilation, separation, and marginalization—to account for the complex dynamics between the L and the L cultures in relation to the migrant’s identification with the host culture and culture of origin. The corresponding categories in terms of bilingual development—competent bilingualism for integration, L monolingualism for assimilation, monolingual L segmentation for segregation, and limited bilingualism for marginalization (Esser, )—are largely determined by the migrant’s motivation to integrate in the host culture. With respect to language attrition, more L attrition is to be expected in speakers veering towards assimilation. According to Yılmaz, these speakers no longer see practical and symbolic value in their L and, therefore, are not motivated to make efforts to maintain the language. Yılmaz also discusses the extent to which the type of language and integration policies put in place in the L country can also inhibit or facilitate attrition, as seen in Ben Rafael & Schmid’s () study comparing levels of L attrition between Russian and Francophone emigrants to Israel. The other theoretical framework reviewed by Yılmaz, Ethnolinguistic Vitality Theory (EVT; Giles et al., ), uses a well-defined quantification of affective variables, taking into account both individual attitudes and community level factors, such as language prestige and institutional support. However, the results from two studies applying this framework indicate that it fails to account for the lack of attrition among Turkish migrants in spite of low measurements of linguistic and cultural vitality (Yağmur, ) or the L attrition of the Dutch participants in Australia in spite of high vitality indicators (Hulsen, ). Yılmaz discusses the role methodology plays in explaining the contradictory results. In those studies where the groups of participants were homogenous, motivational and attitudinal variables were more reliable than in studies with more heterogeneous groups of speakers, where the relative weight of attitudes is measured based on self-reports and questionnaires. Yılmaz concludes by underlining that, in order to arrive at a more accurate understanding of these types of variables, both quantitative and qualitative measurements are needed. The third angle from which Yılmaz explores the relationship between attrition and acquisition looks at the impact the policies set forward by the host country can have on L acquisition and L attrition. The studies reviewed here focus on the integration of nonEuropean minorities in the European migrant context. In order to bridge the cultural and linguistic gap between migrant and host communities, European states have put forward policies that support L acquisition and cultural integration, often at the expense of the L. The underlying belief behind such policies and training programs is that increased L competence is supposed to translate to more successful social and cultural integration in the host country. However, as pointed out by Grosjean (), L acquisition does not necessarily promote cultural and social integration and an overlap between the linguistic and cultural components of the integration is highly dependent on the setting, the community, or the time period. Yılmaz also cites research showing that there is no straightforward correlation between skills in the L and the L, and that in those cases where such a correlation can be established, it is more often positive than negative (e.g., Schmid & Yılmaz, ), indicating that a higher level of L maintenance facilitates, rather than hinders, L development.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
In the last chapter, Chapter , ‘Language Contact and Language Attrition’, Riehl looks at how attrition evolves in language contact situations and identifies the morphological phenomena that influence transformations in the L. In line with the ‘social turn’ in both SLA and L attrition studies, Riehl defines language contact from both a psycholinguistic and a socio-linguistic perspective. In the former, the locus of contact is the bilingual’s brain, while in the latter the locus shifts outside the speaker. In either case, language contact is considered to be a long-term process, which leads to changes in both language systems. Riehl’s attention turns towards the language variety emerging in long-term contact situations. Investigations of L attrition have focused mainly on the impact of the L on the L in first generation migrants who have left the country of origin after puberty. While changes are first and foremost considered at the individual level, attrition research has recognized that within attriting communities, conventionalized language contact phenomena are also beginning to emerge. In time, and across generations, such contact-induced varieties may turn into fully functioning languages, where attrition-specific linguistic behaviours, such as hesitation phenomena and disfluency markers, are actually considerably reduced while contact phenomena increase. Using a distinction from the SLA domain, Riehl introduces the two notions of borrowing (‘incorporation of foreign features into a group’s native language’, Thomason & Kaufman, , p. ) and interference, which refers to the influence of an L on the L (Hickey, ; Matras, ). Riehl discusses how all the nuances of interlinguistic transfer are pertinent to the restructuring of the L under the pressures of a language contact situation. While typological studies indicate that, given enough time and intensity, no part of a language remains completely ‘borrowing proof ’ (Hickey, , p. ), certain structures are simply more easily integrated or lost than others. For example, content items are more susceptible to transfer, since such items are borrowed in a higher number and earlier than functional items. Riehl also discusses how the extent of borrowing is influenced by the properties of the languages in contact. German attriters in an English-speaking environment have a tendency to easily adopt and use adjectives such as happy, beautiful, etc. in a predicative structure, as no morphological interventions are required, rather than in an attributive context (compare da war er happy ‘so he was happy’ with *ein happies Kind ‘a happy child’). While rather easy to go unperceived by speakers, semantic transfer is a powerful phenomenon in language contact. Also known as pattern replication (Matras, ) or conceptual restructuring (Pavlenko, ; Schmid, a), transfer of meaning usually occurs with cognates or homophonous lexemes. Changes similar to those induced by language contact phenomena in an attrition setting can take place in monolingual varieties as well, but, as Riehl underlines, what sets migrant communities apart is the lack of normative control which allows for more norm-deviant usages to enter the variety. For language attrition, such effects on the language can be seen over a long period of time. If the first generation of L speakers inserts a high frequency of transferred items from the L in their L, the next generation will take over a language variety that is not only already limited to restricted domains, but also modified by inserted L material. Given the low norm-monitoring existent in migrant settings, the speaker’s metalinguistic awareness remains the only monitoring device against language transfer.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
. C
.................................................................................................................................. Language attrition does not take place in a vacuum, as the (purely linguistic) outcome of the meeting and interaction of two languages, it has to be seen, investigated, and understood as a process which is driven by context and circumstance, often in ways which are, as yet, poorly understood. The overview presented above and the chapters included in the present section underscore the necessity to approach the factors surrounding and driving the attritional process based on findings from other neighbouring disciplines. The complex interactions of these factors will in future necessitate large-scale studies, meta-analyses, and novel analytical and statistical instruments and tools, in order to further our understanding of the underlying processes of bilingual development.
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
......................................................................................................................
......................................................................................................................
. I
.................................................................................................................................. T is everything in language development. Numerous studies on first language (L) acquisition have repeatedly shown that even short delays in the onset of linguistic exposure may have drastic consequences for attained proficiency (e.g., Nittrouer & Burton, ; Mayberry et al., ). Likewise, evidence from second language (L) acquisition studies shows that if an additional language is learnt later in life, native-like attainment may be compromised (e.g., Johnson & Newport, ; Flege et al., ; Abrahamsson & Hyltenstam, ; Granena & Long, ). This finding is currently being complemented with an increasing number of studies on the effects of cochlear implants on spoken language development in congenitally deaf children (e.g., Sharma et al., ; Niparko et al., ; Houston et al., ). However, one phenomenon that is seldom thought of in the context of age effects in language development is language attrition. Age matters not only for the acquisition of language skills, but also for the retention of them: whereas children’s language skills may undergo extensive attrition within a short period of time, adults’ linguistic knowledge is much less susceptible to loss (e.g., Yeni-Komshian et al., ; Schmid, , ; Pallier et al., ; Köpke, b; Montrul, ; Bylund, a). This finding raises a series of questions: is there a specific age after which attrition becomes less drastic? Can age effects be counterbalanced by other factors? What are the underlying mechanisms of age effects? The aim of the present chapter is to review the available empirical evidence on age effects in language attrition, and discuss their interpretations. In doing so, the chapter seeks to formulate a set of specific principles regarding the role of age in attrition, which may ultimately serve to propel this area of research forward. The scope of the chapter is limited to the behavioural evidence for age effects, leaving the neurological evidence (to the extent that such evidence even exists) aside. Moreover, the chapter deals exclusively with age effects in L attrition, as opposed to L attrition. While the phenomenon of L attrition may provide further insights into the questions discussed here, age effects in L attrition are
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
intimately related to age effects in L acquisition, and, as such, too vast a topic to be covered in the current format. As to nomenclature, the term age of attrition onset, or simply age of attrition (henceforth AA), will be used to refer to the age at which the attrition process begins. This general term is preferable to other terms currently in use due to its lack of denotations about the factors underlying attrition (cf. age of reduced contact), the circumstances of attrition (cf. age of arrival), or the linguistic profile of the attriter (cf. age of bilingualism onset: international adoption shows that bilingualism is not a default outcome in attrition situations). Moreover, the term attrition itself is used here to refer to a non-pathological loss of previously acquired linguistic proficiency (e.g., Andersen, ; Köpke & Schmid, ), irrespective of the attriter’s age of onset. This usage thus contrasts with a more recent definition according to which the term attrition is reserved for speakers with AAs > years, and the term incomplete acquisition is applied to speakers with AA < years (Schmid, a). Lastly, the term seemingly complete loss will be used to describe the state of severe attrition whereby an individual no longer produces, comprehends, or recognizes his/her L, but fragments of intact linguistic knowledge may be detectable through unconscious response patterns and/or retraining activities (hence the qualifier ‘seeming’).
. E
..................................................................................................................................
.. Attrition in the absence of L contact The most severe cases of attrition are found among individuals who lost contact with their L in childhood. Here, international adoptees, who are transplanted from the L setting into a L setting and thus experience a complete cut-off in L input, exhibit the most extreme manifestations of attrition. Studies show that children who were adopted up until years of age may not, as adults, be able to utter or understand a single word of their attrited L, or even identify it in a larger sample of different languages (Hyltenstam et al., ; Park, ). Longitudinal studies have shown that attrition in international adoptees also proceeds at great speed. Isurin (), for example, examined language development in a girl adopted from Russia to the US at years of age. Within one year of adoption, the girl’s overall proficiency had become so poor that she was reluctant to partake in further testing (see also Nicoladis & Grabois, ). A pertinent question in the context of international adoptees is thus whether the L is completely erased from memory, or whether there are L remnants that can be detected through more sensitive methodologies. In an oft cited study, Pallier and associates () assessed this question examining Korean adoptees living in France (AA to years): using an fMRI (functional Magnetic Resonance Imaging) paradigm, they exposed the participants to Korean (along with other languages) to see whether any unconscious traces of this language could be found. However, the adoptees’ haemodynamic response patterns to Korean were identical to the French control group’s, suggesting an absence of such traces. Contrasting evidence was recently presented by Pierce and colleagues (), who, also
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
using fMRI, investigated L remnants in Chinese adoptees arriving in Canada as early as ; to ; years of age. Here, the adoptees were exposed to Chinese tones in a discrimination task, and their performance and response patterns were compared to native French speakers and non-attrited native Chinese speakers. Whereas the adoptees did not differ from either of the control groups in their ability to discriminate the tonal contrasts, their fMRI activation patterns bore a striking resemblance to the Chinese speakers’. Importantly, the degree of activation in the adoptees correlated significantly with their AA, such that participants with higher AA (e.g., years) exhibited greater activation. The fact that Pierce et al. found remnants, as opposed to Pallier et al., may reside in their targeting of specific linguistic features acquired early in life. Another factor that seems to play an important role for detecting L remnants in adoptees is relearning. Ventureyra et al. () and Hyltenstam et al. () studied discrimination of voice onset time contrasts in Korean in adult speakers adopted from South Korea between and years of age. Hyltenstam et al. targeted individuals who had relearnt Korean later in life, and found that they outperformed the novice learners.1 Ventureyra et al.’s study, in contrast, included no relearning sessions, and failed to detect an adoptee advantage. Moreover, Hyltenstam et al. found that the adoptees with the highest AAs (; and ;) achieved the highest scores on the task. However, a potential weakness with Hyltenstam et al.’s study is that there was no control of participants’ proficiency prior to relearning (see also Park, ), rendering conclusions about reactivation versus retention difficult. Similarly, Oh et al. () found that adoptees (mean AA ;) relearning Korean in a summer school outperformed their native English-speaking peers. Here, no AA effects were documented, which may be the result of a heavy bias in the group towards AAs < ;. Oh et al. could not safely attribute the adoptees’ performance to reactivation since no baseline data were collected. The problem of controlling for baseline proficiency was neatly tackled by Singh et al. (), who recorded real-time data of the reactivation process in individuals adopted from India between ; and ; years of age. Here, the adoptees were engaged in a supervised (re)learning task of coronal stops, divided into three blocks over one day. After the first block there was no difference between the adoptees and the controls (novice learners), whereas by the third block the former were outperforming the latter. Singh and associates detected a positive AA effect on relearning success, but the entanglement of this variable with length of residence and age at testing moots the interpretation. Whereas important insights can be gained from studying attrition in international adoptees, it is also a fact that adoption rates decrease with older children. This limits our knowledge as to how older children (AA > years) are affected by complete cut-offs in L input. One of the few studies that, to the best of my knowledge, addresses this question is Schmid’s () investigation of L German proficiency in Kindertransport survivors, who had left Germany at age to , had been adopted by English-speaking families in the UK, and had had no contact with German for upwards of fifty years.2 Such circumstances notwithstanding, this group was still communicatively proficient in German. Their complexity, accuracy, and fluency in German was compared to another group of L German 1 It should be noted, however, that the novice learners outperformed the adoptees on a grammaticality judgement task, suggesting a possible dissociation between phonology and grammar in reactivation. 2 The Kindertransport (Ger. ‘children’s transport’) was a series of efforts to bring predominantly German Jewish children to safety in the UK in 1938–9. The children were placed in British adoptive families.
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
speakers who emigrated at the same age and time with their families, and had remained in contact with German throughout their upbringing. Remarkably, the Kindertransport group exhibited only slightly poorer performance in the area of nominal agreement, but was otherwise comparable to the family migrant groups. This finding is thus in stark contrast with the accounts of adoptees whose AA is situated in the first decade of life. Instead, Schmid’s results find parallels in studies on adult attriters showing that even after decades of minimal input, L knowledge remains surprisingly intact (e.g., Köpke, ).
.. Attrition under reduced L contact Leaving aside young attriters who experience a complete cut-off in L contact, the manifestations of attrition in children who experience reduced L input are usually of a different kind. What characterizes attrition in this group is that it is, first, less drastic, with no cases reported of seemingly complete loss; and second, more variable, ranging from unreciprocal communication strategies (Gal, ), lexical retrieval difficulties and loss (Ammerlaan, ; Hulsen, ), syntactic violations, and morphological disintegration (e.g., Seliger, ; Turian & Altenberg, ; Vago, ; Bylund & Díaz, ), L convergence (Montrul, a), to behaviours that are indistinguishable from non-attrited native speech (e.g., Bylund, ). So far the reviewed evidence has concerned speakers whose attrition onset is either situated before or after years of age. To date, there is unfortunately a scarcity of studies dealing with broader AA ranges (e.g., spanning from childhood through to adulthood), the consequence of which is that unified assessments of attriters with different AAs but otherwise comparable backgrounds are a rarity. The extant evidence from these studies however points to a change in attrition outcomes at around to years of age, with speakers past this AA exhibiting lower levels of attrition. Bylund’s (a) study of conceptualization patterns and Yeni-Komshian et al.’s () investigation of global pronunciation found that speakers with AA > often performed on par with the native controls and, moreover, exhibited less variable outcomes (see also the studies on morphosyntax and lexis by Hakuta & D’Andrea, ; Silva-Corvalán, ). In Yeni-Komshian et al.’s case a negative correlation with L pronunciation proficiency was also found for speakers with AA < , whereas in Bylund’s study L proficiency was held constant. A methodological challenge involved in the study of early attriters concerns controlling for incomplete acquisition: if the break with the L setting took place before the linguistic structure under scrutiny was acquired, then non-convergent behaviour should not be characterized as attrition. Some studies compare the acquisition time line of monolingually developing children against the participants’ AA in an attempt to ascertain whether nonconvergence is a result of attrition or incomplete acquisition (e.g., Bylund, a; Pierce et al., ), whereas others have access to proficiency data (e.g., family videos, diaries) recorded prior to the participant’s break with the L setting (Vago, ; Yukawa, ). A consequence of uncontrolled pre-attrition proficiency is that incomplete acquisition will inevitably account for some of the varying outcomes typical of speakers with AA < .3
3 L incomplete acquisition occurs when contact with the L environment is reduced before native mastery has been attained. See further Part VI for overviews.
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
The levels of attrition documented in speakers whose AA is situated in the teens or later are no way near as drastic as those of younger attriters. Attrition in adults may nonetheless affect the domains of phonetics (e.g., de Leeuw et al., ; de Leeuw, Mennen, & Scobbie, ), syntax (e.g., Schmid, ; Gürel, a; Bylund & Ramírez-Galan, ), morphology (e.g., Jarvis, ), and lexis (e.g., Ecke & Hall, ; Schmid & Jarvis, ). What is interesting about this group is the levelling out of AA effects: whereas correlations between AA and L retention may be found in speakers with AA < (Yukawa, ; Hyltenstam et al., ; Singh et al., ; Pierce et al., ), no such relationships are reported by studies on older attriters (Köpke, ; Gürel, a; Schmid, ; Scherag et al., ; Bylund & Ramírez-Galan, ). One exception to this trend is de Leeuw et al.’s () study on prosodic attrition in ten speakers of L German with AA to , which documented a moderately strong positive correlation between AA and retention of L-specific prosodic elements. The available empirical evidence allows us to firmly conclude that language skills may attrite considerably more severely in children than in adults. Moreover, the evidence suggests that there is a turning point in the beginning of the second decade of life (i.e., around age ) where susceptibility to attrition is reduced. This can be concluded on the basis of, on the one hand, comparing outcomes in children who were adopted before and after this age, and on the other, the handful of studies assessing attriters whose AAs are located before and after this turning point. As to attrition susceptibility up to this turning point, there are some findings pointing to a gradual decline, whereas others fail to do so. Past the turning point, the existing evidence points to a minimal susceptibility to decline, possibly even a flattening out of the AA function. Being a relatively young research field, the evidence in support of these conclusions may in certain regards only be arrived at by comparing outcomes from different studies that use different methodologies to assess different linguistic structures and domains. This could, in many ways, be seen as a shortcoming, because the critical amount of data to formulate more specific hypotheses is lacking. However, the fact that findings from investigations on different linguistic phenomena that were acquired at different ages still point in the same direction can also be seen as further corroboration of the interpretation that age represents an important turning point.
. T
.................................................................................................................................. In what follows some frameworks attempting to explain age-related attrition outcomes will be reviewed and discussed. These are the impediment account, the maturational account, and the psychosocial account. While the first two have been previously formulated in the literature, the third account is presented here under this label for the first time. The intention of introducing the psychosocial account is to bring to the front one dimension that is seldom discussed, with the intention of deriving at testable hypotheses.
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
.. Impediment accounts In the findings reviewed above it is hard not to notice the parallels between age effects in language attrition and age effects in language acquisition. In the case of the latter, the beginning of the second decade of life, that is, puberty, has traditionally been seen as the turning point after which native-like proficiency can no longer be attained (Penfield & Roberts, ; Lenneberg, ). While recognizing that there is a period in life during which there is heightened plasticity in the language processing system, the so-called impediment accounts have suggested that age effects in L acquisition are actually an effect of L entrenchment, rather than neurobiological factors. An underlying assumption of these accounts is that the languages of the bilingual compete for a finite amount of memory and processing space, with the consequence that proficiency in one language impedes proficiency in the other. This logic then also applies to language attrition. The Interference Hypothesis (Pallier et al., ), which is the most clearly defined of the impediment accounts, holds that the L can be fully replaced by the L during the period of heightened attrition susceptibility: ‘when exposure to L ceases, then the [neural] network could somehow “reset” and L would be acquired fully’ (Ventureyra et al., , p. ). The Interference Hypothesis is based on Pallier and associates’ studies reviewed above, in which no traces of the L could be detected in a group of international adoptees. Another impediment account is the Speech Learning Model (e.g., Flege , ), according to which the phonic elements of the L and L exist in a common phonological space, and thus may exert mutual influence on one another. The extent to which the L influences L pronunciation depends on the strength of representation of L categories (in terms of developmental status), in the sense that the more strongly represented the L is, the greater its interference on L pronunciation. The findings on a negative correlation between L and L global pronunciation reported by Yeni-Komshian et al. () thus speak to this idea of interacting proficiency levels.4 Whereas the evidence above could be interpreted as supporting the impediment accounts, other interpretations are also possible. For instance, Yeni-Komshian et al.’s study provides no proof of a causal relationship between L and L proficiency, which was also recognized by the authors. In fact, their findings are just as compatible with the Speech Learning Model as with the possibility that those bilinguals with native-like L pronunciation simply had more contact with the L, whereas those with native-like L pronunciation had more contact with the L. Evidence from Bylund et al. () suggests that L and L proficiency may also be positively correlated, which would weaken claims of a negative causal relationship between the two. As to the Interference Hypothesis, subsequent studies have actually found that adoptees often retain remnants of the L (Hyltenstam et al., ; Oh et al., ; Singh et al., ; Pierce et al., ), which would reject the idea of a total L wipe-out in this population. It has also been found that the L of adoptees is not necessarily native-like (Norrman & Bylund, ), which rejects the other suggestion of the Interference Hypothesis. Interestingly, however, these findings could actually be interpreted in favour of the impediment accounts, since it may be claimed
4 There are additional impediment frameworks, but these focus only on language acquisition, leaving language attrition aside (e.g., Li & Farkas, ; MacWhinney, ).
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
that it is precisely because of the L remnants that the neural networks could not ‘reset’, and therefore nativelike L acquisition was not possible. Such an argument is difficult to falsify, as it could always be claimed that non-native-like L proficiency is due to undetected L remnants. A stronger case could be made for causality if inverse L–L proficiency levels are found whilst controlling for background variables, and/or longitudinally tracking the developing L–L trade-off. The findings to date then provide mixed support for the impediment accounts. While cross-linguistic influence is an undeniable reality of bilingualism at any level, it is questionable whether it can account for age effects in attrition on its own.
.. A psychosocial account As mentioned previously, some of the most important findings on the depth and rate of attrition come from speakers with adoptive backgrounds. The trajectory of this population renders them unique: international adoptees lose contact not only with their L setting, but also with their closest caregivers, whilst at the same time they acquire a L in the context of their adoptive families. This particular type of relocation/emigration is generally regarded as a strenuous, or even traumatic, experience (Berry, ). Moreover, some children have had very distressing pre-adoptive experiences (Johnson, ). There is, in other words, more than just the linguistic trajectory that sets this population apart from the typical subjects in an attrition study. A crucial question—though seldom mentioned in the literature—is whether the particular experiences lived by adoptees may exacerbate their language loss. It may, in other words, be argued that there is a potential confound at stake, since the deepest levels of attrition are documented in a group that not only was very young at the time of the L rupture, but also exhibits a rare and potentially traumatic trajectory. Psychosocial circumstances have indeed been shown to exert an effect on L maintenance, as showcased most clearly by Schmid’s () investigation on German L attrition in Jews who fled from the Nazi regime in the s and s (see also Cherciov, ; Bylund & Ramírez-Galan, ). It was found that participants with more traumatic experiences of ethnic persecution exhibited the highest rates of attrition, most likely as a result of psychologically suppressing their L German and the memories associated with it. The interpretation that traumatic and stressful situations may trigger language loss finds further support in evidence of more extreme forms of attrition. Preliminary reports (Ås, ; Fromm, ; Footnick, ) provide case studies of speakers who, due to negative memories, have suppressed their childhood language, but can retrieve (parts of ) it through hypnosis. In this regard, Footnick () proposes that if a language carries strong negative associations it may become inaccessible through ‘active forgetting’. If psychosocial factors alone can trigger the degrees of attrition found in adoptees, then it should be possible for adults to exhibit the same drastic L loss. Whereas the attrition literature reports no such cases, a look at the literature on psychogenic amnesia nonetheless yields some insights. Glisky and associates () describe a patient, F.F., who was aimlessly walking the streets of Tucson, Texas, with no recollection of who or where he was. When brought to hospital, he was found to speak English with a German accent, but failed to comprehend German. A series of behavioural tests along with fMRI testing revealed that his response to German stimuli was more similar to native English speakers
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
with no knowledge of German, than to a group of German speakers who were instructed to malinger. About a week later F.F. was deported to Germany, where he was immediately arrested and put in jail, allegedly on charges concerning his business. Glisky et al. suggest that the emotional turmoil that F.F. experienced around his business in Germany before going to the US could account for the linguistic blocking. Once back in Germany (approximately ten days post-hospitalization), F.F. was able to speak German again, although with some degree of dysfluency during the first weeks. Another example is the curious case of the US Navy veteran Michael Boatwright, which received attention in the media but was never scholarly investigated. In February of , Boatwright was found in a motel room in California, suffering from severe amnesia. He was unable to produce or comprehend English; instead, he communicated only in Swedish and claimed to have a Swedish name (he had previously lived and worked in Sweden). Preceding his apparent fugue state was a long period of extreme personal and financial stress and a suicide attempt (in Hong Kong and the US). Still seemingly unable to communicate in English, it was arranged that Boatwright move to Sweden, where he reconnected with his old girlfriend and found a job. Eight months later he committed suicide. It is now impossible to assess whether he really suffered from seemingly complete L loss, or whether he was malingering (some fugue patients have been found to simulate an extension of their initially authentic condition; Kritchevsky et al., ). The linguistic behaviour described in the two cases above suggests that psychosocial circumstances may lead to the L being blocked out. The question is, however, whether this is sufficient to claim that the extreme attrition attested in adoptees is not due to their young age, but rather to their unusual circumstances. There are at least two arguments that speak against a psychosocially based interpretation of age effects in attrition. First, in the case of F.F.—the only case that was systematically studied—the patient actually recovered the L within a short time frame, whereas in the case of children, only parts of the L may be recovered, be it through hypnosis or relearning (see above). In other words, there still seems to be an age-related difference in the degree to which the L system is affected by trauma, such that in adults it may render the L temporarily inaccessible, whereas in children it may trigger a (partial but) permanent loss. Second, if severe L attrition may be triggered solely by trauma, there should be more reports of adults having undergone seemingly complete loss. It may of course always be argued that there is no more evidence simply because adults do not often undergo sufficiently or similarly traumatic experiences, but such argument has no falsifiability. Moreover, the Kindertransport survivors studied by Schmid () were adopted under very traumatic circumstances (war and ethnic cleansing), yet the levels of attrition exhibited by this group do not compare to those found in younger adoptees. The strongest evidence in favour of a psychosocial account would be the case of a person who due to psychosocial circumstances exhibits seemingly complete loss.5 Should the psychosocial account prove successful in explaining age effects in attrition, it must be noted that the kind of attrition it is primarily concerned with is not induced by lack of L input (though this factor may play a role in the recovery process). Another consequence is that the qualifier ‘non-pathological’, which is central to current definitions of
5 Note that the case of F.F. does not qualify as such, since he quickly recovered full fluency without retraining engagements.
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
attrition, may not be readily applicable here, since some of the cases above (F.F.) are rather regarded as medical conditions.
.. The maturational account The maturational account bears a family resemblance to models aimed at explaining language acquisition success as a function of sensitive period regulation. This account revolves around the notion that the maturational state of the speaker regulates his/her susceptibility to attrition, as well as the responsiveness to extra-linguistic factors that modulate the attrition outcome. Though not always made explicit, it is often assumed that the neural plasticity in the cortical areas subserving language not only allows for a readily acquisition of linguistic knowledge, but also renders this knowledge vulnerable to loss (Bylund, b). According to the maturational account, the maturational state of the speaker renders him/her extremely vulnerable to attrition susceptibility up until approximately age . During this period, the speaker is highly dependent on advantageous social (e.g., L contact) and cognitive (e.g., language aptitude) factors that may counterweigh the heightened attrition susceptibility. Past this period, however, these factors play a less important role. The differential impact of social factors as a function of AA can be seen by comparing L retention in pre- (e.g., Pallier et al., ; Hyltenstam et al., ; Oh et al., ) and post-pubescent international adoptees (Schmid, ): whereas the former typically exhibit a seemingly complete L loss, the latter maintain communicative fluency. Likewise, the differential impact of cognitive factors is visible from studies showing that language aptitude is a predictor variable for L retention in prepubescent attriters (Bylund et al., ), but not in post-pubescent attriters (Bylund & Ramirez, ). The notion that timing—which is ultimately a proxy for the maturational state of the individual—regulates the impact of a given factor on a developmental event is at the heart of sensitive period theory (Fuhrmann et al., ). A crucial question then concerns the specific temporal and spatial properties of the maturation period. In view of the available evidence, at least three different temporo-spatial combinations can be conceived of. Figure . visualizes these combinations by plotting attrition susceptibility against age. Figure .A shows a minor decline up until puberty, followed by a major decline around puberty, and then a minor decline again. Figure .B is similar to Figure .A but depicts a flattening out of the susceptibility after the major decline. Figure .C shows a plateau of maximum susceptibility up until puberty, then another plateau of minimum susceptibility. Figures .A–C have in common, first, a major susceptibility decline around puberty, which seems uncontroversial in view of the existing data; and, second, this decline as depicted as being somewhat stretched out. This latter characteristic corresponds to the assumption that maturationally induced windows of susceptibility do not usually slam shut, but has a dragged out offset (Knudsen, ; Kral, ). Lastly, Figures .A–C also recognize that susceptibility is not reduced to zero, since attrition still occurs beyond the major decline. What remains to be solved as to the temporo-spatial properties is whether there is indeed a gradient before the major decline (as suggested by Yukawa, ; Hyltenstam et al., ; Singh et al., ; Pierce et al., ), and whether there is one after (as suggested by
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
A
B
C
D
.. Temporal and spatial properties of the maturation period: three different temporospatial combinations of language attrition susceptibility as plotted against age
de Leeuw et al., ). In contrast, should it turn out that there is instead a gradual decline from birth throughout life (Figure .D), this would be evidence of a non-discontinuous susceptibility. Such lack of discontinuity is often taken as counterevidence to maturational constraints (Hakuta et al., ; DeKeyser, ; Granena & Long, ; see however Hyltenstam & Abrahamsson, ). Just as the maturational account predicts heightened vulnerability to attrition up to a certain point in life, it predicts increased resistance to attrition past this point. From this follows that the seemingly complete L loss attested to date in adoptees should not occur past the major decline in attrition susceptibility. An adult who suffers from seemingly complete loss would constitute counterevidence to the idea that a maturational change reduces attrition susceptibility past puberty.
. P
.................................................................................................................................. Against the background of the findings and accounts discussed above, this section concludes the current chapter by outlining a set of principles regarding age effects in attrition that may serve as starting points (alternatively testing points) for future research. Principle . Reduced attrition susceptibility in adults does not imply that the L proficiency of this group is impervious to attrition:
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
An over-interpretation of age effects in attrition would be to suggest that adults are resistant to L attrition (e.g., Muñoz & Singleton, ). Such a reading goes against central findings in the attrition literature, and is reminiscent of the misunderstanding that the critical period hypothesis predicts that adults cannot attain high level proficiency in an additional language (they can, but not to a native-like level). Therefore, a case of an adult speaker exhibiting attrition cannot be taken on its own as evidence against age effects (see further Principle ). Principle . Heightened attrition susceptibility in children does not imply that this group is bound to undergo attrition: It is likewise important to keep in mind that young age equals high susceptibility to attrition, not actual attrition. Heightened susceptibility is, in other words, a necessary but not sufficient condition for attrition. Again, a parallel can be drawn to L acquisition, where a young age of acquisition onset does not guarantee native-like attainment; it simply affords the possibility. Therefore, the case of a child who experiences a reduction in L input but does not attrite cannot on its own be taken as evidence against age effects. Principle . Age effects in attrition entail that children and adults, ceteris paribus, differ in the degree to which they attrite: Age effects in attrition consequently reside in the relative degrees of attrition exhibited by children and adults. All other things being equal, children should exhibit higher degrees of attrition than adults. If such a pattern is not attested, then age effects are lacking. An important question that is open to further research concerns the mechanisms underlying age effects in attrition: whereas the different accounts outlined in the previous section make some attempts at elucidating this question, conclusive evidence is still lacking for either of them. An equally important topic is that of selectivity in age effects. In the acquisition of linguistic knowledge, different subsystems and structures are differently affected by age. It is therefore natural to assume this may also be the case in the attrition of linguistic knowledge. Findings into the matter of selectivity will help avoid hasty conclusions about whether age effects exist or not, and instead contribute to forge a nuanced picture of the age factor. Insights into the effects of age on the stabilization of linguistic knowledge will doubtless contribute to our understanding of the role of timing in language development.
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
......................................................................................................................
......................................................................................................................
.
. I
.................................................................................................................................. T study of first language attrition, like that of other phenomena of bilingual development, is a study of increased variability—within a community, within a speaker, or within a particular linguistic feature. Comparisons of attrited and non-attrited data almost invariably find a wider range of scores among the former (Schmid, a). One of the most fundamental tasks for attrition studies is therefore to investigate what drives this variability, both at the linguistic level (see Part III of this volume) and at the level of the speaker: what exactly is it that causes the L skills of some attriters to deteriorate more strongly than those of others? The intuitively obvious answer to this question would seem to be that the speakers with less attrition phenomena will be the ones with a shorter period of residence, more opportunity to use their L, or both. This assumption is based on theories of memory and activation of knowledge, and it is reflected in the wide range of studies and publications on L attrition who take the impact of these factors to be all but axiomatic. However, the empirical evidence points to a picture that is far more complex and often at odds with this notion. The present chapter will first review the theoretical underpinnings of the view that frequency and recency should be determining factors for language attrition and reduced accessibility. It will then go on to discuss some of the empirical evidence on the role of these factors, and offer some considerations on how the gap between theory and evidence may be reconciled.
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
. M A T H
.................................................................................................................................. The assumption that there should be a strong relationship between amount of use, length of residence (LOR), and attrition is based on the fact that it ‘is a general condition of neurocognitive functions that when they are not being used, they become more difficult to access over time and eventually atrophy’ and that ‘neurocognitive systems require stimulation to remain in good working condition’ (Paradis, , p.). This ‘use it or lose it’ character of memory representations is widely recognized. However, linguistic knowledge—in particular multilinguistic knowledge—is different from other knowledge in many specific ways, and these differences may have unexpected and complex consequences for language attrition. Ecke () provides a detailed outline of theories of memory and their relevance for language attrition research. For the purpose of the current chapter, the most important of these are decay, repression/suppression, and interference. The notion of decay (as Ecke notes, probably the oldest of all approaches to forgetting) underpins early assumptions that language attrition is ‘a kind of forgetting’ (Ammerlaan, , p. ): according to this theory, when information is not used, the ‘trace’ it has left in the brain gradually dissipates so that the information ‘evaporates’ (Ecke, , p. ). This is the most straightforward and common theory of language attrition, and it is echoed widely in particularly the early literature, despite the fact that until around the year , there were almost no attempts to provide an empirical link between amount of use and amount of attrition (see below). The second mechanism discussed by Ecke relevant here is repression/suppression, ‘an (at least initially) intentional mechanism in which the individual refuses recalling/using a memory structure’ (Ecke, , p. ). In psychology, this strategy has most often been discussed in the context of traumatic experiences, but it can also involve the withdrawal of attention from highly active memory structures in order to allow the retrieval of other, less active structures (Ecke, , p. ). Suppression goes hand in hand with interference, where competing bits of information stored in memory can inhibit or block recall of each other. Interference can be retroactive, where newly acquired information blocks information acquired earlier, or proactive, where information already represented in memory can disrupt learning and recall (Ecke, , p. ). These two mechanisms—decay on the one hand and suppression and interference on the other—correspond closely to the two main components of Paradis’ Activation Threshold Hypothesis (ATH): activation and inhibition. According to the Neurolinguistic Theory of Bilingualism (Paradis, , ) the Activation Threshold (AT) of any mental representation (the amount of neural impulses required to activate it) is determined by the frequency and recency of its activation events: Items become more easily accessible the more often they have been recalled, but the threshold of even highly accessible items rises over time if no such recall occurs, making accessing them more effortful after a prolonged period of inactivity (Paradis, , p. f.; see also Köpke & Keijzer, Chapter , this
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
volume. The ATH has frequently been used as a framework for L attrition studies, e.g., Köpke, ; Gürel, ; Schmitt, ). To illustrate the necessity of use and stimulation for the maintenance of skills, Paradis points out that ‘[c]oncert pianists and professional tennis players cannot skip practice with impunity’ (Paradis, , p. )—that is, even extremely finely honed and deeply entrenched skills are liable to degrade fast, unless used. While this basic truth about human cognition and memory is not in dispute, the characteristics of the linguistic system, particularly in bilinguals, make the picture somewhat more complex: as Sharwood Smith & Van Buren () point out, what we find in a language attrition context is typically not a situation of complete non-use of a particular skill, as would be the case if someone was removed entirely from linguistic interaction, for example stranded on a desert island or placed in long-term isolation. Even in such a dramatic scenario (which eludes scientific investigation for obvious reasons) it is doubtful if the absence of other human interlocutors would really entail the complete absence of use of linguistic functions. Humans use their linguistic skills constantly and in various ways—they talk to each other and themselves, they write to each other, they write diaries, they read, they think, they subvocalize, etc., so even hermits, castaways, or prisoners will experience stimulation of their linguistic skills (the Crusoesque character played by Tom Hanks in the film Cast Away talks to a volleyball he calls Wilson). The picture becomes more complex when we look at people who continue to use their linguistic abilities at a normal level, but branch out into two or more languages, so that each of them by necessity is used less than would be the case for someone who uses the same language for all purposes (available time, alas, does not increase to match available languages). Since the multilingual’s linguistic subsystems do not exist in isolation but are connected at multiple levels (e.g., Paradis, ; Schmid & Köpke, a), any kind of use of any language means that all other languages will receive a certain amount of stimulation. This is arguably different from the experience of the concert pianist or tennis player, who probably cannot exercise and maintain their specialized skills through different though related activities in the same fashion. The interconnectedness of the language subsystems means that the activity and accessibility of each subsystem exist in a complex pattern of processes related to memory retrieval, suppression, and interaction. As Paradis further points out, the AT is not only affected by activation of the target linguistic item or language, but also by the activation of competing information, for example translation equivalents or similar syntactic constructions: in order to allow a target representation to be activated, its competitors have to be inhibited, and this inhibition event also raises the AT (Paradis, , p. ). The latter observation points to the importance of contextual factors: Habitual language use in dual language contexts (where code-switching is appropriate and frequent) impacts overall cognitive control as well as language control skills differently from use in single language contexts, where one language is strongly inhibited (cf. Green’s model of the behavioural ecology of bilingualism; Green, ). This suggests that a detailed and fine-grained picture of individual language habits in different contexts, across a range of languages and across time is necessary in order to assess the impact of L use and LOR on language attrition. To make things yet more complex, the linguistic level which is affected should furthermore be taken into account: Paradis () predicts that the lexical level, which is subserved by declarative memory, will be affected more by diminished use than levels subserved by procedural memory, such as grammar or phonetics.
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
. W L ?
.................................................................................................................................. In early research on language attrition, the effect of potential variability of L use within attriting populations was rarely addressed or quantified. The underlying assumption was that the condition of being an immersed bilingual—someone living in an environment where a different language is spoken—and the concurrent reduction in opportunity to use the L would be the main driving force of L attrition (e.g., Andersen, ; Huffines, ; Olshtain & Barzilay, ; Seliger & Vago, a; Major, ). In such approaches, ‘L use’ is thus reduced to a single, dichotomous variable where ‘monolingual use’ corresponds to a % baseline, while use by immersed bilinguals is reduced by an unspecified amount, but assumed to converge towards zero (e.g., ‘individuals whose lives have changed in such a way that they have little or no opportunity to use their native language, perhaps for most of their adult years’; Huffines, , p. ). It is also implicitly assumed to be a homogenous factor among attriters and to represent the main or even sole distinction between attriters and controls (in the initially rare cases where control groups were even used and not merely postulated as an idealized baseline, see Schmid, a). It was only in the s, when the first investigations began to assess the impact of L use as a within-group predictor of variability at the individual level, that it became evident that the link between use and attrition was anything but straightforward. Early studies typically established a dichotomous classification among attriters into high(er) and low(er) levels of use (e.g., Jaspaert & Kroon, ; de Bot et al., ; Köpke, ) which was usually arrived at by conflating a number of variables—the native language of the partner, the number of contacts to other speakers, etc.—into a single, binary factor (in the Köpke, , study, many of these were also analysed separately, but did not yield significant results). These approaches were supplemented by some qualitative investigations (e.g., S ndergaard, ). To the surprise of many, these studies already indicated that the role of frequent L use was much less important than had been assumed, and that the picture might be more complex (see, e.g., the finding that there is a non-linear interaction between LOR and levels of use reported by de Bot et al., , confirmed in Soesman, , and elaborated in Schmid, b). S ndergaard’s conclusion that the common sense based assumption that the immigrants’ competence in Danish in general depends on a) the length of their stay in the new country, b) the extent of their use of Danish in the new environment, has been disproved. (p. )
though probably somewhat overstated, thus holds true across these and many later studies. Similarly, the role played by LOR in the context of attrition in general and the link between LOR and L use in particular has long been underestimated. Originally, the assumption was that attrition would be a very slow process, setting in only in extreme cases after a very long time, as evidenced, for example, by Waas’ surprise to have found attrition ‘even after a stay of only to years’ (, p. ). This view has shifted to the current consensus that attrition is a process which mainly takes place within the first ten years, after which the L will ‘stabilize’ again at whatever level of attrition has been reached during this time (e.g., de Bot & Clyne, , p. ; Opitz, a, p. f.; Schmid, a, p. f.; Baladzhaeva, , p. ). This assumption is rooted in the evidence of an investigation
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
by Clyne in and of a population of long-term Dutch migrants in Australia, which found little change between the two testing moments (de Bot & Clyne, ), as well as on the absence of time effects in most other studies looking at long-term immigrants (see below, Section ..). Based on these findings, Schmid (b, p. ) recommended that attrition studies should focus on periods of residence of fifteen years or more, in order to allow them to find ‘relatively stable effects that are not the outcome of a highly active development of an L’ (Schmid, a, p. ). This well-meaning but misguided piece of advice has had the regrettable effect that to date there is extremely little evidence on how attrition may progress during those first years. The picture is complicated even more by the fact that the circumstances which allow migrants to maintain contact with the language of their home country have changed so dramatically over the past two decades. For most of the migrants who left their country of origin in the years or decades after World War II, usually to relocate to North America or Australia/New Zealand (the target population for a large proportion of studies of attrition), contact with ‘home’ was an extremely rare phenomenon. Travels, phone calls, foreignlanguage books, newspapers, or films, etc. all were prohibitively expensive and enjoyed once in a blue moon, if ever. In the mid-s, this began to change rapidly and this means that the amount of contact and input available to migrants with longer or shorter periods of residence—in particular during those early years which have so often been postulated as ‘crucial’—is at the present time inevitably confounded with the length of time since arrival. If LOR and amount of L use as predictors of the attrition process do indeed exist in nonlinear dependency, as proposed by de Bot et al. (), this presents a challenge to attrition studies which is hard to solve. The lack of insight into the role of these and other factors prompted a group of scholars from onwards to develop a common approach towards eliciting them, in order to allow more finely-grained analyses of larger datasets (see Schmid & Köpke and Schmid & Cherciov, Chapter and Chapter , this volume, respectively). In the context of this network, the Language Attrition Test Battery (LATB) was developed and subsequently applied in a range of investigations, and one of the instruments it contains is a detailed questionnaire for the elicitation of personal background factors (all materials and documentation are available on https://languageattrition.org/resources-for-researchers/ in a variety of languages, and the test battery is introduced and discussed in Schmid, a). An investigation of the responses elicited by these reports by means of Principal Component Analyses (Schmid & Dusseldorp, ) suggested that L use factors should be classified into the following categories: • reported use of the L in informal situations where code-switching is not inhibited (with partner, children, and friends) • reported use of the L in situations where the L is highly active but suppressed and code-switching is inappropriate (e.g., in work-related contexts or in heritage language clubs or churches) • reported (passive) exposure to non-attrited L (reading, media, visits to the L country) These contexts align with Grosjean’s Language Mode model (Grosjean, ) as well as with Green’s model on the behavioural ecology of bilingualism (Green, ) and thus provide a promising approach for studying the impact of L use on L attrition in more detail.
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
. T
.................................................................................................................................. One of the reasons why it has been difficult to attain a comprehensive understanding of the impact of external factors on language attrition is the fact that existing investigations have addressed different populations and different linguistic levels by means of different tasks and elicitation methods, and in addition have often assessed L use in different ways. Since all of these factors can be expected to have an impact on the pattern of results, sweeping generalizations across studies are less than helpful. In order to help gain some more differentiated insights, the present section provides an overview of the current state of empirical findings. This overview is based on those twenty-three studies of which I am aware which empirically and quantitatively investigate the impact of level of L use and/or LOR.1 Together, these studies report on forty-nine tasks or elicitation methods with over , experimental participants and over , controls (see Appendix, Table .). The largest proportion of analyses (seventeen) investigate the lexicon, and these are typically also the largest studies with an average population size of n > . The second largest group is represented by tests of general proficiency (eleven studies, average n = ), fluency and accuracy in free speech (eight studies, average n > ), phonetics (six, average n = ), and morpho-syntax (six, average n = ).2 In line with the proposals made in the LATB, the most popular task to investigate lexical access is the Verbal Fluency Task (VFT), where participants are asked to name as many items as they can from a particular semantic field or conforming to a particular formal criterion (i.e., words sharing the first letter) within a given amount of time (usually one minute). Seven studies employ one or more semantic VFTs and two further investigations use formal VFTs in addition to the semantic ones. Almost as frequent are investigations of lexical diversity in free speech (seven studies, mainly collected through interviews and/or the film retelling task proposed in the LATB), most commonly by means of the vocabulary diversity (VOCD) score (McKee et al., ). General proficiency is most often assessed by means of a C-Test (five), but some other tasks are also employed (for details, see Table ., Appendix). I would like to stress here that the present overview is not a rigorous meta-analysis (nor was it intended to be, as such an analysis would be beyond the scope of a Handbook chapter), but rather tries to identify overall trends and patterns on an impressionistic basis. There are many ways in which one might approach the wealth of data present in these studies. I have chosen here to do this based on (statistical) analyses, not on studies. As such, the overview could legitimately be argued to be skewed: many studies report more than one measure (e.g., a range of proficiency tasks used as predictors and an electroencelphalogram (EEG) experiment, as is the case in Kasparian & Steinhauer, ) and in some cases, the same population is the basis for various studies (e.g., Kasparian & Steinhauer, b, ; 1
I have tried to be as comprehensive as possible in this overview. If I should have overlooked any relevant studies, I apologize to the authors; no offence was intended. 2 The overview furthermore includes one study on semantic processing (n = ; Kasparian et al., ) but since this is the only study investigating this category it will not be considered further in this overview.
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
or Schmid & Dusseldorp, ; and Schmid & Jarvis, ). The findings of the separate analyses are thus influenced by the characteristics of the specific population studied and therefore not independent of each other. However, since it was my aim with this overview to provide some insight into how L use differentially affects L attrition at various levels of language processing and use, I found this approach to be preferable to one which would look more globally at findings reported—across tasks, linguistic levels, and analyses—by different studies. That notwithstanding, all of the insights reported here should be treated with caution and are mainly intended to inform future research agendas.
.. Findings on L use At first glance, the number of analyses reporting an effect of L use () and those which do not ()3 seem to be fairly equally distributed. However, a closer inspection of the data shows that the latter group includes five measures on which attrition and control group did not differ significantly. In the absence of a group effect (that is, a main effect of the diminished exposure characteristic of an attrition setting), it is doubtful to what extent measuring the impact of individual exposure variables may contribute to our understanding. For the purpose of this overview, I count any such studies under the rubric of ‘no impact of L use’, and we therefore arrive at the first impression that L use does not have a significant impact on individual attrition in about two-thirds of the analyses reported here. Looking more closely at the distribution across different areas of language proficiency (Table .) and keeping in mind the small number of studies in some areas, the impression is that L use does not affect any aspect of language use or linguistic level dramatically more or less than any other and generally seems to reach significance in a third of all experiments, with only fluency being affected somewhat more reliably (but still at chance level). An exception might be general proficiency, where the impact looks to be somewhat more restricted than in other areas (possibly because in these areas the overridingly important
Table .. The impact of L use on individual L attrition in the analyses listed in the Appendix no impact of L use
Accuracy & Fluency Lexicon Morpho-syntax Phonetics General proficiency
impact of L use
n
%
n
%
. .
. .
3 Two of the studies included in this overview—Scherag et al. () and Gürel ()—only report the effects of LOR, not of L use, and do thus not enter into these calculations.
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
predictor is generally education). On the other hand, the prediction that the lexicon should be affected more strongly by linguistic habits than morpho-syntax or phonology does not seem to be supported by these findings. The average size of the experimental population in studies in which L use did not reach significance was ., while studies that did find such an effect tested were somewhat larger, with an average of . participants, suggesting that in at least some investigations the absence of an effect might be a Type II error due to insufficient power. A further interesting insight is that, of the fifteen experiments that did report an effect, eight cases do not specify the type of L use any further while the other seven make a classification into contexts in which code-switching or -mixing can be expected (informal use with bilingual friends) and those where it is unlikely (mainly use in professional settings). All seven studies find that the latter type of use facilitates performance in the L while only four find such an effect for the former.
.. Findings on the impact of length of residence Of the forty-nine analyses summarized in this overview, forty-one report LOR as a predictor. In twenty-nine cases, the impact of LOR is not significant, while twelve report a time effect on the outcome variable. In one case (de Leeuw, ) this effect is negative: speakers with more years of residence perform closer to the native norm; while in another three, LOR has an effect despite the fact that there is an absence of any main effect of group. One example of such an effect is the strange finding reported by Schoofs (), who finds a strong and highly significant correlation in her data between LOR and performance on a VFT, where an increase of the former by one year implies a decrease in performance on the fluency task by . named items (p. ). Based on the data Schoofs reports in her appendices, I have plotted the association between the dependent and the independent variables (Figure .). This chart suggests that, while the time effect is indeed strongly visible, it takes some of the attriters over forty years to drop to the range of the native baseline. Such rather puzzling findings aside, an interesting effect emerges from the scrutiny of the available data: in every case where there is both a significant attrition effect and a significant impact of LOR, the minimum LOR represented in the sample is smaller than ten years. There are, of course, also investigations with such short periods of residence which show no effect of LOR for some features (e.g., most of the proficiency tasks reported by Kasparian, , or the lexical access and diversity data investigated by Badstübner, ). On the other hand, none of the investigations which report a minimum LOR of ten years or higher find a significant effect.
. D
.................................................................................................................................. The findings summarized above offer some, albeit limited, insight into the role of L use and LOR for the development of attrition phenomena. First, it is probably fair to say that in order to gain a better understanding of how processes of language maintenance and attrition are shaped, we need more, larger, and more varied studies while still adhering to
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
Group German attriters English attriters English controls Swiss German controls Standard German controls
40.00
VFT
30.00
20.00
10.00
.00 .00
20.00
40.00
60.00
80.00
LoR
.. The impact of length of residence on performance on the Verbal Fluency Task, as reported by Schoofs ()
established frameworks of elicitation and measurement of these predictors. Limited statistical power, for example because of small sample sizes, can prevent us from seeing the larger picture. Furthermore, it is important that we should continue to assess L use in a more differentiated way than has previously been done, and according to a framework which will allow direct comparisons between studies. Different kinds of use and exposure feed differently into L accessibility and maintenance. Lastly, more attention needs to be given to the way in which external factors, including but not limited to the ones discussed here may interact in complex, unpredictable, and possibly non-linear ways. With respect to the first point made above, it is interesting to note that, with few exceptions (such as the study by Schoofs discussed above), in the studies investigated here the control group does outperform the attrition group numerically on most measures, even though the difference in average scores does not always reach significance. In this context, the study by Opitz (a) is relevant: she combines a number of measures relating to both formal assessments (a C-Test) and free speech. While group comparisons for these individual measures come back as non-significant, the overall measure does reliably distinguish the attrition and control group. Furthermore, she finds a moderate correlation between this combined measure and a similarly broad measure of L use. These results indicate the possibility that the low impact of this factor for L maintenance and attrition
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
suggested by the overview above might be an underestimate of the actual picture. It should furthermore be noted that investigations which are limited to a single linguistic community (as is almost invariably the case to date) may simply be dealing with experimental groups that do not have sufficient internal variance on the predictors. For example, Varga () ascribes the lack of a result of L use on L attrition in his investigation on Hungarians in Denmark to the fact that the community is quite homogenous in their levels of L use (or, rather, the absence thereof, p. ). Comparisons across studies, populations, and languages thus might not only provide more statistical power due to the higher numbers, but also more variance in terms of these experiential factors. With respect to the second consideration above, it is important to keep in mind that treating L use as cumulative may obscure the picture, since different kinds of use and different contexts may impact on different skills in different ways. For example, it has been shown that frequent use of the L with other bilinguals in casual, informal interactions may facilitate some forms of L attrition—such as the development of a foreign accent—over time (Schmid, c) in what is likely to be an instance of accelerated contact-induced change not experienced by speakers with less interactions with other bilinguals (see also Grosjean & Py, ). On the other hand, it has been shown that L use for professional purposes may affect very specific parts of the linguistic repertoire: Yılmaz & Schmid () find that, while there are on the whole few differences between their experimental and control groups on lexical access, those attriters who use their L for professional purposes are faster to name low-frequency items in a picture naming task. It makes sense that this kind of language use should facilitate the accessibility of rare items more than, for example, a parent using infant-directed speech a lot. On the other hand, this latter form of L use might have quite different benefits: it is possible, for example, that the exaggerated pronunciation, slower speech rate, and frequent repetition characteristic of this speech style might feed positively into other linguistic features, such as the maintenance of a native-like accent. In a similar vein, the findings presented above on the impact of L use particularly in the first years of residence, while intriguing, need to be taken with more than a pinch of salt. Since all of these studies were published after , it is likely that participants still within their first decade of residence in a foreign context at the time of data collection had emigrated after new communication technologies and cheap travel had made it much easier to maintain contacts with their social networks in the country of origin. Of course, these opportunities became available to migrants with longer periods of residence around the same time. Whether the findings reviewed above support the long-held view that attrition really does take place mainly within the first decade, or whether they are the outcome of interactions with other variables (speakers with shorter periods of residence are likely also younger, which may impact on their levels of use of recent technologies, speakers who have emigrated recently have a larger social network in the L community on which they can draw than speakers who have been cut-off from this network at the time of emigration several decades ago, etc.) remains an open question, which can only be resolved by future investigations of the longer-term developments among speakers with higher levels of exposure during the early years. Lastly, it should be noted that these and other factors exist in a complex web of interdependencies, and how they may combine to either amplify each other or cancel each other out remains to be determined. Recent advances in statistical methods as well as theoretical frameworks may be useful in this respect.
Table .. An overview of twenty-three studies investigating the impact of amount of L use and length of residence on first language attrition area of investigation
linguistic property/ elicitation measure method/task
study
LOR
n exp
n CG
main effect of impact of attrition L use use comments
Accuracy
errors
CC
Opitz a
–
yes
some
cumulative impact of use — variables on cumulative z-scores from variety of tasks
accuracy
errors
CC
Schmid & Dusseldorp,
–
yes
yes
canonical correlation: L at work, in clubs and churches
accuracy
errors
interview & CC
Badstübner,
.–
yes
yes
all three groups different yes from each other for lexicon, semantics and morpho-syntax. but not for function words, morphology and syntax. Reported use significant, but passive input ns
fluency
disfluency markers in free speech
CC
Dostert,
+
yes
no
no
fluency
disfluency markers in free speech
CC
Opitz, a
–
repetitions only
some
fluency
disfluency markers in free speech
CC
Schmid & Beers – Fägersten,
yes
no
no
fluency
disfluency markers in free speech
CC
Varga,
yes
no
no
–
impact LOR of LOR comments
yes
cumulative impact of use — variables on cumulative z-scores from variety of tasks
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
disfluency markers in free speech
interview & CC
Badstübner,
.–
yes
no
— both bilingual groups different from CG but not from each other
lexicon
lexical access
PNT
Yılmaz & Schmid,
–
no
some
no L at work => faster naming of low frequency items
lexicon
lexical access
VFT (semantic & phonetic)
Badstübner,
.–
yes (but not for phonetic fluency)
yes
German instructors same no as CG, other professionals different from both groups. ‘Overall L use’ and ‘Use in spare time’ also significant (but confounded with group difference)
lexicon
lexical access
VFT (semantic & phonetic)
Baladzhaeva,
–
no
no
—
lexicon
lexical access
VFT (semantic)
Dostert,
+
no
no
no
lexicon
lexical access
VFT (semantic)
Kasparian, –
no
yes
(use significant for overall no proficiency but not individual tasks except VFT)
lexicon
lexical access
VFT (semantic)
Schmid & Dusseldorp,
–
yes
yes
impact of L use for professional purposes
lexicon
lexical access
VFT (semantic)
Schmid & Jarvis,
–
yes
yes
weak impact of L use for no professional purposes
lexicon
lexical access
VFT (semantic)
Schoofs,
–
no yes, but opposite: attriters name more items than controls
yes
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
fluency
no
(continued )
Table .. Continued linguistic property/ elicitation measure method/task
study
LOR
n exp
n CG
main effect of impact of attrition L use use comments
impact LOR of LOR comments
lexicon
lexical access
VFT (semantic)
Varga,
–
yes
no
no
lexicon
lexical access
VFT (semantic)
Waas,
–
yes
no
—
lexicon
lexical diversity (VOCD, frequency bands)
interview
Yılmaz & Schmid,
–
yes
no
no
lexicon
lexical diversity (various measures)
interview & CC
Schmid & Jarvis,
–
yes
no
no
lexicon
lexical diversity (VOCD)
interview & CC
Badstübner,
.–
yes (in yes interview but not in CC)
CG different from Other no Professionals, German Instructors intermediate. ‘Use at home’ significant (not confounded btw groups)
lexicon
lexical diversity (VOCD)
CC, picture description
Dostert,
+
no
no
lexicon
lexical diversity (VOCD)
CC
Opitz, a
–
no
some
cumulative impact of use — variables on cumulative z-scores from variety of tasks
lexicon
lexical diversity (VOCD)
interview & CC
Schmid & Dusseldorp,
–
yes
yes
canonical correlation: impact of L use in family, at work and in clubs/churches
lexicon
lexical diversity (VOCD)
CC
Varga,
–
no
no
no
interpretation, picture identification
Gürel,
–
yes
no
morpho-syntax binding in pronoun truth value resoluation judgement,
no
yes
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
area of investigation
Yılmaz,
–
some
morpho-syntax grammatical gender Lexical Decision Task
Scherag et al.,
–
no
morpho-syntax number agreement
GJT + EEG
Kasparian et al., –
yes
yes
no behavioural differences, no but differences in EEG signal; speakers with more Italian exposure/use were more native-like
morpho-syntax relative clauses
GJT + EEG
– Kasparian & Steinhauer,
yes
yes
higher levels of Italian exposure/use influences acceptability rating
morpho-syntax various
GJT
Varga,
–
no some (two out of seven constructions significant)
no
phonetic
foreign accent
global foreign Varga, accent ratings
–
yes
no
no
phonetics
foreign accent
global foreign de Leeuw et al., accent ratings
–
yes
yes
no impact of L use in contacts where there is no language mixing
phonetics
foreign accent
global foreign Hopp & accent ratings Schmid,
–
no
no
no
phonetics
foreign accent
global foreign Bergmann et al., .– accent ratings
yes
yes
yes
morpho-syntax complex embeddings
interview
no
no no
yes
visitors (LOR –) perform identical to long-term attriters (LOR –)
(continued )
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
speakers with more years outside Italy were less native-like in EEG
Table .. Continued linguistic property/ elicitation measure method/task
n exp
n CG
main effect of impact of attrition L use use comments
impact LOR of LOR comments
phonetics
pronunciation of liquids
acoustic Bergmann et al., – measurement
yes
no
yes
phonetics
pronunciation of liquids and vowels
acoustic de Leeuw, – measurement
yes
no
no
study
LOR
proficiency
C-Test
Dostert,
no
no
yes
proficiency
C-Test
Kasparian, –
+
ns
no
(use significant for overall no proficiency but not individual tasks except VFT)
proficiency
C-Test
Opitz, a
–
no
some
cumulative impact of use — variables on cumulative z-scores from variety of tasks
proficiency
C-Test
Schmid & Dusseldorp,
–
yes
yes
impact of L use for professional purposes
proficiency
C-Test
Varga,
–
proficiency
errorKasparian, – detection task
no
yes
no
no
ns
no
(use significant for overall yes proficiency but not individual tasks except VFT)
neagtive impact:longer residence => marginally more nativelike L frequency on liquids
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
area of investigation
FSI
de Bot et al.,
+
proficiency
Scrabble task
Dostert,
+
proficiency
self-reports
Kasparian, –
proficiency
Can-Do Scales
Dostert,
+
proficiency
Can-Do scales
Waas,
GJT + EEG
Kasparian & Steinhauer,
semantic processing
similar sounding words
n/a
yes
correlation FSI rating and yes contact factor
no
no
no
ns
no
(use significant for overall no proficiency but not individual tasks except VFT)
no
yes
no
–
yes
–
some
interaction LOR and contact
— yes
yes no behavioural differences, but differences in EEG signal; speakers with more Italian exposure/use were more native-like
speakers with more years outside Italy were less native-like in EEG
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
proficiency
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
......................................................................................................................
, , ......................................................................................................................
̈
. I
.................................................................................................................................. D to unprecedented transnational movement and globalization across the world, the number of people with diverse forms of multilingual knowledge is constantly growing. Among these, migrant populations who intend to settle in the new country as permanent residents or naturalized citizens or take up employment as migrant workers are of special interest since they are faced with increasing pressures to rapidly and completely shift to L and adopt the cultural norms of their host societies. In this chapter, we seek to explore the extent to which the bilingual’s native language is affected by the development of L knowledge and the attachment and identification to the native versus the host language and culture. Despite the presence of abundant research on integration with a particular focus on L attainment and an increasingly growing literature on L development in attrition contexts, these topics have been largely studied independently from each other. This chapter brings together relevant aspects of the relationship between L attrition and L acquisition, covering psycholinguistic issues (the bilingual mind), socio-linguistic issues (the bicultural mind), and language policy issues as they pertain to the L attrition and L acquisition in order to contribute to the understanding of language development and change in migrant settings.
. T
.................................................................................................................................. It is assumed that the native language, once completely acquired, would be immune to change. However, it has been clearly shown that all speakers who live in bilingual contexts and/or routinely use more than one language experience a certain amount of change in the L partly through non-use and partly through interference from the dominant contact language. Attrition studies have consistently documented that bilinguals tend to experience
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
, ,
a decrease in their productivity and creativity in their native language and may speak with a foreign accent (Cook, a; Schmid, a). Psycholinguistic investigations have greatly contributed to our understanding of how the native language is affected by the dynamics of bilingualism. They have provided corroborating evidence for the presence of a crosslinguistic interaction between the languages in both lexical and syntactic processing starting from the initial stages of learning (see Bergman, Sprenger, & Schmid, ; Coderre, , for review). For example, in word production, cross-language cognates were named faster by Spanish-Catalan bilinguals compared to non-cognate items (Costa et al., ; for reviews, see Friesen et al., ; Hameau & Köpke, ), whereas auditory distractors that are phonologically related to the L translation of the picture significantly delayed the response in the L, indicating that the L translation is active at the same time (Hermans et al., ). In a study on relative clauses, Dussias & Sagarra () found that native Spanish speakers who had extensive immersion experience in their L English environment parsed Spanish sentences in the way native English speakers would. It has also been argued that structures that share the same word order are shared across languages (Bernolet et al., ; for a review of non-selective syntactic activation, see Sanoudaki & Thierry, ). Co-activation is not limited to the lexical and syntactic domain but also affects phonology. It has been proposed that the way in which bilinguals carve up the phonetic space appears to differ from that of monolinguals, affecting both perception and production in either language. In his classical study on phones, Flege () demonstrated that FrenchEnglish bilingual speakers merge L and L categories as evidenced by their intermediate pronunciation of some phones in both languages, and similar findings have often been reported since (for a recent overview, see Bergmann et al., ). The above is only a small set of examples from a large body of research that has consistently shown that cross-linguistic interference between languages is inevitable. This is attributed to the fact that speech production is target language non-specific and both languages are simultaneously activated regardless of the language the person intends to speak (Spivey & Marian, ; Wu & Thierry, ; see Kroll & Sunderman, , for details). All speech production models agree that syntactic and lexical representations are integrated across languages forming a compound language system (de Bot, a; Cook, ). The conceptual system in the mind spreads activation to the relevant representations of both languages and the links within phonetic, syntactic, and lexical features are triggered regardless of the language in which the task is being performed. As a consequence of this constant interaction, processing becomes cognitively more demanding and it may take extra time and effort to manage the cross-linguistic influences, slowing down of the lexical and syntactic processing (Parker Jones et al., ; Sorace, ; see Bialystok et al., , for a review). If the bilingual speaker wishes to maintain both languages simultaneously, more effort will be needed due to cognitive and emotional limits (Herdina & Jessner, ). So, as for direct comparison between L and L proficiency, it has been generally proposed that as a result of sharing the (finite) resources for processing and memory, constant interaction would lead to trade-offs (Green, ; Costa & Santesteban, ) rather than mutual reinforcement of the language systems (Segalowitz, ; Cummins, ). For example, Major () hypothesized that the degree of influence on the affected language would be dependent upon the level of the mastery in the other language, regardless of whether the affected language is the native language or a later learned language (p. ). In support of
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
his hypothesis, his findings from a small sample of L Portuguese speakers of English revealed that greater L attrition implied a higher degree of L proficiency. Yet, it was later demonstrated that the inverse interaction in L and L pronunciation proficiency predominantly applies to bilinguals who are immersed in the L environment before puberty (Yeni-Komshian et al., ). A study on changes in prosody by de Leeuw et al. () very compellingly showed that attrition could also develop independently of the L properties. The participants investigated were a group of German native speakers who had moved to Anglophone Canada in late adolescence to adulthood. Individual results showed that some of the participants’ tonal alignment patterns were significantly different from the monolingual norms but the divergence in the patterns did not resemble English phonetic norms, either. Thus, the authors concluded that variability in language outcome stretches the limits of traditional two-dimensional perspective of language (either like the L or like the L) and L attrition may be a result of processes that go beyond the interaction of subsystems or components of L and L (p. ). In the lexical and grammatical domains too, it is not entirely clear whether the proficiency level attained in the L is associated with deterioration in the L. In two studies that focused on the systematic relations between L and L proficiency, attrition was detected among some of the participants but it was not a result of trade-off effects except at very advanced levels of L proficiency (Cherciov, ; Opitz, a). For example, Cherciov () reported that bilingual Romanian immigrants residing in Canada had reduced language abilities (e.g., vocabulary and fluency) as a group and more variability compared to the monolinguals, but individual level findings revealed that only two out of twenty participants had significant levels of attrition. Her analyses further established that certain bilinguals could retain high proficiency levels in both languages and that trade-off patterns were found only in these participants that were most affected by attrition. Investigations carried out on near-native speakers indeed showed that high L proficiency had persistent influence on L syntax but only interface properties were affected by attrition and not pure syntactic options (Tsimpli, ; Sorace, ). It has been proposed that vulnerability of interfaces leads to optionality in the use of interface properties due to influence from the L in advanced bilinguals. Whenever bilinguals go for the less preferred option in their L (as a result of converging L and L categories), it indicates flexibility or an expansion of the system rather than the deterioration of this category in the L knowledge. This is an expected outcome of the reorganization of the language system as a result of integrating the new learned L structures in to the language system that allows bilinguals to efficiently cope with cross-language competition (Chamorro, Sorrace, & Sturt, ). It is possible that the inability to detect a direct relation between L and L is because attrition is a gradual process taking place over decades, and the symptoms detected at the individual level are often minimal, as most late bilinguals have been found to typically retain their syntactic proficiency and diverge only slightly, or not at all, from the nonimmigrant monolingual norms (Jaspaert & Kroon, ; Gürel, ; Backus, ; Yağmur, ; Tsimpli, ; Gürel & Yılmaz, ; see Schmid, a, for a review). Second, attriters’ performance at the individual level is often inconsistent with the group level findings, with some speakers performing within the native range and others failing to reach native levels. This is typically because attriters themselves tend to form extremely heterogeneous groups with respect to the context and frequency of use of their languages
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
, ,
and the degree of bilingualism, not to mention other contributing/confounding predictors such the age of acquisition—even in post-puberty learners, education, aptitude, length of residence in the L country, and so on (see Opitz, Chapter , and Part IV in this volume). Moreover, some bilinguals appear to have maintained their knowledge across the whole range of measures while other bilinguals’ performance is affected variably at different linguistic levels (e.g., detectable decreased lexical accessibility, but morpho-syntax that appears to be unaltered). Perhaps if future studies can be designed in a way that takes into account these factors (e.g., longitudinal studies using sophisticated designs and elicitation techniques and focusing on all levels of linguistic knowledge), then it may be possible to better unveil the connections between L and L development. Overall findings on L knowledge of bilinguals indicate that the mechanisms involved in accessing and integrating information in real time may become compromised due to sharing memory and processing resources (Green, ; Seliger & Vago, b). It is therefore concluded that attrition constitutes primarily a processing or control phenomenon displayed by performance-related problems (Sharwood Smith & Van Buren, ), which may be temporary. Once re-immersed in the L environment, speakers gradually get tuned to the non-migrant variant of the L (e.g., through re-exposure or relearning; Herdina & Jessner, ), or with ageing and retirement they may resort to their native language and roots, and experience a decrease in L proficiency (i.e., L revitalization or reversion; Schmid & Keijzer, ). To summarize, attrition is partially triggered by psycholinguistic pressures of bilingualism (the growing presence of the L) but it is hardly an automatic consequence of L acquisition. Therefore, the degree to which L attrition can be predictive of L proficiency remains unknown.
. T
.................................................................................................................................. Speaking two languages has consequences that go beyond cross-linguistic interaction. A relatively young research line has proposed that bilingual experience can transform the way people think as a result of restructuring at the conceptual level in some of the classical Whorfian dimensions of categorization of colours, objects, and events (Jarvis & Pavlenko, ; see Bassetti and Cook, , for review). For example, Jameson & Alvarado () reported an enlargement in the distinction between colours in the blue/green category of L Vietnamese–L Russian bilinguals compared to Vietnamese monolinguals. They attributed this shift to the influence of Russian, which has a finer distinction in this category (separate words for green, light blue, and dark blue), on Vietnamese, which has a single word to express this category. Bilinguals were also found to consistently differ from monolinguals in object categorization (Cook et al., ; Athanasopoulos & Kasai, ) and gesturing about events and motion (Brown & Gullberg, ; Cadierno & Ruiz, ) whenever these properties are encoded differently in the L. Clearly, it is not only the languages mixing/merging together but the bilingual’s worldview is blended as a result of thinking and speaking in two languages (see Athanasopoulos, , for a review). In the process of negotiating two languages, bilinguals also need to manage two cultural systems, which can become as interrelated as the languages they speak. A set of studies
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
directly investigated the links between language and cultural values. They contend that language, as a carrier of cultural symbols and values, exerts systematic impact on higher level cognitive functioning and activates corresponding cultural frames (Lee et al., ; Chen, , p. ; see Oyserman and Lee, , for a review). For example, bilinguals randomly assigned to use English or Chinese were found to systematically differ in their recollections of autobiographical events (e.g., a birthday party). In English, they focused on themselves (e.g., what they were doing and thinking), but in Chinese, they focused on other people (e.g., who other guests were and what was happening) (Marian & Neisser, ). The impact of cultural framing within the language associated with it is observed even in speakers’ self-perceptions. Chen et al. () found that Chinese-English bilinguals rated themselves as more competent and conscientious when responding in English than in Chinese. This illustrates that a person can adjust/change her self-presentation according to social contexts shaped by language and culture. Such studies reveal that just like speaking two languages, being exposed to two cultures might entail other cross-cultural interferences. While evidence for such conceptual restructuring has been mainly documented in advanced speakers of L or speakers with prolonged L exposure, it is still remarkable because L perspectives can be internalized to the extent that the speakers come to prefer the L categories over those that were previously favoured, implying that increased proficiency and exposure to L lead to change in the cultural orientation of the bilingual, and therefore may facilitate successful integration into the L society. However, Grosjean () states that just as bilinguals acquire and use languages for different purposes, in different domains, they internalize the cultures to varying degrees. Many of them are in fact bilingual without being bicultural or vice versa, because there is simply no straightforward relation between bilingualism and biculturalism. One of the most comprehensive frameworks invoked in order to account for the complex language and culture outcomes in multilingual and multicultural societies is acculturation (Berry, , ). It refers to the relation between the attitudes towards both the home and the receiving culture and the motivation to learn the host country language and culture versus preserving one’s ethnic roots and L. Depending on the type of attitudes fostered, four acculturation outcomes can be identified: integration, assimilation, separation, and marginalization. Individuals who maintain their own cultural identity and also extend relations in the host society and incorporate elements of the host culture are considered to have an integrated acculturation attitude. At the opposite end is marginalization, which involves weak connections with both the host society and the original culture, where individuals are more oriented towards achieving personal goals. Assimilation is the preference to adopt the collective identity of the host society and abandon that of origin. The last strategy, separation, entails the rejection of the key features of the host culture while preserving one’s own cultural heritage (Sam & Berry, ). These attitudes are also influenced by the host society’s preferred acculturation strategies and the state policies and ideologies (i.e., whether the immigrant group is valued or devalued) (Bourhis et al., ; Yağmur & Van de Vijver, ). In terms of language development, the corresponding categories for the acculturation continuum are: competent bilingualism for integration, monolingual L assimilation for assimilation, monolingual L segmentation for segregation, and linguistic marginality or limited bilingualism for marginalization (Esser, ). It is proposed that the readiness of the individual is largely determined by various motivational attitudes (Schumann, , ).
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
, ,
For instance, shared heritage between migrant and local communities promotes linguistic and social integration, whereas differences (e.g., in socio-economic background, native language, physical appearance, and religion), ethnic concentration, and lower status in relation to the majority group lead to limited social interactions between the communities and hence limited levels of proficiency in the L (e.g., Sniderman et al., ). Results of acculturation studies in L acquisition have systematically indicated that increased motivation is linked with higher levels of success in L acquisition (e.g., Gardner, ). This is particularly evident in investigations of ultimate attainment in pronunciation skills. L speakers were often found to modify their speech to resemble the speech of native speakers with the intention of reducing social distance between them and the host community (Gardner, ; Lybeck, ), and speakers who identified with the L culture more strongly had more target-like pronunciation (Major, ; Graham & Brown, ; Hammer & Dewaele, ). Paradis (, p. ) equates the predictive value of motivation in successful L acquisition with its impact on the rate of attrition. A positive emotional attitude towards one’s native language and culture will be conducive to the maintenance of the native language. So accordingly, more L attrition would be expected in speakers who have a strong desire to immerse themselves in the L culture and for whom L loses its practical and symbolic significance. Consequently, these speakers would not be motivated to make an effort to maintain the L (e.g., Herdina & Jessner, ; Hammer & Dewaele, ). A study carried out within a socio-historical framework revealed a strong link between affiliation and the native language (Schmid, ). The participants were Holocaust survivors who were categorized according to the time at which they migrated to Anglophone countries before the outbreak of World War II. The group that emigrated between the pogrom of November and the beginning of the war had been through the harshest oppression policies and experienced significant loss in vocabulary and syntactic knowledge. No other background factors including age at emigration and length of migration could account for the change in this group’s L, which led Schmid to point out that the strongest predictor for attrition effects was the desire to distance themselves from the L community due to the traumatization experienced by this group. A later study by Ben-Rafael & Schmid () investigated the Francophone and Russian communities who migrated to Israel at two different time periods. The Francophone group was ideologically motivated with the intention of becoming a part of the host society. Their arrival coincided with the state’s overtly monolinguistic ideology in s and s, which required immigrants to learn and use Hebrew exclusively as a sign of full integration. The Russian group had economic and pragmatic interests. They came in the s and s to a more liberal and tolerant environment, when the national policy had changed towards one in favour of multilingualism and cultural diversity. The results showed that the prevailing national language policies and the immigrant groups’ orientations were clearly reflected in the amount and type of L interference in their L. As expected, the French sample had significantly more L interference at both lexical and discourse level, and in the Russian sample, influence of Hebrew was mostly limited of borrowings of isolated lexical elements. While the above studies found that attitudes were indeed linked with attrition, several other studies have been unable to establish such a relation. One of the earliest socio- and ethno-linguistic investigations into native language attrition was carried out by Waas () among German long term immigrants in Australia. She found that the participants’
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
verbal fluency had decreased over time compared to monolinguals, but contrary to expectations, this was not associated with their cultural orientation. It could be because Waas’ measure was based on club/church memberships and citizenships, which could potentially be related to other factors not directly linked to a person’s ethnic affiliation. Building on Waas’ work, two other studies looked for evidence for the impact of attitudes using a more sophisticated quantification of affective variables within the framework of Ethnolinguistic Vitality Theory (EVT; as formulated for example by Giles et al., ). EVT measures not only individual attitudes, but also wider community level factors such as language prestige and institutional support. These two studies are the investigations of L attrition in Turkish bilinguals in Australia (Yağmur, ) and in Dutch bilinguals in New Zealand (Hulsen, ). The hypothesis was that speakers with high vitality perceptions were expected to maintain their language and distinctive cultural traits, whereas ones with low vitality were expected to shift to the dominant language and blend with the host society relatively more. The Dutch immigrant group appeared to be well integrated/ blended both culturally and linguistically into the host community, whereas the Turkish group had a tendency to maintain their native language and culture. Researchers observed that language proficiency of both groups significantly differed from that of the monolinguals. However, neither did high symbolic value attached to Turkish protect its speakers from attrition, nor did affirmative orientations of the Dutch migrants facilitate their L. In short, EVT failed to provide a sufficiently explanatory framework for attrition in both studies. Likewise, in a number of recent studies carried out in different settings, the impact of attitudes on attrition turned out to be considerably limited (Dostert, ; Varga, ; Yılmaz & Schmid, ). Why motivational variables seem to play a role in some studies and not others could be related to the methodology—that is, how attitudes are determined. In Schmid (), the Holocaust survivors were considered to be uniformly affected by external circumstances before migration. Similarly, in Ben-Rafael & Schmid (), the attitudes of French and Russian participants could be determined reliably at the group level at the time of migration. However, in the other studies, the investigators often met rather heterogeneous groups of speakers decade(s) after the migration, when attitudes have most probably changed and identities been reconstructed. They had to base their analyses on questionnaires and self-reports, where the participants are asked questions on the relative importance and the weight of each language and culture in their life. It is possible that type of snapshot measurements cannot reliably determine linguistic and cultural affiliation because (often) the written nature of these tasks limits the participants in explaining how their attitudes have fluctuated across the life span depending on circumstances. It is also possible that written tasks may be somewhat unreliable or not entirely accurate because they compel the participants to write answers that are considered more politically correct or in line with the perceived expectations of the researcher (Schmid, a). A recent investigation that explored the impact of attitude as measured through oral interviews in addition to written questionnaires has interesting results (Cherciov, ). While the quantitative analyses revealed no clear-cut relations, qualitative analyses suggested important links between attitudes and language profiles of the speakers. Therefore, it is important to combine the quantitative and qualitative approaches in order to capture the relation between social and psychological variables and language development.
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
, ,
. T ()
.................................................................................................................................. While the research reviewed so far showed that motivation and cultural affiliation play a role in language development (albeit more implicitly in L attrition), the review also indicated that there is no empirically established inverse relationship between L retention and L learning. We will now turn our focus on European immigrant contexts where policies are based on the assumption that L knowledge, in particular that of minorities with non-European and less prestigious languages, interferes with L learning and hence represents a barrier to successful integration into the host societies of Europe. The presence of several communities with culturally and linguistically distinct backgrounds is often perceived by the host society as a challenge to national identity and social cohesion, while at the same time triggering concerns about the economic well-being of the receiving societies (Semyonov et al., ; Schjerve & Vetter, ). As the inclusion of immigrants in the labour market, in the education system, and popular culture of the host countries has become one of the central policy concerns, successful acquisition of the host country language has been regarded as one of the key components (Shoamy, ; Extra et al., ). Often dissatisfied with the levels of integration, an increasing number of European states have introduced some form of obligatory training programmes on language and culture for both new and settled migrants in order to better equip them with knowledge of the language, society, history, culture, and institutions of the host country (e.g., the Netherlands, Denmark, Germany, France, Austria, Luxembourg, and the UK; Kostakopoulou, ). The unfavourable attitude towards immigrants’ native language, however, has persisted. This has been poignantly communicated in the recent political discourse to the extent that all immigrants should be ordered to speak (only) the language of the host society both in public and at home in order to improve their skills in the host society language. The Christian Social Union (CSU), an ally of the German Chancellor Angela Merkel, drafted a proposal to make the immigrants speak German in public and within the family (‘Speak German at home’, ). In the UK, immigrants who fail to improve their English could lose their chance to stay in Britain under plans of the ex-Prime Minister David Cameron (Lowe, ). An assumption typically associated with this attitude is that individuals whose L skills decline as a result of gaining more L competence would achieve better cultural and social integration, whereas others who retain their native language and culture would achieve relatively limited L competence and are characterized as unwilling to integrate (Bijl & Verweij, ). Research reports that the general trend is complete L integration over two to three generations (Alba, ; Extra et al., ). On the other hand, in an era of globalization, available media technology, and affordable travelling, keeping in touch with the home culture and country has become increasingly easy. As a result, current immigrant generations seem to be integrating at a much slower rate than desired by the receiving countries and asking for more linguistic and cultural recognition (Extra & Yağmur, ; Shohamy, ). Therefore, the study of L development has gained particular momentum
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
in the clash between the current socio-political climate, characterized by high levels of restrictiveness in policies of socio-cultural integration and pressures on L success on the one hand, and the growing importance and affirmation of local identities on the other. Social integration is often assessed based on knowledge of the host society’s culture and history and involvement with the community life and intercultural skills as well as proficiency in the language of the society (Beacco, ). However, in practice, becoming competent in an L in no way guarantees the embracement of the culture and the values of the host community or the acceptance by the host society, due to the simple fact that the ability to speak a language well does not necessarily mean that the language could be used in ways that promote cultural and social integration (see Espinosa and Massey, ; Nesadale, ; among others). For instance, in the US, learning English has enabled many white and Protestant Northern European migrant groups to smoothly and relatively quickly integrate into the Anglo-Protestant American society both linguistically and culturally. However, some other groups from different racial, cultural, and religious backgrounds (i.e., Latinos and Asians) have often been less successful in economic and social integration though they left their native language and cultural connections behind and learned English fluently (McDonald & Balgopal, ; Boyer, ; Tolsma, Lubbers, & Gijsberts, ). There are also examples of immigrant groups that continue to speak their mother tongue over generations and live according to native culture norms, but at the same time attain high levels of linguistic integration (e.g., Mexicans; Citrin et al., ). These examples clearly indicate that the linguistic and cultural components of the integration equation do not necessarily overlap and that no two settings are identical with respect to the role language skills play in bringing the host and the migrant community members together (Grosjean, ). Yet, it is remarkable that it is often the L knowledge that is selected as the (single) strongest variable that predicts integration outcomes (Extra et al., ). It is definitely true that good knowledge of the host language could be highly instrumental for socio-cultural adjustment and becoming acquainted with the host culture through establishing relationships with the host country people, accessing the job opportunities, and educational institutions, while inadequate language ability may curtail educational and professional opportunities. However, this aspect still does not justify the widespread view that cultural integration is primarily a matter of language and knowledge about the host society culture and that L retention is a barrier to integration. A potential reason behind this deterministic attitude could be that language skills are somewhat easier to assess, whereas the wider context of affective variables consisting of attitudes towards language and culture, motivations for language maintenance and L learning, prejudices, involvement, and intercultural understanding are notoriously difficult to operationalize (Berry, ). Ironically, tests, such as citizenship and integration tests, may in fact serve to hinder integration by leading to exclusion and resentment among the members of the migrant communities unable to pass them, because failure means being exempt from certain rights (Strik, ).1
1 Unfortunately, not only are such tests still being used, but ‘pre-admission’ tests are also being widely applied (e.g., the Netherlands, Germany, the UK, Austria, and Denmark; Scholten et al., ). With these tests, newcomers are required to meet certain cultural and linguistic integration criteria in their countries of origin before being admitted into the host country.
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
, ,
. C
.................................................................................................................................. The findings from the studies reported here strongly suggest the presence of an integrated super-system in which both languages affect each other at all levels to the extent that such integration could even lead to conceptual restructuring of cultural values. However, attrition in the native language develops somewhat independent of learning an additional language even in very advanced L speakers. L attrition seems to be a consequence of slowed down mechanisms involved in speech processing, or simply put, a natural part of the developmental process of bilingualism. Degree of attrition or retention of the L is therefore by no means a direct indication of L proficiency. As for the socio-cultural affiliation, positive motivational attitudes certainly facilitate L acquisition, but holding favourable attitudes towards the ethnic language and culture and preserving the native language should not stand in the way of successful L learning and acculturation. In particular, with the L–L trade-off missing in the equation, any inferences about the levels of differential language knowledge and socio-cultural integration would remain incomplete. Most bilinguals appear to have a cultural dominance and switch between cultural norms and behaviours depending on where they spend their time, on their contacts, and also on their preferred identity. At the same time, defining the degree of biculturalism is often hard, as the way people identify themselves culturally may be subject to frequent changes over time and influenced by individual and societal level factors, which in no way makes them less bicultural. In order to explore the complexities and dynamics of bilingualism and biculturalism, there is a great need for studies that integrate language and culture in sophisticated and longitudinal designs. This is particularly important in contexts where the language and the culture of the migrant and host communities are more distant, because such studies would provide better foundations for designing policies that would facilitate L development of migrants and promote intercultural dialogue/mutual acceptance at the same time. Such studies would also be useful in fostering active bilingualism and intergenerational language development in migrant communities and preventing their exclusion and isolation from the host society.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
......................................................................................................................
......................................................................................................................
. W ?
.................................................................................................................................. T term language contact refers to the mutual and bidirectional influence of languages. The crucial question, however, is where language contact actually takes place. In his pioneering work on language contact, Uriel Weinreich gave the following definition: ‘two or more languages will be said to be IN CONTACT if they are used alternately by the same persons. The language-using individuals are thus the locus of the contact’ (Weinreich, , p. ; emphasis in original). This approach is psycholinguistic in nature and means that the proper place of language contact is the brain of the bilingual speaker (Riehl, , p. ). A different, sociolinguistic approach to language contact originated in the s, focusing on societies or social groups. In this framework language contact is ‘the use of more than one language in the same place at the same time’ (Thomason, , p. ). Against this background, language contact is considered a long-term process that leads to language change and has an impact on the inner coherence of language systems (cf. Thomason & Kaufman, ; Winford, ; Hickey, ). While the main focus of language contact research is still on historical linguistic and typological aspects, more recent approaches include language contact in immigrant settings (e.g., Clyne, ; Matras, ; Riehl, ). Some approaches (Myers-Scotton, ; Clyne, ; Winford, ; Matras, ) try to integrate the individual and societal perspectives by assuming that language contact starts at the level of the individual speaker and in course of time is spread throughout the speech community. Whether a specific contact phenomenon is accepted as a norm in a particular language community depends on a variety of factors, for example, how linguistically economic, frequent, or illustrative a certain item is (cf. Riehl, ). There are, however, particular elements in specific language contact constellations that seem to be especially prone to be transferred and emerge in different speakers and in different groups independently. As will be demonstrated in the following sections, this also holds for language attrition settings.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
. L
.................................................................................................................................. The frameworks of language attrition and language contact (in a synchronic perspective) have in common that they investigate the impact of a second language (L) on the first (L). The difference is that both approaches generally focus on different speaker populations. Whereas typical attrition research analyses first generation migrants in an L environment, language contact research focuses more on the use of languages in multilingual speech communities. However, the linguistic features found in individual attrition and typical language contact settings are similar. In both constellations it is difficult to determine whether a certain phenomenon is used only at a particular point, or on a regular basis as a synonym of an L item, or whether the item has already replaced the corresponding L item. Usually—and this holds both for individual speakers and for bilingual speech communities—L and L items are used either alternately or with different functions or in different contexts. In this sense, language contact has to be considered as a bilingual practice: it reflects the processing and use of language in the bilingual speaker’s repertoire (Matras, , p. ). However, a distinction has to be made between speakers who have mainly shifted to their L in the new environment and have no or very limited contact with their L, and speakers who have continuous contact with a bilingual community in the recipient country where they use both L and L alternately and in a bilingual language mode. This means that bilinguals communicate differently when they are with bilinguals who share their languages. Whereas they avoid using their other language with monolinguals, they may call upon it when interacting with bilinguals, either by changing over completely to the other language (= code-switching) or by bringing elements of the other language into the language they are speaking (= transfer) (Grosjean, , p. ). This may lead to a form of mixed speech and conventionalized language contact phenomena in this particular community. Immigrants who raise their children bilingually in the recipient country may also be affected by this practice. Second-generation speakers, who often acquire the system of their parents’ language incompletely (Montrul, ), are more prone to transfer items or structures from their dominant language into the home language and they are also more prone to feed back contact-induced changes to their parents’ speech. The routine use of a language in a bilingual context (i.e., with other migrants in the same setting or with L speakers of the language) causes changes in the language system (see also Köpke, c). This, however, implies that the emerging contact-induced variety is a fully functioning language variety and can be utilized in a number of domains. Typical phenomena of attrition which are caused by lack of usage such as hesitation phenomena and disfluency markers will not occur when this contact-induced variety is used, but only when the speakers are forced to speak in a monolingual language mode in their L (for a discussion, see Riehl, ). Thus, contact-induced changes in a language have to be discerned from other internal processes of language change which are also common in language attrition and language shift—that is, simplification processes of the grammatical system. The difference between these two phenomena can best be illustrated by an observation that was made by Zürrer
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
() in a German language enclave in Valle d’Aosta (Northern Italy). In the first step, the German variety adopted the type of compound personal pronouns from Italian (noialtri lit. ‘we-other’ > wirendri lit. ‘we-other’). In the second step, this new pronoun was fully integrated into the system of the German variety, as it was case-marked according to German case marking patterns (wirendri (NOM), ündschenandru (GEN), ündschenandre (DAT), ündschendri (ACC)). In the third step, it underwent a reduction process—that is, the cases were no longer marked (for a discussion, see Riehl, ).
. K : , ,
.................................................................................................................................. Generally, in language contact studies a distinction had been made between two categories of contact-induced changes: borrowing, which has been defined as ‘incorporation of foreign features into a group’s native language’ (Thomason & Kaufman, , p. ), and interference, which refers to the influence of an L on the L (for a discussion, see Hickey, ; Matras, ).1 Some approaches in both second language acquisition (SLA) research and contact linguistics use the cover term transfer for both types of bringing over an item from one language to the other. In this context, Clyne (, p. ) introduced the notion of transference, which he defines as follows: Transference is employed for the process of bringing over any items, features or rules from one language to another, and for the results of this process. Any instance of transference is a transfer. (Clyne, , p. )
According to this definition transference means taking over elements from one language to the other on different levels of the language system (lexical, phonological, morphological, syntactic, and pragmatic) without defining the direction of this process.2 In doing so, concrete material or structural patterns can be transferred from language B and integrated into the language system of language A or vice versa. This viewpoint is also in line with more recent approaches to multilingualism such as the multi-competence perspective (Cook, b; Jarvis & Pavlenko, ), which considers the different languages a person speaks as one connected system, rather than each language as a separate system. This, in turn, means that in the bilingual or multilingual mind there is always mutual influence between the L and the L (or the third language (L)). 1 Another classification was suggested by Van Coetsem (), who distinguished between borrowing and imposition. In this framework, the direction of transfer is always the same—that is, from a SL to a RL; what differs is the agent of transfer. In the case of borrowing the RL speaker is the agent and transfers material from the SL to the RL (e.g., an English speaker using French words while speaking English); in the case of imposition the SL speaker is the agent (e.g., a French speaker using French articulation while speaking English). 2 SLA researchers also make a distinction between forward transfer (influence of L on L/L) and reverse transfer (influence of L/L on L) as well as lateral and bidirectional transfer (for an overview, see Jarvis & Pavlenko, , p. ).
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
From a structural perspective some scholars (e.g., Heine & Kuteva, ; Matras, ) distinguish between borrowing in a narrower sense as a transfer of forms and in a broader sense as a transfer of structural patterns (meanings, grammatical functions, and syntactic relations). In this context, Matras () introduced the terms matter borrowing for instances where the phonological shape and the morphological form is taken over from one language to another, and pattern replication for those cases where the function of the linguistic feature is taken over and imposed on an indigenous structure. In the latter case, there is no transfer of new elements that have to be integrated into the L system, but a transfer of functions or meanings of single words or constructions and a re-analysis of an L item on the basis of the rules of an L item. When meanings of grammatical or lexical items are transferred some scholars also apply the term restructuring, which refers to the ‘partial modification of already existing language-mediated conceptual categories’ (Jarvis & Pavlenko, , p. ). Another process that differs from transference, although the underlying cognitive processes might be similar, is convergence. It is defined as an increase of similarity between two languages at any level (lexical, morphological, phonological, etc.) (see Matras, , p. ). The term is, however, more often used with reference to linguistic patterns (i.e., mapping relation of meaning to form) rather than to lexical items (Matras, , p. ). To explain the difference between syntactic convergence and syntactic transference, Clyne (, p. ) provides an illustrative example: () Wir haben zu Schule gegangen in Tarrington we aux + have to school go+ PAST.PT in Tarrington () Wir haben gegangen zu Schule in Tarrington we aux + have go+ PAST.PT to school in Tarrington While () is a morpheme-to-morpheme transfer from English, example () partially retains the so-called brace construction (auxiliary and participle embracing the adverbial phrase zu Schule ‘to school’) which is typical for German, but is already diverging from Standard German Wir sind in Tarrington zur Schule gegangen, where the auxiliary sein (‘to be’) is used instead of haben (‘to have’) and where both adverbial phrases are included in the brace construction. Thus, this construction is halfway between the SL and the RL.
. P :
.................................................................................................................................. Typological studies indicate that there is nothing in the structure of a language, be it a single form or a grammatical feature, that is completely ‘borrowing-proof ’ (Aikhenvald, , p. ). Given sufficient duration and intensity of language contact, even subsystems such as core morphology may be affected (see Hickey, , p. ). However, there are linguistic units that are integrated into another language more easily than others. This usually refers to elements that do not require a high degree of restructuring in the system of
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
content item > function word > agglutinating affix > fusional affix
.. Borrowability hierarchy Note: For the difficulties of comparing different contact settings, both from a typological and from a societal perspective, as well as a discussion of more detailed hierarchies of borrowability, see Matras (, pp. –).
the borrowing language. Thus, a hierarchy of borrowability such as that shown in Figure . following can be assumed (cf. Field, ). The indicated hierarchy encompasses both quantity (content items are borrowed in a higher quantity than functional words, etc.), and progression (content words are borrowed earlier in contact history than the other items). This can be explained by the stability of the grammatical system, which means that bound morphemes are usually more resistant to transfer than unbound morphemes. Moreover, content words as an open class are a primary source of borrowing. Typological distance between the languages in contact only plays a role in the speed of transfer in the sense that the typological closeness of languages can accelerate the transfer processes. Furthermore, external factors are also to be taken into consideration (Van Coetsem, ). Different language contact patterns are found in different kind of registers here, and language contact on the grammatical level is generally more advanced in informal speech. In bilingual communities, speakers of the younger generation usually display more instances of language contact phenomena and use them more frequently than the older generation. This means, as already mentioned, that in immigrant communities, contact phenomena are more frequent in speakers of the second and third generation than in canonical first-generation attriters. In the following, examples are typically taken from first-generation attriters. Due to the lack of first-generation corpora for some language pairs or specific contact phenomena, in some examples second-generation utterances are quoted.3
.. Lexical transference As indicated by the hierarchy of borrowability (see Figure .), most instances of transfer occur at the lexical level. The borrowing of lexical items is a frequent phenomenon even in monolingual societies that integrate loans from other languages into their system in order to label new cultural or technical achievements. Immigrants who encounter a new flora and fauna and a new sociocultural environment adopt expressions for natural phenomena, institutions, organizations, etc., from the L even at the very beginning of language contact (cf. Pavlenko & Jarvis, ). Some examples include: creek, gum-tree, highschool, assembly, milk bar, and chemist in immigrant languages in Australia (cf. Clyne, , p. ); and Hauptbahnhof ‘central station’, Arbeitsamt ‘job agency’ in immigrant languages in Germany (cf. Riehl, , p. ). This does not necessarily mean that translation equivalents are not available in the L, but often they have a different connotation. In the course of L acquisition and shift in language dominance, a wide range of other words is transferred. This applies especially for technological developments that did not exist at the time of emigration from the 3
This will be marked in brackets after the example.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
country of origin (e.g., TV, dish-washer, mobile phone). While concrete concepts that have easily perceivable properties are the easiest to transfer, more abstract concepts (e.g., privacy, frustration) can be integrated into the other language, too (cf. Jarvis & Pavlenko, , pp. –). Moreover, for domains that have been encountered for the first time in the new country, a whole repertoire of expressions can be borrowed. Most German immigrants in Australia who immigrated in their early s had never been confronted with building and furnishing a house; they encountered this domain only in Australia and acquired the respective vocabulary together with the construction process, for example, brick-veneer (Germ. Ziegelvorblendung), plasterboard (Germ. Gipskartonwand), plumbing (Germ. Installation), scrapingboard (Germ. Fußleiste). This does not only apply for technical terms, but also affects everyday vocabulary such as tiles (instead of Fliesen), blinds (instead of Rollos).4 While the transfer of nouns is often motivated by denotative needs, verbs are also borrowed for structural reasons; Croatian and German speakers in Australia integrate English verbs, which replace more complex phrasal units in L: ()
[ . . . ] nismo imali telefon da ringamo, nismo imali. (nd gen., Hlavac, , p. ) (‘ . . . we didn’t have at telephone to ring, we didn’t have . . . ’, instead of da se javimo)
() Das Auto habe ich gestern erst servicen lassen (MH, AGC) (‘I had my car serviced only yesterday’, instead of eine Inspektion für das Auto machen lassen)5 ()
dass der D. proposet hat, weißt du ja (MH, AGC). (‘that D. proposed, you know of course’, instead of einen Heiratsantrag machen)
In all three examples the utterances inherits the argument structure of English, which is less complex than the equivalent in the L. In (), the English to ring can be used intransitively and transitively, and is preferred to the equivalent Croatian construction that contains a reflexive particle, se, with the object in dative case—that is, javiti se + DAT. In () the verb servicen substitutes the phrase eine Inspektion machen lassen (lit. ‘to have an inspection made’). In () the possibility that in English the indirect object (‘to whom he has proposed’) can be eliminated is used; in German this would be unusual. So the transferred verb not only replaces a more complex phrasal unit (einen Heiratsantrag machen, lit. ‘to make a proposal of marriage’), but also reduces the valence of the verb. As these examples illustrate, lexical transfer can also have an impact on constructional patterns. The borrowing process is, however, not only influenced by external factors—such as cultural necessity or integration of new domains—but also by properties of the language system. It can be observed that German attriters in an English-speaking environment adopt adjectives such as happy, beautiful, annoying, but only in a predicative use: da war er happy Schmid (a, pp. –) also mentions borrowings in the area of health. The fact that the attriters under investigation are usually comparatively old brings about health issues that can be attributed to age and that have also been experienced by the speakers in the new country. Since L equivalents might not have been used for a long time, L terms are more familiar and can be activated much easier. Schmid also suggests that the L terms—since they are attributed to one’s personal situation—may also acquire a higher emotional component compared to more distant L words which had been used in reference to other people. 5 Most examples from German–English language contact are taken from my corpus of German immigrants in Australia (= AGC). The corpus contains spoken data from free narrative interviews (mean length one hour) from fifteen German-origin immigrants (mean age = , age of arrival = to , length of residence = to years) and participant observation in bilingual networks. 4
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
(‘then he was happy’), das Essen ist beautiful (‘the meal is beautiful’), das war so annoying (‘this was so annoying’), and not in an attributive context (e.g., *ein happies Kind, *ein beautifulles Essen, *ein annoyinges Erlebnis). An explanation could be that the phonematic structure of these adjectives is different from the German one and thus the adjective cannot be fully integrated into the inflectional system (cf. Riehl, , p. ). The same applies for verbs in specific language contact settings. German or Russian are able to integrate other-language verbs quite easily into their language system, in German using -en, or -ieren (e.g., expir-en, farmer-ieren), in Russian using the suffixes of the respective verb classes (and aspectual prefixes) (e.g., zakencelit’ ‘to cancel’, čadžit’ ‘to charge’, zaloginitsa ‘to log in’, kliknyt’ ‘to klick’; cf. Isurin, , p. ). In other languages (e.g., Turkish, Greek, Indian languages) the morphological integration of verbs is not as easy; in German–Turkish language contact, German verbs are integrated into Turkish using a passepartout verb yapmak ‘to do’, which is inflected, and the German loan appears in the infinitive (e.g., tauschen yapmam ‘I do not swap’ (lit. ‘I do not do swapping’) (cf. Riehl, , p. )). This phenomenon also appears in Turkish–Dutch, Turkish– English, and Turkish–Norwegian constellations (see Clyne, , p. ). These observations indicate that obviously similar phonotactic rules or syllable structures facilitate morphological integration into another language system, whereas different phonotactic structures might impede this—unless the form of the transferred item undergoes a major phonological change (see Matras, , p. ).6
.. Semantic transference/conceptual restructuring Another vital type of language contact is the transfer of meaning. This is what Matras () subsumes under the term pattern replication and in some other publications is called conceptual restructuring (Pavlenko, ; Schmid, a). In this case no otherlanguage material is taken over, but existing material gets an additional or different meaning by using it in the other-language context. Semantic transference occurs mainly with cognates (etymologically related words in different languages) or otherwise homophonous lexemes. That is why semantic transfer is most frequent in etymologically related languages such as English, Dutch, and German, for example: ()
Das Krankenhaus wird jetzt als Business gerannt. (AGC, MW) (‘The hospital is now run as a business’; to run a business = Germ. ‘ein Geschäft betreiben’)
()
da bin ich dann zu einem Reifenplatz gefahren (AGC, MH) (‘then I drove to a tyre place’; Germ. Platz is generally not used in the sense of ‘location‘)
()
Siehst du gerade einen Film? Geht der jetzt weiter und du vermisst was? (AGC, MH) (‘Are you just watching a film? Is it going on and you are missing something’, Germ. vermissen = ‘to notice that sb./sth. is lacking (often regretfully)’)
Some examples: topagahai as a replication of the English ‘tape-recorder’ in Pirahã, an indigenous language in Brazil, or banšer, a Palestinian Arabic adoption of the English ‘puncture’. For examples of the integration into native morphophonology of English loans in Japanese, see Winford (, p. ). 6
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
() of course it is easy to overdrive (Dutch overdrijven ‘exaggerate’; example from Sharwood Smith, b, in Schmid, a, p. ) These semantic extensions entail the use of L lexemes with a broader meaning of the L equivalents. Some of these examples are idiosyncratic (e.g., ()), but there is a considerable number of occurrences that are widespread in the community (such as () and ()) and even occur in other German–English contact settings (e.g., in Texas German and German in Namibia, see Riehl, , pp. –). Conceptual restructuring also occurs in typologically more distant language pairs, but here mainly with loanwords, for example, aplikacionnye formy ‘application forms’ in American Russian attriters (Russ. applikacija ‘ornamental work in which fabric is cut out and attached to the surface of another fabric’; cf. Isurin, , p. ). Krefeld (, p. ) observes the tendency in Italian-speaking immigrants in Germany to adopt the meaning of L for similar sounding words (sometimes with completely different roots) in Italian: They use regalo (Ital. ‘gift’) adopting the meaning of German Regal (‘shelf ’), rosino (Ital. ‘little rose‘) with the meaning of German Rosine (‘raisin’) or mappa (Ital. ‘map‘) taking over the meaning of German Mappe (‘folder’). The same applies for Spanish in Anglophone countries, for example, Spanish parientes (‘relatives’) acquires the meaning ‘parents’, and carpeta (‘tablecloth, folder’) the meaning ‘carpet’ (Silva-Corvalán, , p. ). These examples support the assumption that phonologically similar words are interconnected in the mental lexicon independently of the language subset they belong to (cf. Riehl, , ). Semantic transfer between cognates is not only confined to content words but also occurs at the level of function words. Here, again, the German–English language contact is of particular interest as both languages share conjunctions and prepositions with different meanings but the same lexical roots: ()
Der Hase und ich werden jetzt nicht hingucken, weil Nikolas die Karten vermischen wird. (nd gen., AGC) (STG während; ‘The rabbit and I will not watch now, while Nikolas is shuffling the cards’; Germ. weil = ‘because’)
()
Wenn ich ein ganz junges Kind war [ . . . ]. (Clyne, , p. ) (STG als; ‘When I was a very young child’, Engl. when = Germ. wenn, als)
()
Das war bei Gesetz verboten. (AGC, MW) (STG per Gesetz; ‘This was forbidden by law.’)
The adoption of an additional meaning also occurs with translation equivalents that are not etymologically related. Accordingly, it often occurs also in etymologically not related languages, as in the following examples: ()
also die zwei haben eine sehr schöne Verbindung. (AGC, MW) (STG Beziehung, ‘Well, those two have a very nice relationship’, in German, the noun Verbindung is only employed in the sense of ‘connection’)
()
[ . . . ] ne volim ništa što je predebelo. (nd gen., Hlavac, , p. ) (Standard Croatian premasno, ‘I don’t like anything that is too fatty’, in Croatian, predebelo (‘too obese’) can refer to animate referents only, and not to the nutritional content of food.)
Pavlenko (, p. ) also observes instances of semantic restructuring in motion verbs with Russian immigrants in the USA, for example, that the verb pair idti/chodit‘ (‘to walk’) and its derivatives begin to function in the manner of generic verbs to go, to get, and to come.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
.. Lexico-syntactic transference A phenomenon that also occurs frequently in language contact situations is the so-called morpheme-to-morpheme transference of idiomatic expressions or collocations (Clyne, , p. ). This often affects collocations with passepartout verbs that have undergone semantic bleaching (e.g. to take in expressions such as to take a photo/the opportunity/an interview/a shower, etc). German immigrants in Anglophone countries render these expressions word for word in German: ein Foto nehmen, die Gelegenheit nehmen, ein Interview nehmen, eine Dusche nehmen (instead of ein Foto machen, lit. ‘to make a photo‘, die Gelegenheit ergreifen, lit. ‘to capture the opportunity‘, ein Interview machen, lit. ‘to make an interview’, sich duschen, lit. ‘to shower oneself ’). In this case, the same constructions are not only transferred by different speakers independently, but also in different immigrant settings in the USA and Australia (see Schmid, a, pp. –; Riehl, , p. ). Additionally, similar instances are reported from language contact of typologically distant languages, such as English and Finnish (Jarvis, ), for example, ottaaa suihku (lit. ‘take a shower’) instead of mennä suihkuun (lit. ‘go by a shower’), and ottaa bussi (lit. ‘take a bus’) instead of mennä bussilla (lit. ‘go by bus’).
.. Syntactic transference and convergence ... Syntactic transference Syntactic transference in the definition of Clyne (, p. ) means the transference of syntactic rules. As mentioned above (see Section .) this phenomenon differs from syntactic convergence (see Section ...). A typical instance of syntactic transference is the replacement of V in subordinate clauses by the English SVO construction in German or Dutch language attriters: ()
Ich war ja an und für sich sehr froh, dass wir nicht mussten in ein Camp. (AGC, GZ) (STG . . . dass wir nicht in ein Camp mussten, ‘I was actually very glad that we didn’t have to go to a camp.’)
()
Maar als wij praaten in het Hollands [ . . . ] (Clyne, , p. ) (Homeland Dutch: maar als wij in het Hollands praaten ‘but as we speak in Dutch’)
It is difficult to discern whether examples such as in () and () are instances of transference from L English or rather reflect a simplification process. In the latter case, the word order of declarative main clauses in German or Dutch (V) would be applied to subordinate clauses, thus reducing the processing load of the speaker, who has to differentiate between two types of word order (see also Riehl, , pp. –). There are also some occurrences in contact situations of English and Germanic languages where German V in declarative sentences is replaced by English SOV. Compare the following examples from Norwegian (), Icelandic (), and German (): ()
I skolen vi snakte bare engelsk (nd gen., Johannessen, , p. ) (Homeland Norwegian I skolen snakte vi bare engelsk ‘At school we only spoke English’)
()
Dolly stundum talar íslensku (nd gen., Arnbjörnsdóttir, , p. ) (Homeland Icelandic Dolly talar stundum íslensku ‘Dolly sometimes speaks Icelandic’)
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
()
also in meinem Alter ich muss nehmen was mir anbietet (BH, AGZ) (STG also in meinem Alter muss ich nehmen, was sich mir anbietet ‘well, at my age I have to take what I get offered’)
As mentioned above, these instances only occur in first-generation attriters occasionally, but become more frequent in the next generations (see Riehl, , ). Another example of syntactic transference is the use of overt subject pronouns in pro-drop languages in contact with languages where the use of subject pronouns is obligatory: ()
Había una chica que vivía con su mama y ella va a visitar a su abuela (nd gen., Montrul, a, p. ) (‘There was a girl who lived with her mother and she goes to visit her grandmother’)
()
Ők tudták hogy őneki rossz kedve volt (Tóth, , p. ) (‘They knew that he was in a bad mood’)
()
vdrug on č uvstvuet čto on pojmal ogromnuju rybu (Isurin & Ivanova-Sullivan, , p. ) (‘suddenly he feels that he caught an enormous fish’)
In ()–(), the pronouns ella (‘she’), ök (‘they’), and on (‘he’) would be omitted in the respective languages (Spanish, Hungarian, and Russian), but not in the contact language English (the same applies to Turkish, cf. Gürel & Yilmaz, ).7
... Convergence/pivot matching As discussed above (see Section .), convergence can be considered a compromise strategy. In this case the speaker is incorporating in their speech components from the different subsets of their linguistic repertoire. These combinations are usually instances of creative usage in a particular situation, but nevertheless they are of the same nature as those processes that lead to contact-induced language change if they are repeatedly used by a collective of speakers and over a long period of time (cf. Riehl ). Matras (, ), however, maintains that the patterns of replication do not render a source language (SL) construction in a morpheme-to-morpheme translation (see Section ..), but are usually a blend; speakers make use of one or more key features of the model construction and map it to the recipient language (RL) by using various rules of morpho-syntax of the RL and also by using exclusively recipient words. Matras calls this pivot matching; in this framework the speaker deconstructs a certain construction by isolating its ‘pivotal features’ (Matras, , pp. –). The construction pivot is then matched to context-appropriate word forms and their formation and combination rules in L. In doing so, speakers enrich the inventory of constructions they have at their disposal in a given context of interaction (which means a subset of
7
According to Dubinina & Polinsky () this phenomenon could also be considered an instance of simplification. The overt expression of the pronoun requires less effort to establish its co-reference with the previous antecedent.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
a given ‘language’). Compare the following examples from German (), Russian (), and Croatian () attriters: ()
a. er ist eigentlich australisch geboren (AGC, CH) (‘he is actually Australian born’)
()
b. printsessa v lubvi s (Laleko, , p. ) (‘the princess was in love with’)
()
c. pričati s mamom i tatom i—ili kako su njihovi dani bili (nd gen., Hlavac, : ‘chat with mum and dad and—or how their days were’)
Lattey & Tracy () also observed a range of instances of this type of construction— which they called crossover in their corpus of German postwar migrants in the US.
.. Phonological transference and convergence Phonological transferences—that is, the addition or deletion of phonemes under the influence of the phonemic structure of the contact language—usually only occur in secondor third-generation speakers. There is, however, evidence of phonological convergence in first-generation immigrants, concerning the quality of vowels (Waas, , pp. –) or voice onset time (VOT) of voiceless stops (/p/, /t/, /k/). As it turns out, VOTs appear to be predominantly affected by cross-linguistic transfer. In a study on bilingual L speakers of English (L French) and L speakers of French (L English), Flege () investigated the VOT of L voiceless stops in both groups and found a change towards the values of the other language in both their L and their L. Speakers, however, did not reach the values of the respective monolingual baseline. Similar effects have been attested for attriters of L English in American immigrants to Brazil (Major, ). In a recent case study on VOT in immigrant Russians in Germany, Brehmer & Kurbangulova () also observed a tendency towards a longer duration of fortis stops in word-initial and intervocalic positions under the influence of L German in the first generation, which is extended in the second generation. Recent investigations on VOT and the structure of the vowel space in English-Dutch speakers (Mayr et al., ) and the lateral phoneme /l/ in the L and L of long-term German migrants in the Vancouver area (de Leeuw, ) give evidence of an assimilation to the L system. Similar effects were documented for the sounds /aː/ and /l/ in the study by Bergmann et al. () on attriters of German in North America. All the studies, however, also found a great deal of variation between speakers and in one and the same speaker. Influence of L has also been reported in intonation patterns in first language attriters (Waas, , p. ; Clyne, , p. ). These anecdotal impressions have now been empirically proven in a study by de Leeuw et al. (), who found a merged alignment of the prenuclear rising accent in German and English in German-English late bilinguals.
.. Morphological transference As morphological transference usually affects bound morphemes that come late in the borrowing hierarchy (see Figure .), it is much less likely to occur in first-generation language attriters. Indeed, language contact studies (e.g., Thomason & Kaufman, )
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
demonstrate that intense contact with the SL and high cultural pressure is needed to borrow grammatical morphemes and integrate them into the L system. However, there are casual occurrences of morphemic transfer, especially the transfer of plural markers in individual words, in first-generation attrition settings, too. Clyne (, p. ) reports the generalization of the s-plural in Dutch-speaking immigrants in Australia in the following lexemes: klant-s for klanten ‘clients’, stams for stammen ‘stems’, hoofleidings for hoofleidingen ‘mains’.8
.. Pragmatic transference and convergence A different type of transference is the copying of linguistic behaviour in the SL to the behaviour in the RL. Pragmatic patterns include forms of address, the use of reflex responses, discourse markers, and back channel behaviour as well as the use of speech act patterns. It is a typical observation with attriters in English-speaking countries that they extend the use of first name and informal pronouns (German du, Italian tu, French tu, etc.) to formal contexts or to conversations with a stranger where the formal address pronoun (Sie, lei, and vous respectively) would be the norm (cf. Waas, , pp. –; Clyne, , pp. –; and personal observation). Another phenomenon of pragmatic transference is the use of other-language discourse markers and related operators.9 In this context, Matras () introduced the notion of utterance modifiers to describe the different types of lexical items or phrasal units that are responsible for monitoring and directing the hearer’s processing activities. This definition, however, captures only one function of these items. Another explanation also includes the speaker’s monitoring activities—that is, commenting on language processing and bridging gaps in utterance planning (such as well, you know), which are used to guide the hearer but also to gain time for utterance planning, cf.: ()
a. ik vind het ook niet leuk, you know? (Clyne, , p. ) (‘I don’t like it either, you know?’)
()
CMR: Und dann hast du trotzdem schon mit dem Baby angefangen zu unterrichten? RB: Well, zuerst hab ich die Kinderkrippe geleitet und hab ihn mitgenommen. (AGC, RB) (‘And you started teaching, although you had the baby? Well, first I managed the nursery school, and I took him with me.’)
8 The non-target-like case use or reduction of case marking in languages which have a rich casemarking system (e.g., in Russian-English (Schmitt, ), in Hungarian-English (Fenyvesi, ) or German-English (Schmid, )) is not a typical contact phenomenon but an instance of simplification (see Section .). This becomes particularly evident when considering settings where both languages in contact have a case marking system (e.g., Russian-German or Hungarian-German, see Riehl, ). 9 In a narrower definition (e.g., Schiffrin, ), the term discourse marker includes adversative and other coordinating conjunctions and sentence particles, such as well, so, anyway. A broader definition also encompasses focus and modal particles, interjections, speech act enhancers, and phrasal adverbs (such as already, no longer). For a discussion see Matras (, , pp. –).
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Most studies on language contact emphasize the frequency of L insertion of this type of lexical items. Their borrowability is supported by the ‘detachability’ from the core grammar of the language. ‘Grammatical elements that organize the speech event are perceived as gesture-like, situation-bound devices and are therefore detachable from the content message of the utterance’ (Matras, , p. ). Therefore, as suggested by Dik (); Kaltenböck et al. (); and Heine et al. (), a distinction should be made between sentence grammar and discourse grammar, which serves functions such as interaction management, attitude specification, discourse organization, and execution. Another distinction was suggested in Riehl (), who assumes a tripartite system of utterance processing: the content-related level (Darstellungsebene), which refers to the content of the utterance; the evaluation-related level (Bewertungsebene), which refers to the level on which the speaker evaluates the content of the utterance; and the interactionrelated level (Interaktionsebene), where the speaker organizes the speaking activities and monitors language processing. Accordingly, speakers can resort to the repertoire of different languages at the different levels. Speaking Italian as a first-generation immigrant in Australia, the speaker can use Italian grammar in her utterances, but English discourse markers to manage the interaction (), or even switch into the English language to comment on her utterances (): ()
ma well and sperava d’ andare prima dei dei trent’ anni (Clyne, , p. ) (‘Well, there is hope but well and there has been hope of going for thirty years’)
()
Siamo ritornati a Roma e poi l’abbiamo lasciato. It was just amazing. Era proprio perfetto. (Riehl, , p. ) (‘We returned to Rome and afterwards we left it. It was really perfect.’)
In this context, the comparison of different language contact situations gives particular insights into language pragmatics; it becomes evident that the most frequently transferred discourse markers are particles with the meaning of English well, followed by connectors such as but (Clyne, , pp. –; Matras, , pp. –). This is illustrated by the following examples of occurrences of well in different language constellations of immigrants in Australia: ()
Well je kunt haast zeggen, ieder kind op de school ken ik (Dutch, Clyne, , p. ) (‘Well, you can almost say I know every child in the school’)
()
Well, meine Mutter war schon tot. (German, ACG, RB) (‘Well, my mother had already died.’)
()
Well si spera ma well and sperava d’andare prima dei dei trent’ anni (Italian, see ())
As Matras () assumes (see above), in managing interaction utterance modifiers adopt a gesture-like function. Those items whose lexical meaning is opaque (such as Engl. well, Russ. eto, Germ. also, Hung. Hát, all meaning ‘well’) are more gesture-like and are therefore transferred more easily than items with a full lexical meaning such as really. This finding, however, can be explained not only by the structural properties of the items but also by the pragmatic function of the respective units. While particles such as well or connectors such as but have the function of organizing the interaction, particles such as really have an evaluating
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
function. Therefore, they operate at different levels of the utterance processing system (Interaktionsebene vs. Evaluationsebene). In this case, the level of interaction is more often expressed in the more dominant language (Riehl, , p. ).
. E :
.................................................................................................................................. As demonstrated in the previous sections, language contact phenomena may be idiosyncratic in different types of attriters. But there are also transferences that appear, in similar contexts, in the utterances of different speakers. Thus, the question arises whether speakers have initiated this type of transference by themselves or whether they have adopted the item from the speech of other speakers in the bilingual community. Indeed, as Matras () points out, the interlocutors’ reactions are crucial to the effectiveness of a new item or construction. It has a major effect on whether the new element is accepted, used by the speaker again, and in turn replicated by others—which in the long run could lead to language change. This language change works in the same way as language change in monolingual settings. In immigrant communities, however, there is less normative control, which allows spontaneous innovations or norm-deviant use. Most of the norm-deviant forms go unnoticed by the speaker and/or their respective communication partner. If noticed, the utterance is not evaluated negatively and therefore not corrected. Consequently, there is also a certain plurality of different variants in the input delivered to the next generation (cf. Rosenberg, ; Riehl, ). As mentioned in Section ., both of the processes ‘attrition’ at the individual level and ‘language change’ at the community level are closely intertwined. If the first generation uses a high number of transferred items in their L, the next generation acquires a contactinduced variety with a high number of variants. This in turn may lead to more instances of transference in the next generation, due to the fact that the younger speakers are only able to express themselves in limited domains (i.e., in the family). Examples of these types of processes can be found in established bilingual language communities (e.g., in language enclaves (cf. Riehl, )). It has also become evident that in a monolingual language mode, speakers usually avoid inserting other language material (matter borrowing), as they are usually conscious of other-language items. Here, the amount of inserted material from L depends to a large extent on the metalinguistic awareness of individual speakers. Nevertheless, speakers display instances of transference that slip through their monitoring processes. These types of transferences typically occur in different speakers and different communities independently. As has been illustrated above, among those are: • phonologically-related lexemes adopting L meaning • constructions using rules of morpho-syntax and lexical material from L, but following L patterns (pivot matching)
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
• types of restructuring such as mapping collocations (e.g., Foto nehmen, Dusche nehmen) which do not provide a more precise meaning • monosyllabic gesture-like units without a lexical meaning (turn-related discourse markers). Whether in the end an instance of transference is integrated into a speaker’s L system or not depends on a variety of external factors such as language awareness and attitude, size of the bilingual community and amount of contact to native speakers in the country of origin (cf. Riehl, , ).
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
.............................................................................................................
SECOND LANGUAGE ATTRITION ............................................................................................................. Edited by
. ̈
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
......................................................................................................................
......................................................................................................................
. ̈
. I
.................................................................................................................................. T section addresses the topic of second (L) and foreign language (FL) attrition concerned with the attrition of language(s) acquired later in life, a topic which, surprisingly, has received relatively little attention in comparison to L attrition. L attrition research undoubtedly possesses an appeal not least because of the emotional load and impact involved in something as ingrained and intimate to a person as one’s mother tongue. However, L/FL attrition, too, has the potential to affect a huge number of people on a personal and emotional level due to the fact that more and more people learn and acquire FLs not to mention the impact it could have economically on the whole industry of FL learning and teaching. It is the aim of the present section (Part V) of this volume to serve as a starting point for researchers and other individuals interested in furthering the knowledge and understanding of the phenomenon. The chapters included present the issue from different points of view aiming to provide a detailed account of the existing expertise. The current introduction presents the terminology used in the field, focusing on some important differences with respect to the lingo used in L attrition studies. It provides a short overview of the development of the field and outlines the major challenges and obstacles that research on the topic has to deal with. We briefly describe the major findings and the knowledge amassed before proceeding to review some of the most significant and large-scale projects carried out on L/FL attrition. Finally, several theoretical frameworks that could be used to explain the phenomenon of L/FL attrition as well as some ideas for future research are discussed.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
ö
. T
.................................................................................................................................. While L and L/FL attrition have many aspects in common and share terminology, there are also a number of differences that have to be clarified and discussed especially in a volume dedicated mainly to L attrition. One term that is specific to L/FL attrition studies is incubation period (Gardner, ). Incubation period, or length of attrition, refers to the time elapsed between the moment active contact with and use of the L/FL ceases and the moment when proficiency in that language is evaluated again. Mehotcheva & Mytara (Chapter , this volume) provide more information for the role incubation plays in L/FL attrition. Many L attrition studies focus on returnees, a term referring to both children and adults who have spent time in an L environment and have returned to their country of origin, without specification of a duration or minimal amount of time established in order to be considered a returnee (see also Taura, Chapter , this volume). Another important distinction between L and L attrition research concerns the baseline used to establish attrition. While in L attrition studies attriting immigrant participants are usually compared to a similar population in the country of origin, L/FL attrition studies work with people with varying levels of proficiency thus making the detection of attrition (as opposed to lack of acquisition) extremely challenging. One term that is used in both L and L/FL attrition is onset proficiency, also sometimes referred to as attained proficiency, relating to the level of proficiency achieved before active use and contact with the language decrease, marking the beginning of incubation period. Finally, concerning the use of the terms L and FL we are going to adhere to the distinction proposed by Schmid & Mehotcheva () whereby L attrition refers to the attrition of the L in bilingual individuals or in returnees who start using and relying more on their L (e.g., Starren, ; Hansen, a; Taura, ), while FL attrition deals with an instructed language learned in school or a language ‘picked up’ in a naturalistic environment while L attrition refers to immersed acquisition of an L (Murtagh, ; Mehotcheva, ; Wang, ; Xu, ). Here we are going to use the term L/FL attrition where both scenarios could be involved and use the terms separately if so required.
. H
.................................................................................................................................. Interest in L attrition can be traced back as far as , to Cole’s foundational article on ‘The effect of a summer vacation on students’ knowledge of French’. Several other studies were inspired by this paper, all using the summer recess to test the attrition of Latin by American high school students (Kennedy, ), German by American university students (Scherer, ), and French by native English university students in Canada (Smythe et al., ). The first collection of papers dedicated to the phenomenon of attrition by Lambert and Freed (), although focusing on attrition in general, also included five contributions specifically addressing the topic of L/FL attrition. Among these were four theoretical articles—Berko-Gleason (); Dorian (); Obler (); and Valdman () outlining directions for future research and methodological considerations from the point of view of
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
the respective areas of expertise of the authors, and one reporting on ‘The U.S. Government’s FL attrition and maintenance experience’ (Lowe, ). A second volume focusing on language attrition followed in edited by Weltens et al. Its largest section (Part IV) specifically addressed second and FL attrition and included both theoretical articles (Cohen, ; Jordens et al., ; Lambert & Moore, ) and empirical studies exploring the phenomenon from different perspectives—that is, FL attrition in the elderly (de Bot & Lintsen, ), young speakers (Olshtain, ), and high school acquired French (Weltens et al., ). The following years saw a further rise in the number of empirical studies dedicated to the topic (see Schmid, b, for an overview). In , Hansen edited a collection of empirical studies exploring L/FL retention/ attrition within a Japanese context (see annotated bibliography Schmid, b and Taura, Chapter , this volume) and more recently there have been a number of largescale studies dedicated to adult L/FL attrition (see Table . for details on each project) in addition to some case studies (Table .). Despite a certain increase of interest in the topic and the diversity of populations and situations covered, the topic of L/FL attrition remains understudied and one of the reasons why this is so may lie in the obstacles and challenges that any research on the topic has to face.
. T
.................................................................................................................................. A challenge particular to L/FL research is that it usually deals with past experiences and thus has to evaluate a speaker’s knowledge of a language at a past point in time. Since going back in time is not an option, this poses serious methodological problems of how to estimate the linguistic knowledge of L/FL attriters before the incubation period. The importance of establishing a valid baseline of the knowledge and linguistic competence of the attriters before the onset of attrition has been recognized from the early stages of development of the field. Andersen (, p. ) noted that ‘we need to know how normal LCs [linguistically competent users, not necessarily native speakers] use that feature’ in the first place, in order to be able to claim that a certain feature has suffered attrition rather than not having been acquired completely. A further problem for investigations of L attrition is linked to recruitment: unlike investigations of L acquisition, which can draw upon instructed learners in a classroom environment, L attrition studies rely on participants for whom the linguistic experience lies years or even decades in the past. Not only are such participants difficult to locate, it is also a challenge to match them regarding their linguistic experiences, methods of acquisition (in particular since teaching styles and methods tend to change rapidly) time spent learning the language or being exposed to it. This means that estimating the level of proficiency at the onset of the incubation period is extremely complex—something which may explain the lack of large-scale studies and the abundance of case studies in L attrition research. As mentioned earlier, this diversity also poses a challenge when establishing a baseline against which to compare the performance of the attriters. L/FL attriters cannot be compared to native speakers because their proficiency could vary substantially from one
Periods studied/ groups
Materials used
Main findings
Seven groups at , , , , , , and years of attrition
Oral translation task from English into Persian.
General words attrite more than special words.
End of instruction, and months later
Receptive & productive vocabulary achievement tests.
Attrition in both receptive & productive vocabulary; greater in productive.
Japanese/KoreanEnglish
/
– years of attrition
Real words & pseudo-words translation task.
Support for the savings paradigm; the larger the lexicon, the greater the savings benefit.
Hansen & Newbold,
Japanese-English
months to years of attrition
Listening comprehension; classifier & negative elicitation; storytelling.
Advantage for literacy.
Hansen et al.,
Japanese/KoreanEnglish
/
to years— Korean/ to years of attrition.
Real words and pseudo-words translation task.
Support for the savings paradigm; the larger the lexicon, the greater the savings benefit.
Hansen & Chen,
Japanese/MandarinEnglish
/
n.a.
Off-line picture naming.
Attrition both in syntax and semantics; loss of semantic categories in an inverse order.
Marefat & Rouhshad,
English-Iranian
months
Productive & receptive vocabulary, in & out of context; concrete & abstract nouns.
Attrition for non-contextualized nouns.
Morshedian,
English-Iranian
months
Productive & receptive vocabulary.
Advantage for receptive lexicon, and higher initial proficiency.
Study
Attriting—Native lang.
n
Abbasian & Khajavi,
English-Persian
Alharti,
English-Arabic
Hansen,
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Table .. Group studies on adult L/FL attrition, post
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Table .. Case studies on L/FL attrition, post Study
Attriting—Native lang. n
Mehotcheva, Spanish-German
Snape et al.,
English-Japanese
Tomiyama,
English-Japanese
Tomiyama,
English-Japanese
Periods studied Materials used
Main findings
Attrition for and months Storytelling task, productive skills. of attrition self-evaluation & listening, grammar, vocabulary. -month period TOEIC & C-test scores, Acceptability Judgement Task. and Free conversation, months of storytelling, attrition Peabody Picture Vocabulary test, the Bilingual Syntax measure.
Slight attrition in aspect for one participant.
, , , , , , Storytelling. and months of attrition
No difference for age, some attrition in lexical complexity.
Change in syntax & morphology, stability in lexicon & fluency.
person to another and in many cases will not be close to native-like (see, e.g., ReetzKurashige, ; Murtagh & Van der Slik, ). Using longitudinal designs where each individual’s knowledge is used as its own baseline is a possibility, however, longitudinal designs present practical difficulties, such as high drop-out rates and long investigation periods, which can often be difficult to fund. As a way out, Weltens (, p. ) proposed using cross-sectional designs where the baseline group consists of ‘the same—or at least highly comparable—individuals as those whose attrition data are used’ but who are still in active contact with and use the language. Another possibility is to use mixed designs, as was done by Mehotcheva () who complemented her cross-sectional data of Dutch and German university FL Spanish attriters with longitudinal data for some of the participants. A further difficulty is posed by the lack of a unified methodology and standardized data collection materials. There is great variance with respect to these in the existing work, which furthermore focus on different aspects and linguistic features, making cross-study comparisons next to impossible. For example, Murtagh () studied oral production of words and discourse in high school leavers over a period of months; Taura () focused on productive skills in high school students with varying lengths of attrition from one to over five; Tomiyama (a) followed the linguistic development of an -year-old returnee boy during a period of nineteen months, while Weltens () explored the receptive knowledge of university students over a four-year period. The data collection techniques used include specially designed oral tests (Murtagh, ); free conversation (Tomiyama, ); storytelling based on pictures/story book (Cohen, ; Olshtain, ; Taura, ); vocabulary recognition and recall (Bahrick, a,b); classical listening and reading tests (Weltens, ); and multiple choice proficiency tests (Xu, ).
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
ö
Although all of these studies provide valuable information regarding the mechanisms of L/FL attrition, they do not allow insights into the wider picture. In order to be able to make comparisons, a more unified framework for research design (and analysis) is needed. An attempt to provide such a design was one of the objectives of the European Graduate Network on Language Attrition (see Schmid & Cherciov, Chapter , this volume). This group of scholars set out to compile a set of tools, covering different tasks, languages, and methods comprising introspective and self-assessment tools, formal elicitation tasks as well as guidelines for the elicitation, transcription, and analysis of spontaneous speech. These tools, as well as guidelines for their application and data-analysis, were made available in a wide range of languages on https:\\languageattrition.org. The most important shortcoming of L/FL attrition research so far, however, is that many studies remain purely descriptive and are not motivated by theoretical questions. While L attrition studies are often driven by questions about the nature of grammar, the permanence of linguistic knowledge, or the functioning of fundamental mechanisms of language processing (e.g., frequency of activation), such an embedding in a theoretical framework is often absent in L attrition research. The lack of theoretical underpinning can in part be ascribed to the original practical orientation of the field (see Köpke & Schmid, ), with its interest in assessing the vulnerability of instructed languages to attrition in a variety of learner settings and the impact of the extralinguistic factors that play a role in determining extent of retention/attrition of linguistic knowledge and use. However, a stronger embedding of research within theories of SLA is now needed in order to fully understand the base mechanisms of the L attritional process.
. S
.................................................................................................................................. Despite the problems outlined in the previous section, research carried out so far allows a number of insights regarding the nature and rate of L/FL attrition as well as the role of incubation period, the amount of knowledge lost, some of the extralinguistic factors involved, and finally some speculations as to the possible role of linguistic proximity. Whether the nature of attrition is a process that is uniform across all linguistic levels or differential, has long been a subject of speculation and research has aimed at establishing which are the elements and linguistic levels at each end. One of the earliest reported distinctions in that respect is the higher level of vulnerability of productive skills—speaking and writing, over receptive ones such as listening and reading (Bahrick, a; Moorcroft & Gardner, ; Weltens, ; Cohen, ; Olshtain, ; Weltens & Grendel, ; Tomiyama, ; Mehotcheva, ; Wang, ). The idea that receptive vocabulary is more resistant to attrition originates in research on L acquisition which has shown that receptive skills precede productive ones in acquisition (Ellis, ) and that size of receptive vocabulary is a good predictor of productive vocabulary size (Webb, ). Rooted in predictions of frameworks such as the Activation Threshold Hypothesis (ATH; see below), the notion that productive skills are more easily affected by attrition than receptive ones is one of the few uncontroversial assumptions in attrition research. When it comes to rate of attrition, two seemingly opposing views have been put forward: (a) initial attrition followed by levelling-off versus (b) an initial plateau for the duration of
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
which linguistic knowledge remains intact followed by attrition (see Weltens, , for a detailed discussion). While the former view has received ample support from empirical research (Bahrick, a,b; Weltens, ; Weltens et al., ; Taura, ; Mehotcheva, ), the latter has been extensively questioned in recent studies. Yoshitomi () expressed her doubt at the existence of ‘a real initial plateau’ since in its early stages attrition takes ‘the form of erosion in the ability to coordinate various linguistic subsystems simultaneously and spontaneously’ (p. ), something that could be difficult to detect through the measurements used. However, rate of attrition will also vary a lot according to the skills tested and is largely depend on extralinguistic factors, namely age of onset (attrition rate being faster the younger the subject at onset) and the time scales involved (see Bylund and Mehotcheva & Mytara, Chapters and , respectively, in this volume). Regarding the incubation period, although intuitively expected to be crucial, its importance as a cause for a language to attrite has not been confirmed by research. Studies by Weltens (); Murtagh (); Taura (); and Mehotcheva (), have shown that attrition does not occur as a ‘function of incubation’ (Taura, , p. ) and that language attrition is not a linear process (Taura, ; Mehotcheva, ), that is to say, the amount of attrition is not dependent on the length of the incubation period. As to the amount of knowledge lost, the traditional theory on forgetting considers that a definite proportion of one’s knowledge is lost irrespective of the total knowledge. In practical terms this means that the more you know, the more you would be prone to lose. Research by Bahrick (a, b) and Weltens () has challenged that view claiming that attriters lose a constant amount of knowledge across all proficiency levels. However, in relative terms this means that highly proficient individuals are left with a bigger portion of their knowledge intact. With regard to the other extralinguistic factors—linguistic, personal, and external—the main findings concerning their influence can be summarized as follows (see Mehotcheva & Mytara, Chapter , this volume, for more details). Onset proficiency has been outlined as one of the most influential factors (Bahrick, a,b; Weltens, ; Harley, ; Murtagh, ; Mehotcheva, ). Sometimes operationalized as course grades and number of courses taken (Bahrick, a), it might predict the outcome of attrition/retention with higher initial proficiency, higher course grades, and higher number of courses associated with better retention of the language. In L attrition, it has been suggested that once a language is sufficiently established/entrenched in the speaker’s brain, it becomes impervious to deterioration (Schmid, ) and this claim might hold true for L/FL attrition as well. Other factors related to retention/attrition include age—older children fare better than younger ones (Cohen, ; Olshtain, ; Taura, ); literacy—literate speakers and/or those having received formal education in the target language, retain the language better (Hansen & Newbold, ); attitude and motivation—attitudinal factors help speakers retain better a language (Gardner et al., ); rehearsal (frequency and recency) during the period of attrition has been suggested to lead to better retention (Andersen, ) but the findings so far are controversial (Taura, ; Mehotcheva, ; Wang ; Xu, ). Finally, we would like to very briefly mention the relevance of linguistic proximity, that is the similarities and differences between languages and their consequences for L/FL attrition. In situations where only two languages are involved, as Schmid & Mehotcheva () argue, an L can either promote or inhibit the retention of the attriting language. When it comes to multilingual speakers the role and influence of previously or subsequently acquired/learned languages may all come into play. Furthermore, linguistic proximity
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
ö
could influence different linguistic levels independently: it might exert a facilitatory role on one level and an inhibitory one on another (Schmid & Mehotcheva, ). Weltens () reports that linguistic proximity (i.e., cognates) did facilitate language retention and Hansen () observed a facilitatory role of linguistic proximity for vocabulary. Olshtain () and Reetz-Kurashige () report that typological conflict between the languages negatively affects features with high semantic and pragmatic load. But for now, the data gathered on these issues is rather limited and there is need for more studies in order to precise the interplay of the different factors.
. D
.................................................................................................................................. Although it is beyond the scope of the present introduction to provide a detailed overview of all the studies cited previously, we would like to include a brief overview of some of the big-scale projects that we feel have been most influential in the field or we hope would be in the future. Further studies are mentioned in Tables . and .. Bahrick’s (a) study on the retention of L Spanish by adult English L speakers is probably one of the most influential, as well as one of the largest, studies of attrition ever conducted, with a total of participants: of the participants were still learning the language in high school or university language courses or had recently completed such a course (maximum two months prior to being interviewed) and had taken one or more courses of Spanish while still at college or university, one to fifty years prior to the interview. The attriting participants were assigned to one of eight groups depending on the length of their incubation period and further grouped into ten training levels depending on the courses they had previously taken. The remaining forty participants had not received instruction in the target language. They were included in order to discriminate between Spanish learned in class and Spanish picked up incidentally, and between answers that were marked correctly and answers that were just guessed correctly. Bahrick used two instruments to gather data: a test of Spanish and a questionnaire about the method of instruction, the grades obtained on the courses taken, as well as the frequency with which Spanish or other Romance languages were used during the incubation period. The linguistic data comprised ten subtests: reading comprehension; Spanish-English/English-Spanish recall of vocabulary; Spanish-English/English-Spanish recognition of vocabulary; grammar recall and recognition; idiom recognition and recall, and mixed word order sentences. The results showed that during the first three to six years there was a steady and rapid drop in the retention of language knowledge which then stabilized, becoming largely immutable to further attrition for the next fifty years. The grades participants had obtained were valid predictors of performance even several decades later; the more courses a person had taken, the greater the amount of retained knowledge, while having completed only one course did not leave any significant footprint, and the absolute amount of knowledge forgotten during the first five years was equal for individuals across different training levels but proportionally this represented a smaller portion of the total knowledge of participants with high proficiency level.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Weltens () focused on the retention of school-acquired French among learners with L Dutch. This study explored receptive skills, comparing two different training levels: four and six years of studying the language, using both cross-sectional and longitudinal (four years) measurements. The tests consisted of measures of general proficiency (a cloze test), listening and reading comprehension tests, two phonology tests—a phoneme discrimination test and a variant of the rhyme test, two vocabulary translation tests, and two morpho-syntactic tests—multiple choice blank filling, in addition to a questionnaire with self-reported data about the participants’ linguistic background, their attitude towards French, as well as their self-perceived proficiency in listening and reading comprehension measured by Can-Do scales. Furthermore, the linguistic relation between the two languages involved—that is, Dutch and French—was also explored by discriminating between contrasting and similar items in the phonological, lexical, and morphological subtests. The results obtained were quite peculiar. A proficiency effect was registered for all the tests, with the higher training level outperforming the lower one across all tests. Linguistic proximity was found to play a role, with cognates and similar items faring better than noncognates and contrasting items for the reading, and phonological and morpho-syntactic sections respectively. Attrition, however, was registered only at the morpho-syntactic level where a positive correlation with incubation period was registered. Regarding the general proficiency tests, the slight decrease registered was not significant and this was interpreted as a proof that the participants’ ‘general proficiency remained stable over time’ (Weltens, , p. ). However, for the listening, reading, and phonological sections there was a negative correlation with incubation period, meaning that scores had increased rather than decreased. Weltens speculates that the results in the phonological subtests might be due to a ceiling effect and he also reports a positive correlation between test scores on the reading comprehension and the time required to complete the test for the low proficiency group, indicating that participants who took more time were also more accurate. Since the difference in time required between the high and low proficiency students was only moderate, however, Weltens dismisses its possible effect on the results. Finally, a discrepancy was registered between the self-report data where participants claimed heavy attrition in both listening and reading comprehension and the improvement on the actual test results. Lelia Murtagh’s study () explored the retention of school-acquired Irish among high school leavers. Although an official language in Ireland, outside of the traditional Irish speaking areas (where daily use of the language is reported to be at around %) Irish is used only by a very small fraction of the adult population—.% according to Murtagh (, p. ). The author explains that even though bilingualism is promoted and supported by providing Irish tuition from primary education till the end of secondary education, in reality students encounter few opportunities to practise the language once they finish their education due to the limited use of the language in society. Murtagh compared secondary school leavers from three different learning environments: ordinary level Irish, advanced level Irish, and immersion programme students. The longitudinal design included two data-collection sessions over an eighteen-month period involving a cloze test, a communicative test in spoken Irish (both created for the purposes of the study), in addition to a student questionnaire. The latter provided personal information, self-assessment ratings on the ability to speak Irish, and self-estimated use of the language outside of school as well as attitude and motivation towards learning Irish (based on the original AMTB—attitude and motivational test battery—by Gardner, ).
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
ö
The results found no evidence for language attrition for the studied period. With a few individual exceptions, incubation period alone showed no link with the participants’ proficiency. This replicated Weltens’ results regarding the mismatch between the participants’ self-reported feeling of decreased performance and their performance on the tests, even though Murtagh’s study used a census-type choice test with an increasing load rather than a Can-Do scale like Weltens. For all instructional levels the opportunities to use the language were found to be reduced; however, it was the students from the immersion programme who were more likely to continue using the language and also the ones that used it more actively while still at school. Initial proficiency and reading in Irish after leaving school were the only two significant predictors of performance. Taura () looked into the retention and attrition of L and FL English with a focus on the effect of formal education by comparing two groups of high school Japanese returnees, one that received formal education in English and another one that did not (see also Taura, Chapter , this volume). The study also explored the effect of extralinguistic variables such as literacy, education, onset age at which moving to the L environment took place, length of residence in the L environment, and incubation period. Taura employed a crosssectional design and used three different tasks: a receptive vocabulary test that was administered to eighty participants aged to years at the time of testing, and two spontaneous production tasks (oral and written) administered to sixty-four participants aged to . The project was further complemented by a series of five smaller-scale longitudinal studies that additionally probed into the influence of individual differences. The results showed that formal L education between to years of age was central for retaining the acquired knowledge once the intensive use and contact with the language was over. Language attrition was not found to be the result of incubation period. Formal L education lasting four or more years was linked to better retention of some sub-skills. Syntax and complexity both did not seem to be influenced by the processes of attrition and writing skills not only did not show signs of attrition but they actually improved over time. Fluency, however, seemed to be affected as early as beginning of the second incubation year for the participants that received formal education and first year for those who did not, evidenced by a higher filler count in the oral data and a decline in the error free T-units. Mehotcheva () investigated the attrition and retention of university acquired FL Spanish in Dutch and German university students who had also completed a study abroad (SA) period as Erasmus exchange students in a Spanish speaking country. Mehotcheva combined a cross-sectional design—fifty-one informants allocated to one of three attriting groups with an increasing incubation period and a baseline group for whom attrition had not started yet—with a longitudinal subsample of five informants whose linguistic development was tracked for a year after the initial data collection session. Data was gathered by means of a semi-structured interview, a C-test and a picture naming task assessing speed and accuracy of lexical access. In addition, the study also aimed at investigating the role of background and personal factors such as incubation period, contact with the language, attitude and motivation, and initial proficiency. Evidence for attrition was found at both the linguistic level, manifested as an increased number of disfluency markers, reduced lexical diversity, and higher incidence of disfluency markers preceding lexical items in oral speech, and at the psycholinguistic level, as evidenced by slower reaction times and lower accuracy in the naming task. Incubation period alone was not found to affect the linguistic capabilities of the participants. Similarly, contact and use of the language during the
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
incubation period, contrary to what was expected, did not have a beneficial effect on language retention. The results from the exploration of attitude and motivation, established with an adapted version of Gardners’ AMTB (), showed that the different types of motivation—that is, instrumental, integrative, and interest in FL—correlated with different linguistic abilities but there was no overall positive effect. Mehotcheva interpreted this as possibly a lack of sensitivity of the questionnaire or as a consequence of the changing nature of attitudinal variables that could not be reliably measured throughout the different stages of the developmental process—that is, acquisition, use, and then attrition. Overall, initial proficiency was established as the strongest predictor of retention/attrition, with higher onset proficiency related to better retention of the language. Wang (), in a very large-scale project, explored the attrition of English in Chinese college and university English majors. She used a cross-sectional design with five groups of around informants each with zero, two, and four years of college-level English instruction as well as two and four years of attrition. This design allowed her to track both the process of acquisition and attrition of English. The focus of the study was on listening and reading comprehension and lexical knowledge based on tasks such as dictation of sentences, a multiple choice vocabulary task, and a reading comprehension task. In addition, Wang also used a questionnaire based on Gardner’s () AMTB to probe the effect of language use and attitude on the attrition and retention of English. The results showed significant attrition across all the language skills tested, with students faring reasonably well on listening comprehension, lexical knowledge showing the greatest signs of attrition, and reading comprehension falling in between. Regarding the different periods of attrition, a very abrupt and dramatic decline in the knowledge of the two-year attriters was registered, followed by further attrition, although not as drastic, for the four-year attriters whose linguistic competence was found to be equal to that of the freshmen who had not yet started English classes at college/university. The author ascribes this dramatic drop in competence to the typological distance between the two languages involved—English and Chinese—and the consequent lack of cross-language maintenance effects. The study also reports that more complex and more difficult language skills are the first to attrite, thus providing support for the Regression Hypothesis (see below Section .). Finally, personal attitudes and motivation were found to be very important for the maintenance of the language on any level and across all the skills that were explored. Xu’s study () investigated the retention and attrition of English as a FL in Chinese and Dutch university students after the end of the period of instruction. The study questioned whether contact with the language and the participants’ attitudes towards the language influence the outcome of retention/attrition. An interesting feature of the study is the comparison between two learning/attriting contexts: (a) the Netherlands, where English is omnipresent in daily life and is readily available through the TV, radio, books, magazines, etc., and (b) China where English is less available. Xu used a longitudinal design and employed linguistic tests to explore the attrition in vocabulary and grammar, Can-Do statements to evaluate listening and speaking skills, and a combination of both for reading and writing skills. The Chinese sample consisted of over non-English majors and the Dutch sample of eighty-one university students, tested at three and two different times respectively for the two groups over the course of their university studies. Regarding the
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
ö
Chinese sample, the two-year period was found to be critical for the retention of English. Attrition was registered across all skills and linguistic aspects investigated; the reading skills were found to be more vulnerable than the other receptive skills and writing was identified as the skill that was the most affected by attrition as measured in terms of lexical richness and overall quality. As to the Dutch sample, writing was the only skill that registered a decrease in the scores, whereas grammar, vocabulary, and reading scores, as well as the results of the questionnaire were not significantly different. In line with Bahrick’s (a) and Weltens’ () conclusions, initial proficiency explained a large amount of the variance in the performance registered two years later. The contact factor neither explained the English language retention, nor did it distinguish the two different samples as originally expected. Finally, the participants’ attitudes had a different effect in the two groups: while positive attitudes in the Chinese sample were related to better language retention, that was not the case for the Dutch sample where attitudes did not seem to play a role.
. H ( )
.................................................................................................................................. In the beginning of this chapter we outlined several differences between research on L and L attrition, the most important of which is related to the theoretical grounding of research. While L attrition studies have usually been founded in theoretical linguistic frameworks and theories such as Markedness theory (see Hansen & Chen, ), studies on L/FL attrition are most often driven by practical aims and lack solid theoretical underpinnings. Still, there are a number of theories and hypotheses that have been put forward in order to interpret findings from L/FL attrition. Among the most important of these are: the Regression Hypothesis, the Savings paradigm, ATH, Critical Threshold, and the Dynamic Model of Multilingualism (DMM).
.. The Regression Hypothesis Originally formulated by Ribot in the s and later used by Freud (who coined the term regression) in his work with aphasics, the Regression Hypothesis was finally adapted to language by Jakobson who applied it to phonological regression in aphasics as compared to to language acquisition in children (Berko-Gleason, , p.; see also Barkat-Defradas et al., Chapter , this volume). The Regression Hypothesis postulates that language dissolution is the mirror process of language acquisition—that is, what has been acquired last is the first to be gone and vice versa. This hypothesis was taken up by Andersen () in his seminal article on the linguistic traits of attrition, where he expressed the idea that early-acquired forms and structures would be retained better than later acquisitions in attrition. A number of studies (Olshtain, , ; Moorcroft & Gardener, ; Kuhberg, ; Hansen, a, ; Hansen & Chen, ) have provided support for the Regression Hypothesis in relation to L/FL attrition. However, as Hansen (a, p. ) points out, there is still a need for more research on the topic in order to see ‘when and under what conditions its predictions hold
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
true’, especially considering the fact that, as Murtagh () observed, many of the studies finding support for the Regression Hypothesis involved young children whose L system were not yet fully developed. Furthermore, it is likely that order of acquisition leads to differences in entrenchment (Steinkrauss & Schmid, ) and that these and other factors (such as initial level, incubation period, etc.) may therefore be confounded, making a straightforward corroboration of the Regression Hypothesis difficult.
.. The Savings paradigm One of the earliest theories expressed in relation to L/FL attrition is the Savings paradigm (Nelson, ), sometimes also referred to as the Relearning theory (see also Köpke & Keijzer and Larson-Hall, Chapter and , respectively, in this volume). It has its origins in theories on forgetting claiming that stored information is never completely lost from memory. The Savings paradigm is based on the idea that there are residues of knowledge that can be used to reactivate information, with higher level of activation needed for recall than for recognition (de Bot et al., ; see also ATH below). Reduced levels of activation first lead to retrieval and then to recognition failure, but it is assumed that the information is not totally erased and completely forgotten. One way to reactivate it is through relearning, the assumption being that newly learned words cannot reach the same level of activation after only a short period of learning as previously known information. Research by de Bot & Stoessel (), who tried to reactivate the knowledge of a language (Dutch) which their participant had learned thirty years earlier, found signs for residual knowledge. Even though strong cross-linguistic facilitation for cognate words was reported, there was evidence that even non-cognates were retained. Another study by de Bot et al. () found that their participants were better in learning words that they were likely to have learned when studying a FL (German in the USA and French in the Netherlands) in comparison to learning new vocabulary items. Hansen et al. () used a similar design with English L attriters of Japanese and Korean and also found evidence for a savings effect, despite the absence of any cognates. It has to be noted that all these studies focused on lexical reactivation and it is not clear if the benefit of relearning is accessible to other linguistic levels such as syntax and/or morphology. The only study focusing on the syntactic and morphological level (Ioup, ) failed to detect any residual knowledge. As de Bot & Stoessel () conclude, there is need for further research within this paradigm with more attention and control of the individual differences.
.. The Activation Threshold Hypothesis The reference to the activation level is core to another hypothesis: the ATH (Paradis, , ). The ATH is built on the assumption that each linguistic item and subsystem has an activation threshold level that depends on the impulses that are necessary to activate it. When an item has a low threshold level, fewer impulses are needed, while a high threshold requires more (see also Köpke & Keijzer, Chapter , this volume). The more frequently an item is activated, and the more recent the last activation, the lower its threshold level and vice versa. Although frequency (and recency) of use is one of the first variables intuitively
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
ö
expected by researchers and laymen alike to be of importance to L/FL attrition/retention, the existing research has failed to find unequivocal support in that direction. While some studies did not find an effect at all (Taura, ; Mehotcheva, ; Xu, ) others have reported limited effect for only certain areas of linguistic competence (Murtagh, ; Alharthi & Al Fraidan, ). Furthermore, recognition of an item is externally-stimulated usually by a visual or auditory signal, but production of an item is internally driven and requires an impulse from within making it a more challenging process. So, while a speaker might fail to retrieve an item for production, s/he may still be able to recognize and understand it. This is very much in line with the existing findings that have found the receptive skills to be fairly immune to attrition in comparison to the productive skills (Bahrick, a; Weltens, ; Mehotcheva, ).
.. Critical threshold A ‘critical’ threshold level is core in yet another theory on L/FL attrition but this time focused on the level of proficiency that has to be achieved in order to retain the competence in a language. This theory goes back to Bahrick’s findings that linked higher levels of proficiency to the notion of a permastore, that is, linguistic knowledge with a lifespan of over twenty-five years. Neisser (, p. ) in turn suggested that there might be a critical threshold of strength that some items reach and that it is that threshold which makes high proficiency speakers of a language immune to substantial language attrition. This idea is supported by research on L attrition as in the studies of Weltens (); Murtagh (); and Mehotcheva (). Hansen (a) also found support for the critical threshold notion in her research on returned missionaries who had acquired Japanese in naturalistic settings. She found that participants who had spent only two years in Japan did not retain the language as well as participants who had spent three years there. She concluded that in order to retain a naturally acquired language, more than two years ‘of daily language use in the target culture’ are necessary (Hansen, a, p. ). In a similar vein, the idea of the cumulating effect of language use leading to enhanced entrenchment has been discussed recently in the context of L attrition by Steinkrauss & Schmid ().
.. Dynamic Model of Multilingualism The DMM developed by Herdina & Jessner () uses expertise from several fields of research including Dynamic Systems Theory (DST) (see Opitz, Chapter , this volume), general biology and cognitive psychology and applies it to the study of multilingual systems especially relevant to bilingual and/or multilingual individuals. Under DMM, language development is characterized by change of quality, reversibility, stability, complexity, nonlinearity, and interdependence. Proficiency in a language is seen as fluctuating: deteriorating, improving, or remaining stable depending on the time and attention devoted to it. A multilingual language system consists of other smaller, nested sub-systems representing the different languages spoken by a multilingual; these in turn are composed of the subsystems of morphology, syntax, phonology, etc. All subsystems interact between
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
themselves and with the surrounding environment. They are in a process ‘of constant adjustment to the changing environment and internal conditions aiming at maintenance of a state of (dynamic) balance’ (Herdina & Jessner, , p. ). DMM provisions positive and negative growth depending on the time and effort invested into the language system. More time and effort result in a development of the language, that is positive growth, while decrease in the time devoted to the language system results in negative growth, which eventually leads to language attrition or gradual language loss. According to DMM, language attrition very often passes unnoticed, especially in the early stages since its major expression at this point is only less frequent performance. DMM considers the language maintenance effort (LME) that bilinguals and multilinguals have to exert in order to keep their languages healthy. Specifically, the LME combines the use of the language for communication and the verification of hypotheses concerning the language system. Not using a language means there is lack of LME and results in deterioration of the linguistic competence, while at the same time, cross-linguistic influence, and therefore activation, is not ruled out. Although this cross-linguistic activation may not be enough to keep a linguistic system fully functional, it could explain the fact that complete loss is not registered beyond a certain age limit. DMM is one of the few models that explicitly considers language attrition as part of the natural linguistic processes taking place in bilingual and/or multilingual speakers and while it predicts that ‘language loss will affect different linguistic subsystems to an unequal degree’ (Herdina & Jessner, , p. ) it makes no detailed speculations as to which linguistic system—that is, the lexical, grammatical, etc.—will be affected more or to any possible order of attrition.
. F
.................................................................................................................................. One theory that is steadily developing in the field of second language acquisition and seems quite promising with respect to language attrition is Complex Dynamic Systems Theory— CDST (see Opitz, Chapter , this volume, for more details) combining DST and Complex/ Chaos Theory. DST was introduced to the study of the development of language and language acquisition, and consequently language attrition, as a response to the limitations of the traditional, linear view on these fields of research (Evans, ; Larsen-Freeman, Schmid, & Lowie, ). DST (Van Geert, ; de Bot, ; de Bot, Lowie, & Verspoor, a), and Complexity/Chaos Theory (Larsen-Freeman, , , ; Cameron & Larsen-Freeman, ) emerged as a better candidate to account for the flexible and dynamic nature of language attrition. As both theories share similar founding assumptions and principles, they are now referred to jointly as CDST (de Bot, ). A complex dynamic system consists of numerous subsystems where all parts are interconnected and continuously influencing each other. Thus, the system is in a state of constant adjustment to change itself, occasionally settling into ‘an attractor state’ which de Bot & Larsen-Freeman (, p. ) define ‘as the state that the system prefers to be in over other [repeller] states at a particular point in time’. Both ‘attractor’ and ‘repellor’ states come about as a result of the development of the system, which evolves on the basis of selforganization and emergence (Larsen-Freeman, ). Furthermore, a dynamic system is characterized as being open—that is, allowing for external influences and resources to come
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
ö
into it (Cameron & Larsen-Freeman, ) and influence it (de Bot & Larsen-Freeman, )—and non-linear in nature (de Bot & Larsen-Freeman, ; Evans, ; Thelen & Smith, ); also seen in a computerized simulation on L attrition carried out by Meara () (see Larson-Hall, Chapter , this volume). Although it is acknowledged that initial conditions could influence the development of the system (Larsen-Freeman, ; de Bot et al., a; de Bot & Larsen-Freeman, ), it is very difficult, not to say next to impossible, to make predictions for the course of its development. While this is the result of many causes, de Bot & Larsen-Freeman (, p. ) outline three major reasons: the interconnectedness of the system, its non-linear nature, and the large variability in the initial condition. The interconnectedness makes it impossible to isolate individual factors and the great number of interacting elements only ‘makes it more problematic [ . . . ] to predict how the system will change’ (de Bot & Larsen-Freeman, , p. ). Its non-linearity means that ‘the effect is disproportionate to the cause’ (LarsenFreeman, , p. ), implying that an insignificant change can have a large impact on the system (much like a butterfly effect) but also that a large change can have little to no impact at all. Finally, the huge number of initial conditions and their large variability across different populations and among individuals make the task of choosing the relevant ones a Herculean one. It is our sincere hope that similar studies will soon delve into the L/FL attrition universe. There is considerably less research on L/FL attrition than on L attrition research, but recent developments have seen new projects. We suggest that future work should be guided by three main directions. The first one is to attain a commonly agreed methodological framework for the collection and analysis of empirical data in order to allow comparisons between investigations and a wider evidence base. Second, such investigations should be more thoroughly grounded in linguistic, psycholinguistic, and sociolinguistic theories of bilingual development. Third, studies should avail themselves of methodological developments in on-line and neuroimaging technology (see Part II of the present volume).
. O P V
.................................................................................................................................. The chapters assembled in Part V are an attempt to provide first steps in these directions, introducing both innovative methodological approaches and new theoretical interpretations of the existing data. Chapter focuses on one of the topics which has received most attention in L/FL attrition research: the impact of extralinguistic factors in L/FL attrition. Instead of following the traditional division of personal, external, and linguistic factors (Weltens, ; Schmid & Mehotcheva, ), Mehotcheva & Mytara present a new interpretation claiming that the majority of the factors involved are interrelated and influence one another making clear-cut division between individual variables and their study as separate items impossible. Following Köpke’s proposal of ‘cluster factors’ (Köpke, ), the authors develop the idea of interconnectedness further. They claim that the limited explanatory potential of individual factors in previous research can be ascribed to their correlation and interaction, which makes studying and analysing them in isolation problematic. Among the factors
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
discussed in the chapter are age, context, attitude and motivation, length of exposure/ residence, language contact and use, attained proficiency, incubation period, literacy and education. In Chapter , concerned with syntax and phonology in L attrition, Bardovi-Harlig & Stringer focus on the age factor (see also Bylund, Chapter , this volume). Through a thorough comparison of research on both L and L attrition, the authors provide a fresh look at L attrition research. With respect to phonology, their literature review convincingly demonstrates a strong resilience of phonological features in L attrition. Regarding syntax, however, a more complex picture emerges. In order to explain this, the authors argue that the design of most studies does not allow the disassociation of syntax from the retrieval of lexico-semantic representations, discourse features, and working memory constraints, which may cause apparent loss of syntactic complexity. Given the limited evidence of a critical period these data provide, the authors suggest interpreting age effects as a stabilization of the mental lexicon at around to years. In Chapter , Larson-Hall presents research on lexical L attrition with an approach based on the assumption that lexical items are not isolated elements but integrated parts of an extensive network. A large part of the discussion focuses on the computer modelizations proposed by Meara () in order to simulate vocabulary loss in a bilingual lexical network. Such simulations suggest that lexical loss is much less predictable than has been supposed, and that knowledge of vocabulary is never entirely lost, as has also been shown by studies based on the Savings paradigm. The discussion of data from studies on lexical L attrition together with studies on forgetting provided in this chapters allows the author to adopt a new, more theoretical perspective of lexical L attrition taking into account advances from DST and research on memory. In Chapter , Taura provides a detailed overview of the work carried out on Japanese returnees, a research domain with a long tradition in Japan where the initially negative reception of the ‘foreignness’ of the first returnees was replaced by a more positive attitude in an attempt to enrich the traditionally monocultural and monolingual Japanese society. Three distinctive research periods are outlined and then discussed in more detail. The first period, which began in the s, consisted mainly of descriptive research. The second stage, a decade later, was dedicated to testing various theoretical frameworks and models— that is, Yoshitomi’s () neurobiological and psychological model, the Regression Hypothesis (see above), and the influence of various independent variables such as age, literacy, onset proficiency, and attitude and motivation. The third, current stage sees the introduction of brain imaging techniques such as fMRI (functional Magnetic Resonance Imaging) and fNIRS (functional Near Infrared Spectroscopy) into the study of language acquisition and consequently language attrition. In Chapter , Osterhout and collaborators outline the interest of event-related potentials (ERP) data for research on L/FL attrition and present preliminary results of the first study conducted in the context of L attrition with such a technique. They report on a longitudinal study of both FL acquisition and attrition in Anglophone learners of Finnish. The participants performed a lexical decision task involving violations of vowelharmony, based in Finnish on simple and highly regular rules that have no correspondence in English. As in other studies on language comprehension in L learners, the acquisition data show a progression from N to P over the sessions. Interestingly, this pattern is reversed in the attrition phase after the end of the instruction period providing evidence
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
ö
of the vulnerability of the P effect with such a specific grammatical feature. These data clearly demonstrate the high potential of brain-based research paradigms in L/FL attrition research. Taken together, the chapters of the present section provide a number of stimulating suggestions which should allow L/FL attrition research to enter a new decade and develop research based on a larger variety of methodologies and paradigms on the one hand, with a solid embedding in theories of SLA and bilingualism on the other, as for example through the establishment of links with research on the mental lexicon or language control.
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
......................................................................................................................
/ ......................................................................................................................
.
. I
.................................................................................................................................. T role played by extra-linguistic factors in language attrition has long been recognized by researchers. The proceedings of the first conference on the loss of language skills (Lambert & Freed, ), the first edited volume on language attrition, include three contributions on this topic: Freed’s () introduction, and papers by Berko-Gleason () and Oxford () that pointed out the importance not only of linguistic variables, but also of extra-linguistic factors to the study of language attrition. Freed () identified four major categories of extra-linguistic variables liable to affect the attrition of language skills: personal characteristics, motivation (to acquire, retain, and use another language), learning context (e.g., formal vs. informal), and language use (e.g., frequency, recency). Weltens () was the first to specifically address the importance of extra-linguistic factors in the study of second language (L)/foreign language (FL) attrition, distinguishing between three types of variables related to (a) the process of acquisition, (b) the process of attrition, and (c) personal characteristics. Not surprisingly, given that L/FL attrition is a natural process of L or FL learning, many of the factors he outlined were derived from the study of second language acquisition (SLA) and were deemed important for both acquisition and attrition. SLA research had demonstrated that the outcome of language learning is influenced by a number of factors, with some fostering language learning (e.g., positive attitude and motivation (ATM)) and others impeding it (e.g., age) (see Gardner, , for a detailed discussion of their influence in SLA). In L/FL attrition, investigating the influence of extra-linguistic factors could not only provide the key to predicting attrition and reinforcing language maintenance, but could
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
also—and above all—lead to greater insight into the rules governing L learning and processing in general. Now more than ever, this information is of vital importance, with implications on personal, educational, institutional, and business levels. On a personal level, people invest time, effort, and money in mastering FLs, but have no guarantee that the knowledge they acquire will last. On an educational level, individuals are advised to continue using the language (e.g., reading books, watching TV, listening to the radio), in order to maintain their linguistic ability, but the efficiency of these measures has not yet been fully corroborated by linguistic research. On institutional and business levels, disentangling these factors could affect the L/FL teaching field by allowing more effective curricula and L/FL teaching methods to be devised, and by improving L/FL teacher training. It could also lead to the creation of a whole new area of language learning and teaching for maintainers—people who have studied an L/FL and are determined to consolidate the knowledge they have acquired and keep it functional by following specially designed refresher courses and using materials specifically targeted at maintainers. Since Weltens’ work (), many potentially influential factors have been identified and grouped into distinct categories, namely personal, aptitudinal, attitudinal, social/community factors, and others related to cognitive processes (Weltens & Grendel, ; Hansen, , ; Köpke & Schmid, ; Schmid, , ; Köpke, ; Mehotcheva, ; Taura, ; Bardovi-Harlig & Stringer, ; Wang, ; Xu, ; Schmid & Mehotcheva, ). However, in the current chapter we argue that most of these variables, though not all, are interrelated and interconnected, as illustrated by Figure ., meaning that it is neither realistic nor feasible to examine each one in isolation. Instead, we defend Köpke’s () idea of cluster factors, that is, combinations of interacting factors, where each component carries a different load, defined by the speaker’s individual circumstances and experiences. We also discuss the implications of Dynamic Systems Theory (DST) and the Dynamic Model of Multilingualism (DMM) for the study of extralinguistic factors (see Opitz, Chapter , this volume). In Figure ., there is one variable that is influenced by all the others, i.e., attained proficiency—the peak of language proficiency achieved by an L/FL learner before the period of attrition/disuse begins. Another distinctive element, but this time on account of its isolation, is incubation, namely the time during which the language has not been used (see Mehotcheva & Köpke, Chapter , this volume, for a detailed discussion of incubation). Incubation is the only factor that does not affect attained proficiency (or any other factor), nor is it influenced by other variables. Because of their importance, we discuss these two items at the beginning and end of the discussion of factors. We describe the remaining factors in relation to their presumed importance and the attention they have so far received from researchers. We start our presentation with the usual suspects, focusing on those characteristics that have been most extensively debated, including age (during both acquisition and attrition), ATM (during both acquisition/ learning and attrition), and language contact and use (during and after the acquisition process). We then move on to the great unknown, discussing two variables that are considered to be very important, albeit clearly under-researched: context (i.e., the language acquisition/learning setting) and length of exposure/residence (i.e., the time the attriter spent learning the language in a formal setting or immersed in the target language culture). Finally, in odds and ends, we briefly mention three extra-linguistic determinants we know really very little about: aptitude, level of education, and literacy.
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
ATTITUDE AND MOTIVATION
LANGUAGE CONTACT AND USE
CONTEXT
AGE
LITERACY ATTAINED PROFICIENCY
LENGTH OF EXPOSURE/ RESIDENCE
INCUBATION
APTITUDE
LEVEL OF EDUCATION
.. Interrelations and interconnections among extra-linguistic variables
. F : D S T
.................................................................................................................................. As mentioned above (Section .), we believe a multicomponent view (Köpke, ) to be a more appropriate approach for the study of extra-linguistic factors, albeit not an easy one to follow. This scenario provides for much more variability with regard to the importance
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
and influence of the variables, for depending on the situation, different combinations of factors can lead to the same outcome in the retention/attrition of a language. This may explain why studies comparing seemingly similar factors sometimes reach different conclusions about their influence. The role of extra-linguistic factors in language attrition has already been considered by most of the researchers in this field. However, Köpke () was the first to propose the notion of cluster factors. A cluster factor can be described as a combination of factors that are interrelated and interconnected. Each factor contributes differently to the cluster, depending on the speaker’s personal characteristics, situation, and linguistic experience. For example, in the case of first language (L) attrition, Köpke suggests that age has five different components, including brain plasticity, type of memory involved, language use, cognitive grounding, and motivation. As for L/FL attrition, two potential cluster factors are attained proficiency, which is made up of the other factors included in Figure . (excluding incubation), and contact and use of the language. The latter is dependent upon ATM, context, age, and attained proficiency. In turn, ATM, a complex set of variables on both personal and external levels (discussed in Section ..), is influenced by age, context, and attained proficiency. All these factors are in constant interaction and continuously influence each other, forming a different pattern for each individual. Thus, while the rate and nature of attrition may seem the same on the surface, the underlying factors may differ from one individual to another, shaped by their personal circumstances. One relevant approach that seems to account not only for the interconnections and interactions between the different factors, but also for the unpredictability and variability of the outcome of such a system, is DST (Briggs & Peat, ; Van Geert, ), which de Bot () discussed in relation to language, language learning, and attrition. According to this theory, a person’s language(s) represent(s) a complex, Dynamic System (DS) whose different variables or sets of variables change and interact over time. Although a detailed account of the properties of a DS is beyond the scope of the present contribution (but see Opitz, Chapter , this volume, for more details), we briefly summarize some of its main characteristics. First, a DS is completely interconnected, consisting of other, smaller nested subsystems, each of which is itself multilayered. All these levels interact with each other and with the surrounding environment. They have to constantly adjust to changes in internal conditions and the external environment. One very important aspect of a DS is its non-linear development, for whereas one small interaction may not leave any traces in the system, another equally small interaction may have a very significant effect (de Bot, ). Similarly, a major interaction may have no impact whatsoever, whereas a much smaller and seemingly meaningless interaction may profoundly affect the system. This means that when—and if—an extra-linguistic variable changes (e.g., motivation, aptitude, contact with a language), this modification may have an impact on both the remaining variables and the entire system itself. This approach allows for much greater variability in the outcome, where apparently similar conditions (e.g., time spent learning the language, length of residence, length of attrition, attitude) may result in different rates and patterns of attrition, as empirical findings have shown. The DS approach allows the effect of extra-linguistic variables to be studied from an interconnected perspective rather than an isolationist one.
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
. A
.................................................................................................................................. As mentioned earlier, attained proficiency is one of the most prominent factors, with several studies establishing a negative correlation with attrition. Attained proficiency, sometimes also referred to as level of achievement (Schmid, ) or peak level of attainment (BardoviHarlig & Stringer, ), is defined as the highest level of proficiency attained by a speaker before the onset of language attrition. Attained proficiency has consistently been highlighted as the most determining factor in language attrition/retention, and as shown in Figure ., it is linked to almost all the other factors. Attained proficiency is especially relevant to L/FL research, which often deals with populations that have acquired the language to different levels (Reetz-Kurashige, ; Murtagh & Van der Slick, ) which are not necessarily even close to native-like or high proficiency, thus making it very challenging to compare or categorize participants. This is very different from L attrition studies, where the language is generally assumed to have been fully mastered. Thus, as Schmid & Mehotcheva () argue, controlling for attained proficiency provides a means of comparing populations that share similar characteristics in terms of age, length of exposure, context, and so on, but otherwise vary widely in the degree to which they have mastered the attriting language. One of the first to point out a reverse connection between attained proficiency and attrition was Godsall-Myers (), who explored the attrition of German among American students. When Bahrick (a,b) studied the retention of Spanish in adults after fifty years of non-use, he also found a reverse relationship (for a detailed overview of Bahrick’s study, see Mehotcheva & Köpke, Chapter , this volume). This author reported that while attrition affected all populations during the first years of lack of contact, speakers tended to lose equal amounts of knowledge, meaning that high-proficiency speakers were left with a higher proportion of knowledge. Another notable finding was that a large portion of knowledge survived for as long as fifty years. Bahrick explained this finding using the concept of permastore content, which describes a special part of knowledge containing items with a lifespan of over twenty-five years. The permastore idea was later referred to as a critical threshold of L knowledge that made items ‘sharply resistant to forgetting’ (Neisser, , p. ) and in terms of a ‘critical mass of language’ (Pan & Berko Gleason, , p. ). Several studies have highlighted a relationship between higher attained proficiency and better language retention across different learning settings (i.e., school/university-acquired FLs or naturalistically acquired Ls, as in the case of returnees or study-abroad/exchange students). Studies involving FLs instructed in formal settings include Harley () on the attrition of school-acquired French in Canadian L English speakers, Moorcroft & Gardner () on the attrition of school-acquired French in English L Canadian high-school students during the summer vacation, and Xu () on the comparative attrition of English in two different contexts: Dutch L university students in the Netherlands and Chinese L university students in China (see Mehotcheva & Köpke, Chapter , this volume, for more details). A number of researchers have also focused on Ls acquired in naturalistic settings. Nagasawa (), for example, investigated the attrition and retention of Japanese in L English graduate students who returned to their L linguistic environment but subsequently resumed language instruction in the attriting language. Reetz-Kurashige
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
() and Taura () both examined changes in English L in Japanese returnees. More specifically, Reetz-Kurashige () used narrative data (storytelling tasks) elicited from Japanese L elementary and secondary school returnees to study changes in English verb usage over time, while Taura () focused on the attrition and retention of English in Japanese L returnees (discussed at length in Mehotcheva & Köpke and in Taura, Chapters and , respectively, in this volume). Furthermore, Weltens () and Weltens et al. () explored the retention of school-acquired French in Dutch L students. Mehotcheva’s () study concerned university-acquired Spanish that was later used naturalistically by study-abroad Dutch and German L university students. Finally, it is interesting to note the relationship between ATM, attained proficiency, and language contact and use. The connections between these three factors are bidirectional, and they constantly influence and feed on each other. High motivation during the acquisition process may prompt learners to seek more opportunities to use the language, and lead (either on its own or in combination with language contact and use) to higher attained proficiency. Language contact and use may affect attained proficiency, and hence a person’s motivation to either continue or terminate all contact with and use of the language. In turn, attained proficiency may affect a person’s linguistic confidence, having a decisive influence on that person’s motivation or decision to seek contact with and opportunities for using the language.
. T : , , ,
.................................................................................................................................. Age, ATM, and language contact and use are among the most discussed and studied variables regarding not only L/FL attrition, but L attrition, too (see Bylund, Riehl, and Schmid & Cherciov, Chapters , , and , respectively, in this volume). They are also central to two of the theoretical frameworks specifically targeting language attrition—the Activation Threshold Hypothesis (ATH; Paradis, , , ) and the DMM (Herdina & Jessner, ), both discussed below. Furthermore, although the importance of these variables seems indisputable, empirical findings in the field of L/FL attrition are not as unanimous when it comes to the precise nature of their role in attrition/maintenance.
.. Age Age is one of the core factors in L attrition, with some studies pointing to the possibility of total or partial erasure of a mother tongue (for more details, see Bylund, Chapter , this volume) and while it does not have quite such a dramatic impact on L/FL attrition, it is important on several levels. First, age at the time of acquisition/contact with the language may have implications in the way that language is perceived and acquired. For example, while they are abroad, child returnees tend to be immersed more in the target culture, as
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
they are usually educated in kindergarten and/or school in the target language in the same way that native children are. These children often do not have any means other than the target language of communicating with their caregivers/teachers/classmates/friends, possibly contributing to their motivation and need to acquire the native language. Adult learners on the other hand, especially adult immigrants, very often tend to cluster together with fellow compatriots, and either study the language in a formal setting (e.g., in classes organized by the authorities) or simply learn by practising it with native speakers rather than being taught it, though not with the same intensity and exposure as children. Thus, age influences both context of acquisition and use, and ATM. Second, as in L attrition studies, age at onset of L/FL attrition is important, especially given that even a native language can be erased, if contact with the language is lost before puberty (Nicoladis & Grabois, ; Pallier et al., ; Ventureyra, ). Schmid () argues that there is a saturation threshold for L speakers, beyond which loss of contact with the native language no longer influences its maintenance, as the relevant language knowledge is already entrenched in the brain. Researchers have suggested that this cut-off age is to years (i.e., around puberty), but interestingly this period also coincides with the automatization of literacy, another factor that is thought to be important for language resistance after puberty (Hansen, ; Köpke, ). The implications for L/FL learners could be that older children who have been exposed longer to the language and may have reached that saturation threshold, or children who have already developed literacy, may exhibit better L/FL retention than children who have not reached this turning point. However, research comparing child returnees of different ages has so far yielded divergent results. Cohen (), who studied the attrition of L Portuguese in two English Hebrewspeaking children aged and years, registered more attrition in the younger participant, who produced a small number of different words and fewer and shorter T-units. Olshtain () explored the retention of L English in L Hebrew speakers using a combination of written and oral tests, and comparing two different age groups: to and to years. Like Cohen, she found that the younger participants were more affected by attrition, but she also noted that these participants were the ones who were least motivated to maintain their L upon return to their homeland, and had not yet achieved literacy. Finally, Tomiyama (; also discussed in Taura, Chapter , this volume) studied the attrition of L English in two L Japanese returnees aged and years over a period of thirty-one months. She noted that despite the age difference, the participants’ initial proficiency was the same and they were both literate. Whereas there were no differences in performance at the outset, after the second year, grammatical accuracy declined in the younger participant, which the author attributed to cognitive maturation. By contrast, in a study by Kuhberg (), who explored German L attrition in two Turkish girls aged and years, attrition was observed after six months, but rather surprisingly was more pronounced for the older participant than for the younger one. Similarly, when Reetz-Kurashige () investigated the retention of L English in Japanese L returnees aged . to . years, she found no effect of age and concluded that strong retention depended on a combination of factors (i.e., age, length of residence, and attained proficiency), rather than just one isolated factor. Finally, age should be considered as a biological factor in ageing individuals, especially in the light of the work by Schmid & Keijzer (), who studied L attrition and reversion among ageing Dutch and German migrants in Canada and the Netherlands. The authors found an age effect both in the control group (non-significant effect) and in the
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
experimental groups, registering a peak in language attrition between and years. While this effect was not observed among the oldest participants, in whom a revival in linguistic abilities was found, results showed that, as Schmid & Dusseldorp argued (), biological age is a relevant factor as it can affect cognitive processes, and consequently performance on certain tasks (see also Higby et al., Chapter , this volume).
.. Attitude and motivation Another factor that could be linked to a number of other variables is ATM. This has repeatedly been put forward as an important aspect of language attrition, not least because of its importance in SLA (Dörnyei, , ). ATM is significant not only during the acquisition/learning stage (discussed in greater detail below), but also during the period of active use for participants in immersion, exchange or stay-abroad programmes, as well as during the incubation period. Additionally, ATM may influence the length of exposure to the language and/or length of residence in the target-language culture, the amount of language contact and use during both acquisition/learning and incubation, and attained proficiency. To make things even more complicated, ATM encompasses a complex set of variables on both personal and environmental levels. On the personal level, factors include the speaker’s attitude towards the language, its culture and its speakers, the inner motivation to learn that language (i.e., integrative or instrumental), and the attitude towards the language learning situation and its participants (i.e., classroom environment, students, teacher) (Gardner & Lambert, ; Gardner, , ; Gardner, Lalonde, Moorcroft, & Evers, ; Ushioda & Dörnyei, ). External factors relate to variables such as the prestige of the language in question and the context of its acquisition and attrition (e.g., community, learning environment, teacher, fellow students). In the following paragraph, we outline the implications of both the personal factors and the external ones, which are described in greater detail in Section .. (on context). In L development, language acquisition occurs naturally, regardless of ATM, as it is driven by the natural need to communicate. This may explain why L attrition studies have failed to find any evidence of a link between ATM and attrition (Yağmur, ; Hulsen, ; Schmid & Dusseldorp, ; Cherciov, ). In SLA, on the other hand, personal attitudinal factors have long been linked to proficiency level, with positive ATM leading to better results (Schumann, , ; Masgoret & Gardner, ). Gardner (, p. ) therefore hypothesized that these factors are also relevant to the attrition (or maintenance) of linguistic knowledge once the active learning process is over. However, no conclusive evidence for the effect of personal ATM on adult L/FL attrition has been reported. To the authors’ knowledge, only one study has so far found a positive effect: when Wang () investigated the attrition of English in Chinese English college graduates, she concluded that ‘the participants’ good performances on the language test can be accounted for by their favourable attitudes and motivations at any level’. Interestingly, Xu’s study () found that attitude towards English positively influenced the results of the Chinese sample, but the Dutch participants did not seem to benefit from this effect. Concerning the Chinese participants, Xu reported a strong effect of attitudes on their English proficiency, noting the bidirectionality characterizing these two variables, meaning that it could either be the students’ attitudes that positively influenced their proficiency or vice versa. For the Dutch
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
participants, attitudes did not seem to be a significant factor, either during acquisition or attrition. Xu argued that the measurement instruments used in the study may have failed to pick up the possible influence of attitudes on proficiency. Whereas Xu’s data were collected with an adaptation of the attitude questionnaire of the Assessment of Pupils’ Skills in English in Eight European Countries (Bonnet, ), Wang used an adapted version of Gardner’s () Attitude and Motivation Battery Test. Another researcher who applied this battery test to the study of adult FL attrition was Mehotcheva (). She obtained mixed results, with positive attitude and integrative orientation leading to better results on some of the measures (naming latencies and accuracy of responses in a picture-naming task; lexical diversity, filled pauses, false starts, and use of FL words in a one-to-one interview). However, there were no clear-cut patterns that could be generalized. As Schmid & Mehotcheva argued (), researchers have previously emphasized that ATM variables ‘develop dynamically and are subject to change’ (Dörnyei & Ottó, ; Nikitina & Furuoka, ; Schmid, ; Waninge et al., ). Since attitudinal factors cover such a wide spectrum of variables, it might very well be the case that some, if not all, of these factors are subject to change over time, depending on life circumstances, and are thus not truthfully reflected during the interview/testing, which usually measures one specific point in time. It is interesting to note that in the ATH, based on the notion of an accessibility level (or threshold) of items and/or linguistic systems (Paradis, , , ) and applicable to L/FL attrition (for a detailed overview, see Köpke & Keijzer, Chapter , this volume), ATM variables are considered to be of special importance, despite being only indirectly related. According to this hypothesis, the accessibility of an item/language depends on its activation threshold: the more an item/language is used, the lower its activation threshold and the easier it is to access, and vice versa. The role of ATM is an indirect one, stemming from the opportunities to practise the language that a positive attitude and high motivation may encourage speakers to actively seek out. For instance, opportunities to engage with the language, such as reading books and magazines, watching TV/movies, listening to the radio, creating a social network with native speakers, lead to increased contact with and use of that language. Social networks have also been found to be important in the maintenance of L in immigrants (Hulsen, ; Hulsen et al., ; Smyth, ; Stoessel, ). Finally, it should be noted that ATM could be influenced by aptitude, as we discuss in Section .. below.
.. Language contact and use Language contact and use has repeatedly been identified as one of the most significant and predictive factors for linguistic maintenance in L attrition contexts (Köpke, ; Schmid, ). Psycholinguistic research on attrition has often hinted at the importance of language contact and use: according to the above-mentioned ATH, rehearsal is crucial for ensuring that items and/or languages remain accessible, and attrition is defined as the result of a ‘long-term lack of stimulation’ (Paradis, , , p. ). Similarly, the DMM (Herdina & Jessner, ) emphasizes the significance of language use by discussing the language maintenance effort (LME) that bilinguals/multilinguals need to exert to maintain their languages. Lack of LME is considered central to the deterioration of linguistic skills. However, the results of research on both L and L attrition are less clear
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
and straightforward regarding the importance of language contact and use. More specifically, with respect to L attrition, Köpke & Schmid () point out that most studies have so far failed to yield findings corroborating the presence of a correlation between language contact and use and attrition, mainly owing to the lack of a suitable methodology to investigate their relationship. Moreover, Schmid () suggests that researchers should consider the impact not only of language mode (bilingual/monolingual) as per Grosjean’s distinction (), but also of language context (e.g., informal/formal, use with family/ friends/colleagues) when attempting to evaluate the role of a given factor, as this could provide a more comprehensive understanding of the way these variables interact. As far as L/FL attrition is concerned, most of the studies that have explored the importance of language contact and use (Bahrick, a, b; Murtagh, ; Taura, ; Mehotcheva, ; Xu, ; Alharthi & Al Fraidan, ) have found no support for its assumed facilitating role. The only exceptions are Murtagh (), who found that reading during incubation resulted in better general proficiency, and Alharthi & Al Fraidan (, p. ), who found that Internet use was closely related to ‘better retention of productive vocabulary’. These mixed results may stem from differences in the measurement of language contact and use. First, as Schmid & Mehotcheva () noted, there is the challenge of measuring the amount of use in a valid and realistic way. Not only may quantifying the factor (i.e., using minutes/hours per day/week) prove problematic, but participants’ reports may also be easily biased by their own personal perceptions of time. Second, Köpke () argued that when talking about examining language contact and use, the focus is almost always on quantity, rather than quality, even though the latter is equally influential. Another problem is how to establish which kind of subset of contact and use is most influential: active and productive use of the language or passive and receptive input? Studies generally fail to distinguish clearly between the two (Köpke, ), and as Schmid & Dusseldorp () observed, many make no distinction whatsoever. Moreover, we may have unrealistic expectations of the effect of language contact. As Weltens () pointed out, although in principle any contact with the language implies some form of relearning and strengthening of the information, this is not always the case. For instance, if a person watches films in an FL, it does not necessarily mean that this kind of contact with the language will automatically be transformed into knowledge. Finally, it is not unreasonable to expect a person with high ATM and who has achieved high proficiency (as opposed to someone who is less positively disposed towards the language or does not feel confident about using it) to be more inclined to conscientiously look for opportunities to further practise the language (e.g., browse the Internet, find reading materials), even if the language is not readily available in the immediate surroundings.
. T : , /
.................................................................................................................................. Two factors that could play a key role in explaining language attrition, but which have generally been neglected thus far, are acquisition/attrition context and length of residence/ exposure. While these variables seem to be of great importance, insofar as they are linked to
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
a number of other factors, especially attained proficiency, which has been found to have a considerable influence on the attrition process, their impact has seldom been studied. This is often attributed to problems quantifying and efficiently measuring these factors owing to their complexity. More specifically, qualitative analysis may be a more appropriate method for studying context, whereas assessment tools may be more suitable for investigating length of residence/exposure, as these would provide a more detailed insight into its involvement in attrition.
.. Context Research has shown that cultural and social context and life experience can have a dramatic impact on L attrition and retention among international adoptees (Pallier, ), political immigrants (Schmid, ), and ideological or political returnees (Ben-Rafael & Schmid, ). In these situations, a negative association with the L and/or cultural and societal pressure may lead to considerable L attrition, especially among young adoptees (Nicoladis & Grabois, ; Ventureyra & Pallier, ; but see Pierce et al., Chapter , this volume, for the most recent developments in this field). Regarding L/FL attrition, both the context in which attrition takes place and the context in which the language was learned/acquired are important. In both cases, there is the influence of the immediate environment and society and the general attitude towards the language in question. The status of the target language in the community where it is learned/acquired may influence the learner’s ATM, language contact and use, and ultimately his/her attained proficiency. A high-prestige language that is seen as an asset may boost all three of these factors, whereas a language that is seen as strange or unwanted may have a negative impact on its learners. There is also the case of a formally frowned-upon language. In the ex-Soviet bloc, for instance, English was seen as a possible way out and held out the promise of a better future, such that citizens would go to great lengths and even run considerable risk to master it. Opportunities for practice and contact with the language may also depend not only on the prestige of the language itself, but also on the socioeconomic status of the learner/attriter, as a favourable socioeconomic status may be instrumental in maintaining the language because the person can afford to have more contact with the language. In addition, the learning context can be influenced by the particular learning situation, the teaching methods used, and the identity of the protagonists (e.g., teacher and fellow classmates), as demonstrated by Gardner (). To the authors’ knowledge there have not been any recent studies on this topic, but in , Murtagh identified instructional environment as one of the factors related to the better retention of Irish among high-school graduates, with participants in immersion schools obtaining better results than non-immersion participants. Regarding context during the incubation period, the status of the attriting language within the community may be especially important. Depending on historical, geographical, and geopolitical factors, a language may be assigned completely different statuses across different countries. As with immigrants, it is not unreasonable to expect a language that is highly valued and respected in a given society to be better retained by L/FL attriters than by attriters in a country where the same language is less prestigious. Additionally, in countries with varying levels of social standing, an individual’s higher socioeconomic status
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
may not only provide more opportunities for direct contact with the language during the learning phase (thus influencing the student’s desire and motivation to learn the language), but may also mean more opportunities for contact with the language in general and its more frequent use during the incubation period (Xu, ). Although Xu reported more contact with English during the attritional period for participants with high social and economic status in the Chinese sample, neither their attitudes nor their amount of contact had any impact on their maintenance of the language. By contrast, Wang () reported that positive attitudes were related to better retention of the language, while poor performance could be explained by their ‘less favourable or negative attitudes’ (Wang, , p. ).
.. Length of exposure/residence Length of exposure/residence is the time that a speaker spent being actively involved with the language, studying it in a formal setting (e.g., in class, private lessons) for schoolacquired languages and being exposed to it and using it in a country where the language is spoken, in the case of stay-abroad/returnee scenarios. It has been suggested that length of exposure/residence is even more important than attained proficiency (Hansen, ). In a study of the attrition of Japanese by returning missionaries, Hansen found that length of residence was correlated with better performance on the negation task that was used in the study, with even six months making a difference, as participants with thirty-six months of exposure performed better than those with thirty months. Mehotcheva () reported that longer study abroad was correlated with only two elements (i.e., fewer self-corrections in free production and higher percentage of correct responses in a picture-naming task), and did not affect the majority of variables that were investigated. However, as Schmid (), and Bardovi-Harlig & Stringer () observed, and as we show in Figure ., it is not unreasonable to assume that longer exposure to the language and/or longer residence in the country positively influences attained proficiency. Research on studying abroad has found evidence that longer periods abroad lead to greater linguistic gains (Llanes & Muños, ), and while this may not be always the case, as stay-abroad and exchange students often tend to cluster together by nationality and do not always take full advantage of the immersion in the target language and culture, length of residence/exposure is an essential factor that should be taken into consideration. Length of exposure/residence may also be influenced by ATM, especially in the case of adult learners. People with a positive attitude towards a given language presumably look for more opportunities to come into contact with the language and its culture. They are also more willing to dedicate longer periods of their time to mastering the language in a classroom than someone who does not perceive the language and its surrounding culture as something interesting and fascinating.
. O : , ,
.................................................................................................................................. Finally, there are a number of disparate factors that researchers have identified as important, but for which there is a dearth of empirical data.
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
.. Aptitude Gardner () was one of the first to suggest that aptitude, namely a person’s innate ability to learn and process a language, can be of great importance to L/FL attrition. Aptitude has been found to be decisive in both formal and informal language learning and acquisition (Bylund et al., ). Greater language aptitude is generally related to higher language proficiency (Köpke, ), and individuals with higher language aptitude might be expected to retain their L/FL better. Similarly, and as mentioned earlier in Section ., a person with a higher aptitude might be expected to be more motivated to continue studying the language than somebody who is struggling with the whole process. However, to our knowledge, no study has explicitly addressed the issue of aptitude in the context of L/FL attrition.
.. Level of education While no study of L/FL attrition has specifically explored its role, research on L attrition has suggested that level of education is a valid factor that influences the attritional process (Jaspaert & Kroon, ) by promoting retention. Another valid reason to control for the level of education is that it may also cause results to be misinterpreted, by affecting performances on certain tasks rather than the attritional process itself. Multiple-choice tests, for example, are typical of university education, and people with a lower educational level may find them quite challenging, as has been demonstrated by Yağmur ().
.. Literacy Köpke () outlines several sociolinguistic and psycholinguistic arguments as to why literacy might be a factor worth considering in relation to its possible positive effect on language retention. As she points out, it not only enables language maintenance and contact through reading development and increased access written input, but also permits a more rapid establishment of the language in human memory through the creation of orthographic coding. Support for the relevance of literacy comes from studies by Olshtain (, ) and Reetz-Kurashige (), who both report better retention by literate participants. However, as Olshtain (); Hansen (); and Köpke () point out, literacy in children is linked to age and level of proficiency, sometimes making it impossible to isolate this factor. Regarding the effect of literacy on attrition in adults, Hansen () reports on three different studies (Hansen & Newbold, ; Hansen & Chantril, ; Shewell & Hansen, ) that successfully isolated it as a factor for better language retention. Nonetheless, further analysis carried out in the study by Shewell and Hansen () revealed that both higher literacy and better linguistic maintenance were explained more by higher ATM.
. I
.................................................................................................................................. Finally, incubation (Gardner, ), also sometimes referred to as the period of non-use/ disuse (de Bot, ) or length of attrition (Mehotcheva, ), is the amount of time that
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
has elapsed between the cessation of active contact with and use of the L/FL and the retention assessment. It is interesting to note that while all the other factors presented in Figure . were interconnected and influenced each other, incubation was entirely isolated, with no connections to any of the other factors. It was the only factor that neither influenced nor was influenced by the others. This could explain why it is the only factor that has been successfully isolated and studied as a separate variable in studies by Bahrick (a); Taura (); and Mehotcheva (), with Taura describing it simply as a ‘function of incubation’ (, p. ). Although a positive correlation between incubation and attrition (i.e., more attrition for longer periods of non-use) was expected in these studies, results showed that while it was intuitively assumed to be important, incubation alone was not sufficient to explain attrition.
. D
.................................................................................................................................. The principal aim of the present contribution was to draw up an inventory of the most important extra-linguistic factors that seem to either hinder or enhance the process of L/FL attrition. Given the findings of the various studies outlined above, we believe that the greatest challenge for L/FL attrition research is not only to specify the nature and significance of each factor that potentially influences the outcome of the process, but also to provide possible solutions to the difficulties that arise when attempting to examine them in detail. One of the problems encountered in many studies aspiring to explore the impact of extra-linguistic factors is the inability to adequately measure them. This is due in part to the intrinsic nature of some of the factors and the impossibility of quantifying them, given that many of them (e.g., time spent using a different language in a particular context, time spent talking to friends, reading books in the target language) rely on the speaker’s personal perceptions. Another problem is the changing nature of some of the factors. ATM, for instance, can vary across a person’s lifespan, but when researchers try to assess it in attriters, they simply report ATM at that particular point in time, which is not necessarily the same as when the person was learning/acquiring the language and/or was immersed in the target language culture. Yet another challenge is the inability to isolate factors and study them separately, owing to their interconnected and interrelated nature. As we discussed earlier, the only factor that can be effectively isolated and studied on its own is incubation, which is the only one that does not affect or is not affected by the other factors. However, there is no empirical evidence in favour of its effect, one possible explanation for this being its isolated nature and lack of connection to other factors. It is attained proficiency, a factor influenced by all the other factors that has come out as the most influential one in a number of studies (Godsall-Myers, ; Bahrick, a, b; Moorcroft & Gardner, ; Weltens, ; Weltens et al., ; Harley, ; Nagasawa, ; Reetz-Kurashige, ; Taura, ; Mehotcheva, ; Xu, ). Language contact and use and ATM are not only the most influential factors after attained proficiency, with a number of links to other variables, but are also some of the most frequently studied. This supports the idea that extra-linguistic factors do not function in isolation, but rather in clusters (Köpke, ) or groups of factors that constantly interact with and influence each other. It is these clusters
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
and combinations, and the relations between the extra-linguistic factors within them, that we should be aiming to examine. In order to better comprehend which factors or combinations of factors have an impact on L/FL attrition and how their interconnectedness influences this process, future research in the field should consider assessment methods that provide a more extensive understanding of the different variables that come into play, as well as a more precise identification of their characteristics and the consequences they have for attrition. We believe that adopting a DS perspective would help to explain the nature of the extralinguistic factors, which are characterized by dynamic relations, interconnectedness, constant change, and reliance on external resources. According to de Bot (), a dynamic perspective could also explain the considerable individual variability observed under the same attritional conditions, the different rates of attrition, and/or differences in the nature of the attrition (lexical vs. grammatical). Furthermore, undertaking longitudinal studies to track learners/attriters over long periods of time using periodic assessments would make it possible to examine the correlations among the various factors in greater detail, although we acknowledge that this kind of repeated data collection is not always easy or even achievable. As Schmid & Mehotcheva () postulate, two major issues that need to be taken into consideration with respect to the design of longitudinal studies are the selection of appropriate individuals who are willing to participate in long-term research and the effect that this kind of design has on the research itself, mostly due to the retesting carried out during the different phases of the study. Finally, it is crucial to identify the different types of populations that are investigated, as their particular features may affect the L/FL attrition and retention processes to a considerable extent and in a variety of ways. For instance, college or university students who have received formal instruction differ greatly from returnees or even from students who first learned the target language in formal settings but were subsequently immersed into the host country for some period of time.
A We are grateful to Barbara Köpke for her unfailing support throughout the writing of this text. We are also indebted to the anonymous reviewers for their useful and down to the point suggestions. The remaining errors are entirely our own.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
......................................................................................................................
Modularity and resilience ......................................................................................................................
-
. I
.................................................................................................................................. I previous work, we demonstrated that most evidence in support of language loss is fundamentally lexical in nature, on a broad conception of the mental lexicon as containing items both below the word level, such as affixes, and above the word level, such as idioms and conventional expressions (Bardovi-Harlig & Stringer, ). On the assumption of Jackendoff ’s (, ) theory of Representational Modularity, open- and closed-class lexical items can be understood not as unitary entities but as links between syntactic, phonological, and semantic representations housed in distinct computational modules. Thus one possible understanding of the widely attested phenomenon of lexical attrition is that it involves not a fading away of individual items but a breakdown of correspondences between representations in a lexical network. This leaves open the possibility that certain modular aspects of syntax and phonology may be more resilient to attrition, as is occasionally hypothesized in the literature (e.g., Montrul, , p. ; Schmid, a, pp. , –). In this chapter, we ask to what degree syntax and phonology appear to be impervious to second language (L) attrition and examine the related question of whether there is a critical period for either acquisition or attrition in these domains. A second assumption underlying the current inquiry is that, in line with Schmid & Köpke (a), we believe that language transfer, or cross-linguistic influence (CLI), works both ways: just as first language (L) knowledge affects the L, given an appropriate level of activation, the L in turn affects L knowledge. This appears to be true not only in cases where the L becomes the dominant language, but in all cases of functional bilingualism, as an inherent characteristic of the bilingual condition (see also Kecskes & Papp, ; Cook, a; Tsimpli et al., ). Moreover, despite the general age effects attested in both L acquisition (e.g., Johnson & Newport, ) and L attrition (e.g., Bylund, b), the effects of CLI can be attested irrespective of any purported critical period, as shown in
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
studies of phonological perception (e.g., Bosch & Sebastián-Gallés, ) and syntax (e.g., Unsworth, ). This lends credence to the position that CLI in bilingualism may be characterized irrespective of the age of acquisition and the L/L distinction. Therefore, we draw on both the more comprehensive L attrition literature as well as the more sporadic research on child and adult L attrition in order to shed light on this L investigation. Regarding the possibility of a critical period relevant to both L acquisition and L/L attrition, we take as our point of departure Montrul’s () suggestion that the crucial juncture in development might be around to years old. Montrul illustrates this graphically as the intersection of a downward line showing decline in target-like L acquisition, and an upward line showing L retention (and presumably child L retention) in cases of a shift to a new dominant language.1 She describes this intersection as ‘the approximate age at which, hypothetically, the linguistic capacity for language, or the L in monolingual acquisition, is about to reach, or has reached, its mature and steady state’ (p. ), and suggests that this crystallization of the L may result in a subsequent disabling of the capacity for L acquisition, as well as marking the end of the period of vulnerability to L loss. In the next section, we briefly examine evidence for and against critical periods in L acquisition and in L/L attrition. We then evaluate findings on the proposed resilience of phonology and syntax in cases of L attrition. This serves as a foundation for our examination of L attrition studies. To anticipate our conclusions: we find that the question of whether grammar is resilient to attrition following a critical period must be asked separately of phonology and syntax. Regarding phonology, there are indeed cases of amazing resilience that flatly contradict the idea of a critical period for attrition. However, both L and L phonological systems are subject to CLI in cases of cross-linguistic similarity. As for syntax, it proves more difficult to argue for resilience in the absence of CLI, as neither comprehension nor production of syntax can be examined independently of lexical retrieval and the real time constraints of working memory. Our pursuit of these questions leads us to suspect that something may indeed occur around to years old involving nonmodular knowledge, with dramatic effects on the language system, namely network stabilization of the mental lexicon.
. C
.................................................................................................................................. Montrul’s () suggestion of a common developmental phenomenon in middle childhood changing the course of both L acquisition and L/L attrition emerges from independent observations in both domains of research. On one hand, several L acquisition researchers have proposed that convergence on native-like knowledge of the L is possible only until age to , after which there is an inexorable decline in potential (Coppieters, ; Johnson & Newport, ). On the other hand, studies of L attrition in adoptees have 1 The second, solid line in Montrul’s (, p. ) graph is labelled in the original as L loss; however, we interpret this as referring to L retention, as the y-axis measures proficiency.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
-
shown the possibility of complete loss of ability to process the L if attrition sets in before precisely this same age-range (Nicoladis & Grabois, ; Pallier et al., ), even though complete convergence on the L is not guaranteed (Norrman et al., ; see also Pierce et al., Chapter , this volume). While research on L attrition before the proposed critical period is minimal, one of the authors also has first-hand anecdotal evidence for the extraordinary effects of early L attrition. David Stringer’s eldest daughter, Tamsin, was Japanese-dominant at age , after four years of daily immersion in Japanese-only environments outside the home in a rural Japanese community. Japanese was the only language of communication with peers, and the preferred language of play with siblings and of self-talk; school teachers considered her Japanese to be native-like. Yet after just three months of complete absence of input following her arrival in the US (a linguistic transition similar to that experienced by adoptees), syntactic production started to breakdown, and after six months, there was no comprehension of even simple sentences. To all appearances, an entire dominant language had disappeared at age , in just six months. Montrul’s () attempt to link these patterns in acquisition and attrition is conceptually attractive. Perhaps there is a critical period when L acquisition comes easily but during which reduced L or L input can be catastrophic, and after which acquisition of further languages is compromised (Montrul, , pp. –). However, this conceptually neat scenario involves considerable simplification and is only plausible if we consider language holistically. It is worth pausing to consider separately patterns of L acquisition of phonology and syntax. No phonologist has ever suggested that a critical period for L phonological acquisition closes at to years old. Several researchers have argued that capacity for phonological acquisition remains the same in adulthood, and that difficulties arise only in cases of crosslinguistic similarity (e.g., Flege, ; Iverson et al., ). Native-like L phonology has been reported for both aspects of perception (e.g., Werker & Tees, ; Darcy et al., ) and production (e.g., Bongaerts, ; Flege et al., ). Moreover, CLI effects in phonology have been found well in advance of any suggested critical period: there are unambiguous interference effects in simultaneous bilinguals from at least months old (Bosch & Sebastián-Gallés, ). As for syntax, researchers have variously suggested that acquisition becomes compromised at around (Meisel, ) to years old (Johnson & Newport, ). However, the evidence for a critical period in syntax is somewhat thin. When the data from behavioural studies indicating dramatic decline are disaggregated, age effects tend to disappear (Birdsong & Molis, ); conversely, other studies show native-like attainment for some learners (e.g., White & Genesee, ; Montrul & Slabakova, ). Moreover, there is no actual evidence from studies of brain imaging, cognitive ageing, brain volume, or dopamine systems that the brain is subject to a critical period for language acquisition (see Birdsong, , for a review). Rather than there being a biological critical period for syntax, it seems more plausible to consider age in L acquisition as a ‘macro-variable’ (Montrul, , p. ), with wellrecognized general age effects due to other confounding factors, such as quantity and quality of input and frequency of use of the L and L. When Unsworth () carefully controlled for input factors in groups of children acquiring gender agreement in Dutch, differences between monolinguals, simultaneous bilinguals, early successive bilinguals (initial exposure to years old), and child L learners (initial exposure to years
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
old) effectively disappeared. Frequency of use can affect lexical retrieval and processing efficiency. When Hopp () examined knowledge of case, gender, and subject–verb agreement in L German by advanced L speakers of Dutch, English, and Russian, not only did he find native-like L performance in an offline task, but he also found that increasing the speed of an online task for L German native speakers (using speeded grammaticality judgements) reduced their processing efficiency such that there were no longer significant differences between L and L participants. Such findings cast serious doubt on the existence of a critical period for underlying knowledge of syntax.
. P
.................................................................................................................................. In cases of L/L attrition, it does seem clear that there is a dramatic crash in lexical retrieval and syntactic processing ability (in terms of both production and comprehension) if input is ended abruptly sometime before the age of to years (Ammerlaan, ; Pallier et al., ; Montrul, ; Bylund, b). However, it is not clear how the lexicon, phonology, or syntax differentially affect this breakdown, how much is due to disuse rather than CLI, and to what degree lack of real-time processing in test conditions is indicative of a lack of dormant grammatical knowledge. In contrast with this general picture, several studies have shown that aspects of phonology may be surprisingly robust during periods of childhood attrition. Oh et al. () tested for remnants of phonological knowledge in twelve Korean adoptees in Minnesota enrolled in university-level beginner Korean language classes. Five had no Korean input at all between the age of adoption ( s) haemodynamic functions inherent in most functional Magnetic Resonance Imaging (fMRI) research. Furthermore, ERPs are uniquely sensitive to syntactic and semantic aspects of language processing. In many native (L) speakers, semantic anomalies (e.g., The cat will bake the food . . . ) elicit an increase in the amplitude of the N component, a negative deflection that peaks at about ms after onset of the word (the N effect; Kutas & Hillyard, ). N amplitude is a function of a number of lexical properties, including lexicality (with pronounceable nonwords producing the largest Ns), normative word frequency, and the ‘semantic fit’ between the word’s meaning and preceding context (how well the word’s meaning coheres with preceding context; Kutas & Hillyard, ; Kutas & Federmeier, ). By contrast, a variety of syntactic anomalies (e.g., The cat will eating the food . . . ) elicit a large positive wave that starts at about ms and persists for about half a second (the P effect; Osterhout & Holcomb, ). Although much remains unknown about these brain responses, there is compelling evidence that the P is distinct from the P ERP response elicited by statistically unlikely or otherwise unexpected events (Osterhout et al., ; Osterhout & Hagoort, ). Words that are both syntactically and semantically anomalous elicit an ERP response that is equivalent to the summation of the N and P responses to semantically or syntactically anomalous, indicating that these responses are (to a first approximation) independently generated in the brain (Osterhout & Nicol, ). Although no definitive proof exists concerning the cognitive processes underlying the P, one possibility is that P amplitude reflects (directly or indirectly) the degree to which the relevant syntactic rule has been ‘proceduralized’ in FL learners (Morgan-Short et al., ). These generalizations hold well under many conditions (e.g., stimuli with a relatively simple sentence structure and participants who are fluent, native speakers of the language being tested), they do not always transparently generalize to complex stimuli (Kim & Osterhout, ; Kuperberg, ) or to people who do not have native-like proficiency in the language being tested (McLaughlin et al., ). For example, seemingly ‘semantic’ effects sometimes elicit P-like responses in certain types of complex sentences (Kim & Osterhout, ; Kuperberg, ). However, perhaps the most striking (and potentially useful) ‘failure to generalize’ pertains to ERPs recorded from adult learners of an FL. We review this literature in the following section.
. E-R P ‘’
.................................................................................................................................. The fact that syntactic and semantic anomalies elicit qualitatively distinct brain responses suggests that distinct syntactic and semantic processes might exist in the mind/brain of a native speaker. In an L learning context, one implication is that L learners must somehow segregate linguistic input into those aspects of the language that relate to sentence form and those that relate to sentence meaning. That is, part of learning an FL seems to involve ‘grammaticalizing’ (or ‘proceduralizing’; Ullman, , ) some aspects of the FL, but
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
, ̈, ,
not others. In this context, ‘grammaticalization’ refers specifically to the instantiation of that aspect of the FL into the learner’s online, real time language processing system. Prior ERP results have shown that ERPs are perhaps uniquely sensitive to the ‘grammaticalization’ process. For example, Osterhout et al. () reported a longitudinal study of L English college students who were enrolled in a first-year FL French class; learners were tested three times, near the beginning, middle, and end of the instructional period. Stimuli were simple French sentences, some of which contained a semantically anomalous word or a violation of French verbal–person agreement. The semantic anomalies elicited an N effect at each testing sessions. After just one month of instruction, the learners’ brains discriminated between the syntactically well-formed and ill-formed sentences. However, rather than eliciting the P effect (as we saw in native French speakers), the syntactically anomalous words elicited an N-like effect. By four months, the N effect was replaced by a P effect. Our results can be explained by assuming that learners (much like child L learners) initially memorize salient word sequences (e.g., tu adores; Ellis, ; Arnon, & Christiansen, ). Violations of the verbal person rule (e.g., *tu adorez) result in novel word combinations and, hence, elicit an N effect. After more instruction, learners induce a general verbal person rule (tu -s, nous -ons, vous -ez, etc.); violations of the rule elicit a P effect. (For related findings, see Osterhout et al., ; Hahne et al., ; Osterhout et al., ; Steinhauer et al., ; McLaughlin et al., ; Tanner et al., ; Morgan-Short, ; Tanner et al., ). As noted above, FL learning is all too often followed by language attrition, that is, the gradual forgetting of the FL (Schmid, b; Schmid & de Bot, ; Schmid & Mehotcheva, ). How might the loss of some aspect of a language be manifested in brain activity? A particularly salient claim is motivated by the regression hypothesis, which states that attrition is the mirror image of acquisition (Caramazza & Zurif, b; Olshtain, ; de Bot & Weltens, ; Hedgcock, ; Hyltenstam & Viberg, ; Hayashi, ; Keijzer, , a; for a review, see Schmid, b). Much of the supportive evidence takes the form of corpora of language productions, acquired at various time points during the instructional period, and again at varying time points after the end of FL instruction. If the hypothesis applies to the neural processes underlying FL acquisition and attrition, then the brain response observed during the attrition phase should be the inverse of that observed during acquisition.
. A
.................................................................................................................................. In this section, we describe one of the first ERP studies to longitudinally assess the acquisition and subsequent attrition of a classroom-learned FL. We investigated the acquisition and application of Finnish vowel harmony. Vowel harmony is a type of long-distance assimilatory phonological process involving vowels; it requires that all vowels in a word must be members of the same vowel subclass (thus ‘in harmony’). Finnish has two opposing vowel classes, back vowels (u, o, a) and front vowels (y, ö, ä). The vowel harmony rule prevents vowels from different vowel classes from appearing in the same word (but
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
only applies to native, non-colloquial words). If the initial vowel in a word is a front vowel, all subsequent vowels must be either front vowels or neutral vowels. If the initial vowel is a back vowel, only back vowels and neutral vowels are permitted (Van der Hulst & Van de Weijer, ; Suomi et al., ). We chose to focus on the Finnish vowel harmony rule because the rule is simple and highly regular, and because a similar rule does not exist in English or in Indo-European languages typically taught in high school and university settings within the United States. Participants included eighteen native English speakers who were enrolled in an introductory Finnish course at the University of Washington and sixteen native Finnish speakers.
.. Materials Two lists were constructed, each containing letter strings: % () of the letter strings were real Finnish words selected from the learners’ instructional book and % () were non-words; % () of the non-words were orthographically legal in Finnish; and % () of the non-words contained combinations of vowels that rendered the letter strings illegal in Finnish by violating a rule of vowel harmony. Description of this rule was included in the first chapter of the learners’ instructional book. The stimuli of interest in the data reported here represent only a subset of the overall stimulus items (for a full description of the entire set of stimuli and the rules for stimulus selection, see Pitkänen, ). This subset included a set of fifty-six Finnish words in each stimulus list (e.g., tuoli, ‘chair’), and two types of non-words: fifty-two orthographically legal pseudo-words (*louti), and fifty-two vowel harmony violating letter strings derived from the real Finnish words (*tyoli). The stimulus items contained four to eight letters. The orthographically legal pseudo-words were formed by manipulating the letters of a real Finnish word while obeying the vowel harmony rule. For example, the position of consonants in a word was exchanged (koulu ‘school’—*louku), or both the position of consonants and vowels were exchanged (tuoli ‘chair’—*louti) because exchanging only the consonants would create a real Finnish word (luoti ‘bullet’). In some instances the real words were manipulated to a greater degree by adding or deleting a letter and/or mixing the order of letters and syllables of a real word (lautanen ‘plate’—*taunale). The illegal, vowel harmony violating letter strings were formed from real Finnish words by replacing one back vowel with a front vowel, such that the resulting illegal letter strings contained both a front vowel and back vowel(s). The vowel harmony violations involved the replacement of a back vowel ‘u’ with a front vowel ‘y’ which is orthographically familiar to the learners from their native English. These two vowels were chosen for the exchange to control for vowel height. Each of the manipulated real Finnish words contained at least two back vowels, and when one back vowel was replaced with a front vowel, at least one unaltered back vowel remained in the letter string, rendering a vowel harmony violation. The stimuli were counterbalanced across lists such that a single list never contained an orthographically well-formed and ill-formed version of an identical stimulus item (koulu—*koylu). The stimuli were pseudo-randomly ordered in the lists. The experiment was comprised of an acquisition phase and a post-instruction attrition phase. During the acquisition phase, Finnish learners were tested three times in a longitudinal design during their nine months of Finnish instruction (session : mean thirty-six hours of instruction, standard deviation (SD) = .; session : mean seventy-one hours,
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
, ̈, ,
SD = .; session : mean hours, SD = .). Each participant saw only one of the two lists during each testing session; the two lists were alternated for each participant such that session involved a repetition of the stimulus list seen by the participant in session , but the order of the stimulus items within the list was altered from session to session . After the end of Finnish instruction, the learners returned for a single post-instructional attrition session; details are described later in the chapter. The native speakers of Finnish participated in a single testing session and were randomly assigned to one stimulus list. The visual stimuli were displayed for a brief duration on a computer monitor. The subjects made a lexical decision to each letter string; for each trial, subjects indicated their word-non-word decisions by pressing one of two buttons on a response box. Continuous electroencelphalogram (EEG) was recorded using nineteen tin electrodes attached to an elastic cap (Electro-cap International) in accordance with the extended – system (Nuwer et al., ). Electrode positions included three midline sites (Fz, Cz, Pz), five pairs of mediallateral sites (right hemisphere: Fp, F, C, P, O; left hemisphere: Fp, F, C, P, O), and three pairs of lateral-lateral sites (right hemisphere: F, T, P; left hemisphere: F, T, P). These electrodes were referenced to an electrode placed over the left mastoid and were amplified with a bandpass of .– Hz ( db cutoff ) by an SAI (short-latency afferent inhibition) bioamplifier system. Impedances at scalp and mastoid electrodes were held below µΩ and below µΩ at eye electrodes. All participants received a small monetary compensation for their participation and gave their informed consent. Continuous analog-to-digital conversion of the EEG and stimulus trigger codes (initiated by the presentation of each word in a sentence) was performed at a sampling frequency of Hz. Epochs were comprised of the ms preceding and , ms following stimulus presentation. ERPs, time-locked to the onset of the letter string were averaged offline for each subject at each electrode site. Grand average wave forms were created by averaging over participants. Trials characterized by eye blinks, excessive muscle artifact, or amplifier blocking were not included in the averages. N amplitude was quantified as mean voltage within a – ms window for native Finnish speakers and – ms for the learners. P amplitude was quantified as mean voltage within – ms. Repeated-measures ANOVAs with the Greenhouse-Geisser correction were used. Separate ANOVAs were computed for data from midline, medial-lateral, and laterallateral sites to allow for quantitative analysis of hemispheric differences. On the data from midline sites, two-way ANOVAs were performed with repeated measures on two levels of stimulus type (well-formed, ill-formed) and three levels of electrode position (frontal, central, parietal) for the two conditions (orthographically legal pseudo-words, and vowel harmony violations). In addition, three-way ANOVAs were performed on the data from medial-lateral sites for all conditions using two levels of stimulus type, two levels of hemisphere (left, right) and five levels of electrode position (pre-frontal, frontal, central, parietal, and occipital). On data from lateral-lateral sites, three-way ANOVAs were performed using two levels of stimulus type, two levels of hemisphere (left, right) and three levels of electrode position (inferior-frontal, temporal, and posterior-parietal). Lexical sensitivity in the lexical decision task was assessed by using the d-prime sensitivity measure. It was derived by computing the proportion of hits and false alarms for each subject at each session and across subjects for each group. Separate omnibus ANOVAs that included testing session as a variable were performed on the data obtained from the Finnish learners.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Prior work has shown that pronounceable non-words (i.e., pseudo-words) elicit a larger N component than do words. Since vowel harmony violations are non-words only by virtue of violating phonotactic rules, one would expect that in native Finnish speakers they should elicit a P effect rather than an N effect. This prediction is motivated by prior work showing that the P effect is elicited by violations of constraints on linguistic form, including both syntactic and phonetic form. Of particular interest is how the Finnish learners might respond to the vowel harmony violations. If the learner has not acquired the vowel harmony rule, then the vowel harmony violations should be perceived to be pronounceable Finnish pseudo-words—and hence elicit an N effect. However, once the learner has learned the rule, then the vowel harmony violations should elicit a P instead (as described above; cf. Osterhout et al., ; Osterhout et al., ; McLaughlin et al., ).
.. Behavioural responses Table . shows the proportion of non-words identified as words (false alarms) and average d-prime scores for the native Finnish speakers, and for the Finnish learners in the three sessions that occurred during the instructional period (the last row of the table applies to the attrition study presented later in this chapter). Native speakers of Finnish were able to distinguish real Finnish words from the two types of non-words nearly perfectly. They judged the real words to be acceptable on % of the trials, and each type of non-word to be acceptable on only % of the trials. The learners showed greater variability in their performance, but they performed at above chance levels in all sessions. Statistical analyses revealed that learners’ performance improved with instruction (main effect of session: F(,) = ., p = .). Pairwise comparisons revealed that the learners’ performance was significantly better in session than in sessions or , p < .. Sessions and did not reliably differ from each other. For all sessions, the learners’ performance was significantly worse for pseudo-words than for vowel harmony violated words, p < ..
Table .. D-prime (dʹ) and proportion of false alarms (FA) for pseudo-words and vowel harmony violations Pseudo-words
Native Speakers Learners Session Session Session Post-instruction
Vowel harmony violations
Proportion of FA
dʹ
Proportion of FA
dʹ
.
.
.
.
. . . .
. . . .
. . . .
. . . .
Note: We calculated dʹ using a formula provided by Miller () dʹ = z(h) – z(fa), where dʹ represents the likelihood of a real target being recognized, z(h) represents the z-score that corresponds to the proportion of real words that were identified as words (hits), and z(fa) represents the z-score that corresponds to the proportion of non-words that were identified as words (false alarms).
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
, ̈, ,
.. ERPs elicited during the ‘learning phase’ Separate ERP ‘grand averages’ (over both trials and subjects) were created for vowelharmony-obeying and vowel-harmony-violating Finnish words, for both the native Finnish speakers and for the Finnish learners. For the learners, ERPs were computed separately for each of the three testing sessions that occurred during the learning period. As expected, for native Finnish speakers, pseudo-words elicited a larger-amplitude N than did real Finnish words (– ms window: midline electrodes, F(, ) = ., p < .; medial-lateral electrodes, F(, ) = ., p < .; lateral-lateral, F(,) = ., p = .). By contrast, vowel harmony-violating words elicited a P effect (– ms window: midline, F(, ) = ., p < .; medial-lateral, F(, ) = ., p < .; lateral-lateral, F(,) = ., p < .; see top of Figure .). In the Finnish learners, pseudo-words elicited a larger-amplitude N than did words at each testing session (– ms window: midline (F(, ) = ., p < .; medial-lateral, F(, ) = ., p < .; lateral-lateral, F(,) = ., p = .). This effect changed minimally with increasing instruction, as there was no reliable interaction between session and condition (F < ). However, the ERP response to vowel harmony violations changed qualitatively as well as quantitatively across the instructional period (condition x session interaction, – ms window: midline, F(,) = ., p < .; medial-lateral, F(,) = ., p = .; lateral-lateral, F(,) = ., p = .; – ms window: midline, F(,) = ., p < .; medial-lateral, F(,) = ., p < .; lateral-lateral, F(,) = ., p < .). At sessions and , the violations elicited small N effects (– ms window: midline, F(, ) = ., p < .; medial-lateral, F(, ) = ., p < .; laterallateral, F(,) = ., p = .). At session , however, the same violations elicited a large-amplitude P effect (– ms window: midline, F(, ) = ., p < .; medial-lateral, F(, ) = ., p < .; lateral-lateral, F(,) = ., p = .; see bottom of Figure .). To determine the relationship between amount of instruction and the ERP effects, the N and P effects for the vowel harmony violations were correlated with hours of instruction across sessions. Whereas the N decreased as the number of hours of instruction increased (r = –., r2 = ., p < .), the P effect increased with increasing instruction (r = ., r2 = ., p < .). We interpret this change in brain activity to the gradual ‘grammaticalization’ of the vowel harmony rule: once the vowel-harmony rule has been generalized to all perceived instances of vowel harmony, vowel harmony violations stop eliciting N effects and start eliciting P effects.
.. Post-learning ‘attrition phase’ Sixteen of the eighteen Finnish learners returned for a single follow-up post-instruction ERP session. The attrition period was quantified as the number of days since the learners had last received formal Finnish instruction (mean days, range to days). The stimulus lists and procedure were the same as in the acquisition experiment. Each participant was assigned a stimulus list with the same items that they had seen in the second session in the acquisition experiment, but the order of the items in the list had been altered.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Pseudo-words Native Speakers
Vowel Harmony Violations
N400
P600
Learners Session 1:
N400
N400
Session 2:
N400 Session 3:
P600 –3µv 300 600 900
Well-formed words Pseudo-words Vowel harmony violations
.. ERPs elicited by Finnish pseudo-words and vowel harmony violations, for native Finnish speakers (top) and L English FL Finnish learners nearing the end of the first, second, and third quarter of classroom-based FL (Finnish) instruction (bottom) Note: The vertical calibration bar indicates word onset; each tick mark represents ms. Negative voltage is plotted up. Data acquired over the central midline electrode (Cz) are shown.
Thus, no participant was assigned a list that they had seen in their last session during the first year of Finnish instruction. The bottom row of Table . shows average d-prime scores for the learners in the postinstruction session. A pairwise comparison between the third acquisition session and the post-instruction session revealed that the participants’ performance did not significantly decline, p > .. To investigate the relationship between length of the attrition period and the accuracy of learners’ lexical judgements, post-instruction d-prime scores were correlated with length of attrition period. No significant correlations were observed between the number of days since instruction had ended and behavioural judgements for either vowel harmony violations (r = –., p > .) or pseudo-words (r = –., p > .). Although the behavioural data did not show any change in sensitivity from the acquisition stage to the attrition stage, the ERP data revealed a qualitative change in the neural representation of the vowel harmony rule. Our first priority was to test predictions that follow from a ‘processing-based’ version of the regression hypothesis. During the acquisition phase described above, vowel harmony violations elicited an N effect during early stages of learning, and a P effect after additional instruction (hypothetically reflecting ‘pre-grammatical’ and ‘post-grammatical’
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
, ̈, , N400 N400 Amplitude (μV)
P600 Amplitude (μV)
P600 5 4 3 2 1 0 –1 –2 –3 –4 0
100 200 300 400 500 600 700 800 Days Since Instruction
5 4 3 2 1 0 –1 –2 –3 –4 0
100 200 300 400 500 600 700 800 Days Since Instruction
.. P amplitude (left panel) and N amplitude (right panel) elicited by Finnish (FL) vowel harmony violations, plotted as a function of days since the end of FL (Finnish) instruction
representations of the vowel harmony rule). The primary question of interest is whether this sequence would repeat itself during the attrition phase, but in reverse. More specifically, the prediction is that the vowel harmony violations should elicit a P effect for as long as a learner maintains a ‘grammaticalized’ representation of the vowel harmony rule and applies it during real-time language comprehension. However, once the rule has attrited, the violations would ‘regain’ their status as apparent pronounceable non-words, and hence elicit N effects instead. To test this prediction, we calculated a correlation between mean amplitudes within the N and P windows for ERPs elicited by the vowel harmony violations and days since instruction. As noted above, during the acquisition phase, N amplitude to the vowel harmony violations decreased as a function of instruction, whereas P amplitude increased to a similar degree. Remarkably, a similar ‘anti-correlation’ between the N and P effects was observed during the attrition phase—but in reverse. That is, the longer the time since instruction had ended, the smaller the P response to the vowel harmony violations (r () = –., r2 = ., p = .), and the larger the N response (r () = ., r2 = ., p < .; see Figure .). To further investigate the relationship between these ERP effects and time since instruction, the participants were divided in two groups by a median split based on the length of the attrition period. A subset of the learners (N = ) were tested less than seven months after the end of Finnish instruction (‘short attrition group’, mean eighty-seven days since instruction, range to days), and another subset (N = ) were tested more than nine months after instruction (‘long attrition group’, mean days, range to days). Whereas in the short attrition group vowel harmony violations elicited a reliable P effect (main effect of condition, – ms window: midline, F(,) = ., p = .; medial-lateral, F(,) = ., p = .; lateral-lateral, F(,) = ., p = .), in the long attrition group they elicited a reliable N effect (main effect of condition, – ms window: midline, F(,) = ., p = .; medial-lateral, F(,) = ., p = .; laterallateral, F(,) = ., p = .; see Figure .).
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Short Attrition Group Fz
Long Attrition Group Fz N400
P600 Cz
Cz
Pz
Pz
–3µv 300
600
900
.. Post-instruction grand average ERPs to Finnish words (solid line) and vowel harmony violated words (dashed line) recorded from participants with a post-instruction interval of less than seven months (short attrition group, N = ) and greater than nine months (long attrition group, N = ) Note: Negative voltage is plotted up. The vertical calibration bar indicates word onset; each tick mark represents ms. Data acquired over midline electrode sites are shown.
. D
.................................................................................................................................. The study reported here was designed to explore the acquisition and subsequent attrition of a classroom-learned FL (Finnish), in native English speakers. We focused specifically on changes in the learners’ ability to discriminate between real Finnish words and Finnish pronounceable non-words (assessing learners’ lexical learning), and between real Finnish words and pronounceable pseudo-words that violated the Finnish vowel harmony rule (assessing learners’ acquisition of a grammatical/phonotactic rule). We assessed learning by recording learners’ lexicality judgements to Finnish letter strings, and by recording ERPs elicited by the letter strings. Our results show that systematic changes in brain activity accompany both the acquisition of a FL, and the subsequent post-instruction attrition of that language. During the acquisition phase, vowel harmony violations elicited an N effect early during the instructional period and a P effect after further instruction. During the post-instruction attrition phase of the experiment, the pattern was reversed: vowel harmony violations elicited a P effect after a relatively short post-instruction period, and an N effect after a longer period. This pattern of results can be interpreted as reflecting the gradual
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
, ̈, ,
acquisition and subsequent attrition of the Finnish vowel harmony rule: Before the rule is learned, the vowel harmony violations are effectively Finnish pseudo-words (i.e., both legal and pronounceable letter strings)—and hence elicit N effects. After the rule is learned, the same stimuli represent violations of the vowel harmony rule—and hence elicit a P effect. Conversely, once the vowel harmony rule has been ‘forgotten’ during the attrition period (or can no longer be applied effectively in real time during comprehension), the vowel harmony violations regain (in the mind of the learner/attriter) their status as pronounceable pseudo-words—and hence elicit an N effect. Our results are therefore consistent with a ‘processing’ version of the regression hypothesis, in which the brain response to the pseudo-words is a function of the learners’ ability to ‘apply the rule’ in real time, as they process the stimulus. The interpretation outlined above is also consistent with relevant prior ERP work examining the acquisition and use of linguistic rules at different levels of linguistic structure. For example, Tanner et al. () recorded ERPs from FL German learners. Participants were presented with German sentences that were either well-formed (Ich wohne in Berlin ‘I live in Berlin’) or contained verb agreement violations (*Ich wohnt in Berlin ‘I lives in Berlin’). The results revealed some striking individual differences: Whereas some learners showed primarily an N effect to the anomaly, others showed primarily a P effect. Critically, the ‘quality’ of the response (N or P) was predicted by the learners’ accuracy in the sentence-acceptability task, with better performance predicting P effects. In the present study, the accuracy of learners’ lexicality judgements declined moderately during the attrition period (relative to the final testing session during the instructional period), with a larger reduction in accuracy (as assessed by d-prime) for the vowel harmony violations (% reduction) than for the pseudo-words (% reduction). This result is consistent with findings from prior studies reporting less attrition for FL lexical aspects than for grammatical aspects of an FL (Moorcroft & Gardner, ; Weltens et al., ; Tomiyama, ). At the same time, the learners were better at detecting the vowel harmony violations than the pronounceable non-words at each testing session, and during the attrition period. Such an outcome is to be expected, given that the vowel harmony decision requires knowing a general rule that applies to all strings, whereas the lexical decision must be made independently for each item. However, the magnitude of the ERP response to vowel harmony violations was not robustly correlated with participants’ accuracy on the lexical decision task (r < .). This pattern differs from the relationship of task performance and P effect sizes observed during the acquisition phase, during which lexical decision accuracy and P amplitude were correlated. Although learners’ lexical decision accuracy did not change as a function of the length of the attrition period, the ERP response to the vowel harmony violations did change. Specifically, as the post-instruction period increased, the amplitude of the P effect decreased, whereas the amplitude of the N effect increased. This pattern was the reverse of the pattern observed during the acquisition phase, during which N amplitude decreased in amplitude over time while P amplitude increased. Thus, the ERP response to phonotactically anomalous letter strings ‘regressed’, in that it increasingly (over time) approximated the ERP response seen near the beginning of the instructional period. The results from the attrition experiment show that neural processing changes are observed not only during FL acquisition but also during FL attrition. These observations
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
would seem to suggest that some ‘restructuring’ of linguistic information takes place after a period of active learning has ended. The amplitude of the P effect to vowel harmony violations was robustly correlated with increasing accuracy in the learners’ acceptability judgements, across the extent of the instructional period. Therefore, one might reasonably propose that a reduction in P amplitude might represent a neural correlate of the vulnerability of grammar to attrition, specifically in beginning learners. The observed pattern may reflect the earliest stage of an attrition process that will ultimately lead to loss of grammatical knowledge, even if no behavioural sign of the attrition is evident in the learners’ explicit judgements. It is possible that the relatively simple task of making lexical decisions on isolated letter strings may not have revealed as much behavioural erosion as more demanding and subtle tasks might have disclosed. For example, language production tasks or reaction time measurements might have shown more attrition after long periods of no instruction. Prior behavioural evidence indicates that grammatical aspects of an FL may be more vulnerable to loss than the lexicon, particularly in low-proficiency learners (Moorcroft & Gardner, ; Weltens et al., ). Our data are consistent with this hypothesis: the neural indicator of grammatical processing (the P effect) decreased in magnitude during the attrition period, whereas lexical sensitivity (reflected by the N effect) increased over time. This qualitative change in neural processing did not correlate with changes in behavioural sensitivity in the simple lexical decision task, mirroring a dissociation between behavioural judgements and brain responses observed in prior ERP studies (McLaughlin et al., ). The small number of participants and lack of a longitudinal design in the attrition experiment are reasons for caution in the interpretation of the results. Moreover, the experiment only investigated learning and attrition of words that were manipulated with a single salient word-formation rule. Such simple manipulations may be more likely to elicit consistent results from a group of learners. The simplicity of the experiment was an advantage in this instance, but it may also suggest that these effects might not be easily reproduced with more complex tasks and grammatical rules. Even the longitudinal ERP studies examining learning of morpho-syntactic FL rules have found that some learners do not show the pattern of an N replaced by a P during the first year of FL instruction (Osterhout et al., ), whereas the learners in the current study showed more consistent (less variable across learners) brain responses during the instructional period. The use of more complicated grammatical violations might make it less likely to see such well-ordered attrition functions. Even if we assume that the pattern of neural-processing changes observed during the attrition phase reflects the erosion of grammatical knowledge, it is possible that the vulnerability of the P effect to attrition is more pronounced for grammatical features that are not shared by the L and FL. Word-formation rules of Finnish, including the vowel harmony rule, are not useful for Finnish learners living in an English-dominant environment. In the case of vowel harmony, it may even be detrimental to process combinations of vowels as ungrammatical when those combinations are perfectly acceptable in English words. Levy et al. () have shown that native-language phonology is suppressed during FL use. Perhaps suppression of FL phonotactic rules occurs after active FL exposure and use has ended, and performing an FL lexical decision task after a period of FL disuse is not sufficient to overcome the effects of prolonged inhibition. The length of the attrition period
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
, ̈, ,
may correlate with the degree of grammatical knowledge suppression. Levy et al. () also showed that unlike phonology, access to semantic knowledge was not inhibited. Similarly, the learners did not lose their lexical knowledge of Finnish during attrition. Linguistic features that are similar in L and FL, by contrast, might more permanently ‘preserve’ the P response to FL anomalies. In summary, the vast majority of FL experiments have investigated either language acquisition or language attrition, but not both. The design of the present study provided an opportunity to observe relationships between the learning/acquisition phase and the post-learning retention/attrition phase, in adult classroom-based learners. Our results revealed a striking symmetry in the acquisition and subsequent attrition of a specific FL grammatical feature, when that feature was not present in the learners’ L. These findings highlight the dynamic and systematic process of learning and ‘unlearning’ structural restrictions of FL words as manifested in brain activity, and associate different patterns of neural activity with the acquisition and attrition of structural and lexical aspects of language. Additional research is needed to determine whether this finding can be generalized across different aspects of an FL, different languages, and different language learning contexts. The findings presented here demonstrate the viability of a brain-based research paradigm for studying these and other questions about FL attrition. Conceivably, the gained knowledge might lead to pedagogical strategies that minimize FL attrition, thereby maximizing the return on the societal investment in FL instruction.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
.............................................................................................................
HERITAGE LANGUAGES ............................................................................................................. Edited by
, , ̆
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
......................................................................................................................
......................................................................................................................
. I
.................................................................................................................................. 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 the larger (national) society. Like the acquisition of a primary language in monolingual situations and the acquisition of two or more languages in situations of societal bilingualism/ multilingualism, the heritage language is acquired on the basis of an interaction with naturalistic input and whatever in-born linguistic mechanisms are at play in any instance of child language acquisition. (Rothman, , p. )
Given this definition, heritage speakers are bilinguals who speak a minority language (their heritage language) and another language that is societally more dominant. To quote Tanja Kupisch (Chapter , this volume), a heritage language (HL) is considered to be a minority language that is not an official language in a speaker’s geographical area, and that was acquired at home, regardless of whether the HL was spoken by one or both parents. (p. )
There may be different reasons for the dominance of the non-heritage language, and we cannot do them justice in this short chapter; in what follows we will be referring to the nonheritage language as the dominant/majority language. In Guatemala, for instance, where Spanish is the dominant societal language, any other language used in a specific community may be considered a heritage language, including immigrant, indigenous, or even sign language communities. Similarly, in the United States, immigrant (and colonial) languages, as well as ethnic (Fishman, ) and sign languages, are considered heritage languages in
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
an English-dominant bilingual context. Because of the nature of their language acquisition, heritage speakers are often more competent speakers of their second language (the socially dominant language) than their first language (the heritage language). The imperfect control that heritage speakers exhibit over their first language (L) often differs in significant and consistent ways from the imperfect control of second-language learners; frequently, heritage learners are understood to experience ‘attrition’ of their original language system, not just incomplete mastery of aspects of that system (Polinsky, ; Polinsky & Kagan, ; Rothman, ; Benmamoun et al., a; Scontras et al., ; Montrul, a). Although we find a great degree of variation in heritage populations, it is typical for the heritage language to be the weaker one of the bilingual dyad; differences between heritage speakers lie in the degree of that weakness, as some heritage speakers are quite close to the monolingual baseline while others may be at the level of receptive bilinguals. The degree of dissonance between the two languages in the dyad varies, but the general balance in favour of the dominant language is a hallmark of heritage speakers. Heritage speakers often lack confidence in their home language and feel that they are not as good at speaking it as one may expect them to be—especially since the high expectations are based on their native-like control of the sound system. Heritage speakers can shed light upon the current theoretical discussion about the nature of language, allowing us to adopt a novel approach to an old question: what do we know when we know a language? We may have intuitions about what a native speaker knows about language, but a heritage language forces us to consider more deeply the question of what exactly it means for a person to be a native speaker. There is a consensus that native speakers/signers differ from non-native speakers of a language because the native speakers acquired their language from a very early age within a natural input environment; this makes native speakers different from second language (L) speakers but identical to heritage speakers. Heritage speakers, like native speakers, acquire a home language naturally, at an early age—the difference is that they also acquire a community language at the same time, which they gradually come to rely on as their primary language outside the home (e.g., Chinese at home and English at school and elsewhere). Heritage speakers are exposed to the heritage language exclusively during early childhood or simultaneously with the majority language. During the school age period the majority language tends to become dominant and by early adulthood heritage speakers display a wide range of proficiencies in the heritage language, which vary significantly and span the entire spectrum from low/receptive to full productive ability. As a result, heritage speakers display systematic gaps in their linguistic knowledge in several structural components of language (phonology, morphology, syntax, semantics, discourse). Still they are native speakers of their heritage language, in many aspects different from L learners (Montrul, ; Rothman & Treffer-Dallers, ), which is particularly apparent in their target-like sound system. Given that they are native speakers (or at least they start as such), why do they vary so much in the levels of proficiency eventually attained in the heritage language? In what follows, we argue that it is largely due to how the heritage language was acquired and nurtured from birth to adolescence in a bilingual environment, with many individual and contextual factors playing a role in its development.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
. B :
.................................................................................................................................. A bilingual child needs to figure out the components and rules of each of their input languages based on input in two languages. Hoff () provides considerable evidence that many social factors affect language development in children, ranging from the amount of time children spend in conversation with interlocutors to the contexts in which they talk and hear the language. How many opportunities children have to exercise their communicative skills, and how well their language model matches their abilities for analysis, will have an effect on the rate of their particular language development (Shatz, ). Although some approaches to child language acquisition stress the rapidity and universality of the process (Crain & Lillo-Martin, ; Guasti, ), language acquisition develops over time: it begins at birth, if not in the womb, and it takes several years for a child to become an adult native speaker of the language. There are well-attested linguistic milestones in both monolingual and bilingual acquisition that children reach at specific ages. Children who are a few days old can discriminate their own language from another, and bilingual children recognize the two languages spoken in their ambient environment (SebastiánGallés, ). By months of age, infants recognize the sounds of their native language(s). By about to months, infants begin to babble, producing reduplicated CV syllables, and by to months children produce their first words (excluding ‘mama’). They understand the meaning of several words several months before they begin to produce any words (Huttenlocher & Smiley, ). By about months, there is rapid acquisition and use of new words, and a vocabulary spurt is observed in most children. Carey () estimated that children acquire about nine new words a day from the age of months to years. At about the same time, children start to combine words into two-word phrases (big toy, mommy go)—the two-word stage—but do not produce inflectional morphology. Studies of monolingual and bilingual children show that at the two-word stage, children respect the word order of the languages they are exposed to. For example, if the language, or one of the languages, is SVO, the child will tend to produce words that stand for SV (mommy go) or VO (eat cookie), but not for OV (cookie eat). Brown (), Pinker (), Braine (), and Bloom (), among many others, have noted that word order ‘errors’ are very infrequent in L acquisition. These two-word utterances are characterized as ‘telegraphic’ speech because they are composed of lexical categories only (nouns, verbs, adjectives, prepositions) and lack function words and grammatical morphemes. English-speaking children, for example, frequently omit subjects at this stage (*eating soup ‘I am eating soup’). This is not a grammatical option in the adult language, which is the main source of input to the child. Children at first omit grammatical morphology, especially in a language like English, and when inflectional morphemes appear, they emerge and develop in a fixed order. Children learning two languages follow the developmental sequence established in each of the languages (Meisel, , ). The omission of a morpheme in an obligatory context and the use of the wrong morpheme in a given context are referred to as developmental ‘errors’ in child language. Developmental errors eventually go away, naturally. By months, children’s utterances become longer and more complex and the children gradually produce the required morphemes in the contexts where they are required. This, of course, depends on the
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
particular language being acquired, since not all features exist in all languages and not all features are acquired at the same speed from one language to another. For example, gender in nouns and definite determiners in Dutch is acquired quite late by Dutch-speaking children, after age (Gillis & De Houwer, ) or even by age (Unsworth & Hulk, ), whereas in Spanish gender agreement is mastered by age (Montrul, a). Comprehension-production dissociations are common during early syntactic and phonological development. Children begin to produce a variety of complex sentences during the pre-school period. The earliest complex sentences in English-speaking children emerge around the second year (bare infinitives, wanna constructions, complements of verbs think, said, know, conjunctions and and but) (Diessel, ). By age , children produce relative clauses, adverbial clauses, participial clauses, and modal verbs. Yet comprehension-based experiments have shown that children do not understand many complex sentences until well into the school years (Chomsky, ; Clark, ; Sheldon, ; Tavakolian, ), perhaps because the acquisition of complex sentences is related to the complexity of relating two clauses on the one hand, and to the frequency of the constructions in the input, on the other. Furthermore, pragmatic and cognitive factors also play a role in the acquisition of complex sentences. For example, presentational relative clauses, which require the use of the indicative in Spanish, as in (), are acquired earlier than presuppositional relative clauses, which require the use of the subjunctive, (): () Juan necesita un empleado que sabe computación. Juan needs an employee who knows. computation ‘Juan needs an employee who knows computers.’ () Juan necesita un empleado que sepa computación. Juan needs an employee who knows. computation ‘Juan needs an employee who would know computers.’ The subjunctive in Spanish is used with complex syntax, and is not mastered until age to (Blake, ). Table . summarizes key milestones in early monolingual and bilingual development in pre-literate children, whose linguistic competence is acquired largely orally
Table .. Developmental milestones in early language development Age
Milestone
Linguistic characteristics
Birth to months Cooing
Early speech perception and phonetic discrimination
to months
Babbling
Production of CV syllables Attunement to sounds of the native language
to months
One-word stage
Production of first word(s) (form-meaning matching)
to months
Two-word stage
Telegraphic speech Memorized chunks No productive use of inflectional morphology
to months
Early multiword speech Basic syntax. Emergence of morphology
+ months
Later multiword speech Complex sentences
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
and implicitly. Many adult heritage speakers display morphological and syntactic errors typical of early language development in monolingual children (Montrul, a). Later language development refers to the school-age period, and this period has not been as intensely investigated in linguistics as the pre-school period, perhaps because the Chomskyan view of language acquisition portrays becoming a native speaker as a fast and efficient process, with a continuous and effortless transition from the initial state to the final state (Crain & McKee, , p. ). In stark contrast to this view, Berman (, ) contends that becoming a proficient native speaker takes a long time because in addition to basic linguistic competence, language use in a variety of contexts must be taken into account. Berman’s studies show that the language of - and -year-old children differs markedly from that of adults, not only in content, but also in morpho-syntax and lexicon. Many linguistic forms, even those that emerge at early preschool age, have a long developmental history to become acquired and mastered, that is, entrenched (Berman, , p. ). Keijzer () showed that in a number of tests of morphology and syntax in Dutch, - and -year-olds were still very different from adult Dutch speakers as measured by a grammaticality judgement task and an elicited production test. Albirini () documented that it takes a while for Jordanian Arabic children to produce plural morphology in Arabic with more than % accuracy. Only at ages and years do the children appear to have internalized the complexity of plural morphology in Arabic. The children use all the forms but make mistakes until about age , when they behave like adult native speakers. Crucial to understanding later language development is the distinction between emergence, acquisition, and mastery. For Berman () a native speaker is somebody who has mastered and become proficient in several dimensions of knowledge and use of the native language through a constant interaction of competence and performance. For example, children may have acquired the grammar of number and gender agreement, yet still be at an ‘item-based’ phase of person marking (Tomasello, ), which for Berman would be ‘partial’ knowledge. Command of linguistic knowledge develops and is successfully reintegrated with increased ability in the domain of language use, which in the case of school-age children is reinforced by reading and writing. With increased age and cognitive and social maturation, the linguistic behaviour of speaker-writers comes to have an increasing effect on their internal linguistic representations. Ravid () shows that Hebrew-speaking children know and use the passive participle -u by age to , but it is not fully mastered until age . In a test administered to school-age children, -year-olds gave consistently nonpassive responses when presented with obligatory contexts for passive formation. Only around years of age did the children regularly provide passive constructions where required on the same test. Therefore, the morphology and syntax acquired during early language development is not always mastered until later. By the time children start elementary school, at age or , they exhibit a vocabulary of , to , words (Carey, ), inflect nouns and verbs with the correct morphology (% accuracy according to Brown, ), and articulate most of their words correctly. Bilingual children develop an amount of concepts equal to or greater than monolinguals, except that they are distributed between the two languages because the acquisition of vocabulary is context-dependent. Bilingual children who attend school in only one language—like many heritage speakers in the United States and in other countries— show faster and more development in the language of school (Merino, ). Input is particularly important for vocabulary development (Menyuk & Brisk, ). During the
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
pre-school period, children also gradually develop awareness of the language, and begin to discriminate between grammatical and ungrammatical forms. Because they cannot verbalize the rules, their grammaticality judgements are based on intuitions. Metalinguistic awareness develops at around age (Doherty & Perner, ) and is crucial for literacy development. At school, children fix, restructure, and expand their basic linguistic competence, gaining more communicative competence as they learn to read and write (Barriga Villanueva, ). They are also exposed to different types of discourse that require the expansion of more abstract vocabulary, and the use of more complex syntactic, semantic, and pragmatic structures. For example, children are encouraged to describe abstract objects or processes, to talk about cause and effect, to formulate hypotheses, and to support arguments. Depending on children’s interests, they develop vocabulary for specific areas of knowledge, and their vocabulary size can range from , to , words by the end of first Grade (Menyuk & Brisk, ). During this period, children also learn semantic and formal relationships between words (synomyms, antonyms, homophones, etc.) and morphological relatedness (rapid-rapidly, amaze-amazement-amazingly). Children learn to use their language as a medium to express their thoughts and experiences in speaking and writing. Syntactically, children develop the ability to use low-frequency structures such as the passive voice, common in scientific reports and writing, as well as generic statements (Dogs have four legs). Table . gives some of the features of later language development.
Table .. Structural and pragmatic development in - to -year-old children Category Grammar
Change Syntax
Morphology
Pragmatics
Sentence length increases (e.g., I see the boy who I played with yesterday) Combining structures becomes more frequent (through complementation, conjunction, subordination) (e.g., She likes me to do homework before watching television) Prefixing and suffixing increases (e.g., unhappiness, disapprove, discussion)
Lexicon
Use of abstract categories increases (e.g., liberty, vast, imagination) Synonyms and antonyms used more widely (e.g., large, big, huge, small, little, minute) Multisyllabic words appear more frequently (e.g., disappointment, unhappiness)
Phonology
Stress rules of language acquired (e.g., history, historical, influence, influential) Morpho-phonological rules (e.g., a car, an apple)
Conversation Begin to take perspective of others Begin to make relevant responses Storytelling
Begin to be listener friendly Begin to follow story grammar
Explanation
Begin to move from personal reference to abstract knowledge
Source: adapted from Menyuk & Brisk ().
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
However, mere exposure to and analysis of complex language at school is not sufficient for learning to occur. It is critical that older children have the opportunity to use the language in academic assignments requiring formal speaking and writing (Nippold, ). Compared to the oral discourse of conversation, expository discourse stimulates the use of longer utterances containing greater clausal density through adverbial, relative, and nominal clauses (Nippold, ). Analysing poetry or writing about controversial topics also stimulates use of complex thought and language. By engaging in these activities, children are called upon to employ the sophisticated lexical and syntactic elements they have been learning. At school, children acquire advanced levels of spoken and written language (Scott, ). Learning to write requires learning to use more abstract language and to move from simple noun phrases (my house) to complex nominalizations (the housing project that was inaugurated last year). It also requires use of modal verbs, conditionals, and adverbial conjunctions (meanwhile, consequently, therefore) in complex sentences. Children must also acquire knowledge of different registers and the ability to recognize when to use formal or informal language. Some languages, such as Korean and Japanese, grammatically encode honorification and use multiple registers to encode the social status of the addressee and speaker. Such registers are acquired in school and are an essential component of literacy. In addition to the knowledge of registers, children learn about other types of variation in the target language at school. The source of such variation may be dialectal, as in the standard language of school compared with the local dialect of the home; and/or contextual, as in differences between everyday colloquial usage, standard-intermediate level usage of the media or of academic discourse, or the normative requirements of the official language establishment. Ready and flexible access to diverse linguistic registers, varied levels of usage, and different types of texts are prerequisites for ‘linguistic literacy’ (Ravid & Tolschinsky, ), and require both protracted cognitive maturation and extensive experience with different communicative settings. A native speaker is someone who can distinguish between various styles and dialects of their language; they may not be able to name them all or explain how they work but they have enough knowledge to recognize the differences. But such knowledge comes from years of experience with different types of speakers and different contexts. It takes years to develop sensitivity to variation in one’s language and that is a type of sensitivity that heritage speakers often lack. Another critical aspect of literacy is written proficiency and spelling. Writing offers important advantages over speaking for linguistic development. Writing allows the learner more time to search the lexicon for the precise vocabulary and to organize the discourse more concisely, and to devote more thought to the formal aspects of language (Nippold, ). To summarize, the process of native language development extends beyond early childhood. Mature and proficient knowledge and use of the native language require several years of cognitive and linguistic maturation and experience with literacy-related schoolbased activities. At the same time, many aspects of later language development continue throughout life, since the language used by adult speaker-writers of a standard dialect differs in significant ways from that of high school seniors (Reilly et al., ). Since many heritage speakers do not receive academic support of their heritage language through schooling, it is not surprising to see that they do not develop many aspects of their heritage language, even those aspects that were supposed to be mastered in the early language development period, like inflectional morphology.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
. L
.................................................................................................................................. From the discussion above, it should be obvious that it takes several years of substantial exposure and use of the language in a variety of social contexts in order for a person to become a proficient native speaker, be that monolingual or bilingual. The basic linguistic competence developed during the pre-school years gets expanded and solidified as the child grows cognitively and socially, and is exposed to written language at school. The two main ingredients for language acquisition are the innate linguistic capacity for language (or language acquisition device (LAD)) (nature) and the input (nurture). However, mere exposure to samples of language is not sufficient: both the comprehension and the production systems need to be engaged for the speaker to use and produce the language in meaningful contexts. Literate children and adults need exposure to written and oral language. Another key factor that interacts with the innate capacity for language and input is timing, conceptualized as the age of the learner, as in Figure .. Successful language acquisition is subject to a maturational schedule: the LAD must be set in motion early, through exposure to the language. The idea that the elements and structure of language must be learned in childhood to be mastered at native levels has been around for many years, and supported by scientific evidence from brain development (Penfield, ; Penfield & Roberts, ; Lenneberg, ; Mayberry, ). The critical period hypothesis for language states that the capacity for implicit and unconscious native language learning is lost if it is not activated during a ‘critical’ period in childhood.
AGE Cognitive and social development
NATIVE LANGUAGE ACQUISITION AND ENTRENCHMENT NATURE
NURTURE
Language Acquisition Device
Exposure to and use of the language
.. Nature, nurture, and age: factors affecting native language acquisition and entrenchment
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Although Lenneberg () set the terminus of the critical period at age (or puberty), other ages were proposed in later research, as well as multiple ages for different components of language (Seliger, ; Long, ; Newport, ; Bialystok & Hakuta, ). Age effects are relevant both for L and L acquisition, as well as for L loss or attrition (Pallier, ; Montrul, ; Bylund, a,b). When attrition occurs in adulthood, the L grammar appears to undergo minor changes (Schmid, ), which in some cases can be induced by the L. In general, individuals undergoing attrition in an immigrant setting retain the ability to understand and use the language at an advanced level. It appears that reduced input and even disuse of the language for several years in adults does not seem to affect the integrity of the native grammar substantially (Schmid, , ). In some reported cases, the effects of L attrition have been minimal: after more than fifty years of language disuse Schmid () found that German Jewish émigrés living in the United States exhibited some transfer from English but very few actual morpho-syntactic errors that could be attributed to L attrition. No adult undergoing attrition in a bilingual environment has been shown to regress to such an extent as to forget how to conjugate verbs, ask questions, or produce and discriminate native sounds (Keijzer, ). The situation is different, however, when intense exposure to the L starts in childhood, as in the case of heritage speakers, although it is not the case that acquisition of an L in itself necessarily causes loss of the L in childhood. After all, there are fully fluent bilinguals who are exposed to the L and L in childhood and who do not exhibit L attrition (Kupisch et al., ). The studies documenting extensive effects of attrition at the lexical, phonological, and morpho-syntactic levels are about children (Kaufman & Aronoff, ; Turian & Altenberg, ) or about adults who immigrated in childhood (Vago, ; Polinsky, ), suggesting important differences due to age of intense exposure to the L and reduced used of the L. What causes severe L attrition is reduced input and lack of consistent and sustained exposure to and use of the L during a time when the native language is not fully fixed in the brain, most likely before and around the closure of the critical period (puberty). The L is used less because children growing up in an L environment spend most of their waking hours using the L at school and with peers, at the expense of the L. Age of emigration and intense exposure to the L affect L loss in a bilingual environment (Montrul, ; Bylund, a,b; Flores, a). The younger the individual when reduction of input and lack of use of the L take place, the more severe the extent of language loss at the grammatical level, such that the effects of L attrition in childhood are more dramatic than in adulthood. Several studies have shown that child and adult heritage speakers who are sequential bilinguals and who experienced a period of monolingualism or language dominance in their heritage language tend to have higher proficiency in the heritage language than children who are simultaneous bilinguals (Montrul, , ; Allen et al., ; Allen, ; Montrul & Sánchez-Walker, ). Montrul () shows that within childhood and in a minority language context, simultaneous bilingual children are more vulnerable to attrition than sequential bilingual children, because sequential bilinguals were exposed to their L for a longer period of time than simultaneous bilinguals (Montrul, ). However, it is important to bear in mind that L attrition also happens in adult immigrants, and such attrition may happen even without extensive knowledge of an L (Кöpke, c; Schmid, a; Baladzhaeva & Laufer, ; among others). In addition to age and timing of L input, the quantity and quality of first language input play a significant role in the extent of L attrition: reduced input in childhood is not the
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
same as completely interrupted input (Montrul, ; Hyltenstam et al., ). Immigrant children continue to have exposure to the family language, even if they do not use the language very often and may end up being receptive bilinguals or overhearers (Au et al., ; Oh et al., Chapter , this volume). In general, bilingual children who immigrate with their parents have some productive ability in their family language even though they may exhibit different degrees of acquisition and attrition (Montrul, , ; Polinsky, ; Polinsky & Kagan, ). By contrast, internationally adopted children are often adopted by families who do not speak the child’s language, although some families make efforts to keep the culture of the child present in some way (Di Gregorio, ). As a result input in the L is interrupted abruptly right after adoption in international adoptees. The extent and speed of attrition and actual total loss of the L is even more severe in adoptees than in immigrant children (Montrul, b). To summarize, for a language to develop, stabilize, and not regress, a critical mass of input and use is required during an extended period of time, including the span of the language development period, which does not automatically end at age and but continues throughout the period of schooling and later language development. Reduced input interacts with age: it affects the developing grammars of bilingual children more than that of bilingual adults.
. O
.................................................................................................................................. Native language development is a long process, if we take into account early emergence, acquisition, and mastery of simple and complex structures in production and comprehension, in addition to their subtle semantic and pragmatic implications. It takes at least thirteen to fourteen years, if not more, to achieve adult native levels of proficiency, and schooling plays a role in morpho-syntactic development in adolescence (Keijzer, ). A bilingual environment presents additional challenges since all these linguistic milestones must be achieved in two languages when the quantity of input in each language is not a hundred percent. When input is not optimal in quantity, many heritage speakers exhibit acquisition without target mastery of several aspects of their heritage grammars. Input factors and use of the heritage language in the immediate family and school context and in the broader socio-linguistic context contribute to the acquisition and development of specific grammatical properties of the heritage language grammar. How can the quantity of input be operationalized? In the previous section we noted that, in general, young adult sequential bilingual heritage speakers show stronger language acquisition and maintenance in several areas than heritage speakers who are simultaneous bilinguals, and this was operationalized as an age of onset of bilingualism effect (early in simultaneous bilinguals and later in sequential bilinguals). Naturally, the time and cumulative amount of exposure as a simultaneous or sequential bilingual cannot be disentangled from the quantity of input and the type of input required to reach full linguistic proficiency in the two languages. In a study of French-English bilingual children in Montreal, where
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
both English and French are valued and used widely in the community, Thordardottir () found that % exposure to each language is sufficient to develop the language at monolingual levels in pre-school age bilingual children. But the reality is that the amount of time or proportion of daily input a bilingual child is exposed to in the two languages can range from % to % in each language depending on the circumstances. The quality of the input and the contexts of language use also matter. The quality of input (and of output) refers to the richness of the language the child is exposed to in terms of diversity and complexity of structures and vocabulary (Jia & Paradis, ). This includes the type of vocabulary the bilingual is exposed to and actually uses, the specific syntactic structures used when speaking in a particular context or about a particular topic, and the type of discourse required depending on topic/context; that is, familiar and presentational versus descriptive, hypothetical, argumentative, etc. Examples of situations and activities that contribute to the quantity and quality of input in child heritage speakers are the percentage of time spent speaking the heritage language versus the non-heritage language; the number of different people with whom the heritage language is spoken; the percentage of time that the heritage language is used in leisure activities (i.e., playing games, reading books and magazines, watching TV and movies, or using the computer in the heritage language); and the frequency with which children attend activities conducted in the heritage language, such as playing with children who speak the heritage language, extracurricular activities, or weekend heritage language school. If a language is not needed in some context or for some purpose, the vocabulary and linguistic properties associated with the context or purpose may not develop, and if reading and writing skills are not needed in one of the languages, they will not be developed either. Because many heritage speakers do not receive schooling in their heritage language, they do not develop their language beyond basic, concrete vocabulary and the syntactic structures required to talk about past and present events. Some heritage speakers may later lose many of the basics learned. Even if heritage speakers continue to use the language throughout their life, the language is used in restricted contexts and may lack lexical and structural variety. Both the quantity and the quality of input are crucial for heritage language acquisition and eventual maintenance (Jia, ). Furthermore, the quality of input in a language contact situation can vary depending on the linguistic proficiency of the interlocutors. If the parents, for example, have lived in an immigration context for more than ten years, they may experience attrition in some aspects of their grammar. One area that has been shown to be subject to change in first generation speakers, for example, is the use of null and overt subjects in null subject languages (Nagy et al., ; Benmamoun et al., a; Benmamoun et al., b; Sorace, a; Tsimpli et al., ; among many others). Therefore, some heritage speakers may also be exposed to input different from the homeland variety, and in this input variety some changes may even arise due to attrition; the source of this input can be the language of their parents or other interlocutors in their social networks (Montrul, ). Another possibility is that some heritage speakers may be exposed to heritage speakers from the same language but a different dialect, and that the dialects may differ in a particular property. Otheguy & Zentella’s () study of pronoun expression in the Spanish of New York City is an example of this situation, because speakers of non-Caribbean dialects develop higher frequency of use of subject pronouns due to contact with speakers of Caribbean varieties and English. Although all heritage speakers report using the heritage language mostly with
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
their parents or grandparents, in some families heritage speakers use the heritage language with siblings and friends, who are also heritage speakers themselves. Therefore, the social networks of the heritage speakers, the density of the networks (number of interlocutors), and the degree of proficiency of the speakers in the network also contribute in important ways to the quality of input heritage speakers are exposed to. While the immediate family plays a significant role in heritage language acquisition, after children start school the family is not sufficient for continued language development (Kerswill, ). The broader socio-cultural and political atmosphere contributes to language development as well. According to Armon-Lotem et al. (), political forces, identity, attitudes, and socio-cultural and socio-linguistic preferences of the particular heritage language community also influence in significant ways the heritage language acquisition process. Even if the parents make an effort to use and transmit the heritage language to their children, the minority status and socio-political prestige status of the language in the broader society also play a critical role in the degree of bilingual development, and can influence the degree of proficiency achieved in the heritage language. As children grow, their peer groups and society at large become the main sources of input and attitudes towards the heritage language. Associated government and educational language policies directly impact the eventual level of ultimate attainment of the minority language in heritage speakers. Because minority languages do not typically have official status, do not have the same public presence as majority languages, and are often not used as the medium of instruction in schools, there are few opportunities to use the language beyond the home. In the modern day, more and more heritage speakers choose to re-learn their home language when they enter adulthood, for example through college classes. Heritage speakers seem to have some advantages in language classes as compared to L learners, but those advantages are typically limited to phonetic and lexical levels (Au & Romo, ; Au et al., ; see also Oh et al., Chapter , this volume). Heritage speakers of course excel in listening and some speaking but their literacy skills are lacking and sometimes take longer to develop than the skills of L learners. It appears that their re-learning process can be greatly enhanced by tapping into their existing knowledge, helping them systematize it and by exposing them to the full range of variation in their home language, which takes longer for L speakers to grasp. A good understanding of heritage language development, structure, and use is crucial for the development of heritage language re-learning methodologies, and new research has been emerging that calls for a direct link between the two fields of linguistic research and evidence-based pedagogical approaches to heritage languages in a classroom setting and beyond (Polinsky, ; Bayram et al., ). Other major factors related to socio-political status that also contribute to heritage language development are internal to the individual and have to do with attitudes and identity. When a language is not important in a particular society, parents may hold unfavourable attitudes toward their own language, leading them to neglect the language at home, impacting the heritage language development of their children. This is common with speakers of indigenous languages in parts of Mexico (Barriga Villanueva, ). On the other hand, often parents do not have negative attitudes toward their language, consider that the heritage language is valuable, and insist on its exclusive use at home. Still, the children may refuse to speak the language or even feel ashamed of the language because they are aware of its minority status in the broader society. For many immigrant children, reduced exposure and use of the heritage language begin as soon as they enter
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
kindergarten (Wong Fillmore, , ; Ellis et al., ; Shin, ), but the steepest decline is observed in early adolescence, ages to (Portes & Rumbaut, ). During this time, rejection of the home language is accompanied by feelings of embarrassment, frustration over the widening cultural gap with parents, and cultural isolation (Tse, , a,b). A move away from the heritage language and culture toward the majority language and culture is common once children start schooling and their main peer group consists of other children. In addition to their own feelings about their identity, some heritage speakers are often judged by elder speakers with higher proficiency in the language. Sherkina-Lieber et al. () report that one difficult feature of Labrador Inuit communities has been the negative attitude of fluent speakers towards non-fluent speech in Inuktitut. Older fluent speakers have been described as producing negative reactions to non-fluent speakers’ attempts to speak Inuktitut, so that non-fluent speakers reported being discouraged from trying. Embarrassment at their language abilities and the realization that their proficiency is not very high also affects heritage speakers’ willingness to use the language at home during the period of later language development. As learners’ identification with the heritage language and culture evolves, so do their language choices and competencies, which in turn may change how they relate to parents, siblings, neighbours, teachers, and friends as part of the socialization process. He () articulates how the ebb and flow of heritage language competencies throughout the lifespan are intrinsically linked to ideologies and language choices that change as heritage language learners grow up. The changes are conditioned by the bilingual learners’ motivations to use the languages, social networks, and opportunities to use the languages. The quantity and quality of input in the heritage language, which feeds the LAD and determines the degree of acquisition and proficiency in the heritage language, is ultimately linked to all these sociopolitical, ideological, affective, and situational factors, as exemplified in Figure .. Once we establish the relationship between the biological, cognitive, political, social, and affective factors that contribute to language development, we can see that there are several
Sociopolitical status Vitality of the language Access to schooling Affective factors (attitudes, identity) Linguistic practices at home and of social networks (peer groups) Input and use (quantity and quality)
Heritage language proficiency
.. Interrelated factors that play a role in heritage language acquisition and proficiency
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
causes of lower proficiency in the heritage language than in the majority language in many heritage speakers, and that these causes ultimately drive the structural changes at all linguistic levels. In previous work (Polinsky, ; Montrul, ) we have referred to heritage language grammars as ‘incomplete grammars’, the result of incomplete acquisition. We acknowledge that referring to a grammar as incomplete can be theoretically problematic if one considers that languages are always changing in some way (see Kupisch & Rothman, , for discussion). It is also hard to tell when individual grammars can actually be acquired completely, as pointed out by Otheguy & Zentella () and Meisel (). Even more unfortunate is the fact that the term ‘incomplete grammar’ can also unintentionally lead to a negative portrayal of the ethnic minorities who speak these languages.1 Perhaps more appropriate is to suggest that the process of incomplete acquisition can lead to a divergent or innovative grammar (see Flores, ; Scontras et al., , for the terminology and further discussion), different from the target grammar or baseline, due in part to insufficient input and use necessary to match the monolingual baseline (see also Chapter by Bayram et al., this volume, for a discussion of this issue). Before concluding, we would also like to offer some observations on the bicultural competence of heritage speakers. It is often assumed without much question that heritage language speakers are fully bicultural. Indeed, they may be limited in their knowledge of language, but they are culturally proficient. A couple of decades ago, when research on heritage languages was just starting, researchers were equally confident that heritage speakers were not much different from monolinguals. We now know that this is not the case; it is time to re-examine the bi-culturality assumption as well. Heritage speakers may know a great deal more about their heritage culture than an average outsider, but they are also limited in their knowledge of cultural norms, especially if those norms are connected with verbal communication. If they do not know how to use different registers they may not be aware of politeness norms (cf. Dubinina, ); they may lack culture-specific gestures or use them inappropriately; they often do not know cultural references of the home culture; they may be unaware of the conversation turn-taking rules in the homeland—this list can be continued indefinitely. As we get better at understanding the subtleties of heritage-language bilingualism, we should examine the bicultural status of heritage language speakers as well. In conclusion, because heritage speakers are a highly heterogeneous population from both a psycholinguistic and socio-linguistic point of view, age of onset of bilingualism, and quantity and quality of input lead to acquisition and mastery of some grammatical areas and attrition or incomplete mastery of others. All of these factors, or a combination of them, contribute to the particular language development of these individuals. The chapters in this section discuss in more detail theoretical approaches and research designs aimed at identifying and isolating these factors. For example, Chamorro, Sturt, & Sorace (, p. ) refer to ‘incomplete heritage speakers’, an unintentional but unfortunate description of these bilinguals. Of course that echoes the notion of ‘semispeakers’ or ‘quasi-speakers’, introduced by Dorian () and still used in studies of endangered languages. Some degree of attrition or reduced mastery of grammar in endangered languages is often taken for granted, but parallels between endangered languages and heritage languages as spoken by bilinguals in large societies still remain to be explored. 1
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Three chapters in this section (Part VI) address the nature of the quality and quantity of input received by heritage speakers. Sharon Unsworth (Chapter ) emphasizes that the quantity of input may vary in a significant way, and that this in turn has an effect on the linguistic development of a heritage speaker. Echoing the comments made earlier in this chapter, Fatih Bayram and his co-authors (Chapter ) call the reader’s attention to the fact that immigrant language (the language of the baseline, which serves as input for heritage language acquisition) already has a number of differences from the language in the homeland, including some differences due to genuine attrition. Heritage speakers only amplify these differences, which results in observable changes in their grammar and usage. In her chapter (Chapter ), Tanja Kupisch also considers the quantity and quality of input. Crucially, she points out that the factors responsible for shaping heritage grammars also include the age of onset effects caused by a loss of brain plasticity and lateralization, distance to the homeland, similarities and differences between the two languages in the bilingual dyad, and the degree of language dominance. Her compelling conclusion is that all of these factors need to be taken into account together; otherwise, we stand the risk of oversimplifying heritage language data and missing important generalizations. Three other chapters focus on particular populations that need to be taken into account in heritage language research: early childhood speakers (overhearers), international adoptees, and international returnees. The chapter by Janet Oh and co-authors (Chapter ) examines the potential benefits of early childhood exposure to language for later language (re)learning among immigrant-background adults. Lara Pierce et al. (Chapter ) address the degree to which international adoptees retain their first language (if at all). Understanding what aspects of language undergo attrition and what is retained by international adoptees sets the stage for determining the neural substrates available for language learning/processing and the constraints under which language is acquired at any given stage of development. Cristina Flores (Chapter ) provides an overview of research on bilingual returnees: ex-migrants, who have lived for an extended period in a migration context and have, at some point in their life, returned to their homeland. Flores highlights the research potential that has yet to be explored in the study of their language. In particular, their data can shed light on the role of extra-linguistic variables, such as quantity and quality of exposure, age, and literacy skills on language development.
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
......................................................................................................................
......................................................................................................................
. I
.................................................................................................................................. B children’s heritage language (HL) experience may vary in many different ways. For example, some children hear the HL from both parents whereas others from one parent only, some children grow up in a wider HL-speaking community whereas other families live in relative isolation, some children mostly hear input from attrited HL speakers whereas others do not, and some children have access to schooling in the HL whereas others do not. To understand the impact of these and other different experiences on children’s HL outcomes, researchers typically collate and quantify specific aspects of children’s language input, transforming or reducing them into other, more general, variables, such as language richness as a measure of input quality and amount of language exposure as a measure of input quantity. As Montrul (, p. ) notes, ‘a problem with using input as a key variable to explain much of language acquisition is that it is difficult to operationalise’. This chapter presents an overview of the most frequently used method of operationalizing language experience in bilingual language acquisition research, namely the parental questionnaire. It is organized as follows: after providing a brief summary of the main experiential factors known to affect bilingual/HL development (Section .), I outline some conceptual and practical issues surrounding parental questionnaires as a means of quantifying bilingual language experience (Section .), before reviewing a number of questionnaires used in recent studies in more detail (Section .). The focus is on bilingual/HL development in childhood and as such, questionnaires designed for use with adult HL speakers (see, e.g., work by Naomi Nagy) are not included.
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
. T ()
.................................................................................................................................. In recent years, a wealth of studies have demonstrated the impact of various experiential variables on bilingual children’s (rate of) language development. Most studies on this topic, however, focus on bilingual children’s development in the societal (i.e., majority) rather than (or at least in addition to) the HL (see Unsworth, a, for a detailed review). In this section, I highlight the main findings relevant for HL development in order to contextualize the questionnaires discussed below and highlight potential avenues for further research. Various recent studies have shown that HL children’s rate of acquisition is related to the amount of HL input to which they are exposed. This has been observed across different linguistic properties (e.g., verbal and nominal morphology, vocabulary) and for different populations (e.g., Spanish-English, Welsh-English) (e.g., Gathercole & Thomas, ; Hoff et al., ). The positive contribution of parents’ HL at home to children’s development in that language has also been observed; continued parental HL input seems especially important once children start schooling in the majority language (e.g., Dixon et al., ; Duursma et al., ), although it may not necessarily guarantee children’s continuing HL development (Sheng et al., ). The amount of societal language input at home has been found to have a greater, negative impact on the HL than it has a positive impact on the acquisition of the societal language (De Houwer, ; Hammer, Davison, Lawrence, & Miccio, ; Hoff et al., ). Furthermore, the effect of parental use of the societal language at home may be modulated by the proficiency of the input provider (Goldberg et al., ; Chondrogianni & Marinis, ). Whilst the differential effect of variation in input quantity on the (rate of) acquisition in many domains of language is undisputed, it is also clear that this relationship is non-linear in nature. In other words, threshold effects exist (e.g., Gutiérrez-Clellen & Kreiter, ; Cattani et al., ; Thordardottir, ) and not all linguistic properties are affected equally (Oller et al., ; Chondrogianni & Marinis, ; Unsworth, , b; cf. Paradis, ; Hoff et al., ). Research on the impact of variation in input quality has for the most part been restricted to studies on societal language development, with the notable exception of the studies by Place & Hoff (, ) and La Morgia (). For example, it has been shown that input from native speakers is associated with more advanced vocabulary in the societal language, even after controlling for amount of input in that language (Place & Hoff, , ; see also Driessen et al., ). Furthermore, linguistically rich input (e.g., from different sources, including TV and reading) predicts rate of acquisition (e.g., Paradis, ), and the number of speakers providing input is also a significant predictor of children’s (HL) development (Place & Hoff, ). One other important aspect of HL children’s language experience is how much they use the language themselves. Indeed, children’s own HL use has been found to predict language outcomes in a number of studies (Montrul, ; Bohman et al., ). Furthermore, whilst
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
output and input are clearly related to each other, they may nevertheless relate to outcomes differently (Bohman et al., ; Unsworth, ). To summarize, recent research has shown that variation in input quantity impacts on rate of acquisition in vocabulary and morpho-syntax, though not for all age groups, or all contexts (see Unsworth, a, for an overview). Furthermore, input beyond a certain level does not facilitate further development and more qualitative properties of language input, such as whether it is from native speakers or from different speakers, have also been found to predict variation in children’s outcomes. On the whole, however, most of the research on input effects in child bilinguals is concerned with development in the societal language, with only a handful of studies focusing on children’s HL development.1 In other words, whilst there are a number of relatively robust findings concerning the impact of experiential factors on bilingual children’s language development, there is certainly scope for further research, especially with respect to language use and the more quality-oriented factors, and with respect to development in the HL. The remainder of this chapter is concerned with the tools one might use to conduct such research.
. H /HL
.................................................................................................................................. There are a number of conceptual and practical issues which the researcher needs to consider when deciding how best to quantify bilingual/HL experience.
.. Conceptual issues From a conceptual point of view, the first step is to decide where experiential variables— whichever of the specific variables discussed above these turn out to be—fit into the research design. Are they independent variables (i.e., predictors) or control variables? For example, if the locus of interest in a given study is whether input quantity or input quality has a greater impact on children’s HL development, experiential variables such as amount of weekly exposure and contact with native speakers could be included as independent variables. On the other hand, if the study’s goal is for example to examine cross-linguistic influence across two groups of bilingual children with different language combinations (e.g., Spanish-Swedish with German-Swedish), experiential variables such as amount and richness of input should be included as control variables. In this latter example, then, bilingual experience is used to ensure that the two bilingual groups are comparable such that any differences between the two on the study’s dependent variable may plausibly be attributed to the independent variable of interest (i.e., the role of the ‘other’ language). In pinpointing the experiential variables of interest, it is important to bear in mind that many such variables often strongly correlated with each other or with other variables of There are of course many studies in the field of bilingual first language acquisition more generally in which both languages are assessed (see, e.g., work by Meisel, Müller, and Hulk, amongst many others). Typically, however, these have not focused on the role of input as discussed here. 1
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
interest. Perhaps the most clear-cut case of this problem of multicollinearity is the interdependence between age of testing, age of onset, and length of exposure. This is a well-known issue in second language (L) acquisition research (e.g., Muñoz, ), but it is also relevant to HL acquisition. For example, in HL children, age of onset to the L will likely be correlated with the extent of early HL exposure, and hence an effect of (extensive) early HL experience may just as easily be interpreted as an effect of later L onset. Similarly, and just as in monolinguals, experiential variables such as (absolute) amount of input may correlate with other external factors, such as socio-economic status (SES) (Hoff, ). Researchers in bilingual child language research have addressed the problem of multicollinearity in different ways.2 For example, in a study on the impact of various internal and external factors on the child L acquisition of English, Paradis () observed a strong correlation between age of onset and months of exposure (r = –.), and consequently decided to include only the latter variable in her regression analyses. In the results, months of exposure turned out to be one of the significant predictors for children’s vocabulary and verb morphology scores; Paradis (, p. ) notes that this is in line with earlier research finding an effect of length of exposure. However, given the multicollinearity with age of onset, we cannot rule out that these results may in fact reflect the effect of this latter variable instead. Blom and Bosma () adopt a different approach: in their study on the acquisition of vocabulary and verb morphology in Frisian-Dutch bilingual children, they regressed length of exposure on age of onset and used the residualized (or decorrelated) values in their subsequent analyses (cf. Wurm & Fisicaro, , for the problems and pitfalls in using residualization as a means of addressing multicollinearity). One final conceptual issue to consider when deciding how to quantify bilingual/HL experience using parental questionnaires is whether relative or absolute measures of input are required. Most of the studies reviewed in Section . correlated relative measures of language experience (e.g., % English input, % Spanish input) with absolute measures of language development (e.g., standardized language scores). Grüter et al. () illustrate how this relationship may change considerably when absolute measures of language experience (e.g., number of words heard each day) are used instead. For example, two children may have the same relative exposure to their HL, but the number of words they hear may differ, depending on—amongst other things—the talkativeness of their inputproviders, and this in turn will likely impact on their rate of acquisition for certain properties of that language. As Carroll (, p. ) notes, deriving estimates of language input from how many hours children spend with various interlocutors and the languages these interlocutors speak to children is problematic because such ‘[t]emporal units are crude measures of exposure and they tell us nothing about input’. Whilst the tone of her conjecture is probably too categorical, Carroll raises an important point, pertinent to the present discussion, namely that individual variation in terms of how much input is heard in a given hour cannot be captured using methods such as parental questionnaires. This is something that should be borne in mind when evaluating the results of studies using parental questionnaire data and the claims being made (see also De Houwer, , for discussion of absolute and relative measures).
2 Multicollinearity exists when two or more independent variables are strongly correlated with each other such that the variance they explain in a multiple regression analysis is largely shared.
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
.. Practical issues One of the most practical considerations in selecting a parental questionnaire relates to the resources available, not only with respect to money but also in terms of time, and not only for data collection but also for analysis. Like anything else, the potential costs of collecting detailed data on children’s HL experience should be weighed up against the benefits in terms of the relative contribution said data will make to the overall research design. In order to make such a cost-benefit analysis, it is important to consider whether more general or specific measures of HL experience are needed, and what the minimum level of detail is that is required to answer the research question. For example, in studies where amount of HL experience is a control variable, a general measure will largely suffice, whereas in studies focusing on the impact of variation in HL input more detailed measures will clearly be required. Other factors that may contribute to data collection costs are whether the questionnaire can be completed by parents only, and whether a research assistant is needed. In this latter case, it may be necessary for the research assistant to be a member of the relevant HL community in order for data collection to be a success. Whether the questionnaire is completed on paper, online, or via an interview (either face to face, via telephone, or through the Internet) will not only contribute to data collection costs, it will also impact the time and money needed for data analysis. For example, many online survey services deliver data in an analysis-ready, spreadsheet format. Finally, it is of course important to consider whether a questionnaire is available in the target language(s) or whether it will need translating, something that will again impact overall (time/money) costs. Another important practical consideration when selecting a parental questionnaire is the characteristics of the target population. This may include factors such as parents’ level of education, literacy, numeracy skills, and access to the Internet, their proximity to you, and whether they will be present during the child’s test session, in which case parent and child data collection can be combined.
. S
.................................................................................................................................. This section provides an overview of a number of available parental questionnaires.3 All are freely accessible (either in appendices, online, or directly from the authors) and all have been used in published studies. They have each been designed to collect information about children’s bilingual language experience in general, that is, their experience with the L as well as the HL. There are two broad types: (i) questionnaires that involve (automatic) calculations and usually result in composite scores, and (ii) questionnaires which capture raw data only. Table . provides information about how each questionnaire is administered, its output, the population for which it was designed, and the language(s) in which it is available. 3 This overview is not intended to be exhaustive. Other questionnaires are for example available via the National Heritage Language Resource Centre (http://nhlrc.ucla.edu/nhlrc/data/questionnaires).
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
In the remainder of this section, I discuss the relative advantages and disadvantages of the questionnaires given in Table .. Whilst the questionnaires differ in emphasis, most are designed for use with young (toddler/pre-schooler) children and use a series of questions about the child’s daily routine to derive composite measures of current exposure to and/or use of their two (or more) languages. Most, if not all, have been used in studies exploring bilingual children’s exposure in their two languages and its relation to language development, although these studies rarely explicitly refer to children as HL speakers. The exception to this is Cattani et al.’s () questionnaire, which focuses on early sequential bilinguals’ L acquisition; however, given that the percentage exposure score is relative, their questionnaire could quite easily be adapted to address HL development.
.. Administration With respect to how the questionnaire is administered, most can be distributed to participating families for completion at home, whereas others (e.g., ALEQ, BiLEC, and LEQ at least in their current forms) involve the parent being interviewed by a research assistant, cultural broker, or speech-language therapist (SLP). The Language Input Diary (LID; De Houwer, ) differs from the others in that it involves parents keeping a record of children’s patterns of language exposure one day a week over a seven-week period. Using a paper booklet, parents note down every minutes who is with the child and which language they are using. By collecting information from Monday in week , Tuesday in week , etc., the researcher builds up a picture of an ‘average’ week’s language exposure. Parents can be reminded via text messaging, email, or WhatsApp when they need to fill in the diary. Face-to-face administration of a questionnaire is time-consuming and may incur additional costs. It is however likely to result in more accurate data and it furthermore allows the researcher to check that parents understand the questions being asked and to ensure that the questionnaires are completed in the same way across participants. For some populations, for example parents with poor literacy skills, an interview may in fact be the only way to collect these data. Needless to say, the questionnaires designed for the parent to self-complete could also be adapted to the interview format, should this be preferred. Whilst such interviews can, if necessary, be conducted via the telephone or computer (e.g., using Skype), for some parents simply having to arrange a time for an interview to take place can be an obstacle to participate in a study. Most of the paper questionnaires in Table . could be adapted for digital administration using one of the many available online survey tools (e.g., Google Forms, SurveyMonkey, Qualtrics). One advantage of online administration is that it often delivers data that is readily usable for stats. It also reduces the risk of paper versions not being returned, getting lost, or damaged. Furthermore, from the point of view of the parents, it is more flexible, although it is quite often the case that participants need chasing up to enter their answers, and thus—in my experience—the completion rate is often lower than for face-to-face administration. Online administration is also of course only suitable for computer-literate parents and parents with reliable Internet access. One way of combining the advantages of both approaches is to have research assistants conduct an interview with the parents but enter data into an online (rather than paper) version of the questionnaire.
Questionnaire
Administration
(Main) output
Population
Language
ALEQ (Alberta Language Exposure Questionnaire; Paradis, )
Interview with research assistant Paper
Current language use, incorporating child’s input and output Language richness scores in both languages Calculated by hand
Primary school children, but could be used with toddlers and preschoolers Sequential bilinguals but could be used with simultaneous bilinguals
English
Cattani et al. ()
Parent Paper Optional Excel file, available from the authors
% exposure in typical week in last year of child’s life Calculated by hand or using Excel file
Toddlers, but could be used with older children Simultaneous and sequential bilinguals
English
Gathercole & Thomas ()a
Interview with research assistant Similar versions also available for parent to complete Paper
Current language use by parents and child Previous language use by parents and child Parents’ current language skills Frequency of contact with languages from other sources
Primary school children
English Welsh Spanish
Gutiérrez-Clellen & Kreiter ()
Parent Paper
No. of years’ exposure Current % exposure at home No. of hours of reading and other literacy activities in L Calculated by hand
Up to years, but could be used with older children Simultaneous and sequential bilinguals
English
Language Exposure Questionnaire (LEQ) (Hoff et al., )b
Interview with research assistant Paper
Current % exposure at home No. of hours per week of input in each language for range of activities and input-providers No. of monolingual / bilingual / native input-providers
Toddlers/pre-schoolers
English Spanish
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
Table .. Overview of parental questionnaires measuring bilingual language experience in children
Composite scores for various language/ literacy activities (how often, in which language) Calculated by hand
Toddlers/pre-schoolers Simultaneous and sequential bilinguals
Dutch Parts available in English
Parent or interview with research assistant Paper
Current language use, incorporating child’s input and output Language richness scores in both/all languages Current language skills in both/all languages Calculated by hand
Toddlers and preschoolers
English Dutch Others available via COST Action ISO
Interview with research assistant Excel file, available via IRISc
% current exposure Cumulative length of exposure % current output Average quality of exposure (nativelikeness) No. of different (native/non-native) speakers providing input at home No. of single language conversational partners Calculated automatically
Toddlers through years Simultaneous and sequential bilinguals
English Dutch
Mayo & Leseman ()
Parent Paper
PABiQ (Questionnaire for Parents of Bilingual Children; Tuller, )
BiLEC (Bilingual Language Experience Calculator; Unsworth, )
Various similar but not identical versions of this questionnaire are available from Ginny Gathercole’s webpage. The LEQ is available directly from Erika Hoff. c Go to http://www.iris-database.org and search for BiLEC. BiLEC was originally dubbed the Utrecht Bilingual Language Exposure Calculator (UBiLEC). Its name has since been changed to better reflect its content (and my current affiliation); its contents are essentially the same, however. b
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
Any
Parent Paper
a
Calculated by hand Who is with child and language(s) spoken in min slots for each day in the week
Toddlers Simultaneous and early sequential bilinguals Need to hear both languages at home
Language Input Diary (LID; De Houwer, )
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
The BiLEC and Cattani et al.’s () questionnaire both come with Excel spreadsheets where parents’ responses are entered and composite measures are calculated automatically. Cattani et al. () provide instructions on how to make the calculations by hand in the appendix to their paper, whereas the Excel file is the only means of using the BiLEC (available together with further instructions, a demo about how to administer it, as well as completed files for a range of example families via the online instrument repository, IRIS, cf. footnote (c) for Table ). It is in principle possible to collect parents’ responses using a paper (or online) version, and this has been done in the past, but the data still need to be entered into the Excel file in order to derive the composite measures for input quantity and quality.
.. Output The main output for almost all the questionnaires are relative measures of current exposure. The ALEQ and its derivative, the PABiQ, combine children’s input (i.e., language(s) spoken to them by the parents, siblings, and other family members) and children’s own language use (i.e., language(s) spoken by them to the parents, siblings and other family members) to create a composite measure (%) of current language use, whereas the questionnaires by Gutiérrez-Clellen & Kreiter () and Cattani et al. () do not include children’s own language use as part of their output. These two variables—children’s language input and use—are separated in BiLEC, as there is some evidence to suggest that they may be differentially related to children’s absolute and relative language proficiency (Bohman et al., ; Unsworth, ; see also Paradis, ). In addition to a measure of current language exposure, the BiLEC also includes a measure of language exposure in the past, cumulative length of exposure (inspired by questions in Gutiérrez-Clellen & Kreiter’s () questionnaire). Unlike traditional measures of length of exposure, which are based on how many years’ contact the child has had with a given language, cumulative length of exposure takes into account how much exposure a child had with said language during each of these years. For example, if we take two -year-old children who have been exposed to their HL at birth, we would traditionally say that they have had five years of exposure to the HL. These two children’s language learning histories may, however, have been very different. Child A may have stayed at home with her HL-speaking mother until starting school at age , whereas child B attended day-care for four days a week in the societal language from age . Clearly, these two HL children have had quantitatively different HL experiences; calculating previous language exposure cumulatively captures (some of) these differences.4 Gathercole & Thomas’ () questionnaire also asks about children’s language input and use before entering school and in the early primary school years.
4 In her studies on simultaneous French/English bilinguals, Thordardottir () calculates overall exposure in a similar way, such that the resulting percentage reflects the proportion of the child’s time since birth spent in an English (or French) environment (i.e., comparable to what Unsworth () calls cumulative length of exposure), rather than the proportion of current language input.
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
Several of the questionnaires include measures of language richness. In the ALEQ (and similarly, its derivative, the PABiQ), parents are asked about the frequency of children’s participation in language-related activities, such as reading, using a computer and watching TV, their involvement in any extra-curricular activities in the HL, and their language use with friends. The answers are then combined to calculate a richness score in the HL and the L. Gathercole & Thomas’ () questionnaire, the LEQ and Gutiérrez-Clellen and Kreiter’s () questionnaire all collect comparable information and the latter also adds this up to create a total number of hours rather than a composite score. The types of extracurricular activities and other sources of language exposure covered in ALEQ’s language richness measure are outputted separately in the BiLEC (as the relative proportion of other sources of language exposure in a given language) or factored into the overall proportion of weekly language exposure.5 Both the BiLEC and the LEQ directly include a measure of language quality in their output, namely native-likeness. In the BiLEC, the parent being interviewed is asked to evaluate his/her proficiency in the languages in question on a scale of (no fluency) to (native fluency), using a number of descriptors to illustrate each point on the scale. The parent is also asked to estimate his/her partner’s proficiency as well as the proficiency of all other significant input-providers. Each input provider’s contribution to the overall score is weighted according to the time they spend with the child in an average week to arrive at a composite measure indicating the overall nativelikeness of children’s input on the same scale ( to ). Other qualitative measures in the BiLEC include the relative proportion of input at home from native speakers and the number of different native and non-native speakers providing input at home, both following Place & Hoff (), who use Hoff ’s LEQ to derive the same variables (by adding up the total number of native/non-native/bilingual speakers inside and/or outside the home providing input to the child). In addition, the BiLEC also calculates the number of single-language conversational partners, that is, the number of input-providers at home with whom the child the child interacts in a given language only (following Place & Hoff, ). Mayo & Leseman’s questionnaire, based on the HOME observation scheme by Bradley & Caldwell () and used in a study on Turkish-Dutch and Moroccan-Dutch bilingual children by Scheele et al. (), differs slightly from the others in that its focus lies on children’s bilingual language experience via reading activities and oral interactions rather than an overall picture of language exposure. In a sense, then, this constitutes a more elaborate version of the richness measure in for example the ALEQ. The measures of language exposure outputted from most of the questionnaires in Table are relative in nature. If desired, however, it should however be possible to extract absolute measures from these questionnaires as well. The output for De Houwer’s () LID is essentially a (rich) set of raw data rather than a composite measure of language exposure. Examples of possible calculations one could make with these data—absolute and/or relative—can be found in the literature (e.g., Place & Hoff, ).
5 In fact, the BiLEC includes four different measures of current language exposure in its output, namely: percentage current exposure at home, percentage current exposure at home and at daycare/ school, percentage current exposure at home and at daycare/school plus other sources, and percentage current exposure at home and at daycare/school plus other sources and holidays.
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
.. Population As noted above, most of the questionnaires are designed for use with toddlers and preschoolers, with some extending to primary school age and beyond. Because the LID is solely concerned with language input at home, it is only suitable for use with toddlers/preschoolers who do not attend institutional care. Whilst some of the questionnaires (e.g., Cattani et al., ) are clearly designed for use with (early) sequential bilingual children, focusing on L rather than HL development, most collect information about children’s overall language experience and hence would be suitable for research concentrating on HL rather than L development, and for use with both sequential and simultaneous bilinguals. Even when the primary focus of a study is on development in HL only, it is always sensible to collect information about children’s patterns of language exposure more generally, including to the L, as this will likely be of importance. In their current form, almost all of the questionnaires are designed for use with children growing up with two languages only, though most could be adapted for trilinguals (the PABIQ for example illustrates how the ALEQ could be modified for this population). The LID can in principle be used to collect data from children growing up with any number of languages, and the automatic calculations performed by the BiLEC are for up to two home languages in addition to the societal language. Given that the amount of information gathered about the home language is limited in Cattani et al. (), their questionnaire is not suitable for use with trilingual children.
.. Languages All the questionnaires are available in English. Early versions of the BiLEC have been translated into Dutch as well as a number of other languages, such as French, Italian, and Greek. As part of the COST Action ISO, the PABiQ has been translated and adapted for use in a number of different languages and linguistic contexts. Most of the other questionnaires should be relatively easy to translate, with changes made to make them culturally appropriate where necessary. For example, the age at which children are likely to attend day-care will depend on a country’s standard maternity provision and what is considered socially acceptable in terms of how often children attend, making questions on this topic superfluous in some contexts.
. A
.................................................................................................................................. All of the questionnaires discussed here rely on parental report, some over longer periods than others. Parental report has been shown to be a valid and reliable means of collecting data about language and language-related behaviour (e.g., Marchman et al., ; Rodriguez et al., ). Retrospective parental report, that is, recalling child-related information over a longer period of time, has been observed to be a good proxy (e.g., for birth weight) according to some (Walton et al., ) but not according to others (e.g., Jaspers et al., ). It is likely that parental recall will be better for younger than for older children.
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
However, Gilger () found that parents’ ability to retrospectively report academic achievements were not adversely affected by children’s age, that is, reports were just as accurate in predicting actual test scores for older children as for younger children. This ability will no doubt vary from parent to parent and to a certain extent depend on the parent’s (active) involvement in children’s early years. For this reason, it is important to ensure that a questionnaire is completed by the parent most likely to provide accurate information about language exposure and use in the early years. Most of the questionnaires ask parents to estimate how much they use each language with their children and several also ask for self-reported proficiency. Clearly, parents’ ability to make such an evaluation will in part depend on their own level of linguistic awareness and proficiency in the language. For this reason, the ALEQ (and based on this, the BiLEC), makes use of descriptors which are easy to evaluate, such as ‘can answer the phone in English’ for limited fluency, or ‘can go to the doctor and explain what is wrong’ for the category somewhat fluent. In a recent study, we collected production data from non-native speaker parents (n = , range of home languages and SES), had these evaluated by native-speaker judges for accent, grammar, vocabulary, fluency, and overall proficiency, and compared the results with the same parents’ self-reported proficiency. The results (not yet published) were quite positive: the correlation between the native-speaker judgements and parental self-report was moderate (and approaching strong), r = ., p < .. In other words, for this sample at least, self-report appeared to be a reasonably accurate estimation of parental proficiency in a non-native language. One other factor to consider when using parental questionnaires to gather information about bilingual children’s patterns of language use is the danger of parents providing socially desirable answers. For example, when asked whether they speak the societal language to their child, parents who are non-native speakers of that language may respond positively, not because this is really the case, but because they feel under pressure to do so (e.g., from teachers, healthcare professionals, family members, wider society). Whilst this social desirability bias is a general problem with self-report, careful consideration of the norms and pressures within the target population, as well as neutrally formulated questions may reduce its impact.
. C
.................................................................................................................................. To the extent that they do not do so already, all studies on children’s HL development should include basic level of information about children’s current and previous patterns of exposure. Which of the available parental questionnaires should be selected to gather these data, and the level of detail to be obtained, depends on the specific goals of the study and the population involved. It is hoped that the overview provided in this chapter, whilst not exhaustive, will be sufficient for researchers working with HL children to make an informed decision in this regard.
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
......................................................................................................................
- Contributions to heritage speaker competence ......................................................................................................................
, ,
. I
.................................................................................................................................. F at least two decades, scholars interested in the field of bilingualism have paid increasing attention to a particular subset of bilinguals, namely heritage speakers (HSs; see Benmamoun et al., a; Kupisch & Rothman, ; Montrul, a; Polinsky, , for updated reviews of the literature). This chapter focuses on the potential role intragenerational attrition may contribute to the ultimate attainment outcomes of HS linguistic competence in adulthood (e.g., Sorace, ; Rothman, ; Pascual y Cabo & Rothman, ; Pascual y Cabo, , ; Montrul & Sánchez-Walker, ; Schmid & Karayayla, ; Karayayla, ). By definition, all HSs are bilingual but not all bilinguals are heritage speakers. That is, HSs have—on a continuum—grammatical and communicative competencies in more than one language, but the socio-linguistic environment under which HSs develop these competencies is ultimately what qualifies an individual as a HS or not. Although there are many definitions of what a heritage language is, who heritage speakers are, and how/why they most often differ from baseline monolingualism, we adopt the one below precisely because of its neutrality with respect to what differences entail a priori. 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 the larger (national) society. Like the acquisition of a primary language in monolingual situations and the acquisition of two or more languages in situations of societal bilingualism/ multilingualism, the heritage language is acquired on the basis of an interaction with naturalistic input and whatever in-born linguistic mechanisms are at play in any instance of child language acquisition. Differently [from monolingual acquisition], there is the possibility that quantitative and qualitative differences in heritage language input, the
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
-
introduction and influence of the societal majority language and differences in literacy and formal education can result in what on the surface seems to be arrested development of the heritage language or attrition in adult bilingual knowledge. (Rothman, , p. )
On this view, a HS is a bilingual individual for whom the language of the home environment— the heritage language (HL)—differs from what is spoken in the society as the main language. Montrul (a, p. ) divides HSs into the following three subcategories: (a) simultaneous bilinguals, those exposed to the HL and the majority language before the age of 3 to 4; (b) sequential bilinguals or child L learners, those exposed to the HL (virtually) exclusively at home and later to the majority language once they start preschool at the ages of 4 to 5; (c) late child L learners, children monolingual in the HL, who receive some elementary schooling in their home country and immigrate around ages 7 to 8. It is important to keep in mind that under any of the above scenarios (a) to (c), the HL is always the first language (L; potentially one of two as in (a)). However, even though the HL can therefore be considered the native language of these speakers (Rothman & TreffersDaller, ), the outcomes of HL acquisition show a wide range of language dominance, with the majority of HSs becoming dominant in the majority language, not their native L (e.g., Montrul, a). Although there are some notable exceptions (see Kupisch & Rothman, ), where proficiency in the HL is concerned most studies find clear differences between HSs and age-matched monolinguals (e.g., Montrul, , a; Benmamoun et al., a). As in other areas of L attrition research, the fact that the native language of certain bilingual populations can exist on a gradient of language dominance is what lends the study of HLs its unique potential to provide insights into bilingual development. A particular point of interest, then, concerns identifying and understanding the conditions and key variables related to the context of HL acquisition that lead individual HSs to end up at particular points on the dominance-gradient. The observation that HS competence and performance outcomes exist on such a large continuum of variability forms the basis of what we refer to as the Heritage Bilingual Paradox (HBP). At its crux, the HBP rests on the following question: what accounts for the fact that, in a HL context, native naturalistic child acquisition can result in large disparities in grammatical competence between HSs and monolinguals as well as across subsets of HSs? At present, there are four main, partially overlapping proposals debated in the literature that likely contribute to HS outcomes. The first relates to so-called arrested development or incomplete acquisition (Montrul, , a). Essentially, it is not coincidental that HSs tend to switch language dominance roughly around the onset of their schooling since this timing coincides with sharp decreases in exposure to and shifts in preferences for using the HL. School will be in the majority language and new relationships for the child are typically formed in the L from that point moving forward. Since several aspects of the language are not fully mastered by the age of to , the idea is that this sharp decrease in access to HL input and in HL use marks a turning point where the L/HL may cease to develop or lag behind monolingual development and thus may remain stunted or incomplete.
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
, ,
Another proposed contribution to the HBP is individual attrition, that is, the erosion or loss of previously acquired linguistic properties as reflected in the competence of adult HSs (e.g., Polinsky, ). On this view, HS children acquire (at least some domains of ) the HL age-appropriately in early childhood but have difficulty in maintaining parts of the system into adulthood. This approach is supported by evidence of adult HSs who do not have native-like competencies of properties that HS children on comparable conditions have, as was the case in Polinsky’s () study on relative clauses in HL Russian in the United States. In this study, monolingual children and adults as well as HS children and adults listened to questions consisting of a relative clause and were asked to match them with pictures. The results showed that only adult HSs differed whereby monolingual children, monolingual adults and child HSs performed similarly. To the extent that a cross-sectional design can capture this, child HSs of Russian seem to have acquired Russian relative clauses completely but evidence from the adult HSs seem to suggest that the property is not fully maintained into adulthood. Putnam & Sánchez () offer a distinct model of HL development. While they take very seriously the observable data that form the basis of incomplete acquisition and individual attrition as primary explanatory factors for HS competence differences, they claim that such approaches fall short in modelling the actual trajectories of HL development. Alternatively, they suggest that HS outcomes largely reflect a path of alternative, yet complete development. Essentially, the idea is that there is a discernable point in the HS developmental continuum—and this will vary from individual to individual depending on exposure to input and ability to intake input—where the path diverges from that of monolinguals. Along that path, it could be the case that previously acquired linguistic representations become altered in a way that is commensurable with exposure to the L; features of the L begin to affect the representations of the L, either in the sense of promoting reconfiguration of features (or feature reassembly) that had already been acquired or by altering what the shape of the intake is for the continued acquisition of properties in the HL variety being developed. Putnam and Sánchez (: pp. –) offer four specific stages to the process as they model it: Stage : Transfer or re-assembly of some functional features (FFs) from the L grammar to L phonological form (PF) and semantic features which may coincide with the activation of L lexical items on a more frequent basis from the standpoint of linguistic production; Stage : Transfer or re-assembly of massive sets of FFs from the L to L PF and semantic features, while concurrently showing significantly higher rates of activation of L lexical items than L lexical items for production purposes (i.e., they might code-switch more than bilinguals in the previous situation); Stage : Exhibit difficulties in activating PF and semantic features (as well as other FFs) in the L for production purposes but are able to do so for comprehension of some high frequency lexical items; and, Stage : Have difficulties activating PF features and semantic features (as well as other FFs) in the L for both production and comprehension purposes.
The four stages are not an inevitable outcome for all HSs, of course. In their view, Stages to are structured in an implicational hierarchical manner (those who wind up in Stage
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
-
outcomes should have passed through Stages and first) and reflect consequences of input qualities and HL intake potentials. And so, the stages offer testable predictions as to what trajectories of alternative HL development paths should be depending on relative opportunity for monolingual-like acquisition in the first place. The outcome of HL grammar, however different from monolingual outcomes, are thus taken to be complete grammars par excellence. The stages reflect what observation demands; they describe both the actual continuum of successes and abilities we witness in the variation of HS outcomes presented in the literature as well as a model that gives some explanatory adequacy regarding how the continuum emerges. A fourth proposal puts a special emphasis on understanding (as much) HL differential outcomes (as possible) on the basis of input differences HSs inevitably have compared to monolinguals. Research such as Rothman () and Pires & Rothman () concur in principle with the general premise of Putnam & Sánchez ()—complete acquisition along an alternative path—offering, however, a more specific discussion related to some specific variables that determine the change in path. Crucially, they assume that the input HSs receive is different not only in quantity, but also in quality from that of monolinguals, and that it is the interaction of these qualitative and quantitative differences that can explain at least some of the divergent outcomes observed among HSs. Their argument, labelled Missing Input Competence Divergence (MICD), essentially states that HSs cannot be expected to acquire properties for which the input they receive, differently to monolinguals, does not provide the same level of unambiguous cues (or cue validity) argued to give rise to monolingual representations. There are three implications from the MICD approach: (a) HSs will not differ from monolinguals for competence that reflects truly universal knowledge (because this is provided by the language faculty), (b) HSs should differ from monolinguals for domains of grammar that are only part of the standard variety imparted via schooling (Rothman, ; Pires & Rothman, ), and (c) HSs should show significant differences from baselines for properties that have undergone attrition in the grammars of their parents’ generation (Pascual y Cabo & Rothman, ; Pascual y Cabo, , ), because such changes necessarily affect the quality of input HSs have at their disposal. There is little question that all of these approaches have offered insights into the nuances and dynamics that underlie the HBP. Despite significant differences in how each one of these approaches understands the acquisition process, the terminology associated with it, and other matters of non-trivial importance (see Kupisch & Rothman, ), they are not mutually exclusive from one another and thus each approach likely explains—in isolation or in combination with other factors— some HL outcomes. It is important to acknowledge that none of the proposals are likely to explain the entirety of variation documented in HL outcomes as much as it is important to keep in mind that none have actually intended or claimed to be able to do so. The question is how much and specifically which HL data are best covered by each proposal. The remainder of this chapter, given the location of its appearance, focuses on one small area that contributes to our understanding of the HBP, namely the domain of intra-generational attrition in HS competence that has emerged from an increasing focus on the quality of input available to HSs (Sorace, ; Rothman, ; Pascual y Cabo & Rothman, ; Pascual y Cabo, ; Montrul & Sánchez Walker, ; Schmid & Karayayla, ; Karayayla, ).
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
, ,
. A
.................................................................................................................................. The definition of HSs given above specifies that they grow up learning a language in the home which is different from the majority language. Inevitably, this means that the (adult) speakers providing input will themselves be attriters or HSs. There may thus be a tricklingdown effect of individual attrition in older generations, as well as language change across generations in the immigrant community, affecting the input that HSs hear as their primary source of linguistic data and from which their grammars for the heritage language develop. In the case of individual attrition, for a linguistic property to be lost it has been argued that that same property must have been fully acquired by the individual before the process of attrition begins. Attrition in immigrant language communities, however, may be due to an inherited consequence of attrition in previous generations in what Gonzo & Saltarelli () referred to as the ‘cascade’ model of language attrition. Variation in HL competence is directly, although not uniquely, related to both qualitatively and quantitatively reduced input conditions (Rothman, ; Pires & Rothman, ; Kupisch & Rothman, ). In a heritage speaker community, the primary linguistic input for L acquisition comes from speakers who are themselves (descendants of) immigrants and have been in contact with the societal language for a long time. The input-providing generation is thus likely to diverge from the monolingual baseline to various degrees and in various domains. This means that the input provided to HSs may be different (in the case of immigrant providers, what they themselves were exposed to as children) from what children in monolingual environments receive. In such a setting, HSs are provided with qualitatively different building blocks from which their HS grammars can develop. In other words, although HSs’ linguistic competence may superficially resemble what has been found in language attrition, in some cases their knowledge and performance in HL may only be a reflection of their exposure to affected L primary linguistic input, which itself is an attrited version of the so-called standard form of the HL. The above possibility was first alluded to, albeit indirectly, by Sorace () in a short commentary to a keynote article by Montrul (a) appearing in the journal Bilingualism: Language and Cognition in which it was argued that differences in HL outcomes should be viewed as stemming from incomplete language acquisition. Sorace () pointed out that properties of the input which a learner receives will inevitably be reflected in their eventual knowledge and performance, an argument that is relevant for HSs who likely grow up listening to attrited grammars. Since , others converted her initial query/suggestion into an explicit link between HS competence and the linguistic context HSs are in as it pertains to trickling down attrition intra-generationally. Rothman () was the first to explicitly make the connection between Sorace’s observation—that is, properties of the input which a learner receives will inevitably be reflected in their eventual knowledge and performance—feeding into intra-generational attrition as a likely, primary cause of some— though not all—differences noted in HS competence outcomes. Entertaining the idea that we can explain some HS outcomes differences on the basis of input exposure asymmetries and that the implications of this can be meaningfully tested empirically provides yet
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
-
another way to get at the how and why of specific outcomes, or what underlies particular HS differences. Approaches that consider intra-generational attrition entertain the possibility that HSs acquire grammars completely for the HL—at least complete in some domains that are nonetheless different from monolingual norms—of an emerging contact variety type. All theoretical approaches to language acquisition, from the most stringently innatist to the most functional usage based, acknowledge the indispensability of input as the main driving factor for acquisition. Although some approaches place a stronger emphasis on the determinism of input, there is no question under any approach that differences in input will have consequences for the formation of ensuing grammars. For example, no one would expect a child growing up in the United States to acquire a British English grammar because the input provided to the child in the United States will not provide the cues that generate such a grammar. Although a British and an American child will both develop descriptively accurate grammars of English, they will differ from each other significantly across all domains (from phonology to syntax). It should not be controversial to state that comparing one to the other and claiming, arbitrarily, that the British child is incomplete because s/he did not acquire the American variety, or vice versa, does not do justice to the acquisition process and its outcome. In this light it might seem questionable that for several decades we have effectively done this in HS studies (see Pascual y Cabo & Rothman, ; Kupisch & Rothman, ) by comparing HSs to a control group of age-matched monolinguals knowing that the conditions under which each group acquires the same language is different, not only sociologically but potentially to significant degrees as it relates to the nature of the input they receive. But again, it is also clear that issues pertaining to the quality of input alone are unlikely to explain the full gamut of differences that characterize HS grammars. In the next section, we will review some recent studies that have sought to empirically test the idea of intra-generation attrition affecting heritage speaker competence in an effort to see what the evidence thus far indicates.
. E
.................................................................................................................................. First, we discuss two studies concerning Portuguese as a HL in the context of the US (Rothman, ; Pires & Rothman, ). Motivated by the idea that input quantity, input quality, and literacy play deterministic roles in HS versus monolingual differences, Rothman () attempted to test whether such a hypothesis could be supported via empirical experimentation. To this end, he examined knowledge of inflected infinitives in HSs of Brazilian Portuguese (BP) who grew up in the United States. Inflected infinitives, unlike uninflected infinitives, are typologically rare structures. As their name suggests, they are non-finite verbs, meaning they carry no tense, aspect, or modal information, yet they do carry overt person/number agreement, as seen in examples () and () below: () Eu acho (eles) conhecerem a cidade. [I (NOM st-SING) believe [CP they (NOM) to know (rd-PLU) the city]]. ‘I believe that they are familiar with the city.’
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
, ,
() O Scott vai comprar a pizza para (nós) podermos comer logo. The Scott (NOM rd-SING) goes to buy the pizza so that pro (NOM st-PLU) to be able (rd-PLU) to eat later. ‘Scott goes to buy the pizza for us to eat later.’ Inflected infinitives can only occur in complement or prepositional clauses (i.e., they are not possible in root or matrix clauses), licensed either by the prepositional head or the selectional properties of the lexical verbal head (the matrix verb). Although they must occur in complement clauses, as non-finite verbs they are incompatible with an overt complementizer (note the lack of a complementizer introducing the embedded CP clause in ()). Not all matrix verbs can select an inflected infinitive complement: Declaratives, factives, and epistemics can freely do so, while volitional verbs can only select finite subjunctive complement clauses to convey similar meaning types. As such, inflected infinitives are subject to the semantic interpretive properties of what is referred to as non-obligatory control (see Pires, ), whereas morphological infinitives are subject to obligatory control. In obligatory control contexts, either the subject or object of the matrix verb serves necessarily as the semantic subject of the controlled verb. As in () below, the infinitival clause itself is the accusative object of the matrix verb conhecer (‘know’), however, the understood subject is identical as the matrix subject because the PRO subject of the infinitive is necessarily controlled. () Eu acho conhecer a cidade. Ii (NOM st-SING) believe PROi to know the city ‘I believe I know the city.’ In the case of inflected infinitives, as in () and () and corresponding to the overt person/ number morphology on the embedded infinitival form, there is obligatorily a split reference between the matrix subject and the subject of the embedded inflected infinitive. When the subject of an inflected infinitive is null, it is taken to be a small pro with nominative case that has the same formal features as the overt morphology on the inflected infinitive. As a result, inflected and non-inflected infinitives have only partially overlapping distributions in use (and always with some interpretive differences). Both are restricted in use compared to finite forms, rendering a testable three-way distinction between uninflected infinitives, inflected infinitives, and finite forms. Inflected infinitives were chosen because, while they are commonly found in all dialects of European Portuguese (EP), they do not seem to form part of the colloquial BP, remaining only in high registers of standard BP (the language of education). In other words, inflected infinitives do appear in formal BP writing and in high register speech (e.g., newscaster talk) but they are not used in informal, quotidian speech where other structures are used to convey the same meanings. Rothman’s hypothesis was that since BP HSs are exposed almost exclusively to colloquial dialects of BP they would, therefore, not receive sufficient input related to inflected infinitives to acquire knowledge of them as part of their grammatical competence for BP. Accordingly, BP HSs should show a lack of knowledge related to constraints on inflected infinitival distribution as opposed to partial knowledge. Rothman () examined adult BP HSs without prior formal education in the HL and compared their data to that of educated BP monolinguals and formally trained advanced L
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
-
learners of BP. The purpose of the L group was to be able to exclude English-Portuguese bilingualism—some influence from English—as the main reason for differences between the monolingual and HS groups. Since the L group is exposed almost exclusively to formal, standard BP, unlike the HSs, the prediction following from the main line of argumentation was that they should show knowledge of the inflected infinitives. As groups, the results showed that BP natives and non-native L learners had knowledge of inflected infinitives, corresponding to the formal linguistic descriptions in the literature (see Raposo, ; Pires, ). The BP HSs, however, did not. In fact, they showed no apparent knowledge of any of the syntactic and semantic constraints that differentiate inflected infinitives from non-inflected infinitives and finite forms. The results indicated that the HSs did notice and process the person/number morphology of the inflected infinitive exemplars, but corrections they offered in the grammaticality judgement task (e.g., insertion of complementizers (que) and the like) reveal that they take these forms to be finite in nature. Instead of assuming a position of incomplete acquisition or attrition to explain the data, HSs’ undetermined knowledge was argued to result from the fact that the property itself was not sufficiently available in the input. The performance of the L group makes such a conclusion all the more likely. Why would non-native adults be able to acquire a property that HSs seemingly could not or could not maintain otherwise? Such a position is further supported by evidence of child L BP acquisition where knowledge of inflected infinitives is developed extremely late, around age , when significant exposure to this property is made available through a critical mass of education in the standard variety. In this sense, it is useful to highlight that the claim is not that BP children acquire inflected infinitives so late because education explicitly teaches them this structure. In fact, they are not taught this structure explicitly at all much like monolingual education of English for English native speakers tends not to teach such explicit structures. The main idea then is that the context of education where the standard dialect is the preferred lingua franca and is supported by published material written exclusively in the standard dialect provides a unique type of input monolinguals receive after school age. Properties lacking in colloquial dialects yet that form part of the standard can then be acquired by those, the typical monolingual and successful L learners, with sufficient exposure to this input. Such claims are not so different from what Dąbrowska (, ) argues for differences across monolingual English speakers that correlate to education level for properties such as passives and relative clauses; essentially, these structures are perhaps best understood as mainly pertaining to the standard variety and thus exposure to them via education contexts is necessary. While Rothman’s () data are compatible with the explanation described above, compatibility does not mean alternative explanations are impossible. In principle, it could be the case that the BP HS knowledge more accurately reflected arrested development of a property exemplified in the HS input or a later stage of individual level attrition of previously acquired inflected infinitives. In an effort to further test the hypothesis, Pires & Rothman () examined knowledge of the same property in a group of US-based EP HSs and compared the results to that of Rothman’s () BP HSs. Recall that all dialects of EP, formal or colloquial, instantiate and productively use inflected infinitives. As a result, irrespective of the EP dialect or register spoken at home, EP HSs have access to inflected infinitives in the environmental primary linguistic data. On alternative accounts, there would be no obvious reason why BP and EP HSs (socio-economic background, competence levels, and usage in the HL controlled for) should differ significantly, that is, both should
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
, ,
show differences from monolingual baselines. The same experiments, modified for dialectal differences, were administered to the EP HSs and the findings from both sets of HS groups differed substantially from each other. In stark contrast to the BP HS reviewed above, the data showed that EP HSs exhibited robust knowledge of this property without any statistical differences from EP monolingual controls (Pires & Rothman, , p. ). These results were taken to indicate that the source of variability in HSs competence, at least for this domain, could not be explained solely in terms of arrested development or HL–L attrition in the individual HS. Alternatively, it was claimed that HS differences more generally can also obtain as a result of a compounded effect that necessarily includes the differences in the input (i.e., monolingual vis-à-vis L attrition) HSs receive. Another study that more directly probes the possibility of intra-generational attrition affecting HS competence involves Spanish as a HL in the US (Pascual y Cabo, , ). While it is generally accepted that long-term Spanish speaking immigrants exhibit simplification tendencies in their speech and that subsequent generations of HSs tend to exhibit even more drastic innovations (e.g., Silva-Corvalán, ; Montrul, ; Cuza, ; Otheguy & Zentella, ), only a handful of studies have attempted to examine the extent to which the nature of the input HSs are exposed to can, at least in part, explain their path of linguistic development and ultimate attainment. Pascual y Cabo’s () study of Spanish dative experiencer verbs, also known as reverse psychological predicates stands out in this respect. Dative experiencers, despite being very frequent in the input, embody a possible domain for vulnerability due to the differences they present with regard to their word order and argument structure. Details aside (e.g., see Parodi, ; Montrul, ; Landau, , for complete, formal analysis/description), such verbs are unaccusatives and unlike Spanish agentive verbs, they have a post-verbal subject (i.e., ) that probes verb agreement. Additionally, the preverbal argument, an indirect object (i.e., ), is always (obligatorily) followed by a dative clitic whether or not the full DP object is fully spelled out. When the full DP is, however, it is obligatorily preceded by the dative marker a (as in a–b). () a. A Pau y a Teo les gusta el pan. To Pau and to Teo ( rd-) to them (.) like the bread. ‘Pau and Teo like bread.’ b. Les gusta el pan. to them (.) like the bread. ‘(they) like bread’ English differs from Spanish in that the verbal argument structure of like, the semantic equivalent of gustar. The theta roles are mapped onto canonical Subject–Verb–Object word order and verb agreement is controlled by the preverbal argument, as seen in (a–b). () a. Laurie likes cookies b. *Laurie like cookies Torrens and colleagues () found that gustar-like verbs are relatively late-acquired by monolingual Spanish children, whereas others have suggested that reversed psychological predicated impose difficulty to master by Spanish L learners (e.g., Quesada, ; Gómez Soler, ). Similarly, in the context of the US, these verbs have been typically found to
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
-
obtain differently among adult Spanish HSs (e.g., Toribio & Nye, ; de Prada Pérez & Pascual y Cabo, ). For example, among other findings, previous work had shown that Spanish HSs are not only prone to having the preverbal control verbal agreement (e.g., Silva-Corvalán, ), but also that they tend to omit the dative marker a in obligatory environments (e.g., Dvorak & Kirschner, ; Toribio & Nye, ). With the goal of determining the source of said outcomes, Pascual y Cabo (, ) examined production and comprehension data of dative experiencers in both child and adult HSs. Following the group pairing used in Polinsky’s () study on Russian relative clauses, Pascual y Cabo compared them to that of age-matched control child and adult monolingual speakers of the same dialect (i.e., Cuban (tested in Cuba) and CubanAmericans (tested in the US)). The idea is, as pointed out by Polinsky (), that in the absence of longitudinal data, such cross-sectional analysis allows (albeit indirectly) probing for child-to-adult developmental paths. In doing so, it is possible to draw conclusions regarding incomplete acquisition and attrition in the following ways: ‘if a child and an adult deviate from the baseline in the same way, it can be assumed that the feature has not been acquired’ (Polinsky, , p. ), that is, the HS difference is due to incomplete acquisition. On the other hand, ‘if a child performs as his or her age-matched baseline control but the adult does not, the feature can be assumed to have been acquired but may have subsequently been lost or reanalyzed’ (Polinsky, , p. ) or have attrited at the individual level. Polinsky simply was not considering other factors at the time, namely intra-generational attrition as the root cause. And so, to test, in addition to the other two, this possibility, Pascual y Cabo added a fifth participant group, namely first generation immigrants or the potentially attriting population of speakers who are the primary input providers to HSs. This methodological decision is an important and necessary point of comparison if one aims to more directly address the issues of input quality in HS bilingual development. Inclusion in this group required that participants had been born and raised in Cuba, had moved to the US after the age of , and had been in the US for at least ten years. In addition to a comprehensive language background questionnaire, participants completed three experiments: two grammaticality judgement tasks and a production task. The grammaticality judgement tasks were used to assess the participants’ sensitivity regarding the use of gustar-like verbs in different contexts and forms. In general terms, the findings from these tasks revealed not only that HSs have access to the canonical structure of gustarlike verbs in the same way that monolingual speakers do, but that they also avail themselves of options that are unavailable in monolingual grammars (i.e., case and verb agreement innovations). Particularly relevant for our discussion are the results from the elicited production task, which specifically tapped into case marking (the dative ‘a’ marker) as well as agreement configurations. The results of this task revealed that adult HSs’ production differed from both the monolingual and bilingual control groups in terms of case marking and subject–verb agreement, and crucially, that the adult first generation immigrants differed significantly from the adult monolingual speakers of the same dialect regarding production of the case marker a in obligatory contexts. It was precisely this difference that was used to explain the outcome obtained. In the absence of this obligatory dative marker, the child HS has no reliable cue in the input that s/he should not assign the canonical nominative case to the first argument of the sentence (i.e., the semantic subject). In turn, the absence of the overt marker a can induce a reanalysis of verbs like gustar that
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
, ,
culminates in the semantic subject being analysed as the syntactic subject, that is, a linguistically constrained reanalysis which accounts for all the so-called innovations in HSs’ treatment of gustar-like predicates. And so, what is noted in these data is a three-way distinction between monolinguals ➔ L immigrant input providers ➔ HSs, but one in which the changing (L attrition) shape of the middleman’s group production provides cues in HL input for reanalysis. As discussed earlier, because this variation/innovation (i.e., absence of dative marker a) is part of the input HSs have at their disposal, which is qualitatively different from the input children receive in monolingual environments, it follows that HSs might show significant differences from monolingual baselines without the need to postulate incomplete acquisition. In other words, insofar as the dropping of the dative marker a by the immigrant input providers allows for a reanalysis of the structural properties of this verbal type, shifting for instance from a strictly stative verb to an ergative/ transitive alternating verb in HS grammars these data lend empirical support to Putnam & Sánchez’s () approach as well as the very idea that intra-generational attrition has rippling effects in HS outcomes. Taken together the above-detailed studies support our hypothesis that, where the input to which HSs are exposed is qualitatively different from the input to which monolingual children are exposed—whether due to attrition in the parental generation or to a restriction in styles or registers represented in the input—this difference has inevitable consequences for ensuing HS competence. Similar observations do, of course, obtain even in monolingual settings (e.g., Dąbrowska, , ). That said, the sort of innovation processes discussed here becomes more pronounced and the linguistic changes more accelerated in language contact situations (e.g., Silva-Corvalán, ). At least one important difference between the two environments helps us to understand why this may be so. While in monolingual environments education/literacy functions as a means of linguistic standardization, in the bilingual environment, education in the HL is often not existent (Rothman et al., ). Literacy, thus, can have an effect on grammatical systems providing learners with input properties that are not exemplified in colloquial dialects, as was originally argued explicitly to explain the case for inflected infinitives in BP (Rothman, ) and more recently extended to bilingual outcomes more generally (Tsimpli, ; Kupisch & Rothman, ; Bayram et al., ).
. C
.................................................................................................................................. Studies in heritage speaker bilingualism have generally focused on the HBP, that is, on describing the state of HS competence as well as explaining how ultimate attainment in HSs often differs from that of monolingual speakers, despite both groups being native childhood acquirers of the same language. We demonstrate above that some competence divergences found between HSs and monolinguals may simply stem from the fact that the primary linguistic input to which HSs are exposed is qualitatively different from what monolingual speakers have at their disposal. Among other variables that contribute to qualitative differences in the shape of input available to HSs, we focused on the potential role that trickled-down L attrition (intra-generational attrition) and a limit in stylistic richness of the input may play. Acknowledging that HSs develop their grammars from
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
-
linguistic sources that differ from that of monolinguals naturally leads to the expectation of ultimate attainment competencies that diverge from one another. Accordingly, we maintain that any ultimate answer to the HBP will minimally have to factor in the role of attrited input affecting the potential scope of HS competence, that is, the extent to which attrited input changes the path of development in the case of HS bilingualism in the sense of Putnam & Sánchez (). That said, the account proposed here is by no means the only possible explanation of the HBP. There is no question that multiple variables contribute to the competence divergences that characterize HS grammars as compared to monolingual ones, among them arrested/ delayed development and individual level attrition. As previous research also demonstrates, there are domains of grammar in which HSs show significant differences from comparable monolingual speakers in the absence of L attrition in the older generation (see Montrul, a). There is also no question that some domains of grammar are inherently more vulnerable than others, as shown by the consistency of differences irrespective of the HL under scrutiny. Our main point, then, is to highlight that a trickling down effect of previous linguistic change has even greater implications for HS grammar than in contexts where the HL and the majority language are one and the same. Being as nuanced as possible as the field grows is of critical importance, not only because we endeavour to be increasingly accurate in explaining the processes we study, but also with regards to the connections that theoretical approaches to studying HS bilingualism can and should have with the more practical side of HL studies.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
......................................................................................................................
......................................................................................................................
. I
.................................................................................................................................. I this contribution, I will assume that simultaneous bilinguals (L bilinguals) are also heritage speakers (HSs), following Kupisch (), Meisel (), and Kupisch & Rothman (a). I will summarize major findings on L development and acquisition outcomes, and compare them to heritage language (HL) research more generally, focusing on the following aspects that I take to be insightful for the general understanding of heritage grammars: . . . .
cross-linguistic influence (CLI) role of age of onset (AoO) input and language dominance distance to the homeland.
There are different views on who counts as an HS (e.g., Montrul, , a; Rothman, ; Benmamoun et al., a; Kupisch & Rothman, forthcoming). Contemporary work tends to include both bilinguals from homes where only the minority language was spoken and bilinguals from homes where both minority and majority language were spoken. The latter situation includes families following the one parent one language strategy, provided parents have different native languages. In the past, some researchers, have restricted the label HS to the former type (e.g., Polinsky & Kagan, ), that is, bilinguals whose HL is the first language (L) and who are early L (eL) learners of the majority language. In what follows, I will refer to the latter group as HSs ‘in the traditional sense’. Thus, traditional HL research has sometimes excluded bilinguals raised with two languages from birth, while L acquisition research has focused on precisely this type.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
In the present chapter, an HL is considered to be a minority language that is not an official language in a speaker’s geographical area, and that was acquired at home, regardless of whether the HL was spoken by one or both parents.1 This means that HSs include L bilinguals—speakers with exposure to two languages before age —as well as eL learners— speakers who have acquired one language from birth and the other one between the ages of and years. Polinsky & Kagan () further distinguish HSs in a narrow sense from HSs in a broader sense, that is, speakers with genuine HL proficiency versus HL learners who study a language of their own heritage (say, in a university context) but without necessarily having used this language when growing up. Herein, I will be concerned with HSs in the narrow sense. I will distinguish, where relevant, between L (heritage) bilingual and eL (heritage) bilingual; early bilingual will be used as a cover term for both. In both cases, the HL is a native language (L) acquired from birth. Note that there already exist comprehensive overviews of L acquisition, including De Houwer (, ); Meisel (); Kehoe (); and Lleó (). Although some repetition cannot be avoided, the primary goal here is to motivate the importance of L research for understanding HS grammars more generally. As mentioned above, L bilinguals and HSs in the traditional sense have been studied separately for some time, and the impression when comparing research in the two traditions is indeed that they have been concerned with different types of bilinguals, although both meet existing definitions of HSs. Crucially, developing Ls were typically portrayed as being monolingual-like, while HSs in the traditional sense were often portrayed as deviant from monolinguals or even ‘incomplete’.2 However, it might just be the case that the two groups have been studied in two different contexts: namely Lers in the context of smaller minorities in Europe speaking more prestigious languages and coming from more privileged social classes (e.g., French in Germany or Sweden) on the one hand, and HSs in the traditional sense in the context of larger minorities in the US speaking comparatively less prestigious language and coming from disadvantaged social classes on the other (e.g., Russian or Spanish in the US). Moreover, L studies have typically focused on development, while HS studies have typically focused on end states (see also Kupisch, ; Kupisch & Rothman, a). Herein, I will argue that previous findings from L acquisition research are insightful when explaining acquisition outcomes in HSs more generally, and that at least some of the differences across HL acquisition outcomes are related to the factors listed under Section . above, to be discussed in the following sections. Given space limitations, the focus is on production studies of morpho-syntax and phonology.
1 Typically, early bilinguals who grow up in multilingual societies, for example, speakers of autochthonous languages like Basque speakers in Spain and France, are not included under the label of HSs. However, these speakers are similar to the HSs described above in many respects. 2 The idea that heritage grammars might be ‘incomplete’ or ‘non-native’ has recently received substantial criticism (see Pascual y Cabo & Rothman, ; Otheguy, ; Meisel, ; Rothman & Treffers-Daller, ; Kupisch & Rothman, forthcoming, for discussion). If native speakers are defined with regard to the AoO, as the term ‘native’ suggests, then HSs are native speakers of their HL by definition. If anything, their grammars are monolingual-like to different degrees, or different from monolinguals in various respects.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
. L CLI
.................................................................................................................................. One of the most intensely debated topics in the early L research was language separation. Perhaps until the s, some scholars still believed that monolingualism represented the norm and that deviations from this norm were potentially risky. Seeking to establish whether such concerns are justified, it has been discussed whether children can develop two languages at the same time. In this context, Volterra & Taeschner () proposed the Single System Hypothesis, arguing in favour of an initial period during which children have only one language system (Stage I) before differentiating two lexicons (Stage II) and eventually two grammars (Stage III). Over the years, this proposal led to debate, and subsequent work provided substantial empirical and theoretical evidence that L acquisition resulted in separate lexicons as early as during the one-word utterance stage (e.g., Quay, ), and separate syntactic systems as soon as children produce multi-word utterances (e.g., Meisel, ), that is, between and months (Radford, ). Evidence of this type gave rise to the Autonomy Hypothesis, Separate (Dual) System Hypothesis, or Separate Development Hypothesis,3 that is, the idea that the two languages of bilingual children develop separately. The separation of phonology was not explicitly discussed by Volterra & Taeschner but taken up, for example, by Vogel (); Ingram (); Deuchar & Clark (). By and large, the early separation of phonologies is taken for granted today, with some sceptical voices remaining (e.g., Vihman, ). Data from adult bilinguals also suggests that phonology might be somewhat different from grammar despite its relatively early acquisition (see Section .). In studies on adult HSs, the issue of language separation is rarely addressed. Instead, these studies tend to focus on the HL and on whether it is affected by majority language influence, while it is taken for granted that the majority language is spoken in a monolingual-like fashion and unaffected by potential HL influence. This seems to be justified, since L research has already shown bilingual children’s ability to separate the languages, and it is not plausible to assume that early separation will lead to fusion at a later stage. Of course, HSs also include eL learners, but for these learners language separation is even more plausible, since they have already built up grammatical representations of the HL before being exposed to the majority language.4 One study discussing language separation in adult HSs investigated the two languages of Turkish HSs in Germany (Kupisch et al., ). The study focused on the definiteness effect in existential sentences, where Turkish and German differ. In line with monolingual speakers of the respective languages, the HSs judged equivalent sentences in Turkish and German differently, accepting negative existential sentences with strong DPs in Turkish, while rejecting them in German. The claim that the simultaneous acquisition of two (or more) languages qualifies as a form of first language (L) development, that is, as ‘L acquisition’, is grounded in the assumption that the developmental trajectory in each of the bilinguals’ languages is similar to that of monolinguals, but not precisely the same. It is obvious that the two languages may 3
These terms imply slightly different things, depending on personal views and frameworks. The differences are not relevant here. 4 Of course, language separation does not exclude the occurrence of L influence into the L.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
influence each other, a process that has come to be called CLI. Since the turn of the century most research has therefore turned to investigating CLI, often based on ideas originally proposed by Paradis & Genesee (). These authors proposed that CLI is ‘systemic’, occurring ‘at the level of representation or competence, sustained over a period of time’. Moreover, CLI differs from code-mixing, which is conceived of as ‘online interaction’ occurring with various structures and frequency. CLI can occur in three different forms: acceleration, delay, and transfer. The former two imply that certain phenomena are acquired earlier or later when compared to monolinguals, while transfer is defined as ‘the incorporation of a grammatical phenomenon into one language from the other’ (p. ). Both acceleration and transfer may occur if a child has reached a more advanced level of syntactic complexity in one language.5 Although the authors’ own data did not yield conclusive evidence for CLI, their account has inspired many subsequent studies providing evidence for accelerated or delayed development. Delayed development was reported, for example, in many studies on the acquisition of clitic object pronouns in Ls acquiring French, Italian, and Portuguese along with a Germanic language (Müller & Hulk, ; Flores & Barbosa, ; Pérez-Leroux et al., ). The bilingual children omitted such pronouns at higher rates and over a longer period of time compared to monolinguals. Delays were also found in the acquisition of gender in French, Italian, and Spanish bilingual children who acquired a Germanic language simultaneously (e.g., Kupisch et al., ; Hulk & Van der Linden, ) and for the use of articles in Romance languages when acquired along with a Germanic language (Kupisch, ). However, delay and acceleration do not exclude each other. The same children who showed delays in the acquisition of gender and articles in their Romance language were more advanced compared to monolinguals in their Germanic languages (e.g., Hulk & Van der Linden, , for Dutch gender; Kupisch, , for Italian articles; see also Egger et al., forthcoming, for Dutch gender with Greek). Similar to their grammars, bilingual children’s phonological systems too may be affected by CLI. For instance, delays have been found in the acquisition of vowel length in German with Spanish as the other language (Kehoe, ) and with long lag voice onset time (VOT; see Kehoe, , for an overview). Acceleration has been documented for the development of coda consonants in Spanish with German as the other language (Lleó et al., ) and for the acquisition of Spanish r-sounds (Kehoe, ). Besides establishing empirical facts, researchers tried to strengthen the argument that CLI is systematic—as expected under the assumption of separate systems by seeking to predict the occurrence of CLI as well as its direction. The factors influencing CLI include partial overlap of the target grammars (see discussions in Döpke, ; Müller, ; Hulk & Müller, ; Müller & Hulk, ; Serratrice et al., ), the interface between syntax and pragmatics (Müller & Hulk, ), language dominance (Yip & Matthews, ; see Section .) and markedness (Lleó, ). A decade ago the question was posed whether CLI in L acquisition is merely quantitative or whether it can also be qualitative. Meisel (a) pointed out that is crucial to determine whether common patterns of grammatical development can be altered under the influence of the respective other language, for example, whether certain developmental phases can be changed as a result of interaction between grammatical knowledge systems,
5
Paradis and Genesee themselves used the terms ‘interdependence’ and ‘influence’ instead of CLI.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
as has been argued for eL acquisition with a later AoO. In that case, CLI could be considered qualitative. By contrast, changes in rate of development, that is, accelerated or delayed development, cannot normally be construed as qualitative, unless the change is significant enough for the normally invariable acquisition order to be reversed, or if the acquisition process is delayed so severely that monolingual-like competence cannot be reached. Based on the evidence available at the time, Meisel (a) concluded that CLI in L acquisition never results in qualitative differences compared to monolinguals, but in the meantime, evidence has accrued that CLI can also be qualitative. For instance, Rodina & Westergaard () have shown that some children acquiring Russian as HL do not acquire the three-way gender system of Russian, having a reduced system instead, or no gender system at all. For phonology, Kehoe et al. () showed that a German-dominant child transferred his long lag VOT from German to Spanish, where long lag VOT does not exist. Lleó () provided evidence for what she calls ‘zero transfer’: Spanish bilinguals fail to acquire the spirantization rules for Spanish due to CLI from German where spirantization does not exist. At this point, a word of caution is in order: The reference groups in the aforementioned studies were always monolinguals. That is, monolingualism is held up as a standard, which is undoubtedly dubious in a world where most speakers are bilingual, bilectal, or multilingual. Recent exceptions are the studies by Kupisch and colleagues, who have used bilingual instead of monolingual controls. More specifically, they have compared L bilinguals with the same two native languages, but living in different countries, for example, GermanItalian speakers in Germany for whom Italian is the HL versus German-Italian speakers in Italy for whom German is the HL (e.g., Kupisch, , ; Bianchi, ; Kupisch et al., ; Barton, ). In summary, research on CLI in Ls has focused on (i) defining different forms of CLI, (ii) determining the phenomena that are vulnerable to CLI as well as (iii) the factors driving these processes. In some work, a distinction was made between quantitative forms of CLI (acceleration and delay) and qualitative ones (transfer). By contrast, research on adult HSs often reports qualitative differences to monolinguals but it is less clear how these qualitative differences are defined. Moreover, the term ‘CLI’ has sometimes been used synonymously with ‘transfer’, which in some work on developing Ls is taken to be a subtype of CLI. Future work on adult HSs could take L research into account, seeking to eliminate terminological mismatches and to establish parallels between development and outcomes. For instance, phenomena that are vulnerable to qualitative differences in development might be more sincerely affected throughout a bilinguals’ lifespan and thus more likely to result in divergent grammatical representations.
. AO
.................................................................................................................................. AoO effects, caused by a loss of brain plasticity and lateralization, have been intensively debated with respect to both developmental patterns and acquisition outcomes. Relevant studies have traditionally focused on late Ls, that is, second languages (Ls) acquired at school or in a naturalistic environment at age or later. It is typically assumed that chances of obtaining monolingual-like proficiency decrease with growing age, and
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
numerous studies have provided empirical evidence that this is indeed the case (see Hyltenstam & Abrahamsson, , for an overview). Research on the acquisition outcomes of HSs challenge these assumptions in so far as many of these speakers end up with HL grammars that are deviant from those of monolingual speakers despite acquiring the HL from birth. Since HSs start acquiring the majority language at different ages, a relevant question is: can differences in AoO between the ages of to determine acquisition outcomes and, if so, under which conditions? Crucially, these questions should be posed for both the majority language and the minority language. For the majority language, we can deduce from L studies that earlier is better. For the minority language, however, the reverse might hold, that is, that a later AoO in the majority language could be beneficial. In Section ., we have seen plenty of evidence for the mutual influence of the two languages during bilingual development. Thus, one possibility is that Ls are disadvantaged compared to eLs because their minority language is affected by simultaneous input from another language from birth, while eLs have more time to develop their minority independently before majority language input sets in. I will discuss these possibilities below based on relevant studies. To begin with, we should recapitulate once more why we call someone a L or an eL learner. Despite the continuing debates on (i) the existence of sensitive age periods, (ii) what causes them, and (iii) which age they are associated with, many researchers speak of L bilingualism with ages of onset in the two languages before age , while eL acquisition refers to cases with an AoO of or later. Indeed, Meisel claimed that some changes affecting morpho-syntax already occur around age (e.g., Meisel, , ). Data in favour of these claims come from comparisons of children who started acquiring their L after to years with L bilinguals on the one hand and adult Ls on the other hand. Sopata (), for instance, has shown that Polish children acquiring German (AoO ; to ;) behave like adult Lers with respect to verb placement. The children in her study failed to place verbs in the required OV order, used target deviant V orders as well as non-finite verbs in V position. These patterns are never found in monolingual German children, while being typical of adult L German learners. Adult L-like patterns have also been observed in the production of grammatical morphology in eL learners of French (see, e.g., Hulk, ; Granfeldt, ; and Meisel, , for an overview), though sometimes with an AoO later than years. This includes cases where the L is a minority language (e.g., French in Sweden) and cases where the L is a majority language (e.g., German in Germany). Thus, both minority and majority language may develop differently from what is typical of monolingual L development. Crucially, the aforementioned studies made claims about developmental paths, not acquisition outcomes. In other words, there is evidence for age effects but the implications for the learners’ ultimate attainment are unclear. Moreover, not all morpho-syntactic phenomena are affected at the same age, there appears to be variation depending on the language combinations and there exist phenomena that are entirely resistant to age effects. Nevertheless, the L versus eL distinction, which is at least partially grounded in assumptions about sensitive periods around age to , has vernacularized. Simultaneous bilinguals are native speakers by definition and there is abundant empirical evidence that many such children develop in a qualitatively similar way to monolingual children (see Section .). However, Schlyter and colleagues (Schlyter, , ; Schlyter & Håkansson, ) provided data casting doubt on these ideas. She investigated
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
data from L French-Swedish children with an extremely unbalanced development in their two languages, demonstrating that their weaker language shows characteristics otherwise typical for adult Ls. For example, the children occasionally failed to place Swedish finite verbs in second position, thus violating the verb-second (V) rule according to which finite verbs follow the subject if an adverb is in sentence-initial position. Such errors are characteristic of adult Ls of Swedish but do not normally occur in the speech of monolinguals or balanced bilinguals. Interestingly, the children in this study grew up in Sweden. Thus, Swedish was not their HL but their majority language. Schlyter’s data have been discussed controversially (Meisel, b) and there are indeed cases demonstrating monolingual-like, albeit slower, development of the weaker language (e.g., Bonnesen, ). Nevertheless, data of the type Schlyter provided suggest that despite the earliest possible onset of language acquisition, monolingual-like development is not guaranteed if language input and/or use are severely limited. These findings might provide a link to data from adult HSs whose acquisition outcomes are different from those of monolinguals. Similarities between early bilinguals and late Ls in acquisition outcomes and/or error patterns indicate that AoO alone cannot be the crucial factor in language acquisition since the two learner types acquire their languages at very different ages. Another important question concerning AoO and HL development is whether the minority language is more vulnerable to CLI or to attrition if acquired in a L setting versus an eL setting. In the L setting, the two languages are in contact for a longer period of time and they are also in contact during a potentially more vulnerable stage when many features of both languages are still developing. By contrast eL HSs have the chance to stabilize at least some aspects of language in the HL before starting to acquire the majority language. At the same time, their majority language is still acquired before potential sensitive periods have faded out. Therefore, an overall advantage of sequential HSs over simultaneous HSs is plausible for the HL. There are a few recent studies on acquisition outcomes that speak to this. Rodina & Westergaard () show that in the acquisition of gender in Norwegian–Russian sequential bilingual children from households where the two parents speak Russian outperform simultaneous bilinguals from mixed language households. Flores et al. () come to similar conclusions for the acquisition of the subjunctive mood in heritage children acquiring European Portuguese in Germany. These findings too suggest that an early AoO alone is insufficient for monolingual-like acquisition outcomes and, moreover, that, everything else being equal, L bilinguals may benefit less from early exposure if the minority language does not receive sufficient support. Finally, it is worthwhile considering acquisition outcomes in the majority language. Since many HSs are eL learners, observations on acquisition outcomes in HSs’ majority language should be related to claims in eL acquisition, according to which the ages of to years correspond to a sensitive phase at least for some aspects of grammar and phonology. Contrary to claims on early sensitive periods, HSs, including eLs, often consider themselves to be more proficient and to have native-like proficiencies in the majority language. Although reservations about self-assessments appear justified, HSs’ subjective selfassessments can be backed up by objective data. For example, Kupisch et al. () investigated Turkish HSs in Germany, Ls and eLs, most of whom claimed to have native-like proficiencies in German but not in Turkish. In line with their self-assessments, an acceptability judgements task showed that these speakers had acquired definiteness restrictions in German akin to monolinguals, regardless of when they were first exposed
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
to German. However, results were less clear for phonological aspects of language. When the speakers’ perceived accents in German were assessed, some of them were deemed to have a mild foreign accent in German, although it was more subtle and harder to perceive than their accents when speaking Turkish (Stangen et al., ; Lloyd-Smith et al., ). Crucially, the speakers’ German accent was unrelated to their first intensive contact with German. That is, simultaneous HSs (AoO before ) did not necessarily sound ‘more German’ than sequential HSs (AoO after ). A comparable study of perceived foreign accent looking at L bilinguals (German-French, German-Italian)—AoO in both languages from birth—showed that these speakers were deemed native-like in their majority language, while often having a perceived foreign accent in their HS (Kupisch, Barton, Klaschik, et al., ). While the two studies compromise the role of AoO, it is unclear whether the differences between them in terms of majority language outcomes are due to language use at home. With their parents, the German-French/Italian used their HL as well as German, but the German-Turkish HSs used mostly their HL. An alternative explanation for the different outcomes of the two studies could be the size of the respective HL communities. The German-Turkish community is larger and there is a greater chance of being exposed to (accented) L varieties of the majority language. In summary, comparison of L and eL HSs suggest that AoO alone is not crucial for acquisition outcomes, and that Ls are at least as vulnerable to CLI as eLers. Since there are some indications that AoO in the majority language can be consequential for minority and majority language development, it is important to systematically include this variable in future studies.
. I
.................................................................................................................................. It has become increasingly clear that the amount and quality of input are crucial for HL development. In recent accounts (see Unsworth, , Chapter , this volume, for overviews), the amount of input has been determined, amongst other things, by the language spoken and chosen by the main caretakers, number of siblings, friends, and relatives speaking the same language, number of visits to the homeland and activities performed in the HL. Quality of input, by contrast, is related to whether HSs are exposed to input from L speakers of the HL, whether their parents’ language has been influenced by long-term L exposure (potentially leading to attrition), whether the HSs’ interlocutors speak a dialect or a standard variety, whether the HSs’ input is exclusively spoken or also written, and, relatedly, whether HSs are exposed to the HL in a school context. Input quantity and quality often determine whether a bilingual develops his/her two languages at a more similar pace, that is, in a balanced fashion or whether one language develops faster. The term language dominance refers to the latter scenario. There is no uniform definition of language dominance (see Silva-Corvalán & TreffersDaller, , for an overview). Most definitions incorporate one or more of the following dimensions: (a) proficiency, (b) quantity and/or quality of input, (c) language of the environment. The three dimensions are obviously related. Since exposure to the two languages varies between bilingual children, this may result in varying proficiencies. Normally, the child receives more input in the language of the national environment,
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
which is also the more proficient language, but there are children, including L bilinguals, who are initially (from to years) dominant in the minority language. Research on child development has used the term ‘language balance’ to highlight the relativeness of the concept. For example, a speaker may develop relatively fast or relatively slowly in both languages, yet with small differences (fairly balanced); s/he can be highly proficient in one language but much less proficient in the other (fairly unbalanced); or relative proficiency may differ across domains (e.g., syntax vs. phonology) or modes (e.g., spoken vs. written).6 Investigating language dominance can be useful, for instance, when determining the causes for CLI during development. While it is common in research on adults to use vocabulary tasks, cloze tests, and self-assessments to determine relative proficiency, finding appropriate measures for language dominance has been a central concern in studies of early bilingual language development (e.g., Genesee et al., ; Ingram, ). There has been some discussion of whether qualitative measures (often seen as related to proficiency) should be distinguished from quantitative ones, because a child may prefer to speak one language, yet without necessarily being more proficient in this language. Typical proficiency measurements for bilingual children are mean length of utterances (morpheme- or word-based), upper bound (the longest utterance), vocabulary size, and, for very young children, for example, the complexity of syllables and the mean number of correctly produced vowels and consonants. Quantitative measurements, that may or may not reflect proficiency, would include the total number of utterances within a certain time, language mixing, or language choice in a monolingual setting. Although there are good reasons for separating quantitative from qualitative measures, the two often coincide. As mentioned in the beginning, language dominance is closely linked to input, and the latter has always played a role in theories on language acquisition, regardless of theoretical orientations. Studies in bilingual development have convincingly shown that changes in the linguistic environment of children can have a major impact on rate of development and relative proficiency in the two languages. Several such cases were documented in the early literature on bilingual children, for example, the case of Hildegard, whose parents spoke to her in their respective native languages (German and English), but who moved between Germany and the US (Leopold, ), or Shelli, who grew up between Israel and the US with two English-speaking parents (Berman, ). Both children witnessed shifting language dominance along with a change in residency. Dominance shifts are very common in sequential HSs who do not change the country of residence, and they also occur in L bilinguals from homes where two languages are spoken and where, depending on language policies at home, the HL may initially be dominant. The four L German-Italian children studied in Kupisch (), for instance, all grew up in Germany with German-speaking fathers and Italian-speaking mothers. Nevertheless, they exemplified three different dominance profiles, that is, balanced, German-dominant and Italian-dominant. Finally, it is important to keep in mind that linguistic structures are differentially (in)sensitive to input frequency, for example, lexically based gender assignment is more substantially affected by reduced input than the syntactic operation of gender agreement (Kupisch et al., ).
6 Variation in rate of development is also common among monolinguals, especially during the early years, and is not necessarily related to bilingualism.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
It is commonly assumed in research on adult HSs that heritage bilinguals undergo dominance shifts during their childhood, often coinciding with school entry, and sometimes also with language shifts at home (Benmamoun et al., a) so that eventually the majority language is spoken with a higher proficiency. By adulthood, the language of the environment tends to be dominant. For example, there are several studies comparing adult bilingual German-Italian speakers in Germany and in Italy in both languages, focusing on a variety of different properties. The bilinguals in Italy performed monolingual-like in Italian but not necessarily in German. Vice versa, the bilinguals in Germany performed monolingual-like in German but not necessarily in Italian (see Stöhr et al., ; Bianchi, , on gender; Kupisch, ; Kupisch & Barton, , on determiner use; Kupisch, , on adjective placement). In other words, all these studies dealing with morpho-syntax or interface phenomena have indicated unidirectional influence from dominant to HL. In phonology, the HL also tends to undergo CLI, but CLI is not necessarily unidirectional. For example, long lag VOT, which is comparatively more marked than short lag VOT, can be affected in both the stronger and weaker (heritage) language, while short lag VOT is only affected in the HL (Kupisch & Lleó, ). Dominance shifts—though common in developing bilinguals—are less likely to occur in adult bilinguals. Kupisch & Van de Weijer () studied two groups of German-French bilingual speakers during adulthood. Both groups grew up with one German-speaking and one French-speaking parent and all were highly proficient in both languages, but the two groups differed in that one group grew up in France and the other in Germany. With respect to all aspects tested, speakers from France behaved monolingual-like in French, but not all performed equally well in comparable tasks in German. Conversely, speakers who grew up in Germany performed monolingual-like in German but not always in French. However, unlike in the ItalianGerman studies summarized above, the speakers from France had all moved to Germany as adults and had been living there between ten and twenty years. Nonetheless, French had remained their dominant language. The study showed that dominance-shifts are less likely to occur in adulthood, even with highly proficient speakers who have been in contact with both languages from birth. Thus, the majority language of Ls is not easily subject to CLI or attrition, even under drastic changes in input conditions. In summary, dominance shifts may occur throughout the childhood of both L and eL HSs. Eventually, the majority language tends to be dominant, provided that speakers do not change their country of residence during childhood. Dominance is relatively more stable during adulthood and does not easily shift even if bilinguals change their country of residence. Given the research reported in Section ., there are some indications that Ls may end up being more unbalanced than eLers, since they will have had less exposure to the HL during their early years. It is an open question whether this makes them more vulnerable to attrition.
. D
.................................................................................................................................. As outlined in the previous sections, many studies on bilingual development have highlighted the fact that L bilinguals separate their languages from early on and develop akin to monolinguals in their two languages. At the same time, numerous studies on adult HSs
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
suggest that their mature grammars differ from those of monolinguals (e.g., Montrul, ). It is noteworthy that many of the early case studies arguing in favour of monolingual-like development have originated in a European context (see, e.g., the overviews in De Houwer, , ), while studies on adult HSs stressing divergent outcomes have often originated in the US, and have only recently become more frequent in the European context. Despite the different foci, that is, development in the former case versus outcomes in the latter, the question arises whether the national environment in which HSs grow up makes a difference and why. To approach this question, I will compare selected work on heritage Romance languages since the Romance language family has been intensively studied on both the European and the American continent, mostly in contact with a Germanic language, which facilitates comparison (see also Kupisch & Rothman, b; Flores et al., , for overviews of studies including Romance). There are many studies on the development of morpho-syntax and properties involving interfaces with French, Italian, and Spanish as one of the languages, demonstrating that bilinguals develop in a qualitatively similar way to monolinguals (see, e.g., Meisel, ; De Houwer, ), although they might show accelerated or decelerated acquisition rates (see, e.g., Serratrice, , for an overview). Often, these studies originated in Germany, Holland, Sweden, and in Canada mostly with English-French bilinguals. Similar findings could be reported for the development of phonology, although there seem to be comparatively more instances of qualitative differences with respect to monolinguals (Kehoe, ; Lleó, , for an overview). Data from adults of the same speaker type indicate that whether or not end states are deviant depend on the property under investigation and possibly on input (see, e.g., Kupisch, , ; Bianchi, ; Kupisch et al., ; Flores & Barbosa, ; Flores, b; Santos & Flores, ). The research on French seems to indicate that developing bilinguals may profit (in both languages) from attending schools in their minority language (Kupisch et al., ). In sharp contrast to the studies mentioned above, numerous studies on Spanish in the US have documented differences between adult HSs and monolingual speakers with respect to many different properties (e.g., Montrul, , ; Montrul & Sanchez-Walker, ). Of course, the studies cited above are not perfectly comparable because they are based on different people, methodologies, and phenomena. Yet, if we put aside some exceptions, it seems that studies in Europe have provided us with a slightly different picture than the studies originating in the US context. Why could this be the case? Language policies and the social prestige of languages are likely to play a role: notably, whether or not a country ensures that HL instruction is offered to children of certain linguistic minorities, whether certain languages are taught as foreign languages at school, and what prestige is attached to learning and speaking this language. For instance, French has traditionally been taught as a foreign language at German schools, Spanish has become increasingly popular over the past three decades and there are also schools offering Italian and Portuguese classes. Whether a language is taught as a school subject not only creates opportunities for HSs to relearn or study their HS but also affects the prestige of this language positively. Relatedly, it may affect the HSs’ perceived distance to their homeland. As mentioned above, many studies with a Romance HS have originated in Germany, which has been a popular destination for migrants since the s, when Germany was in need of manpower to support the growing industrial sector. Between and , Germany had bilateral agreements with Italy, Spain, and Portugal to facilitate immigration and the
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
subsequent return of guest workers. Although the German government encouraged returns, only few workers did in fact return to their home countries. This resulted in continuous migration and remigration, and while immigration to Germany had originally been a means of escaping poorer living conditions, it eventually also attracted more skilled employees and academics. Today, there are three generations of migrants in Germany, many of German nationality. Although perfectly integrated into the German-speaking society, they often keep in close contact with their home countries, which is facilitated by geographical vicinity and the fact that travelling has become easier and more affordable. Many of these immigrants in Germany tend to perceive the relation between their two countries as being characterized by cultural and political closeness, which is possibly often absent in immigration flows in non-European countries. Finally, the EU fosters maintenance of the home language, and an increasing number of bilingual schools can be found in larger cities such as Berlin and Hamburg. In summary, the extent to which HSs are separated from their home countries, mentally and geographically, may determine how often they use the language, which has consequences for how closely the preserved language resembles that of a monolingual in the home country. A better understanding of the role of distance to the homeland might be gained by comparisons of HSs in the traditional sense and HSs of moribund languages who have been isolated from their home countries for many decades or even entire centuries.
. C
.................................................................................................................................. I have provided an overview of L research with respect to specific factors that I conceive to be relevant for HS development and maintenance. Both L HSs and HSs in the traditional sense are affected similarly by CLI. The two groups may differ in terms of their AoO in the majority language. Given assumptions about sensitive periods, there are reasons to assume that HSs in the traditional sense are more affected in their majority language, but there seems to be little evidence, so far, to support this view. Vice versa, Ls may be comparatively more affected in their minority language, but there are only a few studies so far supporting this view. The amount and quality of input affects all HSs and will determine language balance during development and adulthood. Balance is more easily affected during development, and while dominance shifts may occur in both Ls and HSs in the traditional sense, they are more frequently reported with the latter. Finally, it is likely for both types that their maintenance of the HS is affected by the geographical and the mentally perceived distance to the homeland.
A I wish to thank Anika Lloyd-Smith and Simone Waitz as well as two anonymous reviewers for comments on previous versions of this chapter. Many thanks also to my friends and colleagues who have helped me develop the ideas presented in this chapter through joint work and/or discussions: Cristina Flores, Mike Iverson, Margaret Kehoe, Conxita Lleó, Jürgen Meisel, Diego Pascual y Cabo, Esther Rinke, Jason Rothman. All remaining errors are mine.
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
......................................................................................................................
Evidence from behaviour and the brain ......................................................................................................................
. , ,
. I
.................................................................................................................................. I adopted (IA) children are exposed to one language from birth and, in most cases, completely discontinue exposure to and acquisition of that language when they are adopted into families whose language differs from their own. After adoption, IA children are typically exposed to and acquire only the adoption language. This pattern of acquisition distinguishes them from other second language (L) learners who continue to learn and use the birth language to some extent, as well as from L learners who discontinue their first language (L) when they immigrate with their families and learn the dominant language of the community, but who may still have exposure to the L through family and/or by living in neighbourhoods with immigrants from the same language community. IA children typically acquire their adoption language relatively quickly and score within norms relative to age-matched peers on standardized tests of language development within the first year or two following adoption (Scott & Roberts, ), although there is some evidence that a larger than expected subgroup shows delays (Delcenserie, ), and poorer performance on certain aspects of language (e.g., phonology, verbal working memory) has been reported into the school years (Delcenserie & Genesee, a) and beyond (Norrman et al., ). The language outcomes of IA children may be influenced by factors such as pre-adoption conditions (Rice et al., ) and, of particular relevance to the present chapter, age of adoption (Scott & Roberts, ), with IA children adopted at younger ages often showing better outcomes in the adoption language. Age of adoption might influence language outcomes through delayed exposure to the adoption language, through the influence
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
of previous exposure to the birth language, or both. Thus, while, understandably, IA children’s acquisition of their adoption language has received the majority of research attention, a complete understanding of the language development of IA children must include an examination of issues related to attrition of the birth language. A focus on birth language attrition is critical because IA children have months to years of exposure to their birth language during an optimal period for phonological development (e.g., Werker & Hensch, ). Prior research has convincingly argued that a strong foundation in an L supports acquisition of subsequent languages. For example, Mayberry and colleagues have shown that children who were exposed to spoken language from birth and subsequently learned American Sign Language (ASL) as an L perform better on several aspects of ASL than learners who began learning ASL at the same age as the late L learners but who, because they were born deaf, had no early exposure to language in any modality (i.e., they learned ASL as a late L) (Mayberry, , ). The implication is that learning a language early in life supports the acquisition of a subsequent language, even though that language may differ in modality. Similarly, whether aspects of a birth language that are acquired during this early period are lost or retained has implications for the development and processing of IA children’s adoption language—that is, do IA children appear more like L learners owing to complete attrition of the birth language following adoption, or are some aspects of the birth language retained leading them to resemble L learners in some ways? The present chapter addresses this question by first reviewing research on behavioural and neural language attrition in IA children, followed by a discussion of the implications of attrition versus retention of the birth language on the acquisition of a subsequently acquired language. As well, because IA children present a unique situation in which the birth language is completely discontinued following adoption, their patterns of language development after adoption can elucidate the unique role of the earliest stages of language acquisition on later language outcomes. Therefore, the present chapter also highlights how patterns in IA children’s language attrition might reveal more general principles underlying neural plasticity and language acquisition in general.
.. Behavioural attrition Upon adoption, IA children rapidly begin to acquire the language of their adoptive families. They typically receive no further input from their birth language and, as a result, experience a concomitant decline in their ability to speak or understand that language (see Scott & Roberts, , for a review). Indeed, behavioural evidence demonstrates a rapid loss of the birth language, usually within the first few months or even weeks following adoption (e.g., Glennen & Masters, ; Nicoladis & Grabois, ; Price et al., ; Snedeker et al., ), even if some attempt is made to continue exposure to the birth language (e.g., Isurin, ). For example, in a large sample (n = ) of children adopted between birth and months of age from Eastern European countries, Glennen & Masters () found, using parent report surveys, that only eight of the children were still using the birth language by months post-adoption—six of whom were using only simple words and just two produced basic sentences. Of those eight children, all were adopted at older ages, between to months of age, meaning that they were likely speaking the birth language prior to adoption. By five months post-adoption, however, parents reported that none of
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
, ,
the IA children used their birth language any longer, regardless of age of adoption. Gindis () similarly reported that children adopted prior to age or years of age lost the ability to speak their birth language in as little as six to twelve weeks post-adoption, and their receptive abilities in that language disappeared by sixteen to twenty-two weeks. Nicoladis & Grabois () assessed the Chinese language retention of an IA child adopted from China at months of age during play sessions with Chinese-speaking children, a context that would maximize the likelihood of the child producing and understanding Chinese, if she were able. Native speakers of Cantonese coded five sessions, ranging from one to three months post-adoption, for evidence of production and comprehension of Chinese. During the first session, the child used three Chinese words. During the second and third sessions, she was found to understand four different Chinese phrases; however, by the fourth session, no production or comprehension of Chinese vocabulary was evident. While studies reported so far demonstrate rapid attrition in IA children who were relatively young at the time of adoption, similar patterns have been observed in children adopted at older ages as well. For example, Snedeker and colleagues () examined IA children who were as old as ; at the time of adoption and found that even they exhibited a dramatic reduction in the use of their birth language by two months post-adoption. By five months post-adoption, they had functionally stopped using their birth language entirely. In a case study of an adult who was adopted from Guatemala into an English-speaking family in the United States at age , significant attrition of the discontinued L was also evident (Montrul, b). While attrition was not as extensive in this case as in children adopted at younger ages, features such as lexical retrieval and grammatical accuracy were affected considerably. Despite evidence for rapid attrition of IA children’s birth language post-adoption, there is some evidence that certain aspects of IA children’s early language experience remain, even if they are not immediately accessible. For example, Oh et al. () examined English-speaking adult IA participants who were adopted from Korea when they were between months and years of age and who sought to relearn Korean by enrolling in a Korean language course when they were years old, on average. At the time of testing, they had been enrolled in the course for two weeks. Despite complete discontinuation of the birth language by this group, the IA participants performed better on a phonemic identification task involving lenis and aspiration contrasts, which are phonemic in Korean but not in English, than English-speaking participants enrolled in the same course who had never been exposed to Korean. Similarly, Bowers et al. () tested English-speaking adults who were exposed to either Hindi or Zulu for four to ten years as children, but were not exposed to, and demonstrated no knowledge of, these languages after this point. By the end of thirty ten-minute sessions, participants were not only better at discriminating the phonetic contrasts present in the language they had heard as children compared to the phonetic contrasts in the other, unfamiliar language, but their performance on this phoneme discrimination task was also similar to the performance of native speakers of that language. Several other studies have found similar advantages for IA participants in their discontinued birth language on a variety of tasks, including voice onset time (VOT) and phoneme perception, discrimination, identification, and production when compared to participants who had never been exposed to the target language. This has been observed for IA adults from Korea who were enrolled in Korean language courses (Hyltenstam et al., ; Park,
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
) or who completed systematic training of Korean phoneme contrasts (Zhou et al., ; Choi, Broersma, & Cutler, ) and for IA children and adolescents adopted from India who completed a brief retraining session in Hindi (Singh et al., ). Moreover, Korean adoptees in France were less likely than control participants to confuse tone languages with Korean in a language identification task, indicating some underlying knowledge that tone information was not phonemic in Korean (Ventureyra & Pallier, ). Taken together, these advantages suggest that attrition of the L may not be entirely complete even if behavioural evidence of access to the birth language is not apparent (Hyltenstam et al., ). While there is mounting evidence for re-learning advantages following early exposure, it is important to point out that effects are sometimes subtle; individual differences in performance can be considerable; and not all studies report significant effects. For example, Bowers and colleagues () found a relearning advantage only for participants under years of age at the time of testing, but not older participants. Similarly, Ventureyra et al. () found a significant relearning advantage for IA participants from Korea on only one Korean contrast out of six, although non-significant numerical advantages were observed on other, arguably more difficult contrasts. Findings such as these highlight the complexity of language attrition versus retention and point to the need for further research, particularly at the neural level.
.. Neural attrition Despite evidence for rapid behavioural attrition of the birth language following adoption, evidence from adults suggests that traces for certain elements of the discontinued birth language might be retained over time as reflected by better performance on certain language measures for adults who had early but discontinued exposure to a language compared to those who had not (Bowers et al., ; Hyltenstam et al., ; Oh et al., ; Singh et al., ; Park, ). One way to address whether or not elements of the adoption language are retained is by examining neural responses to a ‘lost’ language. Whether or not a ‘lost’ L is retained in the brain, despite apparent behavioural attrition, is an important question because the answer could reveal critical properties of brain plasticity in general, both early in development and over time (see Pierce et al., , for a review). For example, the hypothesis that the brain undergoes complete, or at least significant attrition of a discontinued L, is based on some accounts of neural plasticity that argue that neural circuits sub-serving language acquisition remain plastic during the childhood years. Plasticity during this early phase of development—as evidenced by significant neural attrition of the L, would allow for replacement of the L by a subsequent language if exposure to the L were discontinued during childhood, as is the case for IA children who rapidly cease speaking or understanding their birth language following adoption (Penfield & Roberts, ; Lenneberg, ; Flege et al., ; Pallier et al., ). In contrast, research examining the neural circuitry underlying learning and memory has claimed that once something has been learned the neural traces sub-serving that memory remain in the brain. That is to say, while new memories may be added, these do not update or overwrite existing representations (e.g., McGaugh, ; Rescorla, ; but
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
, ,
see Dudai, ). While results from the language domain are mixed at this time, there is growing evidence that corroborates findings of maintenance of early-acquired language traces analogous to those observed in other species and in other domains (e.g., Rescorla, ). This would suggest that at least some aspects of the birth language may be retained in IA children post-adoption, perhaps indefinitely, even if they become difficult to access (see Pierce et al., , for a review). Evidence that aspects of a discontinued L are in fact retained in the brains of IA children post-adoption comes from research by Pierce et al. () who tested three groups of - to -year-old participants using BOLD fMRI (functional Magnetic Resonance Imaging): () IA participants from China who were exposed to Chinese from birth but were adopted into French-speaking families between and months of age, had no subsequent exposure to Chinese, and became monolingual speakers of French; () Chinese-French bilinguals who were exposed to Chinese from birth, began acquiring French at about the same ages as the IA participants, on average, but continued to speak both Chinese and French; and () French monolingual speakers who had never been exposed to Chinese. While in the scanner, participants completed a lexical tone discrimination task using Chinese pseudowords. Lexical tone was examined for three reasons: () it is a phonemically significant feature of Chinese, but not French; () it is acquired very early in development, by as young as months of age, and thus it is very likely that the IA children had acquired these tonal contrasts prior to adoption (Mattock et al., ; Yeung et al., ); and () neural responses to Chinese lexical tones have been shown to differ between individuals who speak Chinese and those who do not (e.g., Gandour et al., ; Klein et al., ). While in the scanner, participants were required to discriminate Chinese pseudosentences that differed based only on the tone of the final syllable. During this task both the IA participants, who had early but discontinued exposure to Chinese, and the bilingual participants, who had ongoing exposure to Chinese from birth, activated the same location in left superior temporal gyrus/planum temporale. This region has been shown to be involved in tonal language processing in native speakers of tonal languages. Critically, monolingual speakers of French who had never been exposed to Chinese did not show convergent activation with IA or bilingual participants in this or any other region. Moreover, the same region that was activated by both the Chinese-French bilinguals and the IA participants was positively correlated with age of adoption in the IA group. Thus, later age of adoption, and a longer duration of exposure to Chinese, was associated with more ‘native-like’ activity within this region. These results from Pierce and colleagues () are in contrast to previous findings reported by Pallier and colleagues () in which the neural responses of IA adults to their birth language were no different from individuals who had had no exposure to that language at all. In that study, Pallier used fMRI to examine neural activation patterns in adult participants (aged to years of age) who had been adopted from Korea into French-speaking families when they were ; to years of age, and who reported no memory of their birth language. The IA participants were compared to native monolingual French-speaking adults who had never been exposed to Korean. Accuracy did not differ across groups on a sentence recognition task in Korean in which participants were required to match sentence fragments to previously presented sentences, indicating no functional advantage for IA participants despite their previous Korean exposure. Similarly, the fMRI analysis revealed activation in similar brain regions in both IA and control participants
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
during the presentation of French sentences—although the IA group exhibited somewhat less activation—as well as when French and Korean sentences were directly compared. The authors interpreted these results as an indication that, for IA individuals, the later acquired L had completely replaced L representations via plastic changes in the brain and, thus, complete attrition of the birth language had occurred. However, because Pallier and colleagues () did not compare IA participants’ activation patterns to native Korean speakers it was not clear whether the patterns of the IA participants mirrored native-like Korean patterns in any respect. Moreover, the sentence recognition task used may not have been sensitive to specific features of language that the youngest IA participants in the Pallier study would have acquired prior to adoption, nor elements for which behavioural studies have demonstrated a particular advantage for participants with early but discontinued exposure to the target language (e.g., phonological processing; Au et al., ; Oh et al., ; Bowers et al., ; Hyltenstam et al., ; Oh et al., ; Singh et al., ; Park, ; Pierce et al., ; for a review, see Pierce et al., ). In particular, while tasks from both Pallier and colleagues () and Pierce and colleagues () arguably required attention to speech sounds, the sentences used by Pallier came from several different languages, were more varied and less tightly controlled, and may have required a greater memory component than the stimuli used by Pierce and colleagues (). Furthermore, in the former, statistical comparisons were made between languages as opposed to comparing language with non-language sounds. The latter approach represents an effort to draw out activation specific to phonological information. Stimulus features such as these may have modulated the detection of subtle effects in these studies.
.. Attrition summary As reviewed in previous sections, language attrition in IA participants may depend on what aspects of language are measured. In particular, while adoptees functionally appear to lose their birth language, there may be a difference between what is observed in behaviour and what is observed in the brain and in the tasks used to examine behaviour. Moreover, relearning studies reveal a selective advantage among IA participants on tasks involving phonological aspects of the birth language, and there is also evidence for the retention of phonological aspects of the birth language at the neural level (Au et al., ; Oh et al., , ; Bowers et al., ; Hyltenstam et al., ; Singh et al., ; Park, ; Pierce et al., ). Because phonology is learned very early in development and forms the foundation for acquiring increasingly complex aspects of language (Pierce et al., ), it may be more profoundly entrenched in the brain as it is activated more frequently over the course of development than other linguistic elements (Keijzer, , ). Deeper entrenchment may make representations for phonology less likely to be lost over time. At the same time, phonology might be retained despite loss of more general aspects of language because of the nature of the typical developmental processes that occur during the time that phonology is acquired. Specifically, neural commitment to the sounds of a native language occurs during what is thought to be an early optimal period for the acquisition of native language phonology (e.g., Werker & Tees, ; Kuhl et al., ; Werker & Hensch, ). Thus, IA children may develop categories for the phonology of their birth language that are resistant to attrition, an idea we have reviewed previously (Pierce et al., ;
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
, ,
Pierce et al., ). As presented at the outset of this chapter, whether aspects of language undergo attrition or are retained have implications for later language development. Implications for both the L and the L are presented in the following sections.
. I / /
..................................................................................................................................
.. Implications for L Functionally, IA children’s birth language undergoes significant attrition as demonstrated by behavioural evidence from IA children in the weeks and months following adoption, as discussed earlier. However, studies that have re-exposed IA children to their discontinued L later in life have typically reported behavioural advantages for IA participants in comparison to individuals without early exposure to the target language (Bowers et al., ; Hyltenstam et al., ; Oh et al., ; Singh et al., ; Park, ; but see Ventureyra et al., ). Relearning advantages have been observed following different types of training, including enrollment in university level language courses (Hyltenstam et al., ; Park, ) as well as training designed to systematically highlight particular contrasts in the birth language (e.g., Bowers et al., ; Singh et al., ; Choi, ; Zhou, ), and advantages are similar to those observed in immigrants who moved with their families and discontinued their L (e.g., Werker & Tees, ), as well as childhood overhearers of a language (Au et al., ). However, as in those studies, advantages are often subtle, likely owing to small and heterogeneous samples that may differ on critical factors such as motivation and prior experience with the birth language. Also, where advantages are observed they are typically limited to particular aspects of language (e.g., phonology). While a selective advantage points to the retention of phonological representations for the birth language and attrition of more complex language elements (if they were acquired in the first place), it is unclear whether retention of birth language phonological representations might positively support the acquisition of more complex elements of language over time. Thus, while much of the L is functionally lost, there may be some advantage garnered from early exposure if the L is to be relearned later in life. However, more systematic relearning studies would be helpful in order to control for extraneous factors that may have influenced the results of existing studies, such as re-exposure to the birth language following adoption, motivation, the nature of the training, and others.
.. Implications for L It has been argued that, because IA children lose the ability to speak or understand their birth language, their language acquisition system can be ‘reset’ so that they are able to fully acquire an L in the manner of a typical L learner (e.g., Pallier et al., )—creating a kind of ‘blank slate’ as it were. Indeed, because typically very little language is acquired prior to
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
adoption, IA individuals probably experience less interference than typical L learners who retain their L. Supporting this, evidence suggests that IA learners acquire several aspects of their adoption language, from vocabulary to grammatical morphology, to a similar degree of proficiency and following a similar developmental trajectory as typical L learners who acquire the same language continually from birth (e.g., Price et al., ; Pierce et al., ; Snedecker et al., ). This would suggest that whatever traces remain of the birth language are not sufficient to perturb the acquisition of the adoption language substantially. However, in accordance with findings that phonological representations from the birth language are retained (Pierce et al., ), some evidence suggests that IA children’s acquisition of the adoption language is different from that of typical L learners, particularly when they are compared to carefully matched comparison groups or when specific aspects of their adoption language are assessed (e.g., Delcenserie & Genesee, b). For example, IA children who were compared to monolingual comparison groups matched for age, gender, and socio-economic status, all factors that can influence language development, have been found to score lower on both expressive and receptive vocabulary when they were tested at to (Gauthier & Genesee, ), to (Delcenserie & Genesee, ), and to years of age (Delcenserie & Genesee, a). Thus, although IA children’s vocabularies may develop to the typical or even above average range, carefully matched comparisons have revealed more nuanced effects. This pattern of results has been attributed to a selective difficulty in IA language learners for aspects of verbal memory, which is arguably related to phonological processing (e.g., Pierce et al., for a review). For example, Eigsti et al. () examined - to -year-old IA children who were adopted between and months of age in comparison to matched English-speaking monolingual controls and found that the IA children’s verbal memory scores were average to low, with % scoring within a clinically low range, . standard deviations below the mean, on the CELF (Clinical Evaluation of Language Fundamentals—fourth edition; Semel et al., ). Similarly, % scored at this level on immediate and delayed recall of word lists. These results parallel those observed in a series of studies by Genesee and colleagues (i.e., Gauthier & Genesee, ; Delcenserie et al., ; Delcenserie & Genesee, ) who examined IA children from China compared to non-adopted monolingual controls matched for age, gender, and SES. When these children were tested nineteen to . months postadoption, when they were . to . years of age, they performed significantly less accurately than a monolingual French-speaking comparison group on formulating and recalling sentences (Gauthier & Genesee, ). When a similar sample of IA children, including several of the same children, were tested at ; years of age, after ; years of exposure to French, on average, they scored below both the comparison group and monolingual age-norms on sentence recall (Delcenserie et al., ). When the IA children were between ; and ; years (Delcenserie & Genesee, a) and had had an average of ; years of exposure to French, they were tested on a battery of verbal memory measures. The IA children performed less accurately than matched monolingual controls on recalling sentences, verbal short-term memory (digit recall and non-word repetition), phonological working memory (PWM; word recall and backward digit recall), and verbal long-term memory (delayed recall of word lists). Importantly, they did not score below controls on non-verbal cognitive or memory abilities, indicating that their difficulty was specific to memory for the sounds of their adopted language. Neurocognitive evidence for the retention of elements of the birth language suggests how IA children’s acquisition of the adoption language might be affected by prior exposure to
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
, ,
the birth language. For example, Rajagopal and colleagues () tested IA participants who were adopted from China into English-speaking families between months and years of age, after which they had no exposure to Chinese. When they were to years old the children were scanned using fMRI while performing a verb generation task in English. Significant differences in lateralization were observed between the IA participants and English monolinguals. In particular, more right hemisphere frontal and temporal-parietal activation was observed in the IA children from China compared to the non-adopted control children. The authors suggest that this pattern might be due to exposure to Chinese during infancy; specifically, right hemisphere recruitment during an English-language task might reflect the influence of having learned a tonal language (which may recruit the right hemisphere) on English-language processing. This pattern might also reflect activation of additional neurocognitive systems recruited to support the processing of an L that was acquired following delayed exposure to that language (e.g., Pierce, Genesee, & Klein, ; Pierce, Genesee, Delcenserie, & Morgan, ). Pierce et al. () further demonstrated that the maintenance of early-established representations for L phonology (Chinese) influenced the processing of a subsequently acquired language (French). Using fMRI, it was demonstrated that neural activation in response to a French PWM task differed between French monolinguals, who had heard and used French exclusively since birth, and IA participants and Chinese-French bilinguals who acquired French following a delay of six months to three years after birth. While French monolinguals activated brain regions typically observed during native-language PWM processing, particularly left anterior insula, Chinese-French bilinguals and IA participants activated left insula more weakly than French monolinguals. As well, both groups with early exposure to Chinese and delayed exposure to French activated additional regions that are typically implicated in non-verbal memory, general attention, and cognitive control processes (e.g., Chee et al., ; Abutalebi & Green, ; Rottschy et al., ). This suggests that neural representations from the L, established during early development, affected the way that L language sounds were processed, even though the L had been discontinued. Specifically, IA children’s early language experiences appeared to influence their subsequent language processing in a way that led them to resemble L-learners of their adopted language. The retention of phonological representations from the L, despite behavioural attrition of that language, might explain IA children’s selective difficulty with verbal memory and phonological aspects of the adoption language.
. C
.................................................................................................................................. Following a review of the research to date, there is evidence that language attrition in IA children depends on the aspect of language examined. Evidence that phonological aspects of language are retained over time is reflected by greater accuracy on phonological tasks for participants who had early but discontinued exposure to the target language, particularly following episodes of training or relearning of the birth language (Au et al., ; Bowers et al., ; Hyltenstam et al., ; Oh et al., ; Singh et al., ; Park, ), which may ‘reactivate’ stored representations that are no longer in use, but that appear to have been retained at the neural level (Pierce et al., , ). Understanding what aspects of
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
language undergo attrition and what is retained in the brain in international-adoptees sets the stage for determining the neural substrates available for language learning/processing and the constraints under which language is acquired at any given stage of development. For example, both early and ongoing exposure to multiple languages appears to influence how neural mechanisms are employed not only for language processes, but for general executive function/cognitive control processes as well. There is evidence that both bilingual speakers and IA participants recruit brain regions that are typically involved in attention or cognitive control when they are processing language sounds (Abutalebi & Green, ; Luk et al., ; Rottschy et al., ; Pierce et al., ; Berken et al., ). For bilingual speakers, recruitment of additional brain regions is thought to reflect the need to perform supplementary processes related to control of language, such as suppressing the language not in use, switching between languages, and dealing with representations from multiple languages on an ongoing basis. In fact, cognitive control advantages thought to arise from this type of bilingual experience have been observed for bilinguals from as young as months of age (Bialystok et al., ; Kovacs & Mehler, ) into older age (e.g., Schweizer et al., ). However, IA participants who have discontinued their L do not have experience balancing two languages in their day-to-day interactions, and yet they show neural activation patterns that are the same/similar to bilingual speakers when listening to sounds in their adoption language. This suggests that very early experiences might uniquely influence how the brain becomes organized for language. Specifically, the retention of language specific representations formed during early optimal periods for phonological development might shape the way in which language and language sounds are processed over time (Pierce et al., ). It would be interesting for future research to examine executive functioning in IA children to determine whether and to what extent they benefit from engagement of these general cognitive systems while performing other tasks in comparison to bilinguals with ongoing use of two languages. Results from IA participants also have implications for immigrant groups reported in other chapters in this volume who, despite their L gradually becoming their dominant language, may continue to be exposed to their birth language to some extent. Arguably, the neural language processing of such L learners would be even more influenced by their L than IA participants due to immigrants’ continued L exposure over time, even if that exposure gradually becomes quite insignificant. Growing evidence for a persistent influence from the L over time raises the question of whether a language is ever totally lost and inconsequential for learning and processing other languages. Moreover, it encourages a definition of language attrition/retention that is more nuanced with respect to both the element of language in question as well as how the brain has been, and continues to be, shaped by language experience. While an investigation of IA children’s language attrition has helped to highlight the foundational role of early language experience, there are of course limitations of the work to date and a need for much future research. While behavioural language attrition has been observed in IA children immediately following adoption and neural and behavioural evidence for retention of some aspects of language has been observed in older IA children, adolescents, and adults, there has been no examination, to our knowledge, of the neural processing of the birth language in IA children immediately following adoption, during the period in which they are rapidly undergoing behavioural attrition of that language. Understanding the neural re-organization that takes place during that period would add
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
, ,
to our understanding of the process of language attrition. It would also be informative to examine the extent to which language attrition is influenced by the difference or similarity of the birth and adoption languages. It is possible that more similar languages lead to less attrition and/or more transfer/interference across languages. Alternatively, it may be the case that birth language attrition is more complete if a highly similar language is more readily able to alter existing phonological categories. A similar phenomenon has been observed for episodic memory whereby evoking a similar context will update an existing memory rather than creating a new one (e.g., Hupbach et al., ). In terms of the role of language retention on subsequent language processing, it would also be useful to examine relearning of the adoption language in a systematic way with larger and more carefully controlled groups of participants in order to understand the extent that language retention and reactivation might play in relearning. Relatedly, understanding whether and how the retention of phonological representations for a birth language might influence the learning of more complex aspects of that language (e.g., vocabulary/grammar) later in life remains to be explored.
A We would like to thank the Natural Sciences and Engineering Research Council of Canada (NSERC), the Social Sciences and Humanities Research Council of Canada (SSHRC), the Fonds de recherche du Québec—Société et culture (FRQSC), the G.W. Stairs Foundation, and the Centre for Research on Brain Language and Music (CRBLM) for funding that supported this research. We also greatly appreciate the parents, children, and adoption agencies who generously assisted with this research.
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
......................................................................................................................
( ) ......................................................................................................................
. , - , - , .
. I
.................................................................................................................................. E linguistic experience is widely considered critical for attaining native-like abilities in a language, especially when it comes to phonology and morpho-syntax (Fromkin et al., ; Oyama, ; Johnson & Newport, ; Newport, ). Such childhood language experience seems to have long-lasting benefits, even when it is limited only to early childhood. In this chapter, we first review a series of our investigations into the potential benefits of early childhood experiences with a heritage language on later language (re)learning among immigrant-background adults. We then turn to our newer investigation on how these findings might extend to individuals who were internationally adopted as infants and then re-exposed to their childhood language in adulthood.
. H -
.................................................................................................................................. Many immigrant-background children in countries such as the United States and Canada grow up with a non-majority, heritage language at home. For example, the most recent estimates indicate that % of school-age children in the United States speak a language other than English at home. More specifically, % of Asian American and % of Latino/a school-age children speak a non-English language at home (US Census Bureau, ). However, as they begin schooling, these children often lose their heritage language. Some
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
, , ,
try to relearn their heritage language in adulthood, and they seem able to capitalize on their childhood language memory to acquire proficiency more quickly than their peers without such early experiences. For instance, adult beginning learners of Hindi who had had early experience with the language (during just the first two years of life) were better able to distinguish a difficult Hindi stop consonant distinction (retroflex vs. dental) than their peers who had no such early experience. Moreover, the beginning relearners performed much like adults who had no childhood exposure to Hindi but had been learning Hindi for five years (Tees & Werker, ).
. L
.................................................................................................................................. In one series of studies, we examined the success of childhood overhearers of Spanish in learning Spanish in high school or university classes (Au et al., ; Knightly et al., ; Au et al., ; Au, ). We collected very detailed language background information on all of our research participants, including a specially designed survey, as well as a follow-up interview. This language background assessment revealed a common pattern of childhood language exposure among many adult heritage language learners of Spanish: they only overheard the heritage language as children, hence we called them ‘childhood overhearers’. That is, they only had passive exposure to the language until school age (about age ) and very little, if any, such experience thereafter, throughout childhood. Like the typical late learners of a second language, they did not begin learning Spanish formally until high school or university. More specifically, overhearers typically reported that they heard their parents speaking Spanish with their grandparents during early childhood, but that no one spoke directly to them in Spanish. On average, they reported overhearing Spanish for about nine hours per week for at least three years before they started school, and thereafter they experienced a sharp drop in such overhearing experience. At the time of their participation in our study, the overhearers were enrolled in second-year university-level Spanish language classes in Los Angeles, California, USA. We compared childhood overhearers to two other groups: () native Spanish speakers who had spoken Spanish regularly throughout their lives, in order to assess how native-like the childhood overhearers’ accents were, and () novice Spanish learners who had no prior experience with Spanish until high school or university language classes, in order to compare childhood overhearers against typical late first-time Spanish learners in the US. On a number of measures, we found that childhood overhearers speak with a more native-like accent than novice learners of the language who had not had any prior experience with Spanish. More specifically, the overhearers were similar to native Spanish speakers, and more native-like than novice Spanish learners, in their production of Spanish stop consonants, by phonetic measures such as voice onset time (VOT) and the degree of voicing during the consonant. They also sound more native-like according to accent ratings made by native speakers on speech samples, when compared to novice Spanish learners. Together, these findings suggest that even early passive exposure to a language can provide lasting benefits to the adult language learner. In particular, it is important to note that
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
( )
although these adult learners did not speak Spanish as children, they nonetheless demonstrated benefits in their production of Spanish.
. L
.................................................................................................................................. In another line of research, we studied adult heritage language learners of Korean (Oh, Au, & Jun, ; Oh, Jun, et al., ). In this group, the most common language exposure pattern involved adults who had spoken their heritage language regularly during early childhood, until they started school (about age or ), who we refer to as ‘childhood speakers’. That is, they were monolingual or virtually monolingual speakers of Korean as their first language, but when they started school, they quickly shifted to become virtually monolingual speakers of English. At the time of our study, participants were enrolled in first-year university-level Korean language classes in Los Angeles. Childhood speakers of Korean were compared to native Korean speakers, childhood addressees (those who had only heard but never spoke Korean as children), as well as novice learners of Korean on their perception and production of Korean stop consonants. Korean utilizes a three-way laryngeal contrast in stop consonants, unlike English, which utilizes a twoway voicing contrast. These distinctions are, therefore, difficult for English monolingual speakers to hear and produce. Childhood speakers of Korean in our studies not only outperformed childhood addressees and novice learners of Korean in their production of these stop consonants, they were actually quite native-like in both their perception and production. Our analogous research in Spanish revealed that both childhood speakers and childhood overhearers of Spanish were more native-like in their accent than typical late learners of Spanish as a second language. With regard to areas of grammar particularly difficult for late language learners to master (namely, morpho-syntax), childhood speakers of Spanish— although not quite native-like—outperformed their peers who were childhood overhearers or late learners of Spanish (Au et al., ).
. E
.................................................................................................................................. Our prior research therefore indicates that early childhood language memory can be accessed after a long hiatus, through the process of relearning the language later in life. However, individuals growing up in immigrant families, even if they stop speaking or hearing a heritage language regularly, may continue to overhear the language, or at least the phonemes of the language in the accented speech of their family or community members.
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
, , ,
Such continued exposure could potentially help maintain their childhood language memory. However, if childhood language exposure is completely discontinued, its memory may become inaccessible by adulthood. Indeed, international adoptees who had been adopted from Korea to France between ages and years were no better than native French speakers without prior exposure to Korean at discriminating Korean phonemes or identifying Korean sentences among sentences in several unknown languages (Pallier et al., ; Ventureyra et al., ). In this case, the childhood language memory of Korean adoptees seemed inaccessible according to both behavioural and fMRI (functional Magnetic Resonance Imaging) brain activation assessments. On the other hand, research on children who had been adopted internationally from China to France as infants revealed that, when hearing Chinese lexical tones, the French monolingual adopted children showed brain activation patterns matching those of Chinese-French bilingual children, despite the adopted children having no conscious memory of their childhood language (Pierce et al., , Chapter , this volume). Why was early memory detectable for Chinese lexical tones but not Korean phonemes or the Korean language in general (Pallier et al., ; Ventureyra et al., )? Further research is needed to sort out these contrasting findings. For example, researchers need to better understand how such access to early language memory may be associated with age of adoption. Further, some of these differences may have to do with when suprasegmentals, such as lexical tone, versus segments, such as consonant and vowel phonemes, are acquired; there is some research indicating that infants acquire lexical tones earlier than segments (Li & Thompson, ; Yeung et al., ). One way to render long-ago memory more accessible is to build up its retrieval strength via relearning (Ebbinghaus, ; Bjork & Bjork, ). Indeed, there is growing evidence that although childhood language memory may become inaccessible in adulthood due to lack of use, it can become more accessible through relearning. For instance, after substantial perceptual training of target phonemes, a subsample of English-monolingual adults (those under age ) who had been exposed to Hindi or Zulu during childhood while living abroad with their parents relearned the phonemic distinctions in these languages, whereas those without childhood exposure showed little learning (Bowers et al., ). International adoptees can also successfully access their childhood language memory through relearning. Hyltenstam et al. () compared grammaticality judgements and phoneme discrimination in two groups of adults in Sweden who had been studying Korean: one group had been adopted from Korea to Sweden as young children (between months and years) and one group had been born and raised in Sweden without prior exposure to Korean until adulthood. Although the native-born Swedes had been studying Korean longer, on average, than the adoptee group (. vs. . years) and were better at Korean grammaticality judgements, they were no better at phoneme discrimination. In fact, one third of the adoptees actually outperformed the highest-performing native Swede in Korean phoneme discrimination. Short-term intensive relearning can also help activate childhood language memory. With about , perceptual training trials spread over ten to twelve days focusing on Korean consonants, Dutch adults who had been adopted from Korea as infants and had not learned Korean after adoption improved significantly more across the training period for both perception and production of Korean consonants, when compared to Dutch adults without any prior exposure to Korean (Choi, Cutler, & Broersma, ). Moreover, relearning may not have to be extensive to activate early language memory. Children (mean age = years)
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
( )
who had been adopted from India to the US between months and years were comparable to age-matched non-adoptee English speakers in distinguishing a difficult Hindi contrast (retroflex vs. dental stop consonants) at pretest, but adoptees improved significantly after just sixty-four perceptual training trials with feedback, whereas nonadoptees showed no improvement (Singh et al., ). Not only does relearning not have to be extensive, it may not even require focused perceptual training. In an earlier study (Oh et al., ), we have shown that just two weeks into a beginning college Korean language class, students who had been adopted as infants (all but one prior to age year) from Korea to the US outperformed their classmates who had had no such early experience in Korean phoneme perception. In this case, re-exposure did not focus exclusively on phoneme discrimination as with the perceptual trainings conducted by Bowers et al. (), Singh et al. (), and Choi, Cutler, & Broersma (). However, although the adoptees in our study showed a sizable advantage in phoneme perception, we did not include baseline measurements. It therefore remains unclear whether adoptees’ advantage existed even prior to starting their Korean language class.
.. Re-accessing childhood language memory through relearning: the case of Korean-American adopted adults in Minnesota We have built on these prior findings to better understand whether a brief re-exposure to the target language, in the form of a language class, can help adults who were internationally adopted as infants to access their early childhood language memory. We compared the phoneme perception among four groups: () learner-adoptees: adults adopted from Korea to Minnesota as infants and taking beginning university-level Korean language classes (using an expanded sample that included most of the adoptees in Oh et al., ); () comparison-adoptees: adults from the same community and comparable to the learneradoptees by age and age-of-adoption from Korea but having never taken any Korean language class; () novice learners: classmates of the learner-adoptees in Korean language classes; and () native speakers of Korean. We predicted a phoneme perception advantage for those adoptees relearning Korean over those who were not relearning. This study also set out to break new ground by exploring whether the advantage would extend beyond phoneme perception, to phoneme production. Given the age at adoption of our sample (all but one under age months), it is unlikely that they would have had early speaking experience in Korean, but in light of the evidence that improved phoneme perception would improve phoneme production (e.g., Bradlow et al., ; Choi, Cutler, & Broersma, ), we explored this possibility as well.
... Participants Two groups of adopted Korean Americans were recruited. The target sample (‘learneradoptees’) included nineteen students (mean age = ; years; range: to years; thirteen female) who were enrolled in first-semester Korean language classes at a large public
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
, , ,
university in Minnesota, USA. The language classes met for about five hours each week. Most participants took part in the study during their second week of class (with three participants assessed during the third week of class due to scheduling conflicts). We targeted only those participants who had been adopted before age year (mean age at adoption = months, with one participant adopted at months). Phoneme perception (but not production) data have been reported for eleven of the nineteen target participants by Oh et al. (). The adoptee comparison sample included nineteen participants recruited from the same university as well as the surrounding community who had not taken any Korean language classes. The comparison sample was chosen from a larger sample to roughly match the composition of the target sample on age (mean = ; years; range: to years), gender (fourteen female), and age at adoption (mean = months; range: to months). The two groups did not differ reliably on these variables (age: t () = .; gender: w2 (, N = ) = .; age at adoption: t () = .; all n.s.) Novice learners (n = ) were also recruited from the learner-adoptees’ Korean language classes. These participants were not adopted and had not previously taken Korean language classes. This sample was chosen from a larger sample of novice second-language learners of Korean to match the composition of the learner-adoptee target sample on age (mean = ; years; range: to years) and gender (twelve female) and did not differ reliably from the target sample on these demographic variables (age: t () = .; gender: w2 (, N = ) = .; all n.s.). Like the learner-adoptee sample, the novice learners were assessed during their second week of class (with one participant assessed during the third week due to a scheduling conflict). In addition, twelve native Korean speakers (mean age = ; years, range: to years; eight female) were recruited. They were born in Korea and immigrated to the US in adolescence or later (mean age at immigration = ; years).
... Procedure All participants completed a language background questionnaire, followed by a Korean language abilities test, and then a follow-up interview. The computerized language abilities test was programmed in E-Prime (Psychology Software Tools, Inc.; http:// www.pstnet.com/eprime) and included a phoneme identification task and a phoneme production task.
... Language background assessment The language background questionnaire is a self-report of prior experiences with Korean. In previous research, independent reports from informants who had known the participants as young children largely corroborated the participants’ self-reports using this language background questionnaire (Au et al., , ). The questionnaire includes general questions about language background (e.g., participant’s first language, parents’ languages) as well as specific questions about the participant’s experiences with Korean since birth. Questions address both the quantity (how much they heard, were spoken to, and spoke Korean) and quality (words/short phrases/sentences; extent of mixing with English) of their linguistic experiences during various periods of their lives. A follow-up interview was conducted to clarify and confirm questionnaire responses.
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
( )
... Phoneme identification task During each trial of the phoneme identification task, participants heard one speaker saying two different words (A and B), followed by another speaker saying one of the first two words (X). The participant’s task was to identify whether X matched A or B (ABX technique; Harris, ; Liberman et al., ). A and B always came from a minimal triplet which varied only in the target consonant. The task was presented once with target phonemes in phrase-initial position (i.e., target word only) and once in phrase-medial position (i.e., target word preceded by /i/, ‘this’, or /ne/, ‘my’). As mentioned earlier, the Korean language utilizes a three-way laryngeal distinction in stop consonants (lenis, tense, aspirated), while English utilizes a two-way voicing distinction (voiced, voiceless). Target phonemes were the three denti-alveolar (lenis: /t/, tense: /t*/, aspirated /th/) and the three velar (lenis: /k/, tense: /k*/, aspirated: /kh/) Korean stop consonants. Words were drawn from six minimal triplets of single-syllable Korean words that varied only on the target consonant: three triplets that started with denti-alveolar consonants (e.g., /tal/, /t*al/, /thal/) and three that started with velar consonants (e.g., /kong/, /k*ong/, /khong/).
... Phoneme production task The phoneme production task utilized the same set of minimal triplets from the phoneme identification task. Participants heard a sentence in Korean (‘ige ____’, English gloss: ‘this is a ____’) and were then asked to repeat the sentence after a three-second pause (a beep indicated when they should begin speaking). As in the phoneme identification task, target phonemes appeared in both phrase-initial and phrase-medial positions (separate blocks). Utterances were recorded and speech analysis software (Praat; Boersma & Weenink, ) was used to display spectrograms, pitch tracks, and waveforms of the utterances, from which phonetic measurements of VOT and closure duration (CD) of target consonants, as well as fundamental frequency (fo) of the following vowel, were taken.
... Results . . . . All adoptees were monolingual English speakers (some had learned a language other than Korean in school as adolescents, but none claimed fluency in that language). In the learneradoptee target sample, six participants had no post-adoption experience with Korean until their university Korean language class, six had minimal exposure in childhood, and seven participants had minimal exposure in adolescence only. In the adoptee comparison sample, eleven participants had no post-adoption exposure to Korean, seven had minimal exposure during childhood, and one had minimal exposure in adolescence only. Minimal exposure typically consisted of participation in culture camps for Korean adoptees, ranging from weekly participation to one week per year. The culture camps primarily provide a place for Korean adoptees to socialize and meet other adoptees and their families. They are conducted in English, and some provide minimal instruction in basic Korean vocabulary as part of their curriculum.
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
, , ,
Most of the novice learners had no experience with Korean prior to enrolment in the Korean language class (n = ). The remaining five had some exposure to Korean only in adolescence—usually overhearing Korean American friends in high school speaking Korean. Among the novice learners, eight were monolingual English speakers (some had learned a language other than Korean in school as adolescents, but none claimed fluency in that language); the remaining eleven grew up bilingually, with English and a non-Korean home language (the most common were Hmong, n = , and Chinese dialects, n = ; none of the languages represented are in the same language family as Korean). . . . . A one-way ANOVA with group as the independent variable revealed a main effect of group in overall accuracy on the phoneme identification task, F (,) = ., p < . (see Table . for means). Tukey’s HSD post-hoc tests revealed that native speakers outperformed all of the other groups: learner-adoptee, p < ., d =.; comparison-adoptee, p < ., d = .; novice, p = ., d = .. Importantly, the learner-adoptee sample outperformed the comparison-adoptees, p = ., d = .. They also performed numerically, although not statistically, better than the novice sample, n.s., d = .. The comparisonadoptee sample’s performance was not significantly different from the novice sample, n.s., d = .. The same overall patterns held across consonant types: aspirated, F (,) = ., p = .; lenis, F (,) = ., p < .; tense, F (,) = ., p < .. Post-hoc tests revealed that for aspirated consonants, the native speakers outperformed the comparison-adoptee sample, p = ., d = ., but not the learner-adoptees, n.s., d = ., or the novice sample, n.s. (marginally significant, p = .), d = ., and the learner-adoptee sample outperformed the comparison-adoptee sample, p = ., d = .. No other group differences were statistically
Table .. Mean accuracy on phoneme identification task by group
Overall
Aspirated
Lenis
Tense
Mean
Standard error
Learner-Adoptee Comparison-Adpotee Novice Native
. . . .
. . . .
Learner-Adoptee Comparison-Adoptee Novice Native
. . . .
. . . .
Learner-Adoptee Comparison-Adoptee Novice Native
. . . .
. . . .
Learner-Adoptee Comparison-Adoptee Novice Native
. . . .
. . . .
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
( )
significant for aspirated consonants. For lenis consonants, the native sample outperformed all other groups: learner-adoptee, p = ., d = .; comparison-adoptee, p < ., d = .; novice, p < ., d = .. The learner-adoptee sample again outperformed the comparisonadoptee sample, p = ., d = ., but no other group comparisons were statistically significant. For tense consonants, native participants outperformed all other groups: learneradoptee, d =.; comparison-adoptee, d = .; novice, d = .; all ps < .. Both the learner-adoptee, d = ., and the novice, d = ., samples outperformed the comparisonadoptee sample on tense consonants, ps < .. We also re-ran the analyses excluding adopted participants who had post-adoption experience with Korean during childhood. The pattern of results for overall accuracy was the same, F(,) = ., p < .. Further, as with the full sample, the native sample outperformed all other groups, and the learner-adoptee sample outperformed the comparisonadoptee sample, all ps < .. The one difference was that the novice sample outperformed the comparison-adoptee sample, p = .. . . . . Mean CD, VOT, and f by group for each type of consonant are presented in Table .. As evidenced by the means for the native speakers, CD for tense consonants should be longest, followed by aspirated consonants, then lenis consonants. For VOT, aspirated consonants should be longest, followed by lenis, then tense consonants. As for f, lenis consonants should have lower f as compared with aspirated and tense; f does not consistently distinguish between aspirated and tense consonants (Cho et al., ). Means for all groups numerically followed these patterns. To examine whether each group reliably distinguished between stop consonants on these phonetic features, we conducted a series of planned, paired t-tests comparing consonant pairs on each of these measurements, separately by group. Results and effect sizes are presented in Table .. In all cases, native speakers reliably distinguished lenis, tense, and aspirated stop consonants by CD, VOT, and f (as mentioned above, for f we are comparing only lenis vs. aspirated and lenis vs. tense consonants). The learner-adoptee
Table .. Mean (and SE) CD (in msec), VOT (in msec), and fundamental frequency (f in Hz) of stop consonants by group LearnerAdoptee
ComparisonAdoptee
Novice
Native
CD
Lenis Aspirated Tense
. (.) . (.) . (.)
. (.) . (.) . (.)
. (.) . (.) . (.)
. (.) . (.) . (.)
VOT
Tense Lenis Aspirated
. (.) . (.) . (.)
. (.) . (.) . (.)
. (.) . (.) . (.)
. (.) . (.) . (.)
f
Lenis Tense Aspirated
. (.) . (.) . (.)
. (.) . (.) . (.)
. (.) . (.) . (.)
. (.) . (.) . (.)
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
, , ,
Table .. t statistics (and dfs) and Cohen’s d for Paired t-Test comparisons of phonetic measures of phoneme production by group LearnerAdoptee t (df)
CD
ComparisonAdoptee d
t (df)
d
Novice t (df)
Native d
d
Lenis vs. tense .** () . .* () . .**() . .** () . Lenis vs. aspirated .** () . . () . .*() . . ** () . Tense vs. aspirated .** () . . () . .**() . .** () .
Lenis vs. tense .* () . . () . .**() . VOT Lenis vs. aspirated .** () . .* () . .**() . Tense vs. aspirated .** () . .** () . .**() . f
t (df)
Lenis vs. tense .** () . .** () . Lenis vs. aspirated .* () . .** () .
.** () . .** () . .** () .
.**() . .* () . . () . .** () .
*: p < .; **: p < .
participants likewise reliably distinguished among the stop consonants on all three measures. In contrast, the comparison-adoptee sample only reliably distinguished lenis and tense consonants on CD, only the tense versus aspirated and lenis versus aspirated distinction on VOT, and both lenis versus tense and lenis versus aspirated distinctions on f. Novice participants reliably distinguished among the three stop consonants on CD and VOT, but only reliably distinguished lenis versus tense consonants, and not lenis versus aspirated, on f. We re-ran these analyses excluding adoptees who had post-adoption experience with Korean during childhood. In all cases, the pattern of results was the same and statistical significance was the same in most cases, with the exception of the lenis versus tense distinction on VOT and the lenis versus aspirated distinction on f for the learner-adoptee sample, both n.s., and the lenis versus tense distinction on CD and the lenis-aspirated distinction on VOT for the comparison-adoptee sample, both n.s.
... Relearning as one key to accessing international adoptees’ early childhood language memory The research just reported was to assess whether a brief re-exposure to the target language could help adult learners access their early childhood language memory. We had previously found a sizable advantage for adopted Korean American adults over novice learners in Korean phoneme perception after just two weeks of a beginning university-level Korean language class (Oh et al., ). However, that study did not have baseline data for the adoptee relearners, and so it was difficult to ascertain whether the advantage had already existed before the language class. This study therefore compared a similar group of Korean adoptees with a comparison group of adoptees who did not have post-adoption relearning experience for Korean. Further, we not only examined possible advantages in phoneme perception, but also phoneme production.
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
( )
Overall, there were clear advantages among adoptees after two weeks of re-exposure in both phoneme perception and production, as compared with adoptees who had not experienced such re-exposure. In phoneme perception, the learner-adoptee target sample outperformed the comparison-adoptee sample overall, as well as on each of the three types of stop consonants. It is also worth noting that for all three types of consonants, the comparison-adoptee sample had the lowest mean accuracy of the four groups and was significantly outperformed by even the novice learner sample for tense consonants. As Au & Romo () speculated, based on language instructors’ impressions, novice learners often were the most motivated and diligent learners in language classes, perhaps driven by a perceived advantage of heritage language learners. This might have in part explained the novice learners’ performance on some of the measures in this study. For production, the learner-adoptee sample reliably distinguished among the three consonants on all relevant phonetic measures—just as the native speakers did—whereas the comparison-adoptee sample did so on only some comparisons. Again, even the novice learners did better than the comparison-adoptee sample, at least as measured by stop CD and VOT. We recognize that the comparison-adoptee sample does not represent baseline data per se. Although we matched the learner-adoptee target and the comparison-adoptee samples on key demographic variables, they might nonetheless differ in other potentially important ways. For one thing, the learner-adoptee target sample chose to enrol in Korean language classes whereas the comparison-adoptee sample, at least by the time of this study, had not. Motivation and interest in Korean culture may play a critical role in these group differences. Nonetheless, the present study demonstrated that even very early childhood language memory can become accessible in adulthood. In light of prior findings (e.g., Bowers et al., ; Singh et al., ; Choi, Cutler, & Broersma, ), this is probably due to re-exposure to the childhood language. Although two weeks of relearning in a language class seems a rather brief amount of time to allow for reactivating a long-ago memory, it nonetheless seems sufficient to reveal measurable benefits of early childhood language experience. On some of the phonemic measures, only the learner-adoptee sample, but not the novice-learner sample, outperformed the comparison-adoptee sample. That said, the novice-learner sample also outperformed the comparison-adoptee sample on some measures, despite the former’s lack of childhood experience with Korean, indicating that some of the phonemic contrasts are quite learnable even in adulthood (Flege, ). Although our prior findings supported a rather robust adoptee relearner advantage over novice learners in phoneme perception (Oh et al., ), this group difference was more attenuated in the present study. Nonetheless, the sample means do consistently— numerically if not statistically—support an adoptee relearner advantage, both overall and separately for lenis and aspirated consonants (note that the statistically significant group differences in the Oh et al., , study were also for lenis and aspirated consonants). Perhaps the trend would become statistically significant with a larger sample and hence more statistical power. Further, in production, both groups reliably contrasted the Korean stop consonants on CD and VOT, but adoptee relearners also reliably contrasted consonants on f whereas novice learners did not reliably contrast lenis versus aspirated consonants on this measure. This adoptee-relearner production advantage may reflect their early childhood language experiences: there is some evidence to indicate that the lenis-aspirated f distinction is acquired before the VOT distinction in Korean stop consonants (Jun, ). Hence, adoptee relearners may be tapping into their infant
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
, , ,
language memory in re-acquiring this distinction, revealing their advantage over novice learners. Clearly, further research is needed to clarify the robustness of these findings. Our findings therefore support the growing evidence that childhood language memory can become accessible after re-exposure, and that the re-exposure need not be extensive or intensive as in perceptual training focusing on only phonemic contrasts. The benefits of infant linguistic experience for phoneme perception later in life has been demonstrated previously, but the present study is among the first to demonstrate possible benefits for phoneme production as well (see also Choi, Cutler, & Broersma, ). This is somewhat surprising given the average age of adoption of our sample was months, well before children would have started speaking. Perhaps the re-activated childhood language memory offers phonetic representations of what the phonemes ought to sound like, which in turn helps production of the phonemes (see, e.g., Bradlow et al., ; Choi, Cutler, & Broersma, ). This finding no doubt needs to be examined further, but it does indicate that early linguistic exposure, even limited just to infancy, may provide a variety of benefits to the language learner later in life.
. T -
.................................................................................................................................. The variety of experiences of heritage language learners has brought to light a number of issues related to language experiences. One such issue has to do with an increasingly common experience among children from heritage language backgrounds: the loss of the heritage language early in life, as immigrant-background children often transition to the mainstream language when they begin formal schooling. Much of the existing literature on the importance of early exposure to a language for native-like abilities has assumed that such exposure continues beyond early childhood. The question remained as to whether such exposure, if discontinued after early childhood, could still provide benefits to the adult language (re)learner. There is a growing body of evidence suggesting that the answer is probably yes. Importantly, our latest research breaks new ground in several ways. First, our findings indicate that very early childhood language memory (i.e., from the first year of life) remains accessible in adulthood even after a long period of disuse of the language. Second, along with Choi, Cutler, and Broersma’s () findings, it indicates that such infant language memory can be activated via relearning to help improve not only the perception, but also production, of phonemes. Third, rather than the intensive, focused perceptual training on phonemic contrasts in Choi, Cutler, and Broersma’s () study, it seems that just two weeks of naturalistic learning in typical university-level language classes can also help reactivate such long-ago infant language memory. Research with adult heritage language (re)learners constitutes as a rich arena to better identify the nature of early language exposure that can benefit adult language learners. Given the diversity of linguistic experiences among heritage language learners, this research field is poised to yield more insights into the long-lasting benefits of childhood language memory.
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
......................................................................................................................
......................................................................................................................
. I
.................................................................................................................................. T chapter focuses on a bilingual population that has so far received little attention from the scientific community interested in bilingual language development, in general, and language attrition, in particular: bilingual returnees. Returnees are ex-migrants, also called re-migrants (cf. Daller & Treffers-Daller, ), who have lived for an extended period in a migration context and have, at some point in their life, returned to their (or their parents’) country of origin (Daller, ). Despite the smaller number of studies on bilingual returnees, this group of bilinguals has the potential to give us relevant insights into the changes that may affect the bilinguals’ linguistic competence over their lifetime, due to the significant change that eventually occurs in the speakers’ dominant environment. In fact, the main particularity that characterizes bilingual returnees is related to the substantial input alternation that comes along with the return. In the literature on bilingual development, the term ‘returnee’ is specifically used to refer to second (or third) generation migrants who have spent their childhood (and adolescence) in the host country and moved (back) to the home country either still during childhood, as teenagers or as young adults (Flores, ; Daller & Treffers-Daller, ). In this sense, the term does not include speakers who migrated and re-migrated as adults or who have spent a limited time in the migration country (e.g., the case of exchange students). As pointed out by Daller & Treffers-Daller (), the term ‘returnee’ implies ‘going back’ to a place where the speaker once lived, but this is only true for the returnees who were actually born in their country of origin and immigrated in early childhood. The second or third generation migrants who were born in the host country and only know their parents’ home country from their summer holidays are not true returnees in the literal sense of the word. Thus, ‘returnee’ and ‘moving back’ are commonly used umbrella terms for cases where first generation couples emigrate, live for an extended period of time in the host country, raise their children in the migration context and, at a certain moment, the family (or a part of the family) moves
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
again to the country of origin, therefore not distinguishing between those second generation speakers who were born in the host country and those who moved there in early years. Naturally, this does not mean that this distinction is not relevant, so that the specificities of the migration context that characterizes the speakers have to be described carefully when studying returnee populations. As long as they live in the country of migration, that is, before the moment of return to the homeland, this group of bilingual speakers is indistinguishable from heritage speakers (HSs). They were either born in the host country or moved there at a very young age. As is typical for HSs, they acquired the societal language either as an early second language (L) or as a first language (L) in parallel to the heritage language (HL). Typically, HL bilinguals are exposed to their HL since birth, but throughout their childhood, contact with the societal language increases significantly and exposure to the HL becomes reduced (see the introduction to this chapter). As a consequence, commonly, HL bilinguals become more proficient or dominant in the societal language, which is also the language of their schooling, and they may develop knowledge of their HL that diverges from monolingual norms in some linguistic domains (Rothman, ; Benmamoun et al., a). The moment of return and integration in the homeland society comes along with a significant change in the linguistic environment. The HL, which has been the minority language spoken mainly within the family on a colloquial basis, becomes the societal language. This means that the speaker is now faced with a new input variety that is not restricted to colloquial registers at home but is now widely represented in the environment and in the public sphere. Furthermore, if the returnee is of school age or decides to pursue university studies in the homeland, the HL also becomes the language of formal instruction. Thus, from the moment of return onwards, returnees are exposed to larger amounts and qualitatively different input from the target language (once HL), a situation also known as ‘heritage language reversal’ (Montrul, a). Simultaneously, the returnee is faced with the loss of daily and continued exposure to the language that has been the preferred and, generally, the dominant language until the moment of return, that is, the environmental L (or second L in cases of simultaneous language acquisition). Depending on the families’ efforts to maintain the ex-language of migration and depending on its status in the homeland (e.g., being or not a foreign language taught at school), the degree of exposure to the previously dominant language can vary extremely after return. Contact with this language may be completely lost or just reduced (e.g., to communication with other returnees or to contact in a classroom setting as a foreign language). The two dimensions described above, that is, the loss of contact with the language of migration and the increasing exposure to the HL in the homeland, allow the pursuit of different research questions related to language development. On the one hand, by investigating HSs who experience HL reversal as a consequence of immersion in their HL environment at some point in their development, researchers have the opportunity to evaluate questions related with the long-term effects of divergent language acquisition and its ultimate attainment. Of particular interest is the question of what happens to certain linguistic domains that are shaped in a specific way due to reduced input conditions in childhood, if these conditions change in a post-puberty age. On the other hand, by focusing on the language that stops being the language of daily communication, researchers are faced with situations of isolation from a language that has been acquired naturally in childhood. As in cases of international adoption, this constitutes an optimal context for
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
analysing the loss of language skills. In the following sections research questions related with both dimensions will be reviewed, starting with the latter context, that is, when the societal language becomes the minority one. Since this language may be an L or a second L we will, henceforward, refer to it as an ex-language of migration and not as an L.
. W
.................................................................................................................................. After moving back to their parents’ homeland, second generation returnees are faced with the challenge of inserting themselves in a new cultural environment. Sociologicallyoriented research on second generation returnees has shown that there are several factors, related to questions of identity, belonging, and acceptance, that shape the effort of integration made by this population (Sardinha, ). These factors contribute to the variation found within groups of returnees concerning the wish to continue using the exlanguage of migration. More accentuated needs to fully integrate in the new environment and not to be seen as ‘foreign’ may be reflected in lower efforts to maintain the now minority language. Especially in younger ages, peer pressure may increase the tendency to abandon the other language. Several studies report cases of very restricted or complete loss of contact with this language after moving to the homeland, particularly when the return occurs still in childhood (Kuhberg, ; Flores, , a). This circumstance offers an ideal opportunity for investigating cases of language deprivation, similar to those of international adoptees (see Pierce et al. and Oh et al., Chapters and , respectively, in this volume). From a linguistic point of view, the main difference between adoptees and those child returnees who experience a complete loss of contact with the ex-migration language lies in the age and context of acquisition of the target language. While international adoptees had acquired the lost language as L, generally in a monolingual environment, in the case of returnees the target language was acquired either since birth or in early childhood, when entering (pre)school, always in a bilingual context. The difference lies, thus, in studying language loss in a context of replacement of a first language by a second one, on the one hand, and the loss of an early second language or a second first language, while the other L is maintained, on the other hand. A central research question arising from the observation of child returnees is related with the role of age at return, that is, the age at which the speaker loses continued exposure to the ex-language of migration. There is already a large amount of evidence suggesting that the degree of language attrition varies quite radically between speakers who experienced loss of exposure to the target language before and after puberty (Köpke & Schmid, ; Montrul, ; Bylund, b). This has been recurrently documented in cases of L attrition, that is, when young migrants move to a new linguistic environment and no longer have contact with their L (e.g., Schmitt, ). Several attrition studies focused on child returnees document situations in which children at pre-puberty age had acquired a language naturally in a migration context, but had begun to experience a loss of proficiency in that language once removed from its continuous input (Berman & Olshtain, ; Cohen, ; Olshtain, , ; Seliger, ;
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
Kuhberg, ; Reetz-Kurashige, ; Yoshitomi, ; Tomiyama, , ; Flores, , , a). The degree of proficiency decline reported in these studies varies from a complete loss of productive skills to less significant changes in the speakers’ performance. Overall, the most visible changes are always observed in younger children, which leads to the widespread assumption that the younger the children, the easier they acquire a second language, and the easier they also lose a language that is still not completely established (Köpke & Schmid, ; Montrul, ), that is, young children are ‘better language learners’, but also ‘faster forgetters’. Berman & Olshtain () presented one of the first studies focused on child returnees, setting the ground for this field of inquiry. The authors analysed thirty children from Hebrew-speaking homes, who had lived for an extended period of time in the US and went back to Israel at different ages (from kindergarten to junior-high school age). The informants were assessed during the first year after their return to Israel through different methods of data collection: longitudinal observation based on semi-structured interviews and free conversation and more controlled oral and written language tests. The study aimed to investigate the subjects’ morpho-syntactic and lexical knowledge of English (which was classified as being native-like at the moment of return) during the first year of isolation from the dominant English environment. Results confirmed relevant age-effects in the speakers’ maintenance of English. The youngest children (age to years) showed severe lexical retrieval difficulties and significant effects of attrition in the domains of syntax and morphology. The older children (older than age ), who had already acquired literacy skills and continued to have contact with English in school, manifested fewer retrieval difficulties and a lower degree of morphological and syntactic attrition. Nonetheless, also in their case, and despite further exposure to English, the authors observed increasing effects of transfer from Hebrew as their time away from the US increased. They concluded that ‘before puberty, up until around age or , isolation from a language which they had known quite well before might result in their reaching a “freezing point” beyond which there is no further development’ (Berman & Olshtain, , p. ). The importance of the age at which the child leaves the host country is highlighted in several other studies on child returnees (Olshtain, , ; Flores, , ). Flores (, ), for instance, analysed attrition effects in second-generation returnees who grew up in a German-speaking country and moved to Portugal between the age of and years. Based on a cross-sectional methodology, these studies compared different groups of speakers according to their age of return, focusing on their knowledge of German verb placement and object expression. The results showed that attrition effects in the domain of verb placement were very strong in speakers who had come to Portugal between age and years, but not in returnees who left the host country after age /, even if their length of residence (LOR) in Portugal was significantly longer. It is argued that acquired knowledge needs to stabilize in the child’s mind in order to become resistant to attrition. Language deprivation during a period where knowledge is still not stabilized will, therefore, lead to an unstable linguistic competence, an assumption that also has been advocated by other scholars investigating language attrition (see Köpke & Schmid, ; Montrul, ). From a psycholinguistic perspective the idea of ‘(de)stabilisation of linguistic knowledge’ can be best accounted for by neurolinguistic models such as Yoshitomi’s proposal of a ‘psychological model of the process of language attrition’ (Yoshitomi, , p. ). The author proposes that, during the process of language acquisition, linguistic knowledge is
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
consecutively stored in different levels of memory (working memory, intermediate memory, permanent memory) through synaptic neuronal connections. The consolidation of neuronal ties and subsequent storage in deeper layers of memory (i.e., stabilization of linguistic knowledge) depends, among other factors, on frequent input and language use. This means that in cases of lack of input and use, like the ones reported for bilingual returnees, existent connections are weakened or even eliminated. Linguistic knowledge that has not been sufficiently consolidated begins, then, to deteriorate (i.e., to become unstable). More studies are needed to compare returnees who move back at different ages in order to explore this assumption of a stabilization period and, particularly, the linguistic domains that may be susceptible to stabilization. The fact that the return during childhood may lead to significant changes in the child’s bilingual competence is confirmed by case studies, which periodically observed young child returnees, younger than years, during the first stage of isolation from the ex-language of migration. Two very similar studies report cases of the return of second-generation children of migrant families from Germany, either to Portugal (Flores, a) or to Turkey (Kuhberg, ). In both cases, the return is followed by an almost complete loss of contact with German after re-migration. Both studies report the emergence of attrition effects in several linguistic domains before the first year of re-migration was completed. Kuhberg (), who analysed two Turkish returnees at the ages of and for a period of fifteen to twenty months, reports three different stages of decline. The initial period, marked by native-like fluency in German, is followed by a second stage (five months later), in which the subjects start to produce errors in the use of verbs, articles, and prepositions. Their speech is less fluent and contains several instances of code-mixing. In a third stage, more revealing attrition effects are observed in the subjects’ lexicon (severe word retrieval difficulties), morphology (e.g., use of Turkish suffixes to mark number and case and the use of uninflected verbs), and syntax (transfer of the Turkish word order to German root sentences). At the end of the data collection, the subjects revealed such severe lexical retrieval difficulties that free conversation became impossible. Flores (a) observed very similar stages of decline in several domains of linguistic knowledge in Ana, a girl who returned to Portugal at the age of . Also in this case, after eighteen months of re-immersion in the Portuguese environment, the subject was no longer able to maintain a conversation in German. Interestingly, Seliger () had also reported on a three-stage attrition process (in the domain of word order in dative sentences) in a bilingual girl, who grew up in the US and immigrated to Israel at the age of . The reported research shows that at pre-puberty age attrition effects may appear in different stages some months after the loss of input, affecting different linguistic domains gradually (see also Tomiyama, , for similar conclusions). Although ascertaining stages of attrition, no study on returnees has, however, given clear evidence showing a structured decline process that may be explained by the attrition hypotheses proposed in the literature, that is, the critical threshold hypothesis, proposed by Neisser () and reformulated, for instance, by Hansen & Chen (), or the regression hypothesis, initially proposed by Jakobson () and discussed in many studies on attrition (e.g., Schmid, ; Keijzer, a). According to these hypotheses, the attrition process follows a path of language decline that is the reverse of the path observed in acquisition, that is, the first or the best learned properties will be the last forgotten (see also Andersen, ). Such patterns have been observed in a few linguistic domains in some studies on L attrition, for instance by
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
Schmid () with respect to nominal inflection in German (plural and gender marking). This regression pattern is, however, not consistently found, particularly not in studies on bilingual returnees. What appears to be less controversial, also in the case of returnees, is that the attrition process is selective (Wei, ). Lexical retrieval difficulties tend to be the first sign of attrition, followed by changes in morphology and syntax (Kuhberg, ), whereas the phonological domain remains more robust (Flores & Rauber, ). Selectivity also affects language skills, productive skills being much more robust than receptive ones (Tomiyama, a, ). In contrast to the above-mentioned studies on German as the attrited language, most research focused on children who returned from English-speaking countries reports less significant attrition effects. A substantial number of these studies have been conducted with Japanese or Korean returnees (Reetz-Kurashige, ; Tomiyama, a; , ; Yoshitomi, ; Taura, ; Kang, ; Snape, Matthews, et al., ; see also discussion in Hayashi, , and the studies in the volume edited by Hansen, b), most of them assessing the speakers’ language development in the first two years after return. Interestingly, although describing effects of proficiency decline, no study reports cases of severe attrition of productive skills as shown, for instance, by Kuhberg (). These differences are unlikely to be related with the languages per se, but with the degree of exposure to the ex-language of migration after return. This is an additional and crucial factor that we need to bear in mind in studies on returnees. In addition to being part of the school curriculum, as a compulsory foreign language in most countries, English is also the lingua franca of the globalized world, present in children’s and adolescents’ life through music, films, video games, and social networks. This is obviously not the case with German. Therefore, the frequent opportunity to have contact with English and the desire to maintain the speaking skills of a high-status language eventually prevent a complete loss of exposure. This shows that, along with age, the degree of exposure to the ex-language of migration is a crucial variable in language maintenance after return. Studies on language attrition in situations of re-migration have, therefore, the potential to contribute significantly to the ongoing debate on the role of input and experience in bilingual development (see, e.g., the volume edited by Grüter & Paradis, ). So far, the studies on bilingual returnees have not systematically accounted for variables related with the quality and quantity of input in the same way these variables are carefully analysed in studies on bilingual language acquisition (see, e.g., Unsworth, a), that is, through quantitative measures of a set of variables that express input quantity and quality and their correlation with the degree of attrition observed in the domains under investigation. In the case of child returnees, apart from the age of return and the LOR in the homeland, these variables may include the number of people who speak the ex-migration language to the child, the number of hours s/he has contact with these people, how many of them are native speakers, the number of hours s/he watches TV in the target-language and, importantly, the degree of formal instruction. As for the importance of having received literacy education in the ex-language of migration before and/or after the return, the results of several studies on child returnees show that there is an inextricable connection between age, literacy skills, and degree of attrition/retention (Taura, ; Hayashi, ). Taura (), for instance, shows that Japanese returnees from English-speaking countries who have received their first four years of education in the migration environment retain the ex-language of migration, English,
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
longer than returnees who have not attended school for four years before returning to Japan. Additionally, classroom instruction in the ex-language of migration is also essential for language maintenance after returning to the homeland. After reviewing several studies focused on the role of instruction for language retention in returnees, Hayashi () concludes that instruction in the target language may not be enough to fully prevent attrition if the ex-language of migration is not used in daily life, but it seems to be helpful to slow down the process of decline. Finally, a crucial variable, which needs to be taken into account, is the level of proficiency that the returnee has attained in the target language before the onset of attrition (i.e., the ‘peak attainment’, as proposed by Bardovi-Harlig & Stringer, ). Actually, only knowledge that has been acquired before the change of input happened can actually be lost, as stated by Sorace (): In order to determine the effects of attrition, it is essential to ascertain what speakers knew when the attrition process began, since by definition attrition can only affect what was within the speaker’s knowledge. (Sorace, , p. )
. W
.................................................................................................................................. The return to the homeland does not only imply input changes related to the ex-language of migration, but also to the HL. In this case the reverse situation occurs. From a certain moment on, the language that was acquired under reduced input conditions becomes the majority language. If the number of studies focused on the returnees’ ex-language of migration is scarce, investigation centred on the development of the HL after return is even more limited (Treffers-Daller et al., ; Daller & Treffers-Daller, ; Flores & Rato, ; Treffers-Daller et al., ). Still, this line of investigation has the potential to make an important contribution to the research on HL development. Assuming that HSs’ linguistic competence may diverge in some domains from monolinguals’ knowledge, as has been widely shown in an extended number of studies (see introductory chapter to this section), a crucial question that arises in the context of return is whether the HL returnees’ competence converges towards the monolingual norm after their full immersion in the HL environment (see discussion in Daller & Treffers-Daller, ). This general question is linked with several specific research questions, which are all still understudied. First, it allows investigation into which domain speakers’ competence may change upon return, that is, which will show convergence towards monolinguals’ competence after some time, and which will remain unaffected by input changes, that is, which domains or properties are shaped in early years of development and resist further changes. Ultimately this question addresses the malleability of the HSs’ language competence and allows insight into the domains that may be subject to the effects of maturation (and those that are malleable over the lifespan). Second, it may provide evidence for the source of difference between HSs and their monolingual peers, namely if potential divergence detected in the HSs’ performance is the
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
consequence of a process of language acquisition that per se diverges from monolingual acquisition due to reduced exposure to the HL (in line with Montrul’s () proposal of incomplete acquisition of a HL), or if it stems from qualitative differences in their input (as stated, e.g., by the missing-input competence divergence hypothesis, proposed by Pires & Rothman, ) that may be overcome in the case of return, that is, full immersion in the HL environment. Furthermore, it addresses questions related with LOR, that is, how long HSs need to be immersed in the HL environment in order to experience competence changes and if there are parallels or not with L acquisition. Finally, it may provide valuable insight into the issue of reacquisition of a language that was acquired naturally in early childhood but stopped being used by the speaker. These are the cases where HSs have lost their HL before the moment of return and have to relearn it afterwards. Two studies carried out in Turkey on Turkish-German bilinguals who grew up in Germany and returned to Turkey as adolescents (see Daller & Yıldız, ) show that the returnees’ overall language proficiency in Turkish, measured by a C-test, developed positively up to eight years after return. The returnees’ results were significantly lower than the monolinguals’ scores about two years after return, but returnees were statistically comparable to their monolingual, age-matched peers, after living (and studying) for eight years in Turkey (note that, still, returnees showed a larger spread of scores than the monolinguals, revealing more variable knowledge). This early study gives preliminary insights into the question of whether HSs’ competence matches that of monolinguals after some time spent in the HL environment. However, in future research, specific domains of the HSs’ knowledge should be investigated. Particularly, it is necessary to establish a direct link with results on the linguistic competence of HSs who still live in the L environment, in order to analyse precisely those properties that have been shown to be vulnerable in the acquisition of HL (for the discussion of these linguistic domains see, for instance, Montrul, a). An interesting contribution to this field is provided by Treffers-Daller et al. () in a study also focused on Turkish returnees from Germany. Similarly to the above-mentioned study, the authors analyse two groups of returned HSs of Turkish, secondary school-aged adolescents who had been living back in Turkey for about one year and university students who had returned to Turkey seven years previously. The focus of their research is a language domain that is very hard to acquire and to be fully mastered by late L learners, namely collocations and fixed expressions (in this case with the verb yap- ‘to do’). By showing that the HL returnees start to use target-like collocations seven years after return (while avoiding or using them in a target-deviant way before or right after the moment of return), the authors clearly distinguish between HL and L acquisition. They argue that competence changes after immersion in the HL environment are triggered by increasing input, contrasting with the difficulties age-matched L learners display in mastering these structures. Research on HL returnees may, furthermore, provide important insight into the phonetic-phonologic competence of bilingual speakers. Although it is frequently argued that early exposure to a language enhances a native-like accent, research into HSs’ pronunciation and perception skills indicate that exposure to the HL since birth may not be a guarantee per se for the development of a native-like competence. The dominant presence of the majority language from early on may trigger contact-induced changes in the HL sound system, leading to the development of accented pronunciation (e.g., Godson, ; Kupisch, Barton, Klaschik, et al., ). Additionally, studies on L attrition indicate
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
that these changes may even occur when immigration takes place in adulthood and lasts over an extended period of time (De Leeuw et al., ). This suggests that pronunciation may also be susceptible to input changes and malleable over the lifetime. Research on HL returnees may help to complement this picture by analysing the development of their perception and production skills after immersion in the environment where the once minority language becomes the societal language. The first results of a study on Portuguese returnees, who grew up in Germany and Switzerland and returned to Portugal between the ages of and years, show that the age at which the HSs were immersed in the L environment influences their HL accent more than the age at which they returned to the homeland and the length of stay after their return (see Flores & Rato, ). The later the speakers were immersed in the German environment (at ages between and years), the more native-like they were rated by Portuguese monolinguals in a foreign accent rating task. However, no correlation between native-like ratings and a longer LOR in Portugal or a younger age at return were found. This indicates that accent is predominantly shaped in the early years of development and that non-native traces in the returnees’ HL accent may last over a lifetime and never be overcome. These results obviously need to be complemented by the investigation of other language pairs, taking into account variables such as ‘age of migration’, ‘age of return’, ‘length of residence in the L and back in the L environment’, and various factors regarding input quantity and quality.
. C
.................................................................................................................................. This chapter provided a comprehensive overview of the research conducted so far on bilingual returnees and highlighted the research potential that has yet to be explored in this domain. Essentially, the questions addressed in studies focused on this population replicate hypotheses and research aims that generally constitute the research programme of studies in the domain of L and L attrition and on HL development, specifically the relationship between the change of the speakers’ competence and the role of extra-linguistic variables such as quantity and quality of exposure, age and literacy skills, the nature of the change process, and its theoretical foundations. The particular case of losing continued contact with the so-far dominant language and, in parallel, experiencing increasing exposure to a language that was acquired under conditions of reduced contact, which characterizes this population, allows researchers to adopt a two-sided perspective on the process of competence change, by exploring the path of loss of the ex-language of migration on the one hand, and the development of a HL in circumstances of reinforced input on the other. Overall, research on these two dimensions has shown that the returnees’ language competence is malleable in many domains and tends to develop positively in the case of the reinforced HL and negatively in the case of the ex-language of migration, especially in child returnees in situations of severe isolation from the target language. Commonly these two dimensions are explored separately by different studies; however, ideally, investigation into bilingual returnees should adopt a unified approach, which addresses both sides and, thus, contributes to a complete picture of language development in cases of re-migration.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
C
..............................................................................
T
.................................................................................................................................. This volume shows how the field of language attrition has reached adulthood and acceptance as a relevant research topic and an area of meaningful applications. The fact that so many established researchers have contributed to this volume is a sign that research on various forms of attrition has begun to impact on other research areas and paradigms. A strong feature of the current attrition field is that the various theoretical approaches in Applied Linguistics (AL), in particular with respect to innateness and Universal Grammar (UG) versus emergentism and usage based approaches, seem to live peacefully alongside each other. There are still borders not to be crossed, but there may be no other subfield of linguistics in which there is such harmonious cohabitation. This is a very positive result from the editors’ systematic refusal to take sides in various contentious discussions. While in the past this may have been seen as a weakness, this volume shows that the editors were right in steering clear from prioritizing specific approaches to language and language development over others. In a way the early attrition research could be called theory-light, with Ebbinghaus’ still unsurpassed contribution and ideas on regression and recapitulation worth our attention as a challenging perspective even today. Things have changed, with more prominent researchers taking on the challenge to study attrition empirically based on theoretical notions mainly from cognitive science (see MacWhinney, Chapter , this volume). The Regression Hypothesis (see Barkat-Defradas et al., Chapter , this volume; Mehotcheva & Köpke, Chapter , this volume) has seen a remarkable comeback in the work by Osterhout et al. (Chapter , this volume), who used Event-Related Potentials to assess rule sensitivity by second language (L) learners of Finnish and found different patterns at different stages of acquisition and attrition. They conclude that their results support a processing-based version of the regression hypothesis, with learners gradually transitioning from a lexical processing mode to a grammatical processing mode during the acquisition phase, with the reverse transition occurring once instruction has ended. Whether L learners indeed move from lexical processing to grammatical processing and (L) attriters from grammatical to lexical processing is still to be proven. The fact that they move from a N to a P and back again may suggest this, but without detailed analysis of language use ‘in the wild’, this assumption needs to be seen as interesting but tentative.
Defining language Interestingly, few of the contributors attempt to clearly define what language is, and this has ramifications for their understanding of attrition. Is it a set of rules with a lexicon or is it a set of situation-specific utterances? These perspectives on language differ in what counts as attrition and
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
how attrition can be studied empirically. Larson-Hall (Chapter , this volume) argues that studying the decay of separate words does not do justice to the idea that the lexicon is not a collection of separate words, but an integrated network in which word representations are interconnected within the system but also with elements outside the system. Taking her argument a few steps further, we may even consider the possibility that there are no separate words in the lexicon at all, only structures which are strongly interconnected and can only be activated and retrieved in a relevant context. These structures are not the equivalent of words, but of bundles of syntactic rules and meaning components. The word as a unit in language processing is then essentially a metalinguistic notion which linguists can use to analyse speech and writings. From a practical or didactic perspective this means that learning from wordlists is likely to be the least efficient way to develop the language system. Following notions from communicative language teaching, language only exists in use, and language development is the enhancement of communicative settings the learner can deal with. So it is not the words and rules that define language development but the increase of the number and types of communicative situations mastered. This is where UG and Usage-based (UB) approaches differ fundamentally. The early research into attrition was mostly not situated in a specific theoretical framework and only in the last five to ten years has interest in attrition within the UG community emerged, as the contributions of Putnam, Slabakova, Sorace, Montrul, and several others in this volume show. It turns out that many of the bones of contention in the field of second language development could be fruitfully studied through research on attrition. While many studies on attrition from an UG perspective have led to testable assumptions, a systematic evaluation of the UB approach from an attrition perspective seems to be missing. For a research field to flourish, numbers and weight matter. Leaders in the field need to be involved to make it clear that this is a viable topic of research. In the early days there was a small but dedicated group of researchers, held together by Monika Schmid through her research meetings in which early-career researchers were given advice on their research (the European Graduate Network on Language Attrition, see Schmid & Cherciov, Chapter , this volume). ‘We’ would be present as a group at conferences like AAAL’s annual meetings, AILA and ISB and socialize together and there was a clear identification as an ‘attrition-researcher’. While we can lament the decline of such identification, simply because there are too many researchers now working on attrition, the risk of ‘inbreeding’ that is typical of small research communities has faded. The old guard has seen its position as forerunners fade and now has to work hard to keep up with new paradigms, new developments in paradigms, and new techniques for data collection and analysis.
Theoretical approaches: attrition and Universal Grammar The interest in attrition from the UG community has definitely enhanced the status of attrition research as a viable and relevant area of research. Theories and research topics need to be supported by researchers with some academic weight to be accepted more widely. To quote the Alliance for Useful Evidence statement: ‘Research data only really become information when they attract advocates for the messages they contain’ (Nutley et al., , p. ). The various contributions to the present volume written from a UG perspective show that new types of data lead to new questions. Putnam et al. (Chapter , this volume) state that the core ideas of UG can be extended to research beyond the traditional generative paradigm, including an extension to probabilistic models of linguistic analysis. In a similar vein, over the past years Sorace’s () Interface Hypothesis (IH), which proposes that structures at the interface between syntax and other cognitive domains, such as pragmatics, are more likely to undergo attrition than structures that do not involve such an interface, has gained importance. Slabakova (Chapter , this volume) presents the Bottleneck Hypothesis. Its main tenets are that functional morphology presents the biggest challenge to acquisition and attrition, while syntax and semantics are easier to acquire and maintain because they employ universal operations. This presents
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
a useful approach to assess the relative vulnerability of linguistic elements at different levels. To what extent the Bottleneck Hypothesis is complementary to Sorace’s IH, however, is not clear yet.
The balance hypothesis One of the key issues in the study of multilingualism is whether there is some sort of balance in the sense that adding a new language may go at the expense of a language acquired earlier. This raises a host of questions as Yılmaz’s contribution (Chapter , this volume) shows. To begin with, a distinction needs to be made between storage capacity and attentional resources. The storage capacity of the human brain is almost unlimited. Reber () mentions that there are about a billion neurons in the brain that each have an estimated , connections with other neurons. The amount of information that can be stored in this space, which can be conceptualized as a million gigabytes, translates into three million hours of TV shows. You would have to leave the TV running continuously for more than years to use up all that information. Compared to that, adding another language, a system that is supposed to be efficient and flexible, should not pose any problems in terms of storage. But for language attrition, storage is not the issue, it is retrieval under heavy time pressure that matters. College students are assumed to have knowledge of some , words. In language production we produce about two to three words per second, so the brain has to find the right word out of , in about – ms. (Levelt, , p. ). Yılmaz suggests that enhanced retrieval time for lexical items may be indicative of language attrition. This is not supported by findings by Grendel et al. () who looked at the attrition of French as a foreign language in the Netherlands. While they found a significant effect of years of education in French, no change was found in latencies in a production task and a perception task after two to twenty years of non-use. In other words, we need more data first showing longer latencies in reaction time experiments with non-use before we can conclude lexical retrieval is the problem in attrition. A relevant finding in the Grendel data, but also in earlier data by Weltens, is a discrepancy between measured decline and experienced decline. Selfevaluation data showed that the participants felt that they had lost a lot, while the performance data showed no correlation with the self-evaluation (see Mehotcheva & Köpke, Chapter , this volume, for more details).
The tidal waves of research on foreign language skills As the Mehotcheva & Mytara overview (Chapter , this volume) shows, there is a large set of studies of foreign language attrition from the s and s, but the interest seems to have waned after that only to re-emerge in this volume. That is remarkable, since the very early attrition studies by van Els et al. had an explicit language policy goal that could be summarized as: ‘what is the point of teaching foreign languages for so many hours in high school and college when these skills attrite so rapidly?’ (see Köpke & Schmid, ). The relevance of this type of research has not really declined. With foreign language education being threatened in many countries, attrition research can show what methods lead to better long-term retention of foreign language skills and accordingly a more efficient use of instruction time.
Attrition and international adoptees ‘The new kid in town’ is no doubt the study of attrition in international adoptees. The first studies were done by Pallier and Ventureyra with Korean adoptees in France (e.g., Pallier et al., ; Ventureyra et al., ). The general findings of the early studies is that no traces of retention of Korean could be found, not even in adults who were adopted when they were – years old. The
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
literature and the more recent studies share a disbelief, often unspoken, that no traces can be found. The study by Oh et al. (Chapter , this volume) reflects that perspective. They used a relearning paradigm in the form of classroom instruction to reactivate adoptees’ phoneme production and perception. Their conclusion is that ‘overall, there were clear advantages among adoptees after two weeks of re-exposure, in both phoneme perception and production’ (p. ). Their data support an adoptee relearner advantage in the area of phonology and phonetics, though not in other linguistic areas such as the lexicon or grammar. Pierce et al. (Chapter , this volume) also look at internationally adopted children and their development. They refer to studies using brain-imaging techniques and suggest these data point to ‘retention of birth language representations’ (p. ). This is controversial. We simply do not know what birth language representations are.
The unavoidable age factor Bylund (Chapter , this volume) discusses the role of age in attrition. He summarizes the situation in which many factors are unknown as follows: Age effects in attrition consequently reside in the relative degrees of attrition exhibited by children and adults. All other things being equal, children should exhibit higher degrees of attrition than adults. If such a pattern is not attested, then age effects are lacking. (p. ) Obviously ‘all other things being equal’ is impossible, and as we do not, at present, know much about the brain mechanisms behind language attrition, we also cannot determine which ‘things’ should be ‘equal’. The larger question with respect to this issue is to what extent factors that play a role in acquisition and learning are similarly relevant for attrition. If acquisition and attrition combine into ‘language development’, then we can look at what factors play a role in both processes. Amount of contact/input, personality, attitudes and motivation, age, and aptitude are all likely to play a role, both as independent variables but also as factors that interact over time, which can lead to unpredictable patterns of development (Van Geert & Van Dijk, ). Bardovi-Harlig & Stringer (Chapter , this volume) also discuss age and the critical period. Their meta-analysis of research in first and second language acquisition and attrition reveals remarkable resilience for aspects of phonology as far as it is not subject to cross-linguistic influence. This is a problematic idea, since changes in accents have been reported frequently and it is the cross-linguistic influence that takes centre stage in research on the attrition of phonetics (see Chang, Chapter , this volume; de Leeuw, Chapter , this volume). Though the authors focus on phonology only, the problem of comparisons between levels of description (see below) in quantitative terms does become relevant here.
Some problematic areas Units of analysis in attrition research A commonly reported finding in research that looks at different levels of language processing, such as lexical access, grammatical procedures, and phonological representations, is that words are more vulnerable than units at other levels (see Bardovi-Harlig & Stringer, Chapter , this volume). This is a problematic issue that none of the contributions refer to: what is the unit of analysis for such a comparison? How many lost words are the quantitative equivalent of the loss of one syntactic procedure or the merging of two vowels? This problem is not specific to attrition research, but more generally for research on language development. Is it the case that there are simply more words that can be lost than procedures, pragmatic rules, or phonemes, as far as we can define these as entities?
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Incomplete acquisition and incomplete transmission between generations One of the methodological challenges of attrition is to assess whether knowledge that seems to be lost has actually ever been acquired or not. This relates to the problem of defining a point of reference for measuring attrition. There is no really solid way to assess the highest point of development for a given language in a given individual. Several studies have used self-reports to establish some sort of reference point by asking about self-rated proficiency ‘now and then’ (Grendel, ; Hulsen et al., ). This question is also part of the background questionnaire developed in the Graduate Network and available on languageattrition.org. Montrul () and Riehl (Chapter , this volume) present data on incomplete acquisition, while in studies on minority languages and heritage languages (see Part VI) the term used is incomplete transmission. This idea goes back to Gonzo & Saltarelli’s () early ‘cascade-model’ of the decline of minority languages in which attrition and incomplete transmission within and between generations is combined. The first generation with reduced opportunities to use the language will show attrition. The attrited version of the language will be partly transmitted to the next generation, whose language will thus be affected by both attrition and incomplete transmission which leads to substantive decline within two generations. Similar notions underlie many recent theories on heritage language acquisition (e.g., Bayram et al., Chapter , this volume).
Attrition and heritage languages Has the heritage connection made research on attrition more mainstream and relevant for other subdisciplines? One cannot escape the impression that interest in language attrition, in particular in the US, has been triggered by the massive interest in what came to be known as heritage languages. In Europe, the notion of heritage languages did not really catch on, and more commonly used labels include language shift, language attrition, and partial transmission between generations. It is not clear how the addition of the now very popular notion of heritage languages and accordingly the use of a range of definitions would help to make the situation of minority languages in the European situation clearer. Neither can one escape the impression that in the discussion of heritage languages, some reinventing of the wheel took place, for example in discussions on the point of reference in comparisons over time. Probably the more structuralist perspective typical of the European approach did not match with the language and social identity perspective taken by American researchers (see Bayram et al., Chapter , this volume).
Suprasegmental aspects of language attrition There are now substantial number of studies that looked at phonetic aspects of attrition. Chang (Chapter , this volume) refers to phonetic changes that occur in one’s native language (L) due to recent experience in another language (L), a phenomenon known as phonetic drift. Most of the studies on phonetic drift looked at segmental aspects, but influence of the L has also been reported in intonation patterns in first language attriters (Waas, , p. ; Clyne, , p. ). These anecdotal impressions have now been empirically supported in a study by de Leeuw et al. (), who found a merged alignment of the prenuclear rising accent in German and English in German-English late bilinguals. Mennen () seems to be the only study that looked at attrition of intonation, which continues to be a problematic area of research. Attempts to link intonation patterns with meaning, apart from intonation patters for questions have been largely unsuccessful. L intonation is hard to learn and to unlearn. The well-known tendency of French learners of English to put the stress on the last syllable of a word appears very ingrained. Phonetic changes that occur in one’s native language (L) may be due to recent experience in another language (L). Intonation is one of the aspects of language that are still poorly understood. In Levelt’s A Blueprint of the Speaker () it is unclear how intonation is generated, since it is not linked to specific words, but informs whole utterances, but how
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
that is done remains a mystery. Interestingly, something similar is at stake with respect to the use of gestures. It is probably no coincidence that these two aspects have not been studied systematically with respect to attrition.
The attrition of gestures and sign languages Similar to the discussion on the attrition of intonation, it is unclear at present to what extent gestures may be affected in this development. In the contributions to this volume, no reference is made to the role of gestures and how that may change over time. In research on language development, a considerable number of studies have focused on gestures (see for a collection of studies on this topic Gullberg & de Bot, ). Research on gestures is complex. The primary data require substantial time and effort to carefully transcribe the gestures. An interesting question is whether there is something like gestural drift. Some recent studies have shown that the assessment of cross-linguistic influences, such as a foreign accent in gesturing, is very hard to establish. Attrition assumes some point of reference, and we are still very far from anything like that for gestures. There may be differences for different types of gestures. Some iconic and beat gestures tend to be present in different languages, others are more language and culture specific. Gestures are only one aspect of non-verbal communication which is known to be a crucial aspect of communication. No research seems to exist that looked at attrition of this aspect. There also seems to be little research on the attrition of sign languages. Yoel () reports on the attrition of Russian sign language in Israel. The data suggest that sign languages in contact show the same cross-linguistic patterns as spoken languages. The lack of research on this topic may be explained by the fact that not many native signers are in a situation like the Israeli one in which two sign languages are used by a substantial group of people. Not much can be said about this issue since the research is basically missing.
Attrition as a process rather than a state Research on language attrition is typically done through a cross-sectional design in which the point of reference is often unclear. In her contribution to this volume (Chapter ), Opitz presents an excellent overview of the relevance of Complex Dynamic Systems Theory (CDST) for the study of language attrition. It seems that CDST is the only theoretical perspective in which attrition is seen as a process rather than a state, see also Linck & Kroll (Chapter , this volume). In typical attrition research a specific set of linguistic elements is studied by comparing knowledge or skills at some moment in the past when no attrition is assumed with similar data at a later moment in time. It is typically a ‘one, maybe two shot’ design. Ideally the data are longitudinal in nature and dense, in order to track changes over time on different time scales. But language attrition is a slow process and the chances of getting a thirty-year longitudinal study financed are slim. So most researchers revert to various methods to collect cross-sectional data. Such data do not take into account these changes at different time scales. While authors tend to claim that they are interested in the process of language attrition, they do not actually look at the process, but only at the product in terms of grammaticality judgements, language tests, and various reaction time paradigms. Linck & Kroll argue that existing models of language attrition—particularly those that take a dynamic systems approach—may benefit by incorporating general mechanisms of memory retrieval, such as retrieval induced forgetting (RIF), as one of multiple mechanisms driving language change. (Linck & Kroll, Chapter , this volume, p. ) A major contribution of a CDST perspective is that many types of variables that in traditional research are treated as fixed, static, and not-interacting are in fact variable and interacting with other variables. An example is the notion of language aptitude. This has long been seen as a static and
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
invariant variable, but recent research had shown that aptitude may be influenced by language learning rather than the other way around.
Attrition at different time scales Higby et al. (Chapter , this volume, p. ) raise a very relevant point when they argue that ‘attrition may be much harder to investigate than learning because it happens on a much longer time scale. In fact, it is still unclear what the time scale of attrition is at a behavioral level.’ As argued in de Bot (), language development takes place on many interacting time scales. Some aspects like phonology may change quickly, probably on the scale of weeks, while complex syntax or pragmatics may take months or years. Also the cognitive processes behind the behavioural data may function on time scales that range from the milliseconds of lexical retrieval in tasks like lexical decision or picture naming to the development of narrative skills that have been reported to continue to become stronger even after on a life span scale. Different variables may have different time scales and time scales may interact. For example learning new vocabulary may allow for more complex syntax to be used, but more complex morpho-syntax may make the detection of the stem in a morphological complex lexical element more difficult. This in turn can lead to more periphrastic constructions being used which reduces the frequency of use of such complex units. Input and frequency are core components in development, and development is an iterative process in which skills build up through use and input.
T
.................................................................................................................................. The contributions to this volume show the breadth and depth of current research on language attrition. Articles on attrition are regularly published in the leading journals in the field. It acts as a testbed for various linguistic subdisciplines, and there is a healthy interaction between representatives of different linguistic schools. What started as a minor and, according to some, rather sad topic is now a full-fledged internationally respected subfield with its own conferences, books, and standards. Whether the field will grow, stagnate, or decline depends on many things, in particular Monika Schmid’s willingness to invest a part of her enormous energy in this topic.
Acknowledgement The author is grateful for Monika S. Schmid’s comments on an earlier version.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
ANNOTATED BIBLIOGRAPHY ........................................................................... .
Note: An earlier version of this overview was published under: M. S. Schmid (). Language attrition: timeline. Language Teaching (): –. The original content is reproduced here with the kind permission of Oxford University Press. For the purpose of the current publication, the piece has been expanded and updated. We know a fair amount about how people learn languages; we know remarkably little about how language skills, once learned, are forgotten—whatever that means precisely [ . . . ]. (Lambert, , p. ) This statement opens the first collection of papers that specifically consider the deterioration of linguistic knowledge ( & , ),1 a field which in one of the papers in the volume is referred to as being in an ‘antenatal’ state (Berko-Gleason, , p. ). Thirty years down the line, while some researchers investigating language acquisition might quibble with the claim that ‘a fair amount’ is known in that field, it is still a fact that far less is known about the loss or attrition of language skills. Language attrition research has developed in several relatively clearly delimited phases spanning, roughly, each of the three decades between and (see Köpke & Schmid, , for a more detailed overview and analysis). The first phase was an era of stocktaking, with a number of symposia, collected volumes, and special issues of journals. All of these collections cast their net widely, attempting to cover not only what, under the current terminology, would be considered ‘attrition proper’, namely the nonpathological loss of previously acquired first or second language skills in adult speakers (Köpke & Schmid, , p. ), but also such diverse phenomena as incomplete acquisition, language contact and death, dialect change and death, pathological language loss, and the deterioration of sensitivity to phonetic input among infants (this last aspect is treated by Burnham, ). These were usually accompanied by reports on small-scale or pilot studies which were intended to lay the background for larger and sometimes very ambitious investigations. Unfortunately, most of these were never actually carried out or completed. A clear theoretical orientation was often lacking in these collections and pilot studies (as is, for example, pointed out in Hill’s review of the volume by , , & (eds.)). In , the first collection that specifically focused on changes to the native language appeared ( & (eds.), ), marking the beginning of a decade of a more focused, theoretically and empirically driven approach to language attrition, characterized mainly by a limited number of comparatively large-scale investigations of first language attrition, usually in the form of PhD projects.2 This resulted in more clearly defined theoretical foundations and predictions for the field from a variety of perspectives, such as the impact of socio-/ethnolinguistic factors, psycholinguistic questions of accessibility, and linguistic and grammatical models. What was (largely) lost from Citations in refer to publications which have their own entry in the timeline table below. An overview of PhD theses on language attrition as well as download links or, where known, sources from which they can be obtained can be found on https://languageattrition.org/resources-for-researchers/dissertations/ 1 2
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
the earlier period was the intensive interaction and exchange between researchers that had taken place in the numerous meetings and collected publications, as the work that was carried out was largely confined to the typically solitary existence of the graduate student. From onwards, numerous efforts were made to again create a community and network of attrition researchers and to arrive at commonly agreed methodological approaches as well as a sound theoretical underpinning for this research. This decade saw a number of international symposia dedicated solely to attrition research as well as panel sessions at larger conferences on bilingualism, the formation of a graduate network and the publication of numerous collected volumes and special issues of journals. It is probably fair to say that this has resulted in a better visibility of language attrition research and the recognition that the contribution of research and findings from this area can provide important further insights into highly topical research questions, such as the issue of interfaces in grammar and bilingualism (e.g., ., ; Sorace, ) and the question of maturational constraints in second language acquisition (e.g., , ; , ). Again, a fair number of PhD projects on a variety of languages and settings were carried out in this period, many of which adopted the test battery proposed within the attrition community (made available on https://languageattrition.org and described in detail in , a). The overview presented below focuses on studies investigating first language attrition; articles on second language attrition, incomplete acquisition, or child language attrition, heritage or indigenous minority languages, etc., are not included. I also do not list general overview, handbook or encyclopaedia articles. Where not otherwise indicated, participants in the studies are post-puberty long-term migrants (minimum length of residence (LOR) seven years). The items represented in the timeline represent the following themes: . Theoretical frameworks a. Generative theory b. Psycholinguistics and neurolinguistics c. Language contact and language change d. Regression e. Multicompetence and Dynamic Systems Theory . Comparisons of attrition and other linguistic developments a. L acquisition and attrition b. L acquisition and attrition (in particular studies investigating maturational constraints) c. Attrition and heritage languages d. Attrition and language change . Attrition as a selective process a. Lexical access and fluency b. Restructuring of the semantic and conceptual space c. Morphological and morpho-syntactic attrition d. Syntactic attrition e. Phonetic and phonological attrition . Socio- and ethnolinguistic considerations a. The impact of L use and behaviour b. Attitudes and identity c. Attrition and ageing . Methodological considerations a. The appropriateness and validity of various tasks b. Testing the impact of external factors c. Baseline and point of reference . Overview papers/collections and general considerations
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Table A. Applied studies of L attrition in an L context Study
Findings
Strand
Lambert, R. D. & B. Freed (). The Loss of Language Skills. Rowley, MA: Newbury House.
This collected volume was the outcome of a research meeting at UPenn in . It was the aim of this conference to arrive at a common foundation upon which language attrition research could be conducted. The papers collected in this volume focus on the linguistic and extralinguistic variables which should be considered in attrition research (Andersen), the instruments that might be used to measure it (Clark; Oxford) and the social factors (in particular attitude and motivation) that should be taken into account (Gardner). Researchers from neighbouring areas, such as language death (Dorian), child language acquisition (Berko-Gleason), instructed L learning (Valdman), neurolinguistics (Obler), and language policy (Levy; Lowe) give an account of how the findings from their fields should inform and shape attrition research.
Lambert & Freed put together an extremely useful collection of papers and an excellent starting point still for anyone wishing to begin a research project on language attrition. Unfortunately, it has not been used as widely and consistently as it might have been. A better uptake would no doubt have benefited attrition research at large enormously. van Els, T. (). An overview of European research on language attrition. In WELTENS, DE BOT, & VAN ELS (eds.), –.
This is the paper usually credited with first proposing a taxonomy that is very often cited in the introductory sections of attrition studies. It classifies the type of attrition research in a two-by-two design by attriting and environmental language (L and L, respectively). However, only two of these types—L attrition in an L setting, and vice versa—have developed into actual areas for attrition research, while the other two (dialect loss and L ‘regression’ among older migrants) have never played a consistent role and are considered debatable cases of attrition. The usefulness of the taxonomy has been questioned by, for example, Köpke & Schmid ().
Although van Els is almost invariably cited to support this taxonomy, it was first proposed in a paper by de Bot & Weltens, published in Dutch. Jaspaert, K., S. Kroon, & R. Van Hout (). Points of reference in first-language loss research. In WELTENS, DE BOT, & VAN ELS (eds.), –.
This paper discusses one of the most important methodological considerations for language attrition studies: how to obtain a solid baseline for comparison. Pre-/post-test designs are usually not feasible due to the long incubation period (see DE BOT & CLYNE , ); they also come with the potential of a retraining effect so that the actual investigation might prevent the very phenomenon it is looking for (SCHMID, a). This paper contains the first explicit and detailed discussion
c
(continued )
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Table A. Continued Study
Findings
Strand
of longitudinal studies versus static group comparisons, of their feasibility and of respective advantages and disadvantages. The argument is further developed in JASPAERT & KROON () where the authors address (and criticize) a strategy commonly used in language loss research at that time: The adoption of an idealized baseline that is % in compliance with prescriptive models (dictionaries, grammars)—that is, the assumption that unattrited native speakers make no errors, and that all instances of deviance are thus evidence for attrition. Jaspaert, Kroen, & Van Hout point to the inadequacy of this model and the fallacy of assuming that ‘native’ speech always conforms to such prescriptive norms. Weltens, B., K. de Bot, & T. van Els (eds.) (). Language Attrition in Progress. Dordrecht: Foris.
Like LAMBERT & FREED (eds., ) this volume is the outcome of a workshop meeting. In addition to general overview articles (VAN ELS) it contains a number of papers on dialect loss and language shift, and research reports. These latter consist of mainly reports on small-scale studies (Cohen) or preliminary reports (de Bot & Lintsen; Jordens et al.; Weltens & van Els—unfortunately, with the exception of the last one, none of these were eventually completed). The only larger study that was able to report results (although they remain descriptive and no quantitative analysis is presented) is Olshtain, who investigates L Hebrew returnee children after a period spent in an L English environment and finds an age effect.
de Bot, K. & M. Clyne (). Language reversion revisited. Studies in Second Language Acquisition , –.
Clyne’s investigation of Dutch speakers in Australia remains the only truly longitudinal and empirical study of L attrition among a larger population (as opposed to, for example, collections of letters from a single individual as analysed by Jaspaert & Kroon, and Hutz, ). His research in the early s suggested a phenomenon which came to be known as ‘reversion’ among elderly migrants, through which they would suffer attrition in the environmental language that they had used for decades and come to over-rely on their L. In extreme cases, even communication within the family allegedly broke down where children were not proficient in their parents’ L.
c
de Bot & Clyne discuss self-evaluation data of proficiency in English and Dutch elicited from a subgroup (n=) of the same speakers in , and focus particularly on the ten speakers who report a decrease in English proficiency. They hypothesize that language reversion is linked to the Critical Threshold proposed by Neisser () in that migrants who have reached a relatively high level of proficiency in the L will be protected against reversion. The L Dutch language proficiency of
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
this population, on the other hand, appeared extremely stable in the intervening period (see also DE BOT & CLYNE, ). SCHMID & KEIJZER () shed further doubt on the assumption of widespread regression among long-term elderly migrants. Jaspaert, K. & S. Kroon (). Social determinants of language loss. Review of Applied Linguistics (I.T.L.) /, –.
In the late s and early s, Jaspaert & Kroon launched a highly ambitious investigation of Italian in the Netherlands and Flanders. The study, which was to comprise first, second, and third generation speakers of Italian, was to be concerned largely with the impact of sociological and sociolinguistic factors on language attrition. The paper outlines a number of problematic areas for such studies, such as the appropriateness of various measures (free speech, language tests), the baseline issue already addressed in JASPAERT, KROON, & VAN HOUT () and the confound of language proficiency and metalinguistic knowledge. It then introduces a pilot study comprising thirty speakers, used to assess the validity of five language loss tests (comprising correction, editing, lexical, and comprehension tasks). Their results suggest that education is an important factor in language retention, possibly due to the fact that higher education levels may facilitate maintaining input through the written medium and through a higher socioeconomic status, allowing more travel. The amount of contact with the L in the migration setting, on the other hand, does not play a role. In general, the authors express their surprise at the minimal amount of loss that their tests appear to have detected (which they explicitly ascribe to a very high level of language maintenance in the community, and not to a failure of the tasks to detect attrition).
a, a
A follow-up to this study with a larger population (n=) was published by the same authors in . This paper contains a careful consideration and statistical evaluation of the way in which sociological and sociolinguistic background factors as well as factors pertaining to language use and language choice interact and pattern together. This analysis is not combined with an analysis of actual linguistic data, and no further findings from this very interesting study were published. Sharwood Smith, M. A. (). Crosslinguistic influence in language loss. In K. Hyltenstam & L. K. Obler (eds.), Bilingualism across the Lifespan. Cambridge: Cambridge University Press, –.
This paper discusses theoretical concepts and frameworks that are relevant for language attrition research. ‘Loss’ is conceptualized here as a type of development where forms that diverge from the standard target (e.g., as produced by more advanced ‘language losers’) are used as evidence for grammar building by the attriter (this ‘vicious cycle’ approach is also invoked by GROSJEAN & PY, ). Therefore, ‘developing competence in loss and developing competence in acquisition can be seen as related processes’ (p. ) and the forms developed in
b
(continued )
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Table A. Continued Study
Findings
Strand
this process often lead to an enriched form of the attriting language. Sharwood Smith approaches attrition from a psycholinguistic point of view, speculating about whether it will affect underlying knowledge systems (and will thus manifest itself consistently in the speech of a particular attriter) or rather the control of knowledge that is still represented in the brain (manifested incidentally/ variably) Based on the speculations made by Andersen () and Preston (), he draws up a list of potential outcomes of attrition, classified into ‘processes’ (performance) and ‘results’ (competence). He further lists properties of the two contact languages that he labels as ‘loss-inducing’, such as typological or structural similarity. Altenberg, E. P. (). Assessing first language vulnerability to attrition. In SELIGER & VAGO (eds.), –.
This case-study of a married couple with L German living in the United States compares the attrition of lexical, syntactic, and morphological features based on their acceptability/unacceptability in English and German. No baseline data from unattrited speakers are invoked. Altenberg very properly and painstakingly points out the limitations of her study: ‘Since these three tasks were preliminary studies, their findings must be interpreted cautiously and their limitations are obvious: future research of this type will require more subjects, more tokens, and possibly a monolingual control group’ (p. ). That said, she concludes that attrition is most likely to occur where L and L are similar (p. ). Despite this disclaimer, Altenberg’s conclusion has been widely and largely uncritically cited in support of the claim that attrition is conditioned by typological similarity (see SCHMID, a, p. ).
b
de Bot, K. & B. Weltens (). Recapitulation, regression, and language loss. In SELIGER & VAGO (eds.), –.
This contribution introduces the Regression Hypothesis (first proposed in the context of child language acquisition, aphasia, and phonological universals by Roman Jakobson in ), and discusses its applicability to L as well as L attrition. The predictions that language loss will be the mirror image of acquisition (also known as the LIFO—last in first out—model) are tempered by factors such as cross-linguistic influence, and some restrictions to the applicability of the hypothesis, for example its limitation to linguistic phenomena that can be demonstrated to be acquired gradually and in a fixed sequence, are pointed out. De Bot & Weltens conclude that this hypothesis, which has been widely cited in speculations on attrition, has received surprisingly little attention in actual empirical investigations. This statement holds to this date: aside
d
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
from some initial reports on a smaller-scale investigation in Jordens et al. (, ) there is only one larger study of first language attrition framed explicitly within the regression paradigm, namely KEIJZER (). de Bot, K., P. Gommans, & C. Rossing (). L Loss in an L Environment: Dutch Immigrants in France. In SELIGER & VAGO (eds.) –.
de Bot, Gommans, & Rossing present an investigation of thirty long-term Dutch migrants in France. They assess a number of skills and investigate the interaction of proficiency, LOR, and amount of contact. They conclude that there is indeed an impact of contact and time, but that LOR becomes significant only for those speakers who have few contacts. The approach taken here was later replicated by Soesman () and Schmid (b), and the results were largely confirmed.
a
Grosjean, F. & B. Py (). La restructuration d’une première langue: l’intégration de variantes de contact dans la compétence de migrants bilingues [The restructuring of a first language: the integration of contact variants into the competence of bilingual migrants]. La Linguistique , –.
Grosjean & Py’s investigation of fifteen Spanish migrant workers (long-term residents in a Francophone environment in Switzerland) argues that even stable adult linguistic competence is vulnerable to restructuring under L influence. They use a very interesting variant of the acceptability judgement task: for five linguistic items which they judge to be vulnerable to change under contact conditions they ask their participants not only whether it is acceptable in Spanish but also whether they have heard attestations of this form in their linguistic community (independently of whether or not this form is correct, has the participant heard it used or used it herself?). The standard Spanish structures score higher than the Francophone variants on both tasks, and there is more variance on the judgements for the migrant varieties. Grosjean & Py conclude that even mature, adult competence is open to a certain degree of restructuring for speakers who become bilingual and use both languages on a regular basis.
b, d
Olshtain, E. & M. Barzilay (). Lexical retrieval difficulties in adult language attrition. In SELIGER & VAGO (eds.), –.
Olshtain & Barzilay investigate fifteen native speakers of American English who are long-term residents in Israel. No information on age of migration (AoM) is given, but it is asserted that all speakers had ‘full competence in their first language before immigrating to Israel’ (p. ). A control group of six speakers of American English is also tested, making this the first investigation of L attrition since Sharwood Smith (a) to invoke an unattrited baseline. Free speech data is elicited by means of the Frog story, and individual instances of deviant use of lexical items (in particular over- and underspecifications) are discussed, but the overall results are not quantified or analysed statistically. The authors conclude that attrition leads to problems in vocabulary retrieval (p. ).
a
(continued )
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Table A. Continued Study
Findings
Strand
H. W. Seliger & R. M. Vago (eds.) () First Language Attrition. Cambridge: Cambridge University Press.
This collection is the first to focus on changes to the native language and exclude issues involved in foreign language attrition, and it marks the starting point of the divergence of the two research strands. The theoretical and methodological focus of the contributions is more clearly defined than in the earlier collections. All contributions are framed within a particular theoretical perspective and address a specific research question (given the largely unfocused and pre-theoretical nature of some of the earlier collections, this marks an important step forward). The introductory chapter by Seliger & Vago briefly discusses L attrition from a variety of perspectives (psycho- and sociolinguistic, linguistic, and theoretical). As a fundamental question, they return to the issue of which aspects of attrition are core or competence phenomena, and which can be considered as belonging to the periphery or performance (a question that is taken up again in the contribution by SHARWOOD SMITH & VAN BUREN )
The first part of the volume is dedicated to theoretically oriented meta-analyses which approach attrition from various perspectives. Two contributions concern attrition (SHARWOOD SMITH & VAN BUREN and DE BOT & WELTENS) others include considerations relating to aphasia (Obler & Mahecha) and language contact (Maher). This is followed by quantitative investigations and case studies. Again, a number of these are set within the heritage/language shift context (Dressler; Schmidt; Huffines) or are case-studies investigating the deterioration of a linguistic system among very young individual speakers (Kaufmann & Aronoff; Seliger; Turian & Altenberg; Vago), but for the first time, the volume also includes a number of quantitative and qualitative investigations of L attrition ‘proper’ in the studies by ALTENBERG, DE BOT, GOMMANS, & ROSSING; OLSHTAIN & BARZILAY; and SILVA-CORVALÁN. Sharwood Smith, M. & P. Van Buren (). First language attrition and the parameter setting model. In SELIGER & VAGO (eds.), –.
Sharwood Smith & Van Buren develop the questions on attrition of knowledge versus control posed in SHARWOOD SMITH () and ask to what extent an attriter may come to model the grammar underpinning his or her L on the L. In particular, they consider the practicalities of establishing whether attrition phenomena in an individual language user are indications of competence or performance change and the role that explicit/metalinguistic knowledge (as opposed to tacit/implicit knowledge) can play in this respect.
Silva-Corvalán, C. (). Silva-Corvalán presents an investigation of MexicanSpanish language American Spanish-English bilinguals from three attrition in a contact generations (n= of whom six belong to the first
a
c, c
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
situation with English. generation; there is no control group). She approaches her analysis of Spanish verb forms from a In SELIGER & VAGO simplification/markedness perspective and finds a (eds.), –. ‘clear-cut qualitative difference between first generation immigrants’ who retain a full-fledged variety of spoken Spanish, and speakers from subsequent generations. The latter group has a reduced system of verbal morphology lacking, for example, future morphology and exhibiting simplifications in tenses such as the preterite, the pluperfect indicative and the subjunctive (p. ). She concludes that the markedness hierarchy proposed by Muysken () is a valid predictor for her results, but stresses that this is a descriptive and not an explanatory account of the changes she observed (p. ). Major, R. C. (). Losing English as a first language. Modern Language Journal , –.
In what is the first and for fifteen years (until de Leeuw’s PhD study) remained the only investigation of phonetic L attrition, Major investigates changes to voice onset time (VOT) in the L and the L of five speakers of Brazilian Portuguese who are long-term residents in the US. This study is reminiscent of Flege’s investigation of VOTs in French and English which, although it was not cast within an attrition framework, already established bidirectional influence in bilingual speech which increased with experience. In the present study, VOTs for voiceless plosives are measured in formal and casual speech, and Major concludes that the mutual interaction of L and L can affect L phonetics (p. ).
e
de Bot, K. & M. Clyne (). A year longitudinal study of language attrition in Dutch immigrants in Australia. Journal of Multilingual and Multicultural Development (), –.
In this paper, the population whose self-evaluations in their L (Dutch) and their L (English) had been assessed and tested for indications of language reversion in DE BOT & CLYNE () are investigated linguistically for lexical and morpho-syntactic attrition phenomena in free speech. These results are compared to data from the same speakers dating from . This comparison reveals that very little change has occurred in the intervening fifteen years, and de Bot & Clyne conclude that, after the first decade of emigration, L skills appear to be extremely robust.
c
Håkansson, G. (). Syntax and morphology in language attrition: A study of five bilingual expatriate Swedes. International Journal of Applied Linguistics (), –.
Håkansson presents the first quantitative investigation of L attrition from a parameter-setting perspective. She investigates the V rule as well as noun phrase (NP) agreement in the written Swedish of five heritage speakers (returnees) and five L learners of Swedish from various language background, and also briefly considers reports on monolingual aphasics. Her findings suggest that, while L learners struggle with both V and NP agreement, the heritage attriters have
a, c, d
(continued )
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Table A. Continued Study
Findings
Strand
no problem with the former, but are comparable on the latter features. She points out that, rather surprisingly, violations of the V rule are also common among monolingual aphasics, who do not have a competing grammar that offers alternatives to this rule. Håkansson interprets her findings from the point of view of markedness. Her data are reinterpreted later by Platzack () from a minimalist perspective. Ammerlaan, T. (). You get a bit wobbly. Exploring bilingual lexical retrieval processes in the context of first language attrition. PhD dissertation, Katholieke Universiteit Nijmegen.
Ammerlaan’s PhD study investigates lexical accessibility among Dutch speakers in Australia. He conducts a series of lexical tasks (verbal fluency, picture naming, and picture-word-matching) with eighty-eight Dutch L speakers in order to test attrition processes as they relate to lexical access and retrieval. He finds that these processes are influenced by factors such as AoM and LOR, but also by similarities between L and L items (in particular in the picture-word matching task, distractors that were based on English items were often chosen instead of the target). His conclusion is that, while access and retrieval to the L may become compromised to some extent after extended periods of residence in an L environment, these difficulties are temporal, not structural, and can be overcome. This is expressed in the quote from one of the participants who likens his knowledge of the native language to the skill of riding a bicycle: ‘you never forget it, but you get a bit wobbly’ (cited on the title page). A report of this study is published in Ammerlaan ().
b, a
Waas, M. (). Language Attrition Downunder. Frankfurt: Peter Lang.
Waas supplies the first monograph treatment of language attrition, presenting a socio- and ethnolinguistic investigation of the attrition of German among adult long-term migrants in Australia and the impact of affective variables on this process. In addition to a largely qualitative investigation of free speech she elicits data on language proficiency by means of a controlled task (verbal fluency) as well as by self-assessments, and investigates to what extent ethnic affiliation impacts on the attritional process. To this end, she divides her sample into two groups, Permanent Residents versus Naturalized Citizens, subdivided into speakers with or without ‘Ethnic affiliation’ (ie., membership of a German club, church, etc.). The attriters are outperformed by a monolingual German control group on both the controlled task and the selfassessments. No difference is found for the ethnic affiliation parameter on the controlled task, but people without such affiliations rate their own L skills lower. Waas
b
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
concludes that ‘L attrition in an L environment is inevitable, even after a stay of only to years’ and that ‘socio-demographic factors such as citizenship and (non-)affiliation have an impact on the extent of L attrition’ (p. ). Polinsky, M. (). American Russian: Language loss meets language acquisition. In W. Browne, E. Dornisch, N. Kondrashova, & D. Zed (eds.), Annual Workshop on Formal Approaches to Slavic Linguistics: The Cornell Meeting . Ann Arbor: Michigan Slavic, –.
Polinsky’s work represents the beginning interest in heritage languages and how they differ from the language as spoken in the country of origin. In her earlier work (Polinsky, ) she presented investigations of the development of Eastern European and Asian languages in the US, the present study focuses on American Russian. Polinsky investigates the grammatical knowledge of ‘semi-speakers’ (attriters and heritage speakers) who arrived in the US before they had reached adulthood. She introduces a distinction that does not only take into account the order in which the languages have been acquired (L vs. L) but also their role for the speaker (primary/dominant or secondary). Based on this latter distinction, she distinguishes Émigré Russian and American Russian, concluding that the latter, but not the former, is characterized by structural change, while both varieties use lexical borrowing (p. ).
c, c
Her analysis of the Russian data investigated here is based on speech from nineteen American Russians with AoMs between and and one speaker who was born in the US. She describes a reduced system of case and verbal inflection, and points out that there is a correlation between grammatical variables as well as between grammatical and lexical change in her data. The paper ends with an analysis of the incidence of structural change and AoM, LOR and period of disuse. Yağmur, K. (). First Language Attrition Among Turkish Speakers in Sydney. Tilburg: Tilburg University Press.
In his PhD study, Yağmur builds on WAAS’ () work on the impact of affective variables, framing his investigation within the framework of Ethnolinguistic Vitality Theory (EVT, as formulated, for example, by Giles, Bourhis, & Taylor, ). This framework allows a more sophisticated quantification of the affective and ethnolinguistic variables invoked in earlier investigations such as WAAS () or JASPAERT & KROON (, ) in that it measures not only individual attitudes but also community-level factors such as language prestige and institutional support. Yağmur elicits data by means of several controlled tasks (verbal fluency, relativization) from Turkish speakers in Australia as well as from a control group. He also presents a largely qualitative investigation of spontaneous data elicited from the attriters by means of Frog Story retellings and compared to some earlier
b
(continued )
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Table A. Continued Study
Findings
Strand
findings from Turkish monolinguals. Yağmur acknowledges earlier findings on the role of education for language attrition (as pointed out, for example, by WAAS and JASPAERT & KROON) by dividing both his experimental and his control population into high and low education subsamples of n= (total n=). He finds that the attriters exhibit some reduction of L lexical accessibility and syntactic proficiency, and that attrition effects do not appear to be trade-offs with L proficiency (speakers with low levels of skills in L show more loss in L than more proficient Lers) (p. ). More highly educated attriters generally outperform migrants with lower education levels. EVT, however, does not provide a sufficiently explanatory framework for attrition (p. f.). A report of this study is published in Yağmur, de Bot, & Korzilius (). Köpke, B. (). L’attrition de la première langue chez le bilingue tardif: Implications pour l’étude psycholinguistique du bilinguisme. PhD dissertation, Université de Toulouse-Le Mirail.
Köpke presents a psycholinguistic investigation of L German attriters in two different linguistic settings, an L English environment (Canada) and in France. She investigates attrition at various linguistic levels (lexical and morpho-syntactic), asking to what extent the attritional process can be deemed to have affected underlying knowledge as opposed to being a performance phenomenon, and what the impact of typological (dis) similarity in the contact languages is. Based on three tasks (picture description, sentence generation, and grammaticality judgement) she concludes that in late bilinguals, attrition is confined to the level of access in production and does not affect underlying knowledge. She finds the lexicon to be more affected by attrition than morpho-syntax, and the cross-linguistic similarity of languages to be a determining factor. There is considerable variation between her two bilingual populations, which she accounts for by sociolinguistic factors. Reports on this study are published in Köpke (, ) and Köpke & Nespoulous ().
b
Gross, S. (). The role of abstract lexical structure in first language attrition: Germans in America. PhD dissertation, University of South Carolina.
This is one of a series of PhD investigations, conducted at the University of South Carolina under the supervision of Carol Myers-Scotton and applying her -M model to language contact and change (Myers-Scotton & Jake, , based in turn on Levelt’s () Speaking model) to incomplete acquisition and attrition in migrant children of Hungarian (Bolonyai, ) and Russian (Schmitt, , ) as well as to attriters of German in Gross’ investigation. The -M model predicts a
c
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
hierarchy in vulnerability to processes of change, loss, and attrition such that content morphemes (lemmas) will be affected first, followed by early system morpheme lemmas (such as determiners), which are indirectly elected by the content morpheme. Late system morphemes are structurally assigned and least vulnerable to processes of change and attrition. A report of this study is published in Gross (). Hulsen, M. (). Language loss and language processing. three generations of Dutch migrants in New Zealand. PhD dissertation, Katholike Universiteit Nijmegen.
Hulsen continues the research strand started in AMMERLAAN (), investigating lexical access among New Zealand Dutch. Invoking speakers of three generations, she attempts to assess the impact of L use within a Social Networks framework. Her findings show that while productive skills decrease within generations, receptive skills remain largely unchanged. She argues that the difficulties encountered by attriters are retrieval problems, and do not constitute ‘loss’ in the sense of being ‘erased from memory’ (p. ). The use of Dutch within a speaker’s primary social network is a powerful predictor for the retention of the L. A report of the study is published in Hulsen, de Bot, & Weltens ().
a, a
Gürel, A. (). Linguistic characteristics of second language acquisition and first language attrition: Turkish overt versus null pronouns. PhD dissertation, McGill University, Montreal, Canada.
Gürel assesses bilingual development from the perspective of the Government and Binding framework. Her PhD thesis investigates Turkish pronouns in L acquisition by native speakers of English and L attrition in Turks immersed in an L English environment. Through a series of tasks (written interpretation, truth value judgement, picture identification) she assesses to what extent the binding properties of the Turkish overt and null pronouns have been acquired/retained in her populations. She finds evidence that for both her attriters and her Lers the binding properties of Turkish overt pronouns are being replaced by those of English (p. ) which she interprets as evidence for selective transfer effects in ultimate L and L grammars (p. ). In two later articles, Gürel (a,b) reassesses her findings from a variety of perspectives, including the psycholinguistic framework of the Activation Threshold Hypothesis (Paradis, ). In Gürel & Yılmaz () data elicited by means of the same test battery from first and second generation Turkish-Dutch bilinguals are assessed, and similar patterns of attrition and restructuring are found.
a, b, d
An analysis of a similar grammatical property and its bilingual acquisition as well as L attrition in Korean is presented by Kim, Montrul, & Yoon (), who use a truth value judgement task modelled on the one (continued )
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Table A. Continued Study
Findings
Strand
developed for Gürel’s study. They find that early bilinguals are similar to L learners, while there are no signs of attrition among the late bilinguals. Montrul, S. (). Incomplete acquisition and attrition of Spanish tense/aspect distinctions in adult bilinguals. Bilingualism: Language and Cognition, (), –.
In the first of a series of pioneering contributions on incomplete acquisition and attrition of Spanish in the US (in particular also Montrul, a, b, ), Montrul approaches the phenomenon of the impact of the L/ dominant language on the L/heritage language from the point of view of generative linguistic theory. She argues for a substantial difference in ultimate attainment based on age of onset: while sequential and, in particular, late bilinguals have a L mental grammar that shows little evidence of erosion in comparison with an unattrited reference population, simultaneous and early bilinguals are more similar to Lers, in particular where functional categories are concerned. The earlier the age of onset of bilingualism and the more intense the exposure to the sociolinguistically dominant language, the more incomplete the adult grammar may turn out to be. She concludes that an early age of onset may be a necessary but not a sufficient condition for convergent language acquisition (p. ).
c, a
Schmid, M. S. (). First language attrition, use and maintenance: the case of German Jews in Anglophone countries. Amsterdam/ Philadelphia: John Benjamins.
Schmid approaches the question of attitudes and motivation in L attrition previously investigated by, among others, WAAS () and YAĞMUR (). Her data are drawn from a corpus of Oral History interviews conducted with German-Jewish refugees who escaped Nazi persecution in the s and settled in Englishspeaking countries. She situates her analysis within the historical framework of progressive radicalization of persecution measures and assesses to what extent the L attrition of speakers who emigrated during three clearly delineated periods differs. Her analysis of free speech data comprises both an assessment of errors and of target-like language use. She concludes that, in her sample, persecution appears to be the most important predictor for attrition effects (p. ). A summary of this study is published in Schmid (c).
b
Cook, V. (ed.) (). Effects of the second language on the first.
In this volume, Cook introduces a perspective on language attrition from his framework of multicompetence, pioneering the view that attrition is not an exotic and rare phenomenon exhibited by
e
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Clevedon: Multilingual Matters.
speakers under extreme circumstances (long-term separation from the language environment with little intervening use), but a natural linguistic development common to all bilingual language users. His approach thus differs from earlier investigations of L attrition in that he does not assume long-term residence in an L environment as a necessary condition under which L effects on the L can occur. He posits that ‘the first language of people who know other languages differs from that of their monolingual peers in diverse ways’ (p. ) in that all languages residing within the same mind ‘must form a language super-system at some level rather than be completely isolated systems’ (p. ). The investigations collected in this volume thus comprise both studies carried out within the ‘traditional’ attrition settings of migrants immersed in an L environment (Jarvis; Laufer; Pavlenko; Porte), early bilingual populations (Balcom; Murphy & Pine) and studies of instructed L learners in their country of origin (Cenoz; Dewaele & Pavlenko; Cook, Iarossi, Stellakis, & Tokumaru). From the lexicon through syntax to pragmatics, all linguistic levels are touched upon.
Dussias, P. E. (). Parsing a first language like a second: The erosion of L parsing strategies in Spanish-English bilinguals. International Journal of Bilingualism, (), –.
The investigations that Dussias presents here and in Dussias & Sagarra () constitute the first neurocognitive analyses of L attrition. By means of an eye-tracking investigation of participants reading Spanish sentences containing embedded relative clauses, it is demonstrated that language-specific preferences for relative clause attachment are vulnerable to crosslinguistic interference: while Spanish monolinguals favour high attachment, Spanish-English bilinguals are similar to English monolinguals in that they come to prefer low attachment, even after a relatively short immersion period.
b, d
Mennen, I. (). Bidirectional interference in the intonation of Dutch speakers of Greek. Journal of Phonetics , –.
If investigations of phonetic L attrition are scarce, studies considering suprasegmentals are almost unheard of. Mennen presents the first of only two reports that investigate this phenomenon (the second is DE LEEUW, MENNEN, & SCOBBIE, ). She studies the production of peak alignment among five Dutch-Greek bilinguals and finds that four of them show evidence of cross-linguistic interference in both their L and their L, while a fifth speaker (who was younger than the others at migration) appears to be exempt. Similar interpersonal variation is found in de Leeuw et al. () where one of the ten German-English bilinguals shows no sign of attrition in L German. It thus appears
e
(continued )
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Table A. Continued Study
Findings
Strand
that susceptibility to phonetic attrition may vary considerably among individuals. Pavlenko, A. (). L influence and L attrition in adult bilingualism. In M. S. Schmid, B. Köpke, M. Keijzer, & L. Weilemar (eds.), First Language Attrition: Interdisciplinary Perspectives on Methodological Issues. Amsterdam: John Benjamins, pp. –.
This is one of a series of papers in which Pavlenko investigates the question of transfer from L to L (see also Pavlenko, , ; Pavlenko & Jarvis, , among others). The population under investigation are Russian college students in the US and in Russia. She focuses her analysis on particular semantic fields—in the present study, these are emotion terms; in the paper, she focuses on verbs of motion—and the way in which an entire field may show a certain extent of restructuring in an intensive language contact situation. This restructuring may be semantic in nature—for example, the English term privacy, which has no clear Russian equivalent, is discussed—but it may also concern the way in which an event is framed and conceptualized. In the present paper, a typology of language contact and attrition phenomena is introduced, which SCHMID (a) develops and recommends as a highly suitable taxonomy for attrition studies at large.
b
Scherag, A., L. Demuth, F. Rösler, H. J. Neville, & B. Röder (). The effects of late acquisition of L and the consequences of immigration on L for semantic and morpho-syntactic language aspects. Cognition , B–B.
Scherag et al. present a psycholinguistic investigation of German L attrition and L acquisition using a priming paradigm to assess grammatical gender and semantic congruency effects. They study L speakers who are either short term or long term residents in the US, L speakers of English who are long-term residents in Germany, and a German control group. Their findings show that while all groups benefit from semantic priming, morpho-syntactic priming is absent for the late learners of German, but does not seem to be affected by L attrition, as all three German native groups show processing gains from both semantic and morphosyntactic congruency.
b, c
Schmid, Köpke, Keijzer, & Weilemar (eds.) (). First Language Attrition: Interdisciplinary Perspectives on Methodological Issues. Amsterdam/ Philadelphia: John Benjamins
This collection is a conference volume, comprising contributions presented at the first International Conference on First Language Attrition (ICFLA) in Amsterdam in . The overall focus of the volume is on establishing a common methodological framework for future studies in attrition. The introduction by Köpke & Schmid, giving a historical overview of the field to date, and the annotated bibliography by Schmid can to some extent be considered precursors for the present timeline. The volume is further structured into
,
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
three parts: the first section focuses on theoretical models and methodological aspects, with considerations of how to define (Pavlenko) and measure (Altenberg & Vago) attrition, as well as its embedding in sociocultural theory (Jiménez Jiménez) and frameworks of language dominance and emotions (Dewaele). The second part introduces a number of largely descriptive studies, while the third part assesses how such studies may be used in the building and validation of linguistic theories. The volume concludes with a tentative outline of the Language Attrition Test Battery (which is further developed in SCHMID, a). Tsimpli, I., A. Sorace, C. Heycock, & F. Filiaci, F. (). First language attrition and syntactic subjects: A study of Greek and Italian near-native speakers of English. International Journal of Bilingualism (), –.
Tsimpli et al.’s investigation of the production and interpretation of null and overt pronouns in the L (Greek and Italian) of very advanced L speakers of English introduces the study of attrition from the perspective of linguistic interfaces, testing the hypothesis that ‘the changes in L syntax will be restricted to the interface with the conceptual/intentional cognitive system’ (p. , see also Sorace, ; Tsimpli, ). They conclude that their data support the hypothesis that attrition does not affect purely formal, uninterpretable features, but does show up on syntactic/pragmatic, interpretable features (p. ).
a, d
Keijzer, M. (). Last in first out? An investigation of the regression hypothesis in Dutch migrants in Anglophone Canada. PhD dissertation, VU University Amsterdam (published in under the title First Language Acquisition and First Language Attrition: Parallels and Divergences. Saarbrücken: LAP Lambert Academic Publishing)
Keijzer approaches attrition from the perspective of Jakobson’s Regression Hypothesis (), which is a reformulation of the idea, gleaned from psychoanalysis, that forgetting will be the mirror image of acquiring. To this end she compares a population of long-term adult Dutch residents in Anglophone Canada not only with an unattrited Dutch control population but also with a population of relatively advanced Dutch L learners aged to . Using a Wug-test and a grammaticality judgement task she tests a number of morphological and morpho-syntactic features, concluding that attriters and acquirers do indeed show some interesting parallels in particular in the overgeneralization of regular or default features. (A summary article of this study is presented in Keijzer, a.)
d
Köpke, B., M. S. Schmid, M. Keijzer, & S. Dostert (eds.) (). Language Attrition: Theoretical Perspectives. Amsterdam/
This volume, based largely on the contributions and presentations from the second International Conference of First Language Attrition (Amsterdam, ) is an attempt to integrate the study of language attrition into the mainstream of bilingualism research. A number of theoretically oriented contributions from predominant researchers in the field approach the topic from a variety of perspectives, such as psycho- and
(continued )
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Table A. Continued Study
Findings
Strand
Philadelphia: John Benjamins.
neurolinguistics (Gürel, Köpke, Paradis, Pallier), the MOGUL framework which combines Jackendoff ’s model of the language faculty with a developmental perspective (Sharwood Smith), Dynamic Systems Theory (DST) (de Bot), and the Interface Hypothesis (Tsimpli). The predictions and claims made here are put to the test in research reports, such as Schmid’s investigation of the impact of the use of the L on attrition (formulated within the framework of Paradis’ () Activation Threshold Hypothesis) and Footnick’s case study on the reactivation of a ‘lost’ language by means of hypnosis.
Levy, B. J., N. D. McVeigh, A. Marful, & M. C. Anderson (). Inhibiting your native language. The role of retrievalinduced forgetting during secondlanguage acquisition. Psychological Science (), –.
This study investigates problems of L lexical access under conditions of second language learning (retrieval-induced forgetting) through a picture naming task, with items to be named in either English or Spanish, depending on a colour cue, administered to English native speakers engaged in learning Spanish at the University of Oregon. Levy et al.’s findings show that participants who are less fluent in Spanish have problems accessing the English form of items that they had previously named several times in Spanish, but that there is no such inhibition effect among more fluent speakers. They conclude that ‘native-language words are most vulnerable to forgetting when people struggle to produce foreign vocabulary, as might occur to novices during immersion’, but that this inhibition effect is limited to phonology and does not affect the semantic level (p. ).
b
Brown, A. & Gullberg, M. (). Bidirectional crosslinguistic influence in L–L encoding of manner in speech and gesture: A study of Japanese speakers of English. Studies in second language acquisition (), –.
Brown & Gullberg present the first investigation of how becoming proficient in a second language may affect the use of gestures in the native language. They find that Japanese speakers with intermediate proficiency of English use gestures differently in either language from monolingual speakers, showing evidence of bidirectional interaction in the bilingual mind at this level. They propose that including the use of gestures in research on bilingual development may help provide insights beyond those that can be gained by investigating only features of spoken language.
b
Crezee, Ineke H. M. (). I understand it well, but I cannot say it proper back: language
Crezee’s dissertation reports on an investigation of older Dutch migrants in New Zealand, with a view to testing the Language Reversion Hypothesis (DE BOT & CLYNE, , ). While she found some general effects of ageing in the data in both languages (e.g., increased
a, b
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
use among older Dutch migrants in New Zealand. PhD dissertation, University of Auckland.
response latency), her findings to not lend any support to the general assumption—widely held in many migrant communities—that ageing migrants will experience a decline in their L skills alongside a reversion to using their L.
Kim, S. H. O. & D. Starks (). The role of emotions in L attrition: The case of Korean-English late bilinguals in New Zealand. International Journal of Bilingualism (), –.
Kim & Starks attempt to assess the developing patterns of L and L proficiency among late Korean-English bilinguals in New Zealand in the context of emotionrelated language choice (ERLC). To this end, they relate self-reported data on language choice in a variety of situations to accuracy, fluency, complexity, and lexical diversity in an L story-retelling task. They find a difference between their experimental and control population on accuracy and lexical diversity, but not on fluency or complexity (p. ). The participants exhibit a continued preference for their L in most situations related to emotions, with the exception of language use ‘when angry’ (confirming the findings reported, e.g., by Dewaele, ). There furthermore appears to be a substantial correlation between accuracy in L on the one hand and preference for this language in a variety of emotionally charged contexts on the other, suggesting that ‘as late bilinguals gain confidence in fluency in the L, they lose confidence in using the L due to their decreasing L accuracy’ (p. ).
a, b
Negrisanu, Raluca Mihaela. (). Aspects of First Language Attrition: A Case Study of German Immigrants in East Tennessee. PhD dissertation, University of Tennessee.
Negrisanu presents a study of German migrants to the US. Consistent with findings from other studies, her findings reveal consistent though not dramatic attrition effects, particularly in the areas of lexical retrieval and NP morphology. Her findings also show some impact of personal background factors, such as education and L contact, in particular on lexical items.
a
Bylund Spångberg, E. (). Age differences in first language attrition. PhD dissertation, Stockholm University.
This PhD thesis consists of four articles published separately (Bylund a,b; Hyltenstam et al., ; Bylund et al., ), reporting a series of tests on Spanish-Swedish bilinguals. The focus of the study is on the impact of AoM on L attrition, and Bylund finds that there is a great deal more variability in the retention or attrition of the L Spanish among speakers for whom this took place before the onset of puberty. This variability is related to a general aptitude for language learning for pre- but not for post-puberty migrants. The overall findings are interpreted within a Critical Period account of second language acquisition and L attrition and argued to be consistent with the
b
(continued )
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Table A. Continued Study
Findings
Strand
existence of maturational constraints in bilingual development. Linck, J. A., J. F. Kroll, & G. Sunderman (). Losing access to the native language while immersed in a second language: Evidence for the role of inhibition in second-language learning. Psychological Science (), –.
This study builds on the investigation of the impact of inhibition on L access by LEVY ET AL. () by comparing two populations of L learners of Spanish at similar levels of proficiency: one which was immersed in a Spanish-speaking environment (in a study abroad setting), the other comprised classroom learners in the native English environment. The tasks consists of a verbal fluency task and an experiment where the participant has to determine whether a given English word is the correct translation of a Spanish item. The latter task contained English distractor items that were similar to the Spanish item in either form or meaning. Linck, Kroll, & Sunderman’s findings suggest ‘that access to the L is attenuated during language immersion. In both comprehension and production, the L was less accessible for the immersed learners than for the classroom learners’ (p. ).
b
Schmid, M. S. & M. Keijzer (). First language attrition and reversion among older migrants. International Journal of the Sociology of Language , –.
Schmid & Keijzer set out to test the reversion hypothesis proposed by DE BOT & CLYNE () which proposes that elderly migrants show a tendency to regress in terms of their L proficiency, even if they have used this language predominantly for decades, and to return to their L. In order to assess the L skills of longterm elderly migrants, they test lexical diversity, fluency, and performance on a verbal fluency task among five different age groups. They find that the performance of the oldest groups of migrants is less different from their monolingual age-matched peers than that of the younger groups, but ascribe this to a more marked decline with age among the monolinguals. In essence, these controls appear to be ‘catching up’ with the attriters in terms of lexical access and disfluency problems. Schmid & Keijzer ascribe this to the effect of cognitive decline with ageing against which long-term bilinguals might to some extent be protected (as suggested by, e.g., Bialystok et al., ).
c
Schmid, M. S. (). On L attrition and the linguistic system. EUROSLA Yearbook , –.
In this article, Schmid proposes that the study of L attrition may provide a useful approach in resolving controversial issues in L acquisition studies, such as the question of whether advanced late L learners retain a representational deficit. She presents an investigation of a group of ‘typical’ L attriters of
b
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
German (long-term residents in Anglophone Canada and the Netherlands who emigrated voluntarily after World War II). Against this backdrop, she assesses data from a very strongly attrited L speaker of German (the most severe case of attrition that had emerged among the German-Jewish refugees investigated in SCHMID, ) and from a highly successful L learner of German with English L. She concludes that, in particular on those features where L and L are structurally different, the attriting population in general and the individual case in particular still retain a strong resemblance to the unattrited control population, while the L learner of German appears qualitatively different. This is interpreted as evidence for a representational deficit in L learning. Brown, A. & Gullberg, M. (). Changes in encoding of PATH of motion in a first language during acquisition of a second language. Cognitive Linguistics (), –.
This study is one of several investigating the encoding of motion across different languages and its stability among bilinguals (see also Pavlenko, ; BROWN & GULLBERG, , ; ATHANASOPOULOS ET AL., ; among others), suggesting that event conceptualization is not only language-specific but subject to reconstruction among bilinguals. All of these studies proceed from the observation that there is systematic variation across languages with respect to a preferences for encoding either the source, path, or other aspects of the motion event and its ramifications for clause packaging, and demonstrate that such preferences are not stable in bilinguals but may change over an extended attrition period. Comparisons with monolinguals reveal that bilinguals tend to occupy an intermediate position between their languages with respect to the preferences for encoding both Path and Goal. Interestingly, Brown & Gullberg () find no difference between immersed and non-immersed bilinguals in this respect.
b
Celata, C. & J. Cancila (). Phonological attrition and the perception of geminate consonants in the Lucchese community of San Francisco (CA). International Journal of Bilingualism (), –.
This study, arguably the only investigation to date that considers attrition on the phonological (as opposed to the phonetic) level, investigates the perception of a phonemic contrast present in the L but not in the L, namely that of singleton and geminate consonants in the Lucchese dialect of Italian among long-term migrants and second-generation speakers in the San Francisco area. They perform two phoneme identification tasks and show that the bilingual speakers are clearly outperformed by the unattrited native speakers. Furthermore, there is a marked distinction between first and second generation speakers, with the former achieving higher scores on the identification task than the latter. They conclude
e
(continued )
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Table A. Continued Study
Findings
Strand
that the perception of this phonological feature is becoming progressively impaired in the bilingual populations. de Leeuw, E., M. S. Schmid, & I. Mennen (). Perception of foreign accent in native speech. Bilingualism: Language and Cognition (), –.
de Leeuw et al. present the outcomes of a Foreign Accent Rating experiment in which nineteen native speakers of German listen to short excerpts of free speech collected from long-term German migrants in an L English and Dutch setting as well as from an unattrited control group. The raters assess whether or not each speaker is a native German, and also indicate how confident they are in their judgement. The results indicate that the attriters are more likely not to be perceived as native speakers than the controls. It is further shown that the amount of informal use of the L (with friends and family) do not impact on the degree of perceived foreign accent, but that speakers who use this language for professional purposes (ie., in formal contexts) are less likely to be perceived as foreigners.
e
Kim, S. H. O. & D. Starks (). The role of fathers in language maintenance and language attrition: the case of KoreanEnglish late bilinguals in New Zealand. International Journal of Bilingual Education and Bilingualism (), –.
Based on the same corpus as the one analysed in KIM & STARK (), the authors present an intriguing analysis of the role of the home language for the attrition or maintenance of the L. They demonstrate that these speakers are significantly less accurate in their L than a population of -year-old monolingual Koreans. While the L used by the mother of the participants, who emigrated from Korea to New Zealand around age , is not related to their current L proficiency, language use with the father and the siblings does have an impact. These results are ascribed to the types of interaction within the family setting.
a
Major, R. C. (). First language attrition in foreign accent perception. International Journal of Bilingualism (), –.
Major presents an intriguing new approach to language attrition: instead of assessing, as he did in his earlier work (e.g., MAJOR, ), to what extent attriters may develop a foreign accent in their L he employs attriters and L learners of Brazilian Portuguese as well as unattrited controls and speakers with no experience of Portuguese as raters in a foreign accent detection experiment. Surprisingly, all his populations including the speakers with no knowledge of Portuguese, are able to reliably differentiate native and non-native speakers of that language, and there appears to be no attrition in the ability to perceive a foreign accent in the L among the long-term US residents with Portuguese L.
e
Ribbert, A. & F. Kuiken (). L-induced
This study focuses on a phenomenon in which there are only minor differences between the L (German) and
c
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
changes in the L of Germans living in the Netherlands. Bilingualism: Language and Cognition (), –.
the L (Dutch) of the speakers, namely infinitival constructions using the complementizer um/om which exist in both languages but have a somewhat wider distribution in Dutch. By means of a grammaticality judgement task, Ribbert & Kuiken establish that Germans immersed in an L Dutch environment have trouble recognizing the grammaticality of the use of this complementizer in German sentences, especially in cases where um is ungrammatical in German but optional in Dutch.
Schmid, M. S & K. Beers Fägersten (). Fluency and language attrition. Language Learning (), –.
This study tests the assumption that the attenuation of lexical access that has often been witnessed in the attritional process will lead to an increase in disfluency markers. Schmid & Beers Fägersten analyse free spoken data from German and Dutch L attriters and conclude that disfluency markers indicating cognitive hesitation processes do indeed increase, in particular preceeding lexical items. Disfluency markers with a semantic function show an adaptation to the distribution in the L.
a
Schmid, M. S. & E. Dusseldorp (). Quantitative analyses in a multivariate study of language attrition: The impact of extralinguistic factors. Second Language Research (), –.
In this paper, Schmid & Dusseldorp approach the problem of investigating the impact of external sociolinguistic and attitudinal variables on the attritional process. They acknowledge that the number and variety of factors that can be assumed to play a role in language retention or attrition makes their assessment difficult in applied investigations using real (free or elicited) linguistic data, as opposed to surveys based on self-evaluations, due to the necessarily limited sample size. Based on an investigation of German L attrition in a Dutch- and English-speaking environment, they demonstrate how to reduce predictors by presenting a model of combining factors and assessing their impact in a linear regression model. They conclude that in the present data, the impact of such factors is largely negligible, as the only factor they find consistently influencing the outcome is the use of the L in the workplace. Furthermore, LOR has some impact on lexical diversity. L use in informal situations (with friends and family) plays a very small and largely inconsistent role for the different measures, and attitudinal factors have no impact at all. Errors and disfluency measures (the measures most often studied in attrition research) show no relationship to any of the external variables.
b
Stolberg, D. & A. Münch (). ‘Die Muttersprache vergisst man nicht’—
In a longitudinal case-study, Stolberg & Münch assess the initial attrition and partial recovery of L German in a long-term migrant who had virtually no exposure to her L for close to fifty years until the beginning of the
a
(continued )
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Table A. Continued Study
Findings
Strand
or do you? A case study in L attrition and its (partial) reversal. Bilingualism: Language and Cognition (), –.
study in . Regular conversations in German were held with her every two to three months over a four-year period. The results show that over this period, the speech of the participant becomes more fluent and deviations from the target norm decrease, in particular in the lexical-semantic domain. Syntactic and morphological deviations are more variable and show a less marked decrease over time.
Badstübner, T. (). L Attrition: German Immigrants in the US. PhD dissertation, University of Arizona.
Badstübner’s investigation replicates the research design first proposed in the European Graduate Network (see Schmid & Cherciov, Chapter , this volume). She specifically investigates differential attrition patterns in migrants who use their L (German) professionally against those who do not. Her conclusions suggest that attrition is primarily a phenomenon of access to, not deterioration of, knowledge and that it is not a uniform phenomenon but modulated by other factors. Her dichotomous factor of professional language use proves to be important in this context.
a
Lubińska, Dorota (). Förstaspråksattrition hos vuxna: Exemplet polsktalande i Sverige (Adult First Language Attrition: The Case of Polish Speakers in Sweden). PhD dissertation, University of Stockholm.
Lubińska’s study of highly-educated Polish migrants in Sweden investigates fluency and accuracy across a range of grammatical and lexical features, and attempts to link the variability found on these features to levels of input/ exposure and language attitudes. She finds attrition effects on these features, in particular overuse of overt pronouns, in just over half of her participants. Lubińska attempts to gain a comprehensive picture by conducting analyses at the group as well as at the individual level. Her results show that, while linguistic insecurity can be related to exposure and attitudes, there is no straightforward correspondence between these factors and accuracy or fluency.
a, b, a, b
Pavlenko, A. & B. C. Malt (). Kitchen Russian: Cross-linguistic differences and first-language object naming by Russian–English bilinguals. Bilingualism: Language and Cognition (), –.
This study probes into the intriguing question of cross-linguistic differences in categorization and their vulnerability to cross-linguistic influence. Based on the premise that English and Russian base categorization of household objects such as containers from which to drink on different features (shape, material, purpose, etc.), Pavlenko & Malt conduct a classification task of sixty pictures of more or less typical household objects. Participants are grouped according to the age of onset of bilingualism (simultaneous, early, and late) and the findings are compared to responses from two monlingual (English and Russian) reference groups. The authors conclude that ‘naming patterns for [ . . . ] common objects in one language can be swayed by exposure to patterns in another’ (p. ) and that this process is influenced by age of onset.
b
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Perpiñán, S. (). Optionality in bilingual native grammars. Language, Interaction and Acquisition (), –.
Perpiñán investigates subject–verb inversion in adult Spanish L attrition in the United States. She focuses on two types of construction: relative clauses, in which inversion is governed by extra-syntactic conditions and matrix questions, where it is obligatory. Her findings are in accord with the Interface Hypothesis (e.g., TSIMPLI ET AL., ) in that obligatory inversion is unaffected by attrition, while pragmatic/phonologically governed inversion has changed among the attriters.
a, c
Schmid, M. S. (a). Language Attrition. Cambridge: Cambridge University Press.
This is the first textbook devoted to language attrition, providing an introduction to the way in which attrition can affect the L, as well as to the extra- and sociolinguistic features involved. It also familiarizes the reader with experimental approaches to attrition and data analysis techniques and provides hands-on guidelines on how to apply them.
Yılmaz, G. (). Complex embeddings in free speech production among late Turkish-Dutch bilinguals. Language, Interaction and Acquisition (), –.
Yılmaz approaches the question of the attrition and retention of complex embedding structures previously tested by YAĞMUR (). Her analysis is based on free spoken data produced by long-term Turkish migrants in the Netherlands. Applying the markedness hierarchy of different embedding strategies proposed in earlier studies on Turkish, she finds that relativization is not reduced overall, but that attriters tend to make slightly less use of the most marked of these structures (postpositional clauses) as compared to an unattrited control population.
c
Chang, C. E. (). Rapid and multifaceted effects of second-language learning on firstlanguage speech production. Journal of Phonetics , –.
This study provides evidence for the assumption that language attrition is not a phenomenon confined to speakers in long-term immersion contexts: Chang investigates novice L learners of Korean with English L, and finds that even after only six weeks of intensive instruction in that language at a South Korean university, their L speech has become accoustically assimilated to the L on segmental, subsegmental, and global levels. He concludes that ‘cross-language linkages are established from the onset of secondlanguage learning at multiple levels of phonological structure, allowing for pervasive influence of secondlanguage experience on first-language representations’ (p. ).
e
de Leeuw, E., Mennen, I., & Scobbie, J. (). Singing a different tune in your native language: L attrition of prosody. International Journal of Bilingualism (), –.
These findings replicate the investigation presented by MENNEN () in that they show changes in prosodic contour in both the L and the L of bilingual (GermanEnglish) speakers alongside substantial intrapersonal variation. de Leeuw, Mennen, & Scobbie conclude that bidirectional transfer is impacted on by background variables such as age at learning as well as the phonetic environment.
e
(continued )
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Table A. Continued Study
Findings
Strand
Iverson, Michael (). Advanced language attrition of Spanish in contact with Brazilian Portuguese. PhD Dissertation, University of Iowa.
Iverson challenges the notion of the stability of core grammatical properties of a native language by putting to the test the Interface Hypothesis (Chamorro & Sorace, Chapter , this volume) in language attrition. Based on a single-case study of an advanced attriter of Spanish in contact with Brazilian Portuguese, he shows that both production data and grammaticality intuitions elicited from this individual consistently diverge from those of monolingual native speakers of these languages. Iverson concludes that ‘no [linguistic] level, including the syntax proper, has a privileged status with respect to its preservation’ (p. ).
a
Mayr, R., S. Price, & I. Mennen (). First language attrition in the speech of Dutch–English bilinguals: The case of monozygotic twin sisters. Bilingualism: Language and Cognition (), –.
This study presents an extensive investigation of changes in the phonetic space of a long-term L Dutch migrant in the United Kingdom. In order to assess changes to her vowel-system and to her production of initial plosives, controlled production (word list reading) is compared to the production of her identical twin sister (who has always lived in the Netherlands) on the same task. In this way, individual features pertaining to the articulatory apparatus, as well as social and neurological factors, are kept maximally similar between the experimental and the control subject. Mayr et al. conclude that both for VOT and for the structure of the vowel space, the experimental subject shows evidence of an assimilation to the L system.
e
Varga, Zsolt. (). First language attrition and maintenance among Hungarian speakers in Denmark. PhD dissertation, Aarhus University.
Varga presents a study of native Hungarian speakers in Denmark. His study adopts the methodology proposed in the European Graduate Network on Language Attrition (Schmid & Cherciov, Chapter , this volume). Despite long-term immersion (> years) in a Danish context, he finds little overall attrition in the data collected from his participants. In free speech, the only measurable difference related to an increase in disfluency markers and accentedness. Somewhat stronger effects were found regarding the formal linguistic tasks. These findings were not noticeably affected by background factors, such as age at time of emigration, LOR, or amount of contact.
a
Baus, C., Costa, A., & Carreiras, M. (). On the effects of second language immersion on first language production. Acta Psychologica (), –.
This investigation of a cohort of study-abroad students challenges the notion that L to L effects may only be observed after very long periods of immersion: Baus, Costa, & Carreiras find that performance on both a picture naming and a verbal fluency task declines after a period of residence of six months, and that in particular access to non-cognates is compromised.
a, a
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Bortolato, Claudia (). Language maintenance-attrition among generations of the Venetian-Italian community in Anglophone Canada. PhD dissertation, University of Exeter.
Bortolato traces the development of attritional patterns among Italians in Canada. She compares migrants who arrived in different historical phases as well as first- and subsequent generation speakers. A large amount of data is collected through sociolinguistic questionnaires, and this is supplemented by interviews with a subset of the respondents to those questionnaires. Bortolato shows how different circumstances and migrant populations (who may, for example, differ from each other with respect to educational level and socioeconomic standard), as well as different migrant generations, may differ from each other with respect to their attritional patterns.
a
Cherciov, M. (). Investigating the impact of attitude on first language attrition and second language acquisition from a Dynamic Systems Theory perspective. International Journal of Bilingualism (), –.
This paper is one of a series of studies applying the relatively recent framework of DST (see also de Bot, ; DE LEEUW, ; OPITZ, ) to language attrition. Cherciov investigates the role of attitude on the L development and L attrition among Romanian migrants in the Toronto area. She points out that, from a DST point of view, such analyses are not straightforward as the interaction of factors in a DS are characterized by properties such as non-linearity, interdependence, and complexity, and can thus not easily be predicted or measured. This is underlined by her analysis: while the group analyses reveal little predictive power of the measures of language use and attitude on the outcome variables, a discussion of the more complex features of these measures, as revealed in a quantitative interview, shows how various feelings (a desire to integrate, the wish to impart the L to the next generation, changing identities and identification, and so on) can conflict and interact. Individual migrants may respond in different ways to the complex situations they find themselves in, and this may account for the different patterns found in attrition studies, and the failure of linear, group-based statistical measures to yield explanatory power. This article is a report on the author’s PhD thesis (Cherciov, )
e
de Leeuw, E., I. Mennen, & J. M. Scobbie (). Dynamic systems, maturational constraints and L phonetic attrition. International Journal of Bilingualism (), –.
In this paper, de Leeuw, Mennen, & Scobbie present a re-analysis from a DST perspective of the data from de Leeuw’s () PhD thesis. This study investigated the lateral phoneme /l/ in the L and L of long-term German migrants in the Vancouver area, revealing L attrition at the group level as well as a great deal of variation between and within speakers. The authors address the longstanding problem in attrition research of how to define ‘loss’ and whether or not it can be assumed to be permanent. They argue that language development should not be viewed as ‘semi-static’, that is, as being determined by maturational constraints and
e, e
(continued )
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Table A. Continued Study
Findings
Strand
having discrete states such as a ‘final state’ and ‘fossilization’, but as ‘a complex process that continues throughout life’ (p. f.). This view entails the expectation for ‘language development to progress continuously, variably, and at different rates for different speakers’. Domínguez, L. (). Understanding Interfaces: Second Language Acquisition and First Language Attrition of Spanish Subject Realization and Word Order Variation. Amsterdam/ Philadelphia: John Benjamins.
Domínguez presents a book-length treatment of evidence supporting the Interface Hypothesis (Chamorro & Sorace, Chapter , this volume) in data from bilingual SpanishEnglish speakers. She investigates a range of grammatical phenomena used by populations in different settings and of different characteristics in order to help the reader gain a clearer understanding of the underlying theoretical issues.
c, c
Hopp, H. & M. S. Schmid (). Perceived foreign accent in first language attrition and second language acquisition: The impact of age of acquisition and bilingualism. Applied Psycholinguistics (), –.
Hopp & Schmid investigate to what extent a perceived foreign accent can be ascribed to age of onset versus the bilingual experience. To this end they compare late German-English/German-Dutch and highly advanced late English-German/Dutch-German bilinguals on how native-like they are perceived to be by native speakers of German. They find that, while at group level, the attriters are more similar to the controls than the Lers, the two bilingual populations overlap considerably. In particular, a substantial proportion of the Lers are rated within the range delimited by the attriters. Extralinguistic factors, such as the frequency of use of German in daily life, have no predictive power for the attriters but do impact on perceived accentedness for the Lers, and the amount of time spent in a German environment correlates with the degree of perceived accent for both populations.
b, e
Opitz, C. (). A dynamic perspective on late bilinguals’ linguistic development in an L environment. International Journal of Bilingualism (), –.
Opitz presents an investigation of German L and English L in long-term German migrants in Ireland. She finds that many of her participants are highly successful second language learners, performing within the reference group on a range of variables. For individual variables (C-Test, verbal fluency, free speech measures) there is no difference between the attriting group and the German control population; however, when a compound measure of all these variables is calculated, attrition effects to become visible. These are impacted on by a number of the external variables, such as LOR,
e
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
nationality, use of German, and attitudes. Opitz further discusses two case studies in detail, illustrating the complex, dynamic interaction of factors across a long period of residence. This article is a report on the author’s PhD thesis (Opitz, b) Yılmaz, G. & M. S. Schmid (). L accessibility among Turkish-Dutch bilinguals. The Mental Lexicon (), –.
This study of lexical access among long-term Turkish residents in the Netherlands compares controlled production in a Picture Naming Task and lexical diversity and fluency in free speech within the framework of the Activation Threshold Hypothesis (Paradis, ). Yılmaz and Schmid’s findings show that the attriters use a smaller vocabulary of more frequent items and hesitate more in free speech than the monolingual controls, but that their performance (speed and accuracy) on the naming task is unimpaired. An analysis of extralinguistic variables (frequency of use of L, attitudes towards both L and L) finds no significant correlations between these factors and the lexical and fluency features on which the participants appear to have attrited. The authors thus ascribe the reduction in lexical diversity and increase in hesitation markers not so much to ‘the process that is commonly understood to underlie attrition (a decline associated with lack of practice, i.e., a kind of ‘atrophy’) and more to a simple bilingualism effect: when there is more information to choose from, it takes longer to find it’ (p. ).
a
Alferink, I. & Gullberg, M. (). French–Dutch bilinguals do not maintain obligatory semantic distinctions: Evidence from placement verbs. Bilingualism: Language and Cognition (), –.
Alferink & Gullberg’s investigation of French-Dutch participants addresses a semantic distinction in placement verbs, which in Dutch have to be specified for the endstate (horizontal vs. vertical) while French makes no such distinction. They find that bilinguals in both their languages are less consistent with respect to verb choice than monolinguals of either language, and argue that a merged system has developed in these bilinguals with respect to some semi-obligatory semantic distinctions. They propose that such kinds of convergence may be a natural result of functional bilingualism.
b
Schmid, M. S. & S. Jarvis. (). Lexical access and lexical diversity in first language attrition. Bilingualism: Language and Cognition (), –.
Schmid and Jarvis investigate the decrease in lexical accessibility among long-term migrants and try to establish which measures of lexical diversity and lexical sophistication are most appropriate to capture this change. They conclude that measuring attrition in free speech provides a more accurate impression of lexical attrition than formal tasks or elicited narratives.
a
Schmid, M. S. (). The debate on maturational constraints in
Schmid’s study compares a group of forty L attriters and forty advanced L learners of German matched for general proficiency, LOR, and frequency of use of German. Her findings demonstrate that there are areas of
b
(continued )
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Table A. Continued Study
Findings
Strand
bilingual development: A perspective from first language attrition. Language Acquisition , –.
grammar where the L learners outperform long-term attriters, but that in other areas, in particular NP agreement, they persistently lag behind, and that no single L learner performs within the attriter range on these features. She interprets these findings in the context of maturational constraints on L learning.
Ulbrich, C. & M. Ordin. (). Can LEnglish influence LGerman? The case of post-vocalic /r/. Journal of Phonetics , –.
Ulrbich & Ordin investigate the pronunciation of postvocalic /r/ in speakers of a non-rhotic variety of German immersed in both a rhotic and a non-rhotic variety of English. They conclude that exposure to the former affects the pronunciation of /r/ in the L in some (but not all) phonetic environments.
e
Athanasopoulos, P., Bylund, E., Montero-Melis, G., Damjanovic, L., Schartner, A., Kibbe, A., Riches, N., & Thierry, G. (). Two languages, two minds: Flexible cognitive processing driven by language of operation. Psychological Science (), –.
In a series of papers, Athanasopoulos, Bylund, and colleagues present a compelling argument for the study of cognitive restructuring in bilingual development. They offer a series of investigations of the relationship between grammatical aspect and event cognition in both monolinguals (Athanasopoulos & Bylund, ) and bilinguals of various language combinations, including Afrikaans/English (Bylund, Athanasopoulos, & Oostendorp, ), is Xhosa/English (Bylund & Athanasopoulos, ), and Spanish/Swedish (Bylund, ; Bylund & Jarvis, ), cumulatively suggesting that speakers of languages that grammatically encode aspect show a preference to focus on path and less susceptibility to rely on endpoints when encoding motion events.
b
Gürel, A. (). First language attrition of constraints on wh-scrambling: Does the second language have an effect? International Journal of Bilingualism (), –.
Gürel investigates whether wh-scrambling out of islands, which is allowed in Turkish but disallowed in English, may be subject to change in mature, immersed bilinguals under attrition conditions. Similar properties had previously been found to be stable and wellpreserved under attrition conditions (e.g., PERPIÑÁN, ), and these findings are replicated in Gürel’s tasks, which suggest that the property has remained unaffected by long-term immersed bilingualism in Turkish-English bilinguals.
a, c
Kasparian, K. (). The Case of the Non-native-like First-Language:
Kasparian’s thesis represents a set of electroencelphalogram (EEG) experiments probing morpho-syntactic and lexico-semantic features (relative clauses, cognates/homophones, and similar words) in order to assess whether the neurocognitive
b
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Neurophysiological Investigations of FirstLanguage Attrition and Second-Language Processing. PhD Dissertation, McGill University,
mechanisms underlying L processing may change when a shift in exposure to, and use of, L and L occurs. She finds that there are indeed important differences in the brain responses collected from monolingual natives and attriters, modulated by other factors such as proficiency and amount of exposure. This thesis has been published in three papers: Kasparian & Steinhauer (, ) and Kasparian, Vespigniani, & Steinhauer ().
Laufer, B. & Baladzhaeva, L. (). First language attrition without second language acquisition: An exploratory study. ITL-International Journal of Applied Linguistics (), –.
Laufer & Baladzhaeva investigate the unusual phenomenon of migrants whose L may undergo attrition despite them having limited or no proficiency in the environmental language. They study a population of Russian migrants to Israel whom they divide into groups with and without knowledge of Hebrew. They find equal amounts of attrition in both populations with respect to collocations, while the speakers without knowledge of Hebrew even perform worse than the bilingual participants with respect to the accurate use of grammatical constructions. They conclude that a good level of L proficiency may support L maintenance.
b, a
Chamorro, G., Sorace, A., & Sturt, P. (). What is the source of L attrition? The effect of recent L re-exposure on Spanish speakers under L attrition. Bilingualism: Language and Cognition (), –.
Camorro, Sorace, & Sturt present one of the few investigations of language attrition that test the impact of re-immersion into the native language. They set out to test the assumption, formulated in the Interface Hypothesis (Chamorro & Sorace, Chapter , this volume), that while attrition may affect the ability to process interface structures, it will not change knowledge representations. They collect both on-line and off-line data relating to antecedent preferences in the L Spanish of Spanish-English bilinguals, and find that, while speakers who remain immersed in an English-speaking context for the duration of the fieldwork shows attrition effects, the group who was re-exposed to Spanish converges with the monolinguals. Their conclusion is that, while attrition affects online sensitivity to unacceptable structures, it does not cause a permanent change in knowledge representations.
a, c
Domínguez, L. & Hicks, G. (). Synchronic change in a multidialectal community: evidence from Spanish null and postverbal subjects. In A. Cuza, L. Czerwionka, & D. Olson (eds.),
Domínguez & Hicks approach the question of how the typological relatedness of the two contact languages affects language attrition: they investigate two linguistic communities of Spanish speakers which differ with respect to the proportions of overt and null subjects realized in free speech. They find that the language of Spanish native speakers settling in these communities will adapt to the properties of the input, approximating the distribution of overt and null subjects present in the community. Domínguez & Hicks interpret their
a, c
(continued )
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Table A. Continued Study Inquiries in Hispanic Linguistics: From Theory to Empirical Evidence (pp. –). Amsterdam: John Benjamins.
Findings
Strand
findings within the framework of Minimalist Theory and argue that native-language knowledge does not become stable at the point when linguistic maturity has been reached, but that linguistic experience may shape the speaker’s grammatical knowledge throughout her or his lifespan.
Bergmann, C. (). Facets of nativelikeness. Firstlanguage attrition among German emigrants to Anglophone North America. PhD dissertation, University of Groningen.
Bergmann uses a range of tasks and methodologies in order to obtain comprehensive insights into the processing and production of native speakers of German immersed in English-speaking communities in North America. He combines an analysis using EEG to investigate the processing of violations of gender concord (Bergmann, Meulman, Stowe, Sprenger, & Schmid, ), a phonetic analysis and foreign accent rating of language use in free speech (Bergmann, Nota, Sprenger, & Schmid, ) and an investigation of fluency in free speech (Bergmann, Sprenger, & Schmid, ). His findings show limited evidence of attrition, modulated by external factors.
b, a, c, e
Genevska-Hanke, D. (). Intrapersonal variation in late L attrition and its implications for the competence/ performance debate. In N. Levkovych & A. Urdze (eds.), Linguistik im Nordwesten: Beiträge zum . Nordwestdeutschen Linguistischen Kolloquium, – November . Bochum: Brockmeyer.
In an investigation that is similar to that presented by CHAMORRO, SORACE, & STURT (), Genevska-Hanke assesses the effect of re-immersion in an L context on the use of overt and null pronouns in Bulgarian-German bilingual. In this study, too, attrition effects are found prior to the participant’s return to her home country (i.e., she overuses overt and underuses null pronouns), but during a two-week stay in her home country, the distribution of pronouns returned to native-like settings. Genevska-Hanke concludes that language attrition is a matter of performance, not competence, in particular where aspects of core grammar are concerned.
c
de Leeuw, E., A. Tusha, & M. Schmid (). Phonological L attrition in AlbanianEnglish late bilinguals. Bilingualism: Language and Cognition () –.
de Leeuw, Tusha, & Schmid present what is, to date, the only investigation of the question of whether L attrition in the phonetic domain may eventually have ramifications for the phonological system. Based on de Leeuw’s () findings of bi-directional adaptation in the production of liquids in German-English bilinguals, they investigate the production of the same speech sound in Albanian, a language in which the contrast between light and dark /l/ is phonemic. Their findings show tentative evidence that this contrast may have undergone some changes in the productive data of one of their participants.
e
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Karayayla, Tuğba (). Turkish as an immigrant and heritage language in the UK: Effects of exposure and age at onset of bilingualism on grammatical and lexical development of the first language. PhD dissertation, University of Essex.
Karayayla’s PhD thesis traces various aspects of Turkish grammatical, phonetic, and lexical proficiency in a population of Turkish migrants in the UK. Her study is one of the few which include the full range of age at first exposure, from participants who were born in the UK to those who arrived as adults. She analyses the impact of background factors and concludes that Turkish proficiency remains largely stable for migrants who arrive after late childhood, but is variable for heritage speakers and child bilinguals. This study has been published as a series of papers (Karayayla, in press; Karayayla & Schmid, ).
a, c
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
REFERENCES ................................. Abrahamsson, N., & Hyltenstam, K. (). Age of onset and nativelikeness in a second language: Listener perception versus linguistic scrutiny. Language Learning, (), –. https://doi.org/ ./j.-...x Abutalebi, J., & Green, D. (). Bilingual language production: The neurocognition of language representation and control. Journal of Neurolinguistics, (), –. Abutalebi, J., & Green, D. W. (). Control mechanisms in bilingual language production: Neural evidence from language switching studies. Language and Cognitive Processes, (), –. doi:./ Abutalebi, J., Tettamanti, M., & Perani, D. (). The bilingual brain: Linguistic and non-linguistic skills. Brain and Language, (), –. Aikhenvald, A. Y. (). Grammars in contact: A cross-linguistic perspective. In A. Y. Aikhenvald & R. M. W. Dixon (Eds.) Grammars in Contact: A Cross-linguistic Typology (pp. –). Oxford: Oxford University Press. Alba, R. D. (). Language assimilation today: Bilingualism persists more than in the past, but English still dominates (CCIS Working Paper No. ). San Diego: Lewis Mumford Center for Comparative Urban and Regional Research, University at Albany. Albert, M. S., Heller, H. S., & Milberg, W. (). Changes in naming ability with age. Psychology and Aging, (), –. Albirini, A. (). The role of the colloquial varieties in the acquisition of the Standard Variety: The case of Arabic heritage speakers. Foreign Language Annals, (), –. doi: ./flan. Albirini, A., & Benmamoun, E. (). Concatenative and non-concatenative plural formation in L, L, and heritage speakers. Modern Language Journal, (), –. Albirini, A., Benmamoun, E., & Saadah, E. (). Grammatical features of Egyptian and Palestinian Arabic heritage speakers’ oral production. Studies in Second Language Acquisition, , –. Alexiadou, A., & Anagnostopoulou, E. (). Parameterizing Agr: Word order, V-movement and EPP-checking. Natural Language and Linguistic Theory, , –. Alexiadou, A., Anagnostopoulou, E., & F. Schäfer (). External Arguments in Transitivity Alternations: A Layering Approach. Oxford: Oxford University Press. Alexiadou, A., Lohndal, T., Åfarli, T., & Grimstad, M. B. (). Language mixing: A Distributed Morphology approach. In T. Bui & D. Özyildiz (Eds.), Proceedings of the th Annual Meeting of the North East Linguistic Society (pp. –). Somerville, MA: Cascadilla Press. Alharthi, T., & Al Fraidan, A. (). Language use and lexical attrition: Do they change over time? British Journal of English Linguistics, (), –. Alladi, S., Bak, T. H., Duggirala, V. et al. (). Bilingualism delays age at onset of dementia, independent of education and immigration status. Neurology, , –. Allen, S. (). The future of Inuktitut in the face of majority languages: Bilingualism or language shift? Applied Psycholinguistics, (), –. https://doi.org/./S Allen, S. E. M., Crago, M., & Pesco, D. (). The effect of majority language exposure on minority language skills: The case of Inuktitut. International Journal of Bilingual Education and Bilingualism, (), –. https://doi.org/./beb. Allopenna, P. D., Magnuson, J. S., & Tanenhaus, M. K. (). Tracking the time course of spoken word recognition using eye movements: Evidence for continuous mapping models. Journal of Memory and Language, (), –.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Alonso-Ovalle, L., Fernández-Solera, S., Frazier, L., & Clifton, C. Jr. (). Null vs. overt pronouns and the topic-focus articulation in Spanish. Rivista di Linguistica, , –. Altenberg, E. P. (). Assessing first language vulnerability to attrition. In H. W. Seliger & R. M. Vago (Eds.), First Language Attrition (p. –). New York: Cambridge University Press. Altenberg, E. P., & Vago, R. M. (). The role of grammaticality judgements in investigating first language attrition: A cross-disciplinary perspective. In M. S. Schmid, M. Keijzer, & L. Weilemar (Eds.), First Language Attrition: Interdisciplinary Perspectives on Methodological Issues (pp. –). Amsterdam: John Benjamins. Altmann, G. T. (). Language can mediate eye movement control within milliseconds, regardless of whether there is anything to move the eyes to. Acta Psychologica, , –. Altmann, G. T., & Kamide, Y. (). Incremental interpretation at verbs: Restricting the domain of subsequent reference. Cognition, (), –. Alwin, D. F., & McCammon, R. J. (). Aging, cohorts, and verbal ability. The Journals of Gerontology Series B: Psychological Sciences and Social Sciences, (), S–S. Amaral, L., & Roeper, T. (). Multiple grammars and second language representation, Second Language Research, (), –. Ameel, E., Malt, B. C., Storms, G., & Van Assche, F. (). Semantic convergence in the bilingual lexicon. Journal of Memory and Language, , –. Amengual, M. (). Interlingual influence in bilingual speech: Cognate status effect in a continuum of bilingualism. Bilingualism: Language and Cognition, , –. American Psychiatric Association (). Diagnostic and Statistical Manual of Mental Disorders (th edn). Arlington, VA: American Psychiatric Publishing. Ammerlaan, T. (). ‘You get a bit wobbly . . . ’—Exploring bilingual lexical retrieval processes in the context of first language attrition. PhD thesis, Katholieke Universiteit. Ammerlaan, T. (). ‘Corrosion’ or ‘loss’ of immigrant Dutch in Australia: An experiment on first language attrition. In J. Klatter-Folmer & S. Kroon (Eds.), Dutch Overseas: Studies in Maintenance and Loss of Dutch as an Immigrant Language (pp. –). Tilburg: Tilburg University Press. Andersen, R. W. (). Determining the linguistic attributes of language attrition. In R. D. Lambert & B. F. Freed (Eds.), The Loss of Language Skills (pp. –). Rowley, MA: Newbury House. Anderson, M. C. (). Rethinking interference theory: Executive control and the mechanisms of forgetting. Journal of Memory and Language, , –. Anderson, M. C., & Spellman, B. A. (). On the status of inhibitory mechanisms in cognition: Memory retrieval as a model case. Psychological Review, (), –. Anderson, M. C., Bjork, R. A., & Bjork, E. L. (). Remembering can cause forgetting: Retrieval dynamics in long-term memory. Journal of Experimental Psychology: Learning, Memory and Cognition, (), –. Anderson, M. J. (). A new method for non-parametric multivariate analysis of variance. Austral Ecology, (): –. Anderson, R. (). Impact of first language loss on grammar in a bilingual child. Communication Disorders Quarterly, , –. Anderson, R. (). First language loss in Spanish-speaking children. In B. Goldstein (Ed.), Bilingual Language Development and Disorders in Spanish-English Speaking Children (nd edn) (pp. –). Baltimore: Brookes Publishing. Anderson, R. T. (). Lexical morphology and verb use in child first language loss: A preliminary case study investigation. International Journal of Bilingualism, , –. Anderson, S. (). Dimensions of morphological complexity. In M. Baerman, D. Brown, & G. G. Gorbett (Eds.), Understanding and Measuring Morphological Complexity (pp. –). Oxford: Oxford University Press. Andrews, E., Frigau, L., Voyvodic-Casabo, C., Voyvodic, J., & Wright, J. (). Multilingualism and fMRI: Longitudinal study of second language acquisition. Brain Sciences, (), –. Argyri, E., & Sorace, A. (). Crosslinguistic influence and language dominance in older bilingual children. Bilingualism: Language and Cognition, (), –.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Armon-Lotem, S. (). Introduction: Bilingual children with SLI—the nature of the problem. Bilingualism: Language and Cognition, , –. Armon-Lotem, S., Joffe, S., Abutbul-Oz, H., Altman, C., & Walters, J. (). Language exposure, ethnolinguistic identity and attitudes in the acquisition of Hebrew as a second language among bilingual preschool children from Russian- and English-speaking backgrounds. In T. Grüter & J. Paradis (Eds.), Input and Experience in Bilingual Development (pp. –). Amsterdam: John Benjamins. Arnbjörnsdóttir, B. (). Reexamining Icelandic as a heritage language in North America. In J. B. Johannessen & J. C. Salmons (Eds.), Germanic Heritage Languages in North America: Acquisition, Attrition, and Change (pp. –). Amsterdam: John Benjamins. Arnon, I., & Christiansen, M. H. (). Chunk-based acquisition. In P. Brooks & V. Kempe (Eds.), Encyclopedia of Language Development (pp. –). Thousand Oaks, CA: Sage Publications. Arthur, W., Bennett, W., Stanush, P. L., & McNelly, T. L. (). Factors that influence skill decay and retention: A quantitative review and analysis. Human Performance, (), –. Ås, A. (). The recovery of forgotten language knowledge through hypnotic age regression: A case report. American Journal of Clinical Hypnosis, (), –. https://doi.org/./ .. Ashaie, S., & Obler, L., K. (). Effect of age, education, and bilingualism on confrontation naming in older illiterate and low-educated populations. Behavioural Neurology, , –. Athanasopoulos, P. (). Cognitive restructuring in bilingualism. In A. Pavlenko (Ed.), Thinking and speaking Two Languages (pp. –). Bristol: Multilingual Matters. Athanasopoulos, P., & Bylund, E. (). Does grammatical aspect affect motion event cognition? A cross-linguistic comparison of English and Swedish speakers. Cognitive science, (), –. Athanasopoulos, P., & Kasai, C. (). Language and thought in bilinguals: The case of grammatical number and nonverbal classification preferences. Applied Psycholinguistics, , –. Atterer, M., & Ladd, D. R. (). On the phonetics and phonology of ‘segmental anchoring’ of F: evidence from German. Journal of Phonetics, (), –. https://doi.org/./S()- Au, R., Joung, P., Nicholas, M., Obler, L. K., Kass, R., & Albert, M. L. (). Naming ability across the adult life span. Aging, Neuropsychology, and Cognition, (), –. Au, T. K. (). Access to childhood language memory: Implications for cognitive development. In M. Siegal & L. Surian (Eds.), Access to language and cognitive development (pp. –). New York: Oxford University Press. Au, T. K., & Romo, L. F. (). Does childhood language experience help adult learners? In H. C. Chan (Ed.), The Cognitive Processing of Chinese and Related Asian Languages (pp. –). Hong Kong: Chinese University Press. Au, T. K., Knightly, L. M., Jun, S.-A., & Oh, J. S. (). Overhearing a language during childhood. Psychological Science, (), –. doi:https://doi.org/./-. Au, T. K., Oh, J. S., Knightly, L. M., Jun, S.-A., & Romo, L. F. (). Salvaging a childhood language. Journal of Memory and Language, , –. doi: ./j.jml... Austin, J. (). Delay, interference and bilingual development: The acquisition of verbal morphology in children learning Basque and Spanish. International Journal of Bilingualism, , –. Backus, A. (). Turkish as an immigrant language. In T. K. Bhatia & W. C. Ritchie (Eds.), The Handbook of Bilingualism (pp. –). Oxford: Blackwell. Bäckman, L., Jones, S., Berger, A. K., Laukka, E. J., & Small, B. (). Multiple cognitive deficits during the transition to Alzheimer’s disease. Journal of Internal Medicine, (), –. Baddeley, A. (). Working Memory, Thought, and Action. Oxford: Oxford University Press. Baddeley, A. D. (). Working memory: Theories, models, and controversies. Annual Review of Psychology, , –. Baddeley, A. D., & Hitch, G. (). Working memory. In G. H. Bower (Ed.), The Psychology of Learning and Motivation: Advances in Research and Theory (pp. –). New York: Academic Press. Badstübner, T. (). L attrition: German immigrants in the U.S. PhD dissertation, University of Arizona at Tuscon.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Baerman, M., Brown, D., & Gorbett, G. G. (). Understanding and measuring morphological complexity: An introduction. In M. Baerman, D. Brown, & G. G. Gorbett (Eds.), Understanding and Measuring Morphological Complexity (pp. –). Oxford: Oxford University Press. Baese-Berk, M., & Goldrick, M. (). Mechanisms of interaction in speech production. Language and Cognitive Processes, , –. Bahrick, H. P. (a). Fifty years of second language attrition: Implications for programmatic research. The Modern Language Journal, (), –. Bahrick, H. P. (b). Semantic memory content in permastore: Fifty years of memory for Spanish learned in school. Journal of Experimental Psychology: General, (), –. Bahrick, H. P., & Phelps, E. (). Retention of Spanish vocabulary over years. Journal of Experimental Psychology: Learning, Memory, and Cognition, (), –. Bak, T. H. (). The impact of bilingualism on cognitive aging and dementia: Finding a path through a forest of confounding variables. Linguistic Approaches to Bilingualism, , –. Bak, T. H., Nissan, J. J., Allerhand, M. M., & Deary, I. J. (). Does bilingualism influence cognitive aging? Annals of Neurology, (), –. Baker, C. (). Foundations of Bilingual Education and Bilingualism. Bristol: Multilingual Matters. Baker, W., & Trofimovich, P. (). Interaction of native- and second-language vowel system(s) in early and late bilinguals. Language and Speech, (), –. https://doi.org/./ Bakker, I., Takashima, A., Van Hell, J. G., Janzen, G., & McQueen, J. M. (). Changes in theta and beta oscillations as signatures of novel word consolidation. Journal of Cognitive Neuroscience, (), –. doi:./jocn_a_ Baladzhaeva, L. (). L attrition without L acquisition: A case of Russian-speaking immigrants in Israel. MA dissertation, University of Haifa. Baladzhaeva, L., & Laufer, B. (). Is first language attrition possible without second language knowledge? International Review of Applied Linguistics in Language Teaching. https://doi.org/ ./iral-- Balcom, P. (). Cross-linguistic influence of L English on middle constructions in L French. In V. Cook (Ed.), Effects of the Second Language on the First (pp. –). Bristol: Multilingual Matters. Barbu, S., Martin, N., & Chevrot, J.-P. (). The maintenance of regional dialects: a matter of gender? Boys, but not girls, use local varieties in relation to their friends’ nativeness and local identity. Frontiers in Psychology, , . Bardovi-Harlig, K., & Stringer, D. (). Variables in second language attrition: Advancing the state of the art. Studies in Second Language Acquisition, , –. Bardovi-Harlig, K., & Stringer, D. (). The lexicon in second language attrition: What happens when the cat’s got your tongue? In J. Altarriba & L. Isurin (Eds.), Memory, Language, and Bilingualism: Theoretical and Applied Approaches (pp. –). Cambridge: Cambridge University Press. Barresi, B. A., Nicholas, M., Connor, L. T., Obler, L. K., & Albert, M. L. (). Semantic degradation and lexical access in age-related naming failures. Aging, Neuropsychology, and Cognition, (), –. Barriga Villanueva, R. (). Miradas a la interculturalidad [Views on Interculturality]. Revista Mexicana de Investigación Educativa, (), –. Barton, D. (). Generische Nominalphrasen bei deutsch-französischer Zweisprachigkeit: Zur Verwendung des Definitartikels bei erwachsenen Herkunftssprechern [Generic noun phrases in German-French bilingualism: on the use of the definite article in adult heritage speakers]. PhD thesis, University of Hamburg. Bassetti, B., & Cook, V. J. (). Relating language and cognition: The second language user. In V. J. Cook & B. Bassetti (Eds.), Language and Bilingual cognition (pp. –). New York: Psychology Press. Bastiaanse, R., & Bol, G. (). Verb inflection and verb diversity in three populations: Agrammatic speakers, normally developing children, and children with specific language impairment (SLI). Brain and Language, (), –. doi:./brln..
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Bastiaansen, M., & Hagoort, P. (). Oscillatory neuronal dynamics during language comprehension. Progress in Brain Research, , –. doi:./s-()- Bates, E., McNew, S., MacWhinney, B., Devescovi, A., & Smith, S. (). Functional constraints on sentence processing: A cross-linguistic study. Cognition, (), –. Baus, C., Costa, A., & Carreiras, M. (). On the effects of second language immersion on first language production. Acta Psychologica, (), –. Bayram, F., Prada, J., Pascual y Cabo, D., & Rothman, J. (). Why should formal linguistic approaches to heritage language acquisition be linked to heritage language pedagogies? In P. Trifonas & T. Aravossitas (Eds.), Handbook of Research and Practice in Heritage Language Education (pp. –). Dordrecht: Springer. Bayram, F., Rothman, J., Iverson, M., et al. (). Differences in use without deficiencies in competence: Passives in the Turkish and German of Turkish heritage speakers in Germany. International Journal of Bilingual Education and Bilingualism. doi: ./.. Beacco, J. C. (). The role of languages in policies for the integration of adult migrants. Concept paper presented at The Linguistic Integration of Adult Migrants, Strasbourg. Beard, R. (). Morpheme order in a lexeme/morpheme-based morphology, Lingua, , –. Beard, R. (). On the separation of derivation from morphology toward a lexeme/morphemebased morphology. Quaderni di Semantica, (), –. Beckner, C., Blythe, R., Bybee, J., et al. [The ‘Five Graces Group’] (). Language is a complex adaptive system: Position Paper. Language Learning, (s), –. Bedny, M., Pascual-Leone, A., Dodell-Feder, D., Fedorenko, E., & Saxe, R. (). Language processing in the occipital cortex of congenitally blind adults. Proceedings of the National Academy of Sciences, (), –. Bedore, L. M., Peña, E. D., Griffin, Z. M., & Hixon, J. G. (). Effects of age of English exposure, current input/output, and grade on bilingual language performance. Journal of Child Language, , –. Bel, A., Ortells, M., & Morgan, G. (). Reference control in the narratives of adult sign language learners. International Journal of Bilingualism, (), –. Bell, A. (). Language style as audience design. Language in Society, (), –. https://doi.org/ ./SX Belletti, A., & Guasti, M. T. (). The Acquisition of Italian: Morphosyntax and its Interfaces in Different Modes of Acquisition. Amsterdam: John Benjamins. Belletti, A., Benatti, E., & Sorace, A. (). Theoretical and developmental issues in the syntax of subjects: Evidence from near-native Italian. Natural Language and Linguistic Theory, , –. Ben-Rafael, M., & Schmid, M. S. (). Language attrition and ideology: Two groups of immigrants in Israel. In B. Köpke, M. S. Schmid, M. Keijzer, & S. Dostert (Eds.), Language Attrition: Theoretical Perspectives (pp. –). Amsterdam: John Benjamins. Benmamoun, E., Montrul, S., & Polinsky, M. (a). Heritage languages and their speakers: Opportunities and challenges for linguistics. Theoretical Linguistics, (–), –. doi: ./tl- Benmamoun, E., Montrul, S., & Polinsky, M. (b). Defining an ‘ideal’ heritage speaker: Theoretical and methodological challenges Reply to peer commentaries. Theoretical Linguistics, (–), –. https://doi.org/./tl-- Berends, S. M., Brouwer, S., & Sprenger, S. (). Eye-tracking and the visual world paradigm. In M. S. Schmid, S. M. Berends, C. Bergmann, et al. (Eds.), Designing Research on Bilingual Development: Behavioural and Neurolinguistic Experiments (pp. –). Cham: Springer. Berent, I., Steriade, D., Lennertz, T., & Vaknin, V. (). What we know about what we have never heard: Evidence from perceptual illusions. Cognition, , –. Berg, T., & Schade, U. (). The role of inhibition in a spreading activation model of language production: The psycholinguistic perspective. Journal of Psycholinguistic Research, (), –. Bergmann, C., Sprenger, S., & Schmid, M. S. (). The impact of language co-activation on L and L speech fluency. Acta Psychologica, , –.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Bergmann, C., Meulman, N., Stowe, L. A., Sprenger, S. A., & Schmid, M. S. (). Prolonged L immersion engenders little change in morphosyntactic processing of bilingual natives. Neuroreport, , –. Bergmann, C., Nota, A., Sprenger, S. A., & Schmid, M. S. (). L immersion causes non-native-like L pronunciation in German attriters. Journal of Phonetics, , –. Berken, J. A., Chai, X., Chen, J. K., Gracco, V. L., & Klein, D. (). Effects of early and late bilingualism on resting-state functional connectivity. The Journal of Neuroscience, (), –. doi:https://doi.org/./JNEUROSCI.-. Berko-Gleason, J. (). Insights from child language acquisition for second language loss. In R. D. Lambert & B. F. Freed (Eds.), The Loss of Language Skills (pp. –). Rowley, MA: Newbury House. Berman, R. (). The re-emergence of a bilingual: Case-study of a Hebrew-English child. Working Papers on Bilingualism, , –. Berman, R. (). Crosslinguistic aspects of later language development. In S. Strömqvist (Ed.), The Diversity of Language and Language Learning (pp. –). Lund University: Center for Language and Literature. Berman, R. A. (). Between emergence and mastery: The long developmental route of language acquisition. In R. A. Berman (Ed.), Language Development across Childhood and Adolescence: Trends in Language Acquisition Research (pp. –). Amsterdam: John Benjamins. Berman, R. A., & Olshtain, E. (). Features of first language transfer in second language attrition. Applied Linguistics, (), –. doi:./applin/.. Bernolet, S., Hartsuiker, R. J., & Pickering, M. J. (). Shared syntactic representations in bilinguals: Evidence for the role of word-order repetition. Journal of Experimental Psychology: Learning, Memory, and Cognition, (), –. Berry, J. W. (). Acculturation as varieties of adaptation. In A. Padilla (Ed.), Acculturation Theory, Models and Some New Findings (pp. –). Boulder, CO: Westview Press. Berry, J. W. (). Immigration, acculturation, and adaptation. Applied Psychology: An International Review, (), –. https://doi.org/./j.-..tb.x Berry, J. W. (). Conceptual approaches to acculturation. In K. Chun, P. Balls-Organista, & G. Marin (Eds.), Acculturation: Advances in Theory, Measurement and Application (pp. –). Washington: APA Books. Berwick, R. C. (). The Acquisition of Syntactic Knowledge. Cambridge, MA: MIT Press. Best, C., & Tyler, M. (). Nonnative and second-language speech perception: Commonalities and complementarities. In O.-S. Bohn & M. Munro (Eds.), Language Experience in Second Language Speech Learning. In Honor of James Emil Flege (pp. –). Amsterdam: John Benjamins. Bialystok, E. (). Bilingualism: The good, the bad, and the indifferent. Bilingualism: Language and Cognition, (), –. Bialystok, E., & Hakuta, K. (). Confounded age: Linguistic and cognitive factors in age differences for second language acquisition. In D. Birdsong (Ed.), Second Language Acquisition and the Critical Period Hypothesis (pp. –). Mahwah, NJ: Lawrence Erlbaum. Bialystok, E., & Sharwood Smith, M. (). Interlanguage is not a state of mind: An evaluation of the construct for second-language acquisition. Applied Linguistics, (), –. Bialystok, E., Craik, F. I. M., Klein, R., & Viswanathan, M. (). Bilingualism, aging, and cognitive control: evidence from the Simon task. Psychology and Aging, (), –. Bialystok, E., Craik, F. I. M., & Freedman, M. (). Bilingualism as a protection against the onset of symptoms of dementia. Neuropsychologia, (), –. Bialystok, E., Craik, F. I. M., & Luk, G. (). Cognitive control and lexical access in younger and older bilinguals. Journal of Experimental Psychology: Learning, Memory, and Cognition, (), –. Bialystok, E., Craik, F. I. M., Green, D. W., & Gollan, T. H. (). Bilingual minds. Psyhological Science in the Public Interest, (), –. Bialystok, E., Craik, F. I. M., & Luk, G. (). Bilingualism: Consequences for mind and brain. Trends in Cognitive Sciences, (), –.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Bialystok, E., Abutalebi, J., Bak, T. H., Burke, D. M., & Kroll, J. F. (). Aging in two languages: Implications for public health. Ageing Research Reviews, , –. Bianchi, G. (). Gender in Italian-German bilinguals: A comparison with German L learners of Italian. Bilingualism: Language and Cognition, (), –. doi: ./S Biberauer, T., Holmberg, A., Roberts, I., & Sheehan, M. (). Parametric Variation: Null Subjects in Minimalist Theory. Cambridge: Cambridge University Press. Bice, K., & Kroll, J. F. (). Native language change during early stages of second language learning. NeuroReport, (), –. Bierling, M. R. (). Foreign language vocabulary attrition: A study of college-level Spanish students over an -month period. PhD thesis, Michigan State University. Bijl, R., & Verweij, A. (). Measuring and monitoring immigrant integration in Europe: Facts and views. In R. Bijl & A. Verweij (Eds.), Measuring and Monitoring Immigrant Integration in Europe: Integration Policies and Monitoring Efforts in Countries (pp. –). The Hague: The Netherlands Institute for Social Research. Bini, M. (). La adquisición del italiano: Más allá de las propiedades sintácticas del parámetro prodrop [The acquisition of Italian: beyond the syntactic properties of the pro-drop parameter]. In J. M. Liceras (Ed.), La lingüística y el análisis de los sistemas no nativos [Linguistics and the analysis of non-native systems] (pp. –). Ottawa: Dovehouse. Birdsong, D. (). Age and second language acquisition and processing: A selective overview. Language Learning, -S, –. Birdsong, D. (). Dominance in bilingualism: Foundations of measurement, with insights from the study of handedness. In C. Silva-Corvalán & J. Treffers-Daller (Eds.), Language Dominance in Bilinguals: Issues of Measurement and Operationalization (pp. –). Cambridge: Cambridge University Press. doi:./CBO. Birdsong, D., & Molis, M. (). On the evidence for maturational effects in second language acquisition. Journal of Memory and Language, , –. Bishop, D. V. M. (). What causes Specific Language Impairment in children? Current Directions in Psychological Science, , –. Bishop, D. V. M. (). Ten questions about terminology for children with unexplained language problems. International Journal of Language and Communication Disorders, , –. Bishop, D. V. M. (). Why is it so hard to reach agreement on terminology? The case of developmental language disorder (DLD). International Journal of Language & Communication Disorders, (), –. Bjork, E. L., & Bjork, R. A. (). Continuing influences of to-be-forgotten information. Consciousness and Cognition: An International Journal, , –. Blake, R. (). Mood selection among Spanish-speaking children, ages to . The Bilingual Review, (), –. Block, D. (). The Social Turn in Second Language Acquisition. Washington, DC: Georgetown University Press. Blom, E., & Boerma, T. (). Why do children with language impairment have difficulties with narrative macrostructure? Research in Developmental Disabilities, , –. Blom, E., & Boerma, T. (). Effects of language impairment and bilingualism across domains: Vocabulary, morphology and verbal memory. Linguistic Approaches to Bilingualism, (–), –. Blom, E., & Bosma, E. (). The sooner the better? An investigation into the role of age of onset and its relation with transfer and exposure in bilingual Frisian–Dutch children. Journal of Child Language, (), –. doi: ./S Bloom, P. (). Syntactic distinctions in child language. Journal of Child Language, (), –. http://dx.doi.org/./S Blumenfeld, H. K., & Marian, V. (). Constraints on parallel activation in bilingual spoken language processing: Examining proficiency and lexical status using eye-tracking. Language and Cognitive Processes, , –.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Bobb, S. C., & Wodniecka, Z. (). Language switching in picture naming: What asymmetric switch costs (do not) tell us about inhibition in bilingual speech planning. Journal of Cognitive Psychology, (), –. Bock, J. K. (). Syntactic persistence in language production. Cognitive Psychology, (), –. Bock, K., & Griffin, Z. M. (). The persistence of structural priming: Transient activation or implicit learning? Journal of Experimental Psychology: General, (), –. Bock, K., & Levelt, W. (). Language production. In G. T. Altmann (Ed.), Psycholinguistics: Critical Concepts in Psychology, Vol. (pp. –). London: Routledge. Boeckx, C. (). Elementary Syntactic Structures: Prospects of a Feature-free Syntax. Cambridge: Cambridge University Press. Boeckx, C., & Grohmann, K. K. (). The biolinguistics manifesto. Biolinguistics, , –. Boerma, T., Chiat, S., Leseman, P., Timmermeister, M., Wijnen, F., & Blom, E. (). A quasiuniversal nonword repetition task as a diagnostic tool for bilingual children learning Dutch as a second language. Journal of Speech, Language, and Hearing Research, , –. Boerma, T., Leseman, P., Timmermeister, M., Wijnen, F., & Blom, E. (). Narrative abilities of monolingual and bilingual children with and without language impairment: implications for clinical practice. International Journal of Language and Communication Disorders, (), –. Boersma, P., & Weenink, D. (). Praat: Doing phonetics by computer. Version .. http://www.fon. hum.uva.nl/praat/ Bohman, T. M., Bedore, L. M., Pena, E. D., Mendez-Perez, A., & Gillam, R. B. (). What you hear and what you say: Language performance in Spanish-English bilinguals. International Journal of Bilingual Education and Bilingualism, (), –. doi: ./.. Bolonyai, A. (). The hidden dimensions of language contact: The case of Hungarian-English bilingual children. PhD dissertation, University of South Carolina. Bolonyai, A. (). Elective affinities: Language contact in the abstract lexicon and its structural consequences. International Journal of Bilingualism, , –. Bolonyai, A. (). Case systems in contact: Syntactic and lexical case in bilingual child language. Southwest Journal of Linguistics, (), –. Bolonyai, A. (). Vulnerable morphemes in imperfect bilingual L acquisition. In A. Brugos, M. R. Clark-Cotton, and S. Ha (Eds.), Proceedings of the th Annual Boston University Conference on Language Development (pp. –). Boston, MA: Cascadilla Press. Bolonyai, A. (). (In)vulnerable agreement in incomplete bilingual L learners. International Journal of Bilingualism, , –. Bolonyai, A., & Dutkova-Cope, L. (). L Attrition of verbal morphology in bilingual children and adults. In X. Bonch-Bruevich, W. J. Crawford, J. Hellermann, C. Higgins, & H. Nguyen (Eds.), The Past, Present, and Future of Second Language Research: Selected Proceedings of the Second Language Research Forum (pp. –). Somerville, MA: Cascadilla Press. Bongaerts, T. (). Ultimate attainment in L pronunciation: The case of very advanced late learners. In D. Birdsong (Ed.), Second Language Acquisition and the Critical Period Hypothesis (pp. –). Mahwah, NJ: Lawrence Erlbaum. Bonnesen, M. (). The status of the ‘weaker’ language in unbalanced French/German bilingual language acquisition. Bilingualism: Language and Cognition, (), –. doi: ./ S Bornkessel-Schlesewsky, I., Kretzschmar, F., Tune, S., et al. (). Think globally: Cross-linguistic variation in electrophysiological activity during sentence comprehension. Brain and Language, (), –. Bosch, L., & Sebastián-Gallés, N. (). Simultaneous bilingualism and the perception of a language specific vowel contrast in the first year of life. Language and Speech, , –. Bourhis, R. Y., Moise, L. C., Perreault, S., & Senecal, S. (). Towards an interactive acculturation model: A social psychological approach. International Journal of Psychology, , –.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Bousquette, J., Putnam, M. T., Salmons, J. C., Frey, B., & Nützel, D. (). Multiple grammars, dominance, and optimization. In G. Legendre, M. T. Putnam, H. de Swart, & E. Zaroukian (Eds.), Optimality-theoretic, Syntax, Semantics, and Pragmatics: From Uni- to Bidirectional Optimization (pp. –). Oxford: Oxford University Press. Bowers, J. S., Mattys, S. L., & Gage, S. H. (). Preserved implicit knowledge of a forgotten childhood language. Psychological Science, , –. doi: ./j.-...x Boyer, S. P. (). Learning from each other: The integration of immigrant and minority groups in the United States and Europe. Center for American Progress. http://www.migration-boell.de/ downloads/integration/CAP_immigrant_integration.pdf (accessed June ). Bradley, R. H., & Caldwell, B. M. (). The HOME inventory and family demographics. Developmental Psychology, , –. Bradlow, A. R., Pisoni, D. B., Akahane-Yamada, R., & Tohkura, Y. (). Training Japanese listeners to identify English /r/ and /l/: IV. Some effects of perceptual learning on speech production. Journal of the Acoustical Society of America, , –. doi: ./. Braine, M. (). What is learned in acquiring word classes—A step toward an acquisition theory. In B. MacWhinney (Ed.), Mechanisms of Language Acquisition (pp. –). Hillsdale, NJ: Lawrence Erlbaum. Bramer, A., Ren, X., Henley, M., et al. (). Reduced sensitivity to L-English spelling-sound regularity early in L learning. Poster presented at the th International Symposium on Bilingualism, Limerick, July. Branzi, F. M., Della Rosa, P. A., Canini, M., Costa, A., & Abutalebi, J. (). Language control in bilinguals: Monitoring and response selection. Cerebral Cortex, (), –. doi: ./ cercor/bhv Brehmer, B., & Kurbangulova, T. (). Lost in transmission? Family language input and its role for the development of Russian as a heritage language in Germany. In L. Isurin & C. M. Riehl (Eds.), Russian Migration to Germany (pp. –). Amsterdam: John Benjamins. Brooks, P. J., & Tomasello, M. (). How children constrain their argument structure constructions. Language, , –. Brown, A., & Gullberg, M. (). Bidirectional cross-linguistic influence in L-L encoding of manner in speech and gesture: A study of Japanese speakers of English. Studies in Second Language Acquisition, (), –. Brown, A., & Gullberg, M. (). Bidirectional cross-linguistic influence in event conceptualization? Expressions of Path among Japanese learners of English. Bilingualism: Language and cognition, (), –. Brown, G. G., Perthen, J. E., Liu, T. T., & Buxton, R. B. (). A primer on functional magnetic resonance imaging. Neuropsychological Reviews, , –. doi: ./s--- Brown, R. (). A First Language: The Early Stages. Cambridge, MA: Harvard University Press. Bruhn de Garavito, J. (). Verb raising in Spanish, a comparison of early and late bilinguals. In B. Skarabela, S. Fish, & H.-J. Anna (Eds.), Proceedings of the th Annual Boston University Conference on Language Development (pp. –). Somerville, MA: Cascadilla. Brysbaert, B., & Duyck, W. (). Is it time to leave behind the Revised Hierarchical Model of bilingual language processing after fifteen years of service? Bilingualism: Language and Cognition, (), –. Brysbaert, M., Stevens, M., Mandera, P., & Keuleers, E. (). How many words do we know? Practical estimates of vocabulary size dependent on word definition, the degree of language input and the participant’s age. Frontiers in Psychology, ( July), . https://doi.org/./ fpsyg.. Bullock, B. E., & Gerfen, C. (). Frenchville French: A case study in phonological attrition. International Journal of Bilingualism, (), –. Bullock, B. E., & Toribio, A. J. (). Intra-system variability and change in nominal and verbal morphology. In R. S. Gess & D. Arteaga (Eds.), Historical Romance Linguistics: Retrospective and Perspectives (pp. –). Amsterdam: John Benjamins.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Bullock, B. E., & Toribio, A. J. (). Themes in the study of code-switching. In B. E. Bullock & A. J. Toribio (Eds.), The Cambridge Handbook of Linguistic Code-switching (pp. –). Cambridge: Cambridge University Press. Bullock, B. E., Dalola, A., & Gerfen, C. (). Mapping the patterns of maintenance versus merger in bilingual phonology: The preservation of [a] vs. [ɑ] in Frenchville French. In J.-P. Y. Montreuil (Ed.), New Perspectives on Romance linguistics. Vol. II: Phonetics, Phonology and Dialectology (pp. –). Amsterdam: John Benjamins. Bulté, B., & Housen, A. (). Defining and operationalising L complexity. In A. Housen, F. Kuiken, & I. Vedder (Eds.), Dimensions of L Performance and Proficiency: Complexity, Accuracy and Fluency in SLA (pp. –). Amsterdam: John Benjamins. Burke, D. M. (). Language, aging, and inhibitory deficits: Evaluation of a theory. The Journals of Gerontology Series B: Psychological Sciences and Social Sciences, (), P–P. Burke, D. M., MacKay, D. G., Worthley, J. S., & Wade, E. (). On the tip of the tongue: What causes word finding failures in young and older adults? Journal of Memory and Language, (), –. Burling, R. (). Language development of a Garo and English speaking child. Word, , –. Burnham, D. K. (). Developmental loss of speech perception: Exposure to and experience with a first language. Applied Psycholinguistics, (), –. Bybee, J. (). Phonology and Language Use. Cambridge: Cambridge University Press. Bybee, J. (). From usage to grammar: The mind’s response to repetition. Language, (), –. Bylund, E. (). Age Differences in First Language Attrition. Stockholm: Stockholm University. Bylund, E. (a). Effects of age of L acquisition on L event conceptualization patterns. Bilingualism: Language and Cognition, (), –. doi: ./S Bylund, E. (b). Maturational constraints and first language attrition. Language Learning, (), –. doi: ./j.-...x Bylund, E. (). Segmentation and temporal structuring of events in early Spanish-Swedish bilinguals. International Journal of Bilingualism, (), –. doi: ./ Bylund, E., & Athanasopoulos, P. (). Language and thought in a multilingual context: The case of isiXhosa. Bilingualism: Language and Cognition, (), –. Bylund, E., & Díaz, M. (). The effects of heritage language instruction on first language proficiency: A psycholinguistic perspective. International Journal of Bilingual Education and Bilingualism, (), –. doi: ./.. Bylund, E., & Jarvis, S. (). L effects on L event conceptualization. Bilingualism: Language and Cognition, (), –. Bylund, E., & Ramírez-Galan, P. (). Language aptitude in first language attrition: A study on late Spanish–Swedish bilinguals. Applied Linguistics, (), –. Bylund, E., Abrahamsson, N., & Hyltenstam, K. (). The role of language aptitude in first language attrition: The case of pre-pubescent attriters. Applied Linguistics, (), –. doi: ./ applin/amp Bylund, E., Abrahamsson, N., & Hyltenstam, K. (). Does first language maintenance hamper nativelikeness in a second language? Studies in Second Language Acquisition, (), –. doi: ./S Bylund, E., Athanasopoulos, P., & Oostendorp, M. (). Motion event cognition and grammatical aspect: Evidence from Afrikaans. Linguistics, (), –. Cabrelli Amaro, J. (). Testing the Phonological Permeability Hypothesis: L phonological effects on L versus L systems. International Journal of Bilingualism, (), –. Cabrelli Amaro, J., Luque, A., & Finestrat-Martínez, I. (). Phonotactic restructuring in L Brazilian Portuguese. Journal of Phonetics, , –. Cadierno, T., & Ruiz, L. (). Motion events in Spanish L acquisition. Annual Review of Cognitive Linguistics, , –. Cahana-Amitay, D., & Albert, M. L. (). Brain and language: Evidence for neural multifunctionality. Behavioral Neurology, , .
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Calderone, B., Celata, C., & Laks, B. (). Theoretical and empirical approaches to phonotactics and morphonotactics. Language Sciences, (B), –. Calvo, A., & Bialystok, E. (). Independent effects of bilingualism and socioeconomic status on language ability and executive functioning. Cognition, , –. Calvo, N., García, A.-M., Manoiloff, L., & Ibáñez, A. (). Bilingualism and cognitive reserve: A critical overview and a plea for methodological innovations. Frontiers in Aging Neuroscience, (), . Cameron, L., & Larsen-Freeman, D. (). Complex systems and applied linguistics. International Journal of Applied Linguistics, (), –. Campbell, T., Dollaghan, C., Needleman, H., & Janosky, J. (). Reducing bias in language assessment: Processing-dependent measures. Journal of Speech, Language, and Hearing Research, , –. Caplan, B. (). The Case against Education. Princeton, NJ: Princeton University Press. Caramazza, A., & Zurif, E. B. (Eds.) (a). Language Acquisition and Language Breakdown: Parallels and Divergencies. Baltimore: John Hopkins University Press. Caramazza, A., & Zurif, E. B. (b). Comprehension of complex sentences in children and aphasics: A test of the regression hypothesis. In A. Caramazza & E. B. Zurif (Eds.), Language Acquisition and Language Breakdown: Parallels and Divergencies (pp. –). Baltimore, MD: Johns Hopkins University. Caramazza, A., & Yeni-Komshian, G. H. (). Voice onset time in two French dialects. Journal of Phonetics, (), –. Caramazza, A., Yeni-Komshian, G. H., Zurif, E. B., & Carbone, E. (). The acquisition of a new phonological contrast: The case of stop consonants in French-English bilinguals. The Journal of the Acoustical Society of America, (), –. doi: ./. Cardinaletti, A. (). La traduzione: Un caso di attrito linguistico [A case of linguistic attrition]. In A. Cardinaletti & G. Garzone (Eds.), L’Italiano delle traduzioni [Italian in translation] (pp. –). Milano: Franco Angeli. Carey, S. (). The child as a word learner. In M. Halle, J. Bresnan, & G. Miller (Eds.), Linguistic Theory and Psychological Reality (pp. –). Cambridge, MA: MIT Press. Carlson, M. T., Goldrick, M., Blasingame, M., & Fink, A. (). Navigating conflicting phonotactic constraints in bilingual speech perception. Bilingualism: Language and Cognition, (), –. Carlson, M., Halberstadt, L., & Dussias, P. E. (in prep.). Triggering changes in L parsing preferences with the help of experience. Carminati, M. N. (). The processing of Italian subject pronouns. PhD thesis, University of Massachusetts at Amherst. Carreiras, M., & Clifton, C. (). Another word on parsing relative clauses: Eyetracking evidence from Spanish and English. Memory and Cognition, , –. Carroll, S. (). Input and Evidence: The Raw Material of Second Language Acquisition. Amsterdam: John Benjamins. Carroll, S. (). Exposure and input in bilingual development. Bilingualism: Language and Cognition, (), –. doi: ./S Carter, P. M. (). Quantifying rhythmic differences between Spanish, English, and Hispanic English. In R. S. Gess & E. J. Rubin (Eds.), Theoretical and Experimental Approaches to Romance Linguistics (pp. –). Amsterdam: John Benjamins. Caruso, M. (). Attrition in the verb system of Italian in Australia. Australian Review of Applied Linguistics, , –. Cattani, A., Abbot-Smith, K., Farag, R., et al. (). How much exposure to English is necessary for a bilingual toddler to perform like a monolingual peer in language tests? International Journal of Language & Communication Disorders, (), –. doi: ./-. Celata, C., & Cancila, J. (). Phonological attrition and the perception of geminate consonants in the Lucchese community of San Francisco (CA). International Journal of Bilingualism, (), –. Cenoz, J. (). The intercultural style hypothesis: L and L requesting behavior. In V. Cook (Ed.), Effects of the Second Language on the First (pp. –). Clevedon: Multilingual Matters.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Chamorro, G., Sorace, A., & Sturt, P. (). What is the source of L attrition? The effect of recent L re-exposure on Spanish speakers under L attrition. Bilingualism: Language and Cognition, (), –. Chamorro, G., Sturt, P., & Sorace, A. (). Selectivity in L attrition: Differential object marking in Spanish near-native speakers of English. Journal of Psycholinguistic Research, (), –. Chang, C. B. (). First language phonetic drift during second language acquisition. PhD thesis, University of California, Berkeley. Chang, C. B. (). Systemic drift of L vowels in novice L learners. In W.-S. Lee & E. Zee (Eds.), Proceedings of the th International Congress of Phonetic Sciences (pp. –). Hong Kong: City University of Hong Kong. Chang, C. B. (). Rapid and multifaceted effects of second-language learning on first-language speech production. Journal of Phonetics, (), –. Chang, C. B. (). A novelty effect in phonetic drift of the native language. Journal of Phonetics, (), –. Chang, C. B. (). Accounting for multicompetence and restructuring in the study of speech. Journal of the Acoustical Society of America, (), –. Chang, C. B. (). Language change and linguistic inquiry in a world of multicompetence: Sustained phonetic drift and its implications for behavioral linguistic research. Journal of Phonetics , –. Chang, C. B., & Yao, Y. (). Toward an understanding of heritage prosody: Acoustic and perceptual properties of tone produced by heritage, native, and second language speakers of Mandarin. Heritage Language Journal, (), –. Chang, C. B., Yao, Y., Haynes, E. F., & Rhodes, R. (). Production of phonetic and phonological contrast by heritage speakers of Mandarin. Journal of the Acoustical Society of America, (), –. Chee, M. W., Soon, C. S., Lee, H. L., & Pallier, C. (). Left insula activation: A marker for language attainment in bilinguals. Proceedings of the National Academy of Sciences of the United States of America, (), –. doi: ./pnas. Chemero, A. (). Radical embodied cognitive science. Review of General Psychology, (), –. Chen, S. X. (). Toward a social psychology of bilingualism and biculturalism. Asian Journal of Social Psychology, , –. Chen, S. X., Benet-Martínez, V., Wu, W. C. H., Lam, B. C. P., & Bond, M. H. (). The role of dialectical self and bicultural identity integration in psychological adjustment. Journal of Personality, , –. Chen Pichler, D., & Koulidobrova, E. (). Acquisition of sign language as a second language (L). In M. Marschark & P. E. Spencer (Eds.), The Oxford Handbook of Deaf Studies in Language: Research, Policy, and Practice (pp. –). Oxford: Oxford University Press. Cherciov, M. (). Between attrition and acquisition: The case of Romanian in immigrant contexts. PhD thesis, University of Toronto. Cherciov, M. (). Investigating the impact of attitude on first language attrition and second language acquisition from a Dynamic Systems Theory perspective. International Journal of Bilingualism, (), –. Chertkow, H., Whitehead, V., Phillips, N., Wolfson, C., Atherton, J., & Bergman, H. (). Multilingualism (but not always bilingualism) delays the onset of Alzheimer disease: Evidence from a bilingual community. Alzheimer Disease & Associated Disorders, (), –. Chevrot, J.-P., & Foulkes, P. (). Introduction: Language acquisition and sociolinguistic variation. Linguistics, (), –. Chiat, S. (). Nonword Repetition. In S. Armon-Lotem, J. de Jong, & N. Meir (Eds.) Methods for Assessing Multilingual Children: Disentangling Bilingualism from Language Impairment (pp. –). Bristol: Multilingual Matters. Cho, J. H. (). Remapping nominal features in the second language. PhD thesis, University of Iowa. Cho, M.-H., & Lee, S. (). The impact of different L and L learning experience in the acquisition of L phonological processes. Language Sciences, , –.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Cho, T., & Ladefoged, P. (). Variation and universals in VOT: Evidence from languages. Journal of Phonetics, , –. Cho, T., Jun, S.-A., & Ladefoged, P. (). Acoustic and aerodynamic correlates of Korean stops and fricatives. Journal of Phonetics, , –. Choi, J., Broersma, M., & Cutler, A. (). Early phonology revealed by international adoptees’ birth language retention. Proceedings of the National Academy of Sciences, (), –. doi: ./pnas. Choi, J., Cutler, A., & Broersma, M. (). Early development of abstract language knowledge: Evidence from perception-production transfer of birth-language memory. Royal Society Open Science, , . doi:./rsos. Chomsky, C. (). The Acquisition of Syntax in Children from to . Cambridge, MA: MIT Press. Chomsky, N. (). Lectures in Government and Binding. Dordrecht: Foris. Chomsky, N. (). Knowledge of Language. New York: Praeger. Chomsky, N. (). The Minimalist Program. Cambridge, MA: MIT Press. Chomsky, N. (). Minimalist inquiries: The framework. In R. Martin, D. Michaels, & J. Uriagereka (Eds.), Step by Step: Essays on Minimalist Syntax in Honor of Howard Lasnik (pp. –). Cambridge, MA: MIT Press. Chomsky, N. (). Derivation by phase. In M. Kenstowicz (Ed.), Ken Hale: A Life in Language (pp. –). Cambridge, MA: MIT Press. Chomsky, N. (). Beyond explanatory adequacy. In A. Belletti (Ed.), Structures and Beyond: The Cartography of Syntactic Structures, Vol. III (pp. –). Oxford: Oxford University Press. Chomsky, N., & Miller, G. (). Introduction to the formal analysis of natural languages. In R. Duncan Luce, R. Bush, & E. Galanter (Eds.). The Handbook of Mathematical Psychology, Vol. (pp. –). New York: Wiley. Chondrogianni, V., & Marinis, T. (). Differential effects of internal and external factors on the development of vocabulary, tense morphology and morpho-syntax in successive bilingual children. Linguistic Approaches to Bilingualism, (), –. doi: ./lab...cho Citrin, J., Lerman, A., Murakami, M., & Pearson, K. (). Testing Huntington: Is Hispanic immigration a threat to American identity? Perspectives on Politics, (), –. Clahsen, H., & Felser, C. (). Grammatical processing in language learners. Applied Psycholinguistics, (), –. Clark, E. (). On the acquisition of the meaning of before and after. Journal of Verbal Learning and Verbal Behavior, (), –. https://doi.org/./S-()- Clark, J. L. D. (). Measurement considerations in language attrition research. In R. D. Lambert, & B. Freed (Eds.), The Loss of Language Skills (pp. –.). Rowley, MA: Newbury House. Clifton, C., Jr., & Staub, A. (). Syntactic influences on eye movements in reading. In S. P. Liversedge, I. D. Gilchrist, & S. Everling (Eds.), The Oxford Handbook of Eye Movements (pp. –). Oxford: Oxford University Press. Clyne, M. (). Community Languages: The Australian Experience. Cambridge: Cambridge University Press. Clyne, M. (). What can we learn from Sprachinseln? Some observations on ‘Australian German’. In N. Berend & K. Mattheier (Eds.), Sprachinselforschung. Eine Gedenkschrift für Hugo Jedig (pp. –). Bern: Peter Lang. Clyne, M. (). Dynamics of Language Contact: English and Immigrant Languages. Cambridge: Cambridge University Press. Coderre, E. L. (). The neuroscience of bilingualism: Cross-linguistic influences and cognitive effects. In J. E. Warnick & D. Landis (Eds.), Neuroscience in Intercultural Contexts, International and Cultural Psychology (pp. –). New York: Springer Science+Business Media. Cohen, A. D. (). Forgetting a second language. Language Learning, , –. Cohen, A. D. (). Forgetting foreign language vocabulary. In B. Weltens, K. de Bot, & T. J. M. van Els (Eds.), Language Attrition in Progress (pp. –). Dordrecht: Foris.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Cohen, A. D. (). Attrition in the productive lexicon of two Portuguese third language speakers. Studies in Second Language Acquisition, (), –. doi: ./S Cohen, M. X. (). Analyzing Neural Time Series Data: Theory and Practice. Cambridge, MA: MIT Press. Cole, R. D. (). The effect of a summer vacation on students’ knowledge of French. Modern Language Journal, , –. Colgin, L. L., Denninger, T., Fyhn, M., et al. (). Frequency of gamma oscillations routes flow of information in the hippocampus. Nature, (), –. doi: ./nature Contreras-García, L. (). The syntax-semantics interface in generative, parallel and functional frameworks. Cambridge Occasional Papers in Linguistics, (), –. Cook, V. (). Evidence for multicompetence. Language Learning, (), –. ./j...tb.x Cook, V. (Ed.) (a). Effects of the Second Language on the First. Clevedon: Multilingual Matters. Cook, V. (b). Introduction: The changing L in the L user’s mind. In V. Cook (Ed.), Effects of the Second Language on the First (pp. –). Clevedon: Multilingual Matters. Cook, V. (). Linguistic contributions to bilingualism. In J. Altaribba & R. Heredia (Eds.), An Introduction to Bilingualism: Principles and Practice (pp. –). London: Routledge. Cook, V., E. Iarossi, N. Stellakis, & Y. Tokumaru. (). Effects of the L on the syntactic processing of the L. In V. Cook (Ed.), Effects of the Second Language on the First (pp. –). Clevedon: Multilingual Matters. Cook, V., Bassetti, B., Kasai, C., Sasaki, M., & Takahashi, J. (). Do bilinguals have different concepts? The case of shape and material in Japanese L users of English. International Journal of Bilingualism, , –. Coppieters, R. (). Competence differences between native and near-native speakers. Language, , –. Costa, A., & Santesteban, M. (). Lexical access in bilingual speech production: Evidence from language switching in highly proficient bilinguals and L learners. Journal of Memory and Language, (), –. Costa, A., & Sebastián-Gallés, N. (). How does the bilingual experience sculpt the brain? Nature Reviews Neuroscience, (), –. Costa, A., Sebastián-Gallés, N., Miozzo, M., & Caramazza, A. (). The gender congruity effect: Evidence from Spanish and Catalan. Language and Cognitive Processes, (), –. Costa, A., Caramazza, A., & Sebastián-Gallés, N. (). The cognate facilitation effect: Implications for models of lexical access. Journal of Experimental Psychology: Learning, Memory, and Cognition, (), –. Costa, A., Colomé, À., & Caramazza, A. (). Lexical access in speech production: The bilingual case. Psicológica, , –. Costa, A., Hernández, M., Costa-Faidella, J., & Sebastián-Gallés, N. (). On the bilingual advantage in conflict processing: Now you see it, now you don’t. Cognition, (), –. Costa, A., Calabria, M., Marne, P., et al. (). On the parallel deterioration of lexico-semantic processes in the bilinguals’ two languages: Evidence from Alzheimer’s disease. Neuropsychologia, (), –. Coupland, N. (). Style: Language Variation and Identity. Cambridge University Press. Cowan, N. (). Activation, attention, and short-term memory. Memory and Cognition, , –. Cowan, N. (). Working Memory Capacity. New York: Psychology Press. Crago, M., & Paradis, J. (). Two of a kind? The importance of commonalities and variation across languages and learners. In Y. Levy, & J. Schaeffer (Eds.), Language Competence across Populations: Toward a Definition of Specific Language Impairment (pp. –). Mahwah, NJ: Lawrence Erlbaum Associates. Craik, F. I. M., Bialystok, E., & Freedman, M. (). Delaying the onset of Alzheimer disease: Bilingualism as a form of cognitive reserve. Neurology, (), –.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Crain, S., & Lillo-Martin, D. (). An Introduction to Linguistic Theory and Language Acquisition. Oxford: Blackwell. Crain, S., & McKee, C. (). The acquisition of structural restrictions on anaphora. Proceedings of NELS, Vol. (pp. –). Amherst: University of Massachusetts, GLSA. Crane, P. K., Gruhl, J. C., Erosheva, E. A., et al. (). Use of spoken and written Japanese did not protect Japanese-American men from cognitive decline in late life. The Journals of Gerontology Series B: Psychological Sciences and Social Sciences, (), –. Cumming, G. (). The new statistics: A how-to guide. Australian Psychologist, , –. Cummins, J. (). Language, Power, and Pedagogy: Bilingual Children in the Crossfire. Clevedon: Multilingual Matters. Cutler, A., Mehler, J., Norris, D., & Segui, J. (). The monolingual nature of speech segmentation by bilinguals. Cognitive Psychology, (), –. Cuza, A. (). On the L attrition of the Spanish present tense. Hispania, (), –. Cuza, A., & Pérez-Tattam, R. (). Grammatical gender selection and phrasal word order in child heritage Spanish: A feature re-assembly approach. Bilingualism: Language and Cognition, (), –. Dąbrowska, E. (). The LAD goes to school: A cautionary tale for nativists. Linguistics, (), –. Dąbrowska, E. (). Different speakers, different grammars: Individual differences in native language attainment. Linguistic Approaches to Bilingualism, (), –. Daignan-Fornier, B., & Sagot, I. (). Proliferation/quiescence: The controversial ‘aller-retour’. Cell Division, , . doi: ./--- Daller, H. (). Migration und Mehrsprachigkeit. Der Sprachstand türkischer Rückkehrer aus Deutschland [Migration and Multilingualism: The linguistic situation of Turkish returnees from Germany]. Frankfurt: Peter Lang. Daller, H., & Yıldız, C. (). Die türkischen Sprachfähigkeiten von Rückkehrern aus Deutschland [The Language Proficiency of Turkish Returnees from Germany]. In J. Treffers-Daller & H. Daller (Eds.), Zwischen den Sprachen. Sprachgebrauch, Sprachmischung und Sprachfähigkeiten türkischer Rückkehrer aus Deutschland. [Between the languages. Language use, language mixing and language proficiency of Turkish returnees from Germany] (pp. –), Occasional Papers in Linguistics, Vol. . Bogaziçi University, Turkey: The Language Center. Daller, M. H., & Treffers-Daller, J. (). Moving between languages: Turkish returnees from Germany. In B. Menzel & C. Engel (Eds.), Rückkehr in die Fremde? Etnische Remigration russlanddeutscher Spätaussiedler [Return to a foreign country? Ethnic remigration of RussianGerman repatriates.] (pp. –). Berlin: Frank & Timme GmbH. Dalston, R. M. (). Acoustic characteristics of English /w,r,l/ spoken correctly by young children and adults. Acoustical Society of America Journal, , –. doi: ./. Darcy, I., Peperkamp, S., & Dupoux, E. (). Bilinguals play by the rules: Perceptual compensation for assimilation in late L-learners. In J. Cole & J. I. Hualde (Eds.), Laboratory Phonology, Vol. (pp. –). Berlin: Mouton de Gruyter. Darcy, I., Dekydtspotter, L., Sprouse, R. A., et al. (). Direct mapping of acoustics to phonology: On the lexical encoding of front rounded vowels in L English-L French acquisition. Second Language Research, , –. Datta, H. (). Brain bases for first language lexical attrition in Bengali-English speakers. PhD thesis, City University of New York. de Bot, K. (). Language use as an interface between sociolinguistic and psycholinguistic processes in language attrition and language shift. In J. Klatter-Folmer & P. V. Avermaet (Eds.), Theories on Maintenance and Loss of Minority Languages: Towards an More Integrated Explanatory Framework (pp. –). Münster: Waxmann. de Bot, K. (a). The multilingual lexicon: Modeling selection and control. International Journal of Multilingualism, (), –. de Bot, K. (b). Introduction to the Special issue on language attrition. International Journal of Bilingualism, (), –.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
de Bot, K. (). Dynamic systems theory, lifespan development and language attrition. In B. Köpke, M. S. Schmid, M. Keijzer, & S. Dostert (Eds.), Language Attrition: Theoretical Perspectives (pp. –). Amsterdam: John Benjamins. de Bot, K. (). Timescales in second language development. Dutch Journal of Applied Linguistics, (), –. de Bot, K. (). Not my time of day. Circadian rhythms and second language development. International Journal of Bilingualism, (), –. de Bot, K. (). Rates of change. Timescales in second language development. In Z. Dörnyei, P. D. MacIntyre, & A. Henry (Eds.), Motivational Dynamics in Language Learning (pp. –). Bristol: Multilingual Matters. de Bot, K. (). Complexity Theory and Dynamic Systems Theory: Same or different? In Ortega, L., & Han, Z. (Eds.), Complexity Theory and Language Development: In Celebration of Diane LarsenFreeman (pp. –). Amsterdam: John Benjamins. de Bot, K., & Clyne, M. (). Language reversion revisited. Studies in Second Language Acquisition, (), –. de Bot, K., & Clyne, M. (). A year longitudinal study of language attrition in Dutch immigrants in Australia. Journal of Multilingual and Multicultural Development, (), –. de Bot, K., & Larsen-Freeman, D. (). Researching second language development from a Dynamic Systems Theory perspective. In M. Verspoor, K. de Bot, & W. Lowie (Eds.), A Dynamic Approach to Second-language Development. Methods and Techniques (pp. –). Amsterdam: Benjamins. de Bot, K., & Lintsen, T. (). Foreign-language proficiency in the elderly. In B. Weltens, K. de Bot, & T. van Els (Eds.), Language Attrition in Progress (pp. –). Dordrecht: Foris Publications. de Bot, K., & Lowie, W. M. (). On the stability of representations in the multilingual lexicon. In L. Sicola & M. Putz (Eds.), Cognitive Processing in Second Language Acquisition: Inside the Learner’s Mind (pp. –). Amsterdam: John Benjamins. de Bot, K., & Stoessel, S. (). In search of yesterday’s words: Reactivating a long forgotten language. Applied Linguistics, (), –. de Bot, K., & Weltens, B. (). Taalverlies: Beschrijven versus verklaren [Language Loss: Description vs. explanation.]. Handelingen van het e Nederlands Filologencongres, –. de Bot, K., & Weltens, B. (). Recapitulation, regression, and language loss. In H. W. Seliger & R. M. Vago (Eds.), First Language Attrition (pp. –). Cambridge: Cambridge University Press. de Bot, K., & Weltens, B. (). Foreign language attrition. Annual Review of Applied Linguistics, , –. de Bot, K., Gommans, P., & Rossing, C. (). L loss in an L environment: Dutch immigrants in France. In H. W. Seliger & R. M. Vago (Eds.), First Language Attrition (pp. –). New York: Cambridge University Press. de Bot, K., Martens, V., & Stoessel, S. (). The ‘Savings’ approach to testing vocabulary. The International Journal of Bilingualism, (), –. de Bot, K., Lowie, W., & Verspoor, M. (a). A Dynamic Systems Theory approach to second language acquisition. Bilingualism: Language and Cognition, (), –. de Bot, K., Lowie, W., & Verspoor, M. (b). A dynamic view as a complementary perspective. Bilingualism: Language and Cognition, (), –. de Bot, K., Lowie, W. M., Thorne, S., & Verspoor, M. H. (). Dynamic systems theory as a comprehensive theory of second language development. In P. García Mayo, J. Gutierrez Mangado, & M. Martínez Adrián (Eds.), Contemporary Perspectives on Second Language Acquisition (pp. –). Amsterdam: John Benjamins. de Groot, A., & Keijzer, R. (). What hard to learn is easy to forget: The roles of word concreteness, cognate status, and word frequency in foreign language learning and forgetting. Language Learning, , –. de Groot, A. M. B., & Van Hell, J. G. (). The learning of foreign language vocabulary. In J. Kroll & A. M. B. de Groot (Eds.), Handbook of Bilingualism: Psychological Approaches (pp. –). Oxford: Oxford University Press.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
De Houwer, A. (). Bilingual language acquisition. In P. Fletcher & B. MacWhinney (Eds.), The Handbook of Child Language (pp. –). Oxford: Blackwell. De Houwer, A. (). Parental language input patterns and children’s bilingual use. Applied Psycholinguistics, (), –. doi: ./S De Houwer, A. (). How different are monolingual and bilingual acquisition? Ilha do Desterro, , –. De Houwer, A. (). Bilingual First Language Acquisition. Clevedon: Multilingual Matters. De Houwer, A. (). Language input environments and language development in bilingual acquisition. Applied Linguistics Review, , –. doi: ./. De Houwer, A. (). The absolute frequency of maternal input to bilingual and monolingual children: A first comparison. In T. Grüter & J. Paradis (Eds.), Input and Experience in Bilingual Development (pp. –). Amsterdam: John Benjamins. de Leeuw, E. (). When your native language sounds foreign: A phonetic investigation into first language attrition. PhD thesis, Queen Margaret University. de Leeuw, E. (). Reassessing maturational constraints through evidence of L attrition in the domain of phonetics. In E. Thomas & I. Mennen (Eds.), Advances in the Study of Bilingualism: (pp. –). Bristol: Multilingual Matters. de Leeuw, E. (). How phonetics and phonology inform L attrition (narrowly defined) research. Linguistic Approaches to Bilingualism, (), –. https://doi.org/./lab..lee de Leeuw, E. (). Native speech plasticity in the German-English late bilingual Stefanie Graf: A longitudinal study over four decades. Journal of Phonetics, , –. de Leeuw, E., & Bogulski, C. A. (). Frequent L use enhances executive control in sequential bilinguals. Bilingualism: Language and Cognition. de Leeuw, E., & Celata, C. (). Plasticity of native language phonetic and phonological domains in the context of bilingualism. Journal of Phonetics . de Leeuw, E., Schmid, M. S., & Mennen, I. (). The effects of contact on native language pronunciation in an L migrant setting. Bilingualism: Language and Cognition, (), –. de Leeuw, E., Mennen, I., & Scobbie, J. M. (). Singing a different tune in your native language: First language attrition of prosody. International Journal of Bilingualism, (), –. de Leeuw, E., Mennen, I., & Scobbie, J. M. (). Dynamic systems, maturational constraints and L phonetic attrition. International Journal of Bilingualism, (), –. doi: ./ de Leeuw, E., Opitz, C., & Lubinska, D. (). Dynamics of first language attrition across the lifespan. International Journal of Bilingualism, (), . de Leeuw, E., Tusha, A., & Schmid, M. S. (). Individual phonological attrition in AlbanianEnglish late bilinguals. Bilingualism: Language and Cognition. doi: ./S de Leeuw, E., Tusha, A., Zhao, H., Helke, K., & Greenfield, A. (). A case study of extreme L attrition in the domain of phonetics. In C. Wright, T. Piske, & M. Young-Scholten (Eds.), Mind Matters in SLA (pp. –). Bristol: Multilingual Matters. de Prada Pérez, A. (). Subject expression in Minorcan Spanish: Consequences of contact with Catalan. PhD thesis, Pennsylvania State University. de Prada Pérez, A. (). First person singular subject pronoun expression in Spanish in contact with Catalan. In A. M. Carvalho, R. Orozco, & N. Shin (Eds.), Subject Pronoun Expression in Spanish: A Cross-dialectal Perspective (pp. –). Washington, DC: Georgetown University Press. de Prada Pérez, A., & Pascual y Cabo, D. (). Invariable gusta in the Spanish of Heritage Speakers in the US. In J. Herschensohn & D. Tanner (Eds.), Proceedings of the th Generative Approaches to Second Language Acquisition Conference (pp. –). Somerville, MA: Cascadilla Proceedings Project. De Santi, S., Obler, L. K., Sabo-Abrahamson, H., & Goldberger, J. (). Discourse abilities and deficits in multilingual dementia. In Y. Joanette & H. H. Brownell (Eds.), Discourse Ability and Brain Damage: Theoretical and Empirical Perspectives (pp. –). New York: Springer Verlag. de Saussure, F. (/). Course in General Linguistics. New York: McGraw-Hill. De Vreese, L. P., Motta, M., & Toschi, A. (). Compulsive and paradoxical translation behaviour in a case of presenile dementia of the Alzheimer type. Journal of Neurolinguistics, , –.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
DeKeyser, R. (). Age effects in second language learning: Stepping stones toward better understanding. Language Learning, , –. doi: ./j.-...x Delcenserie, A. (). Language, cognitive and academic abilities of school-age internationallyadopted children (pp. –). In F. Genesee & A. Delcenserie (Eds.), Starting Over: The Language Development of Internationally-adopted Children. Amsterdam: John Benjamins. Delcenserie, A., & Genesee, F. (a). Language and memory abilities of internationally adopted children from China: Evidence for early age effects. Journal of Child Language, (), –. doi: ./SX Delcenserie, A., & Genesee, F. (b). The acquisition of accusative object clitics by IA children from China: Do they still experience difficulties during school-age? Journal of Child Language, (), –. doi: ./S Delcenserie, A., Genesee, F., & Gauthier, K. (). Language abilities of internationally adopted children from China during the early school years: Evidence for early age effects? Applied Psycholinguistics, (), –. doi: ./S Denes, P. B., & Pinson, E. N. (). The Speech Chain: The Physics and Biology of Spoken Language. Oxford: W.H. Freeman and Company. Deuchar, M., & Clark, A. (). Early bilingual acquisition of the voicing contrast in English and Spanish. Journal of Phonetics, (), –. doi: ./jpho.. Dewaele, J.-M. (). Perceived language dominance and language preference for emotional speech: The implications for attrition research. In M. S. Schmid, M. Keijzer, & L. Weilemar (Eds.), First Language Attrition: Interdisciplinary Perspectives on Methodological Issues (pp. –). Amsterdam: John Benjamins. Dewaele, J.-M., & A. Pavlenko. (). Productivity and lexical diversity in native and non-native speech: A study of cross-cultural effects. In V. Cook (Ed.), Effects of the Second Language on the First (pp. –). Bristol: Multilingual Matters. Di Domenico, A., & Di Matteo, R. (). Processing Italian relative clauses: Working memory span and word order effects on RTs. The Journal of General Psychology: Experimental, Psychological, and Comparative Psychology, (), –. Di Gregorio, D. (). The phenomenon of international adoption with a focus on second language acquisition. A case study of internationally adopted children and adolescents from Russia. PhD dissertation, Indiana University of Pennsylvania. Diaz, M. T., Johnson, M. A., Burke, D. M., & Madden, D. J. (). Age-related differences in the neural bases of phonological and semantic processes. Journal of Cognitive Neuroscience, (), –. Dick, M. B., Nielson, K. A., Beth, R. E., Shankle, W. R., & Cotman, C. W. (). Acquisition and longterm retention of a fine motor skill in Alzheimer’s disease. Brain and Cognition, , –. Diessel, H. (). The Acquisition of Complex Sentences. Cambridge: Cambridge University Press. Dijkstra, T. (). Bilingual visual word recognition and lexical access. In J. F. Kroll & A. M. B. de Groot (Eds.), Handbook of Bilingualism: Psycholinguistic Approaches (pp. –). Oxford: Oxford University Press. Dijkstra, T., & Van Heuven, W. J. B. (). The BIA model and bilingual word recognition. In J. Grainger & A. M. Jacobs (Eds.), Localist Connectionist Approaches to Human Cognition (pp. –). Mahwah, NJ: Erlbaum. Dijkstra, T., & Van Heuven, W. J. B. (a). The architecture of the bilingual word recognition system: From identification to decision. Bilingualism: Language and Cognition, (), –. Dijkstra, T., & Van Heuven, W. J. B. (b). Modeling bilingual word recognition: Past, present and future. Bilingualism: Language and Cognition, (), –. Dijkstra, T., Doe Bruijn, E., Schriefers, H., & Ten Brinke, S. (). More on interlingual homograph recognition: Language intermixing versus explicitness of instruction. Bilingualism: Language and Cognition, (), –. Dik, S. C. (). The Theory of Functional Grammar. Part : Complex and Derived Constructions. Berlin: de Gruyter.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Dixon, L. Q., Zhao, J., Quiroz, B. G., & Shin, J. (). Home and community factors influencing bilingual children’s ethnic language vocabulary development. International Journal of Bilingualism, , –. doi: ./ Dmitrieva, O. (). Transferring perceptual cue-weighting from second language into first language: Cues to voicing in Russian speakers of English. Journal of Phonetics, , –. Dmitrieva, O., Jongman, A., & Sereno, J. (). Phonological neutralization by native and non-native speakers: The case of Russian final devoicing. Journal of Phonetics, (), –. Doğruöz, S., & Backus, A. (). Innovative constructions in Dutch Turkish: An assessment of on-going contact-induced change. Bilingualism: Language & Cognition, , –. Doherty, M., & Perner, J. (). Metalinguistic awareness and theory of mind: Just two words for the same thing? Cognitive Development, (), –. https://doi.org/./S-()- Domínguez, L. (). Understanding Interfaces: Second Language Acquisition and First Language Attrition of Spanish Subject Realization and Word Order Variation. Amsterdam: John Benjamins. Domínguez, L., Arche, M. J., & Myles, M. (). Testing the predictions of the feature-assembly hypothesis: Evidence from the L acquisition of Spanish aspect morphology. In N. Danis, K. Mesh, & H. Sung (Eds.), Proceedings of the th Annual Boston University Conference on Language Development (BUCLD) (pp. –). Somerville, MA: Cascadilla Press. Döpke, S. (). Competing language structures: The acquisition of verb placement by bilingual German-English children. Journal of Child Language, , –. Döpke, S. (Ed.) (). Cross-linguistic Structures in Simultaneous Bilingualism. Amsterdam: John Benjamins. Dorian, N. C. (). The problem of the semi-speaker in language death. Linguistics, (), –. Dorian, N. C. (). Social factors in second language retention/language loss and maintenance on language contact situations. In R. D. Lambert, & B. Freed (Eds.), The Loss of Language Skills (pp. –). Rowley, MA: Newbury House. Dörnyei, Z. (). Motivation in second and foreign language learning. Language Teaching, , –. Dörnyei, Z. (). New themes and approaches in second language motivation research. Applied Review of Applied Linguistics, , –. Dörnyei, Z., MacIntyre, P. D., & Henry, A. (Eds.) (). Motivational Dynamics in Language Learning. Second Language Acquisition. Bristol: Multilingual Matters. Dörnyei, Z., & Skehan, P. (). Individual differences in second language learning. In C. J. Doughty & M. H. Long (Eds.), The Handbook of Second Language Acquisition (pp. –). Malden, MA: Blackwell Publishing. Dörnyei, Z., & Ottó, I. (). Motivation in action: A process model of L motivation. Working Papers in Applied Linguistics, Vol. (pp. –). London: Thames Valley University. Dostert, S. (). Multilingualism, L attrition and the concept of ‘native speaker’. PhD thesis, Heinrich-Heine Universität Düsseldorf. Draganski, B., Gaser, C., Kempermann, G., et al. (). Temporal and spatial dynamics of brain structure changes during extensive learning. Journal of Neuroscience, , –. Dressler, W. (). On the predictiveness of Natural Morphology. Journal of Linguistics, , –. Dressler, W. (). The sociolinguistic and patholinguistic attrition of Breton phonology, morphology, and morphophonology. In H. Seliger & R. Vago (Eds.), First Language Attrition (pp. –). Cambridge: Cambridge University Press. Driemeyer, J., Boyke, J., Gaser, C., Büchel, C., & May, A. (). Changes in gray matter induced by learning—revisited. PLoS ONE, (): e. https://doi.org/./journal.pone. Driessen, G., Van der Slik, F., & de Bot, K. (). Home language and language proficiency: A largescale longitudinal study in Dutch primary schools. Journal of Multilingual and Multicultural Development, (), –. doi: ./ Dronkers, N., Koss, E., Friedland, R., & Wertz, R. (). ‘Differential’ language impairment and language mixing in a polyglot with probable Alzheimer’s disease. Paper presented at the International Neuropsychological Society Meeting, Veldhoven (NL), June.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Dubinina, I. (). How to ask for a favor: A pilot study in heritage Russian pragmatics. In A. Mustajoki, E. Protassova, & N. Vakhtin (Eds.), Instrumentarium of Linguistics: Sociolinguistic Approaches to Non-Standard Russian. Slavica Helsingiensia, Vol. (pp. –). Helsinki: University of Helsinki Press. Dubinina, I., & Polinsky, M. (). Russian in the USA. In M. Moser (Ed.), Slavic Languages in Migration (pp. –). Wien: University of Vienna. Duchowski, A. T. (). A breadth-first survey of eye-tracking applications. Behavior Research Methods, Instruments, and Computers, , –. Dudai, Y. (). The restless engram: Consolidations never end. Annual Review of Neuroscience, (), –. doi: ./annurev-neuro-- Dugas, L. G. (). Attrition of pronunciation accuracy among advanced American learners of French. PhD thesis, Indiana University. Dupoux, E., Kakehi, K., Hirose, Y., Pallier, C., & Mehler, J. (). Epenthetic vowels in Japanese: A perceptual illusion? Journal of Experimental Psychology: Human Perception and Performance, , –. Dupoux, E., Parlato, E., Frota, S., Hirose, Y., & Peperkamp, S. (). Where do illusory vowels come from? Journal of Memory and Language, , –. Dussias, P. E. (). Parsing a first language like a second: The erosion of L parsing strategies in Spanish-English bilinguals. International Journal of Bilingualism, (), –. Dussias, P. E., & Sagarra, N. (). Cross-linguistic differences and their impact on L sentence processing. Studies in Second Language Acquisition, , –. Dussias, P. E., & Sagarra, N. (). The effect of exposure on syntactic parsing in Spanish-English bilinguals. Bilingualism: Language and Cognition, (), –. doi: ./S Duursma, E., Romero-Contreras, S., Szuber, A., et al. (). The role of home literacy and language environment on bilinguals’ English and Spanish vocabulary development. Applied Psycholinguistics, (), –. doi: ..S Duyck, W., Van Assche, E., Drieghe, D., & Hartsuiker, R. J. (). Visual word recognition by bilinguals in a sentence context: Evidence for nonselective lexical access. Journal of Experimental Psychology: Learning, Memory, and Cognition, (), –. Ebbinghaus, H. (). Über das Gedächtnis [About memory]. Leipzig: Duncker. Ebbinghaus, H. (). Memory: A Contribution to Experimental Psychology (H. A. Ruger & C. E. Bussenius, translators; original work published ). New York: Dover. Ebert, K. D., & Kohnert, K. (). Sustained attention in children with primary language impairment: A meta-analysis. Journal of Speech, Language, and Hearing Research, , –. Ecke, P. (). Language attrition and theories of forgetting: A cross-disciplinary review. International Journal of Bilingualism, (), –. Ecke, P., & Hall, C. J. (). Tracking tip-of-the-tongue states in a multilingual speaker: Evidence of attrition or instability in lexical systems? International Journal of Bilingualism, (), –. doi: ./ Eckert, P., & Rickford, J. R. (). Style and Sociolinguistic Variation. Cambridge University Press. Eckman, F. R. (). Markedness and the contrastive analysis hypothesis. Language Learning, , –. Eckman, F. R. (). Second language phonology. In S. Gass & A. Mackey (Eds.), The Routledge Handbook of Second Language Acquisition (pp. –). New York: Routledge. Eddington, D. (). Spanish gender assignment in an analogical framework. Journal of Quantitative Lingusitics, , –. Egger, E., Hulk, A., & Tsimpli, I. (). Crosslinguistic influence in the discovery of gender: The case of Greek-Dutch bilingual children. Bilingualism: Language and Cognition, (), –. Eguren, L., Fernández-Soriano, O., & Mendikoetxea, A. (Eds.) (). Rethinking Parameters. Oxford: Oxford University Press.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Ehrlich, S. F., & Rayner, K. (). Contextual effects on word perception and eye movements during reading. Journal of Verbal Learning and Verbal Behavior, , –. Eigsti, I.-M., Weitzman, C., Schuh, J., de Marchena, A., & Casey, B. J. (). Language and cognitive outcomes in internationally adopted children. Development and Psychopathology, (), –. doi: ./S Ekman, S. L. (). Monolingual and bilingual communication between patients with dementia diseases and their caregivers. International Psychogeriatrics, (Suppl. ), –. Ekman, S. L., Wahlin, T. B., Norberg, A., & Winblad, B. (). Relationship between bilingual demented immigrants and bilingual/monolingual caregivers. The International Journal of Aging and Human Development, , –. El Aissati, A. (). Language Loss Among Native Speakers of Moroccan-Arabic in the Netherlands. Tilburg: Tilburg University Press. Ellert, M. (). Resolving ambiguous pronouns in a second language: A visual-world eye-tracking study with Dutch learners of German. International Review of Applied Linguistics in Language Teaching, (), –. Ellis, N. C. (). Constructions, chunking, and connectionism: The emergence of second language structure. In C. Doughty & M. Long (Eds.), Handbook of Second Language Acquisition (pp. –). Oxford: Blackwell. Ellis, N. C. (). Frequency effects in language processing. Studies in Second Language Acquisition, (), –. Ellis, N. C. (). The dynamics of second language emergence: Cycles of language use, language change, and language acquisition. The Modern Language Journal, (), –. Ellis, N. C., O’Donnell, J. J., & Römer, U. (). Usage-based language learning. In B. MacWhinney & W. O’Grady (Eds.), The Handbook of Language Emergence (pp. –). New York: Wiley. Ellis, R. (). The Study of Second Language Acquisition. Oxford: Oxford University Press. Ellis, R. (). The definition and measurement of L explicit knowledge. Language Learning, , –. doi: ./j.-...x Ellis, R., Johnson, K. E., & Shin, S. (). Birth order and the language experience of bilingual children. TESOL Quarterly, (), –. doi:./ Elmer, S., Hänggi, J., & Jäncke, L. (). Processing demands upon cognitive, linguistic, and articulatory functions promote grey matter plasticity in the adult multilingual brain: Insights from simultaneous interpreters. Cortex, , –. Embick, D., & Noyer, D. (). Distributed morphology and the syntax/morphology interface. In G. Ramchand & C. Reiss (Eds.), The Oxford Handbook of Linguistic Interfaces (pp. –). Oxford: Oxford University Press. Emmorey, K., Borinstein, H. B., Thompson, R., & Gollan, T. H. (). Bimodal bilingualism. Bilingualism: Language and Cognition, (), –. Enç, M. (). Topic switching and pronominal subjects in Turkish. In Dan Isaac Slobin and Karl Zimmer (Eds.). Studies in Turkish linguistics (pp. –). Amsterdam: John Benjamins. Engstler, C. (). Language retention and improvement after a study abroad experience. PhD thesis, Northwestern University. Engstler, C., & Goldrick, M. (). Lexical properties modulate phonological attrition: Evidence from German. Presentation at th Annual Mid-Continental Workshop on Phonology (MCWOP), Ohio State University in Columbus, OH, – October. Erdocia, I. (). Complex dynamic systems and second language acquisition. Paper presented at IRAAL, ‘Paradigm shifting in applied linguistics: New theories and new methods’, Dublin, November. Eslinger, P. J., & Damasio, A. R. (). Preserved motor learning in Alzheimer’s disease: Implications for anatomy and behaviour. Journal of Neuroscience, , –. Espinosa, K. E., & Massey, D. S. (). Determinants of English proficiency among Mexican migrants to the United States. International Migration Review, (), –. Esser, H. (). Migration, language, integration. AKI Research Review , Berlin.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
European Commission (). Europeans and their Languages. European Commission Report, Special Eurobarometer . Evans, B. G. (). Plasticity in Speech Perception and Production: A study of accent change in young adults. PhD thesis, University College London. Evans, B. G., & Iverson, P. (). Plasticity in vowel perception and production: A study of accent change in young adults. The Journal of the Acoustical Society of America, (), –. doi: ./. Evans, J. (). The emergence of language: A dynamical systems account. In E. Hoff & M. Shatz (Eds.), Blackwell Handbook of Language Development (pp. –). Malden, MA: Blackwell. Extra, G., & Yağmur, K. (Eds.) (). Urban Multilingualism in Europe: Immigrant Minority Languages at Home and School. Clevedon: Multilingual Matters. Extra, G., Spotti, M., & Van Avermaet, P. (). Testing regimes for newcomers. In G. Extra, M. Spotti & P. Van Avermaet (Eds.), Language Testing, Migration and Citizenship: Cross-national Perspectives on Integration Regimes (pp. –). London: Continuum. Fenyvesi, A. (). The case of American Hungarian case: Morphological change in McKeesport, PA. Acta Linguistica Hungarica, (–), –. Fernández, E. M. (). Bilingual Sentence Processing: Relative Clause Attachment in English and Spanish. Amsterdam: John Benjamins. Ferreira, F., & Patson, N. (). The good enough approach to language comprehension. Language and Linguistic Compass, , –. Ferreira, F., Bailey, K. G. D., & Ferraro, V. (). Good-enough representations in language comprehension. Current Directions in Psychological Science, , –. Ferreira, V. S., Bock, K., Wilson, M. P., & Cohen, N. J. (). Memory for syntax despite amnesia. Psychological Science, (), –. Fiebach, C. J., Friederici, A. D., Muller, K., Von Cramon, D. Y., & Hernandez, A. E. (). Distinct brain representations for early and late learned words. Neuroimage, (), –. Field, F. W. (). Linguistic Borrowing in Bilingual Contexts. Amsterdam: Benjamins. Filiaci, F., Sorace, A., & Carreiras, M. (). Anaphoric biases of null and overt subjects in Italian and Spanish: A cross-linguistic comparison. Language, Cognition and Neuroscience, , –. Fine, A. B., & Jaeger, T. F. (). Evidence for implicit learning in syntactic comprehension. Cognitive Science, (), –. Fishman, J. A. (). Language in Sociocultural Change, Vol. . Redwood City, CA: Stanford University Press. Fishman, J. A. (). What is Reversing Language Shift (RLS) and how can it succeed? Journal of Multilingual and Multicultural Development, (–), –. Fishman, J. A. (). -plus years of heritage language education in the United States. In J. K. Peyton, D. Ranard, & S. McGinnis (Eds.), Heritage Languages in America: Preserving a National Resource (pp. –). Washington, DC, and McHenry, IL: Delta Systems and Center for Applied Linguistics. Flege, J. E. (). The production of ‘new’ and ‘similar’ phones in a foreign language: Evidence for the effect of equivalence classification. Journal of Phonetics, , –. Flege, J. E. (). Second language speech learning: Theory, findings, and problems. In W. Strange (Ed.), Speech Perception and Linguistic Experience: Issues in Cross-Language Research (pp. –). Baltimore, MD: York Press. Flege, J. E. (). English vowel productions by Dutch talkers: More evidence for the ‘similar’ vs ‘new’ distinction. In A. James & J. Leather (Eds.), Second-language Speech: Structure and Process (pp. –). Berlin: Mouton de Gruyter. Flege, J. E. (). Age of learning and second language speech. In D. Birdsong (Ed.), Second Language Acquisition and the Critical Period Hypothesis (pp. –). Mahwah, NJ: Erlbaum. Flege, J. E. (). Interactions between the native and second-language phonetic systems. In P. Burmeister, T. Piske, & A. Rohde (Eds.), An Integrated View of Language Development: Papers in Honor of Henning Wode (pp. –). Trier: Wissenschaftlicher Verlag.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Flege, J. E. (). Assessing constraints on second-language segmental production and perception. In N. Schiller & A. Meyer (Eds.), Phonetics and Phonology in Language Comprehension and Production: Differences and Similarities, Vol. (pp. –). Berlin: Mouton de Gruyter. Flege, J. E. (). Language contact in bilingualism: Phonetic system interactions. In J. Cole & J. Hualde (Eds.), Laboratory Phonology, Vol. (pp. –). Berlin: Mouton de Gruyter. Flege, J. E., & Eefting, W. (a). Cross-language switching in stop consonant perception and production by Dutch speakers of English. Speech Communication, (), –. doi: ./ -()- Flege, J. E., & Eefting, W. (b). Production and perception of English stops by native Spanish speakers. Journal of Phonetics, , –. Flege, J. E., & Hillenbrand, J. (). Limits on phonetic accuracy in foreign language speech production. The Journal of the Acoustical Society of America, (), –. doi: ./. Flege, J. E., MacKay, I. R. A., & Piske, T. (). Assessing bilingual dominance. Applied Psycholinguistics, , –. Flege, J. E., & Munro, M. J. (). The word unit in L speech production and perception. Studies in Language Acquisition, , –. Flege, J. E., Munro, M. J., & MacKay, I. R. (). Factors affecting strength of perceived foreign accent in a second language. Journal of the Acoustical Society of America, ( Pt ), –. Flege, J. E., Yeni-Komshian, G. H., & Liu, S. (). Age constraints on second-language acquisition. Journal of Memory and Language, (), –. doi: ./jmla.. Flores, C. (). The effect of age on language attrition: Evidence from bilingual returnees. Bilingualism: Language and Cognition, (), –. doi: ./SX Flores, C. (). Differential effects of language attrition in the domain of verb placement and object expression. Bilingualism: Language and Cognition, (), –. https://doi.org/./ S Flores, C. (a). Losing a language in childhood: A longitudinal case study on language attrition. Journal of Child Language, , –. Flores, C. (b). Understanding heritage language acquisition: Some contributions from the research on heritage speakers of European Portuguese. Lingua, (B), –. doi: ./j. lingua... Flores, C., & Barbosa, P. (). When reduced input leads to delayed acquisition: A study on the acquisition of clitic placement by Portuguese heritage speakers. International Journal of Bilingualism, (), –. doi: ./ Flores, C., & Rato, A. (). Global accent in the Portuguese speech of heritage returnees. Journal of Heritage Language. Special Issue on the Phonetic and Phonological systems of heritage languages, (), –. Flores, C., & Rauber, A. (). Perception of German vowels by bilingual Portuguese-German returnees: A case of phonological attrition? In E. Rinke & T. Kupisch (Eds.), The Development of Grammar: Language Acquisition and Diachronic Change—Volume in Honor of Jürgen M. Meisel (pp. –). Amsterdam: John Benjamins. Flores, C., Santos, A. L., Marques, R., & Jesus, A. (). Age and input effects in the acquisition of mood in Heritage Portuguese. Journal of Child Language, (), –. doi: ./ S Flores, C., Kupisch, T., & Rinke, E. (). Linguistic foundations of heritage language development from the perspective of Romance languages in Germany. In P. Trifonas, & T. Aravossitas (Eds.), Handbook of Research and Practice in Heritage Language Education (pp. –). Cham: Springer. Fogliato, S., & Testa, M. C. (). Strumenti per l’italiano [Tools for Italian]. Torino: Loescher. Footnick, R. (). A hidden language: Recovery of a ‘lost’ language is triggered by hypnosis. In B. Köpke, M. S. Schmid, M. Keijzer, & S. Dostert (Eds.), Language Attrition: Theoretical Perspectives (pp. –). Amsterdam: John Benjamins.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Fowler, C. A., Sramko, V., Ostry, D. J., Rowland, S. A., & Hallé, P. (). Cross language phonetic influences on the speech of French–English bilinguals. Journal of Phonetics, (), –. doi: ./j.wocn... Fraundorf, S. H., & Jaeger, T. F. (). Readers generalize adaptation to newly-encountered dialectal structures to other unfamiliar structures. Journal of Memory and Language, , –. Fricke, M., Kroll, J. F., & Dussias, P. E. (). Phonetic variation in bilingual speech: A lens for studying the production–comprehension link. Journal of Memory and Language, , –. Friederici, A. D. (). Towards a neural basis of auditory sentence processing. Trends in Cognitive Sciences, (), –. Friederici, A. D. (). The cortical language circuit: From auditory perception to sentence comprehension. Trends in Cognitive Sciences, (), –. Friederici, A. D., Mecklinger, A., Spencer, K. M., Steinhauer, K., & Donchin, E. (). Syntactic parsing preferences and their on-line revisions: A spatio-temporal analysis of event-related brain potentials. Cognitive Brain Research, (), –. Friederici, A. D., Steinhauer, K., & Pfeifer, E. (). Brain signatures of artificial language acquisition: Evidence challenging the critical period hypothesis. Proceedings of the National Academy of Sciences USA, , –. Friedland, D. C. (). Language loss in bilingual speakers with Alzheimer’s disease. PhD thesis, Newcastle University. Friesen, D. C., Jared, D., & Haigh, C. A. (). Phonological processing dynamics in bilingual word naming. Canadian Journal of Experimental Psychology / Revue canadienne de psychologie expérimentale, (), –. Fromkin, V., Krashen, S., Curtiss, S., Rigler, D., & Rigler, M. (). The development of language in Genie: A case of language acquisition beyond the ‘critical period’. Brain and Language, , –. doi: ./-X()- Fromm, E. (). Age regression with unexpected reappearance of a repressed childhood language. International Journal of Clinical and Experimental Hypnosis, (), –. doi: ./ Fuhrmann, D., Knoll, L. J., & Blakemore, S.-J. (). Adolescence as a sensitive period of brain development. Trends in Cognitive Sciences, (), –. doi: ./j.tics... Fujita, M. (). Second language English attrition of Japanese bilingual children. PhD thesis, Temple University, Tokyo. Fuller, J. M. (). Morpheme types in a matrix language turnover: The introduction of system morphemes from English into Pennsylvania German. International Journal of Bilingualism, (), –. Gal, S. (). Language Shift: Social Determinants of Linguistic Change in Bilingual Austria. New York: Academic Press. Gandour, J., Wong, D., Hsieh, L., Weinzapfel, B., Van Lancker, D., & Hutchins, G. D. (). A crosslinguistic PET study of tone perception. Journal of Cognitive Neuroscience, (), –. doi: ./ Garcia-Penton, L., Perez Fernandez, A., Iturria-Medina, Y., Gillon-Dowens, M., & Carreiras, M. (). Anatomical connectivity changes in the bilingual brain. Neuroimage, , –. doi: ./j.neuroimage... Gardner, R. C. (). Social factors in language (and second language) retention. In R. D. Lambert & B. Freed (Eds.), The Loss of Language Skills (pp. –). Rowley, MA: Newbury House. Gardner, R. C. (). The Attitude Motivation Test Battery Manual. Ontario: University of Western Ontario. http://publish.uwo.ca/~gardner/ (accessed February ). Gardner, R. C. (). Integrative motivation and second language acquisition. In Z. Dornyei & R. Schmidt (Eds.), Motivation and Second Language Acquisition (pp. –). Honolulu, HI: University of Hawai’i, Second Language Teaching and Curriculum Centre.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Gardner, R. C. (). Motivation and attitudes in second language learning. In A. Barber (Ed.), Encyclopedia of Language and Linguistics, nd edn (pp. –). Oxford: Elsevier. Gardner, R. C., Lalonde, R. N., Moorcroft, R., & Evers, F. T. (). Second language attrition: The role of motivation and use. Journal of Language and Social Psychology, (), –. Garrett, A., & Johnson, K. (). Phonetic bias in sound change. In A. C. L. Yu (Ed.), Origins of Sound Change: Approaches to Phonologization (pp. –). Oxford: Oxford University Press. Gass, S. N., & Mackey, A. (). Frequency effects in second language acquisition: A complex picture? Studies in Second Language Acquisition, , –. Gathercole, V. C. M., & Thomas, E. M. (). Factors contributing to language transmission in bilingual families: The core study—adult interviews. In V. C. M. Gathercole (Ed.), Language Transmission in Bilingual Families in Wales (pp. –). Cardiff: Welsh Language Board. Gathercole, V. C. M., & Thomas, E. M. (). Bilingual first-language development: Dominant language takeover, threatened minority language take-up. Bilingualism: Language and Cognition, (), –. doi: ./S Gautam, P., Anstey, K. J., Wen, W., Sachdev, P. S., & Cherbuin, N. (). Cortical gyrification and its relationships with cortical volume, cortical thickness, and cognitive performance in healthy midlife adults. Behavioral Brain Research, , –. Gauthier, K., & Genesee, F. (). Language development in internationally adopted children: A special case of early second language learning. Child Development, (), –. doi: ./ j.-...x Gayraud, F., & Martinie, B. (). Does structural complexity necessarily imply processing difficulty? Journal of Psycholinguist Research, , –. Genesee, F., Nicoladis, E., & Paradis, J. (). Language differentiation in early bilingual development. Journal of Child Language, (), –. doi: ./S Genevska-Hanke, D. (). Interpersonal variation in late L attrition and its implications for the competence/performance debate. Poster presented at the Third International Conference on Language Attrition (ICLA ), Colchester. Gharibi, K. (). Incomplete acquisition, attrition and maintenance of heritage speakers’ family language: Iranians in New Zealand. PhD dissertation, Victoria University of Wellington. Ghazel, S. (). Cognitive architectures in morphological processing: Acquisition and attrition. PhD thesis, Carleton University. Ghazi-Saidi, L., & Ansaldo, A. I. (). Can a second language help you in more ways than one? AIMS Neuroscience, (), –. Gick, B., Wilson, I., Koch, K., & Cook, C. (). Language-specific articulatory settings: Evidence from inter-utterance rest position. Phonetica, (), –. Gil, K. H., & Marsden, H. (). Existential quantifiers in second language acquisition: A feature reassembly account. Linguistic Approaches to Bilingualism, (), –. Giles, H., Bourhis, R. Y., & Taylor, D. (). Towards a theory of language in ethnic group relations. In H. Giles (Ed.), Language, Ethnicity and Intergroup Relations (pp. –). London: Academic Press. Gilger, J. W. (). Using self-report and parental-report survey data to assess past and present academic achievement of adults and children. Journal of Applied Development Psychology, (), –. doi: ./-()-C Gillis, S., & De Houwer, A. (Eds.) (). The Acquisition of Dutch. Amsterdam: John Benjamins. Gillon Dowens, M., Vergara, M., Barber, H. A., & Carreiras, M. (). Morphosyntactic processing in late second-language learners. Journal of Cognitive Neuroscience, (), –. Gindis, B. (). Language-related issues for international adoptees and adoptive families. In T. Tepper, L. Hanna, & D. Sandstrom (Eds.), International Adoption: Challenges and Opportunities (pp. –). Meadow Lands, PA: PNPIC. Glennen, S., & Masters, M. G. (). Typical and atypical language development in infants and toddlers adopted from Eastern Europe. American Journal of Speech-Language Pathology, (), –. doi: ./-(/)
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Glisky, E. L., Ryan, L., Reminger, S., Hardt, O., Hayes, S. M., & Hupbach, A. (). A case of psychogenic fugue: I understand, aber ich verstehe nichts. Neuropsychologia, (), –. doi: ./j.neuropsychologia... Goad, H., & White, L. (). Ultimate attainment of L inflection: Effects of L prosodic structure. In S. Foster-Cohen, M. Sharwood Smith, A. Sorace, & M. Ota (Eds.), EUROSLA Yearbook, Vol. (pp. –). Amsterdam: John Benjamins. Godson, L. (). Vowel production in the speech of Western Armenian heritage speakers. Heritage Language Journal, (), –. Gold, B. T., Johnson, N. F., & Powell, D. K. (). Lifelong bilingualism contributes to cognitive reserve against white matter integrity declines in aging. Neuropsychologia, (), –. doi: ./j.neuropsychologia... Goldberg, H., Paradis, J., & Crago, M. (). Lexical acquisition over time in minority first language children learning English as a second language. Applied Psycholinguistics, (), –. Goldinger, S. D. (). Words and voices: Episodic traces in spoken word identification and recognition memory. Journal of Experimental Psychology: Learning, Memory, and Cognition, , –. Goldrick, M., Vaughn, C., & Murphy, A. (). The effects of lexical neighbors on stop articulation. Journal of the Acoustical Society of America, , EL. Goldrick, M., Runnqvist, E., & Costa, A. (). Language switching makes pronunciation less nativelike. Psychological Science, , –. Goldrick, M., Putnam, M., & Schwarz, L. (a). Co-activation in bilingual grammars: A computational account of code-mixing. Bilingualism: Language and Cognition, (), –. Goldrick, M., Putnam, M., & Schwarz, L. (b). The future of code mixing research: Integrating psycholinguistics and formal grammatical theories. Bilingualism: Language and Cognition, (), –. Golestani, N., Molko, N., Dehaene, S., LeBihan, D., & Pallier, C. (). Brain structure predicts the learning of foreign speech sounds. Cerebral Cortex, , –. Gollan, T. H., & Acenas, L.-A. R. (). What is a TOT? Cognate and translation effects on tipof-the-tongue states in Spanish-English and Tagalog-English bilinguals. Journal of Experimental Psychology: Learning, Memory, and Cognition, (), –. Gollan, T. H., & Brown, A. S. (). From tip-of-the-tongue (TOT) data to theoretical implications in two steps: When more TOTs means better retrieval. Journal of Experimental Psychology: General, (), –. Gollan, T. H., Montoya, R. I., Fennema-Notestine, C., & Morris, S. K. (). Bilingualism affects picture naming but not picture classification. Memory & Cognition, (), –. Gollan, T. H., Montoya, R. I., Cera, C., & Sandoval, T. C. (). More use almost always means a smaller frequency effect: Aging, bilingualism, and the weaker links hypothesis. Journal of Memory and Language, , –. Gollan, T. H., Salmon, D. P., Montoya, R. I., & da Pena, E. (). Accessibility of the non dominant language in picture naming: A counterintuitive effect of dementia on bilingual language production. Neuropsychologia, (), –. Gollan, T. H., Salmon, D. P., Montoya, R. I., & Galasko, D. R. (). Degree of bilingualism predicts age of diagnosis of Alzheimer’s disease in low-education but not in highly educated Hispanics. Neuropsychologia, (), –. Gollan, T., Slattery, T., Goldenberg, D., Van Assche, E., Duyck, W., & Rayner, K. (). Frequency drives lexical access in reading but not speaking: The frequency lag-hypothesis. Journal of Experimental Psychology: General, (), –. Gómez-Ruiz, I., Aguilar-Alonso, Á., & Espasa, M. A. (). Language impairment in CatalanSpanish bilinguals with Alzheimer’s disease. Journal of Neurolinguistics, (), –. Gómez Soler, I. (). Acquiring Spanish psych-verbs: Maturation or continuity? In A. Biller, E. Chung, & A. Kimball (Eds.), Online Proceedings of the th Boston University Conference on Language Development (BUCLD ) (pp. –). Somerville, MA: Cascadilla Proceedings Project.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Gonzo, S., & Saltarelli, M. (). Pidginization and linguistic change in emigrant languages. In R. W. Andersen (Ed.), Pidginization and Creolization as Language Acquisition (pp. –). Rowley, MA: Newbury House. Goral, M., Spiro, I. I. I., Albert, M. L., Obler, L. K., & Connor, L. T. (). Change in lexical retrieval skills in adulthood. The Mental Lexicon, (), –. Goral, M., Libben, G., Obler, L. K., Jarema, G., & Ohayon, K. (). Lexical attrition in younger and older bilingual adults. Clinical Linguistics & Phonetics, (), –. Graham, C. R. (). Vocabulary attrition in adult speakers of Spanish as an L. In L. Hansen (Ed.), Second Language Acquisition Abroad: The LDS Missionary Experience (pp. –). Amsterdam: John Benjamins. Graham, C. R., & Brown, C. (). The effects of acculturation on second language proficiency in a community with a two-way bilingual program. The Bilingual Research Journal, , –. Granena, G., & Long, M. H. (). Age of onset, length of residence, language aptitude, and ultimate L attainment in three linguistic domains. Second Language Research, (), –. doi: ./ Granfeldt, J. (). Development of object clitics in child L French: A comparison of developmental sequences in different modes of acquisition. Language, Interaction and Acquisition, (), –. doi: ./bct..gra Gray, S. (). Word-learning by preschoolers with specific language impairment: What predicts success? Journal of Speech, Language, and Hearing Research, , –. Graybiel, A. M. (). The basal ganglia and chunking of action repertoires. Neurobiology of Learning and Memory, (), –. Green, D. W. (). Control, activation and resource: A framework and a model for the control of speech in bilinguals. Brain and Language, , –. Green, D. W. (). Toward a model of L comprehension and production. In R. Schreuder & B. Weltens (Eds.), The Bilingual Lexicon (pp. –). Amsterdam: John Benjamins. Green, D. W. (). Mental control of the bilingual lexico-semantic system. Bilingualism: Language and Cognition, , –. Green, D. W. (). Concepts, experiments and mechanisms. Bilingualism: Language and Cognition, (), –. Green, D. W. (). Exploring the Convergence Hypothesis: Processing and representational implications. Paper presented at The Language Learning Round Table, EuroSLA, Edinburgh. Green, D. W. (). Language control in different contexts: The behavioral ecology of bilingual speakers. Frontiers in Psychology, , . doi: ./fpsyg.. Green, D. W., & Abutalebi, J. (). Language control in bilinguals: The adaptive control hypothesis. Journal of Cognitive Psychology, (), –. Green, D. W., & Wei, L. (). Code-switching and language control. Bilingualism: Language and Cognition, (), –. Greenberg, J. (). Language Universals, with Special Reference to Feature Hierarchies (Janua Linguarum, Series Minor, ). The Hague: Mouton. Gregg, K. (). Shallow draughts: Larsen-Freeman and Cameron on complexity. Second Language Research, (), –. Grendel, M. (). Verlies en herstel van franse woordkennis [Loss and recovery of French vocabulary knowledge]. Gramma, (), –. Grendel, M., Weltens, B., & de Bot, K. (). Language attrition: Rise and fall of a research topic? Ttwia, –(), –. Grillo, F. W. (). Long live the axon. Parallels between ageing and pathology from a presynaptic point of view. Journal of Chemical Neuroanatomy, , –. Grimstad, M. B., Lohndal, T., & Åfarli, T. (). Language mixing and exoskeletal theory: A case study of word-internal mixing in American Norwegian. Nordlyd, (), –. Grosjean, F. (). Neurolinguists, beware! The bilingual is not two monolinguals in one person. Brain and Language, (), –.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Grosjean, F. (). The bilingual individual. Interpreting, , –. doi:./intp..-.gro Grosjean, F. (). Studying bilinguals: Methodological and conceptual issues. Bilingualism: Language and Cognition, (), –. doi: ./SX Grosjean, F. (). The bilingual’s language modes. In J. Nicol (Ed.), One Mind, Two Languages: Bilingual Language Processing (pp. –). Oxford: Blackwell. Grosjean, F. (). Studying Bilinguals. Oxford: Oxford University Press. Grosjean, F. (). Bilingual: Life and Reality. Cambridge, MA: Harvard University Press. Grosjean, F. (). Bicultural bilinguals. International Journal of Bilingualism, (), –. Grosjean, F. (). The Complementarity Principle and its impact on processing, acquisition, and dominance. In C. Silva-Corvalán & J. Treffers-Daller (Eds.), Language Dominance in Bilinguals: Issues of Measurement and Operationalization (pp. –). Cambridge: Cambridge University Press. doi:./CBO. Grosjean, F., & Li, P. (). The Psycholinguistics of Bilingualism. Malden, MA: Blackwell Publishing. Grosjean, F., & Py, B. (). La restructuration d’une première langue: L’intégration de variantes de contact dans la compétence de migrants bilingues [Restructuring of a first language: The integration of contact variants into the competence of bilingual migrants.]. La Linguistique, , –. Gross, S. (). A modest proposal: Explaining language attrition in the context of contact linguistics. In M. S. Schmid, B. Köpke, M. Keijzer, & L. Weilemar (Eds.), First Language Attrition: Interdisciplinary Perspectives on Methodological Issues (pp. –). Amsterdam: John Benjamins. Grüter, T. (). Comprehension and production of French object clitics by child second language learners and children with specific language impairment. Applied Psycholinguistics, (), –. Grüter, T., & Paradis, J. (Eds.) (). Input and Experience in Bilingual Development. Amsterdam: John Benjamins. Grüter, T., Hurtado, N., Marchman, V., & Fernald, A. (). Language exposure and online processing efficiency: Relative versus absolute measures. In T. Grüter & J. Paradis (Eds.), Input and Experience in Bilingual Development (pp. –). Amsterdam: John Benjamins. Guasti, M. T. (). Language Acquisition: The Growth of Grammar. Cambridge, MA: MIT Press. Guijarro-Fuentes, P. (). Feature composition in differential object marking. Eurosla Yearbook, (), –. Guion, S. G. (). The vowel systems of Quichua–Spanish bilinguals: Age of acquisition effects on the mutual influence of the first and second languages. Phonetica, (), –. Guion, S. G., Flege, J. E., Liu, S. H., & Yeni-Komshian, G. H. (). Age of learning effects on the duration of sentences produced in a second language. Applied Psycholinguistics, , –. Gullberg, M., & de Bot, K. (). Gestures in Language Development. Amsterdam: John Benjamins Publishing. Guo, T., Liu, H., Misra, M., & Kroll, J. F. (). Local and global inhibition in bilingual word production: fMRI evidence from Chinese-English bilinguals. NeuroImage, (), –. Gürel, A. (). Linguistic characteristics of second language acquisition and first language attrition: Overt versus null pronouns. PhD thesis, McGill University. Gürel, A. (a). Selectivity in L-incuded L attrition: A psycholinguistic account. Journal of Neurolinguistics, (), –. Gürel, A. (b). Attrition in L competence: the case of Turkish. In M. S. Schmid, M. Keijzer, & L. Weilemar (Eds.), First Language Attrition: Interdisciplinary Perspectives on Methodological Issues (pp. –). Amsterdam: John Benjamins. Gürel, A. (). L acquisition of pragmatic and syntactic constraints in the use of overt and null subject pronouns. In R. Slabakova, P. Prévost, & S. Montrul (Eds.), Inquiries in Linguistic Development (pp. –). Amsterdam: John Benjamins. Gürel, A. (). (Psycho)linguistic determinants of L attrition. In B. Köpke, M. S. Schmid, M. Keijzer, & S. Dostert (Eds.), Language Attrition: A Theoretical Perspective (pp. –). Amsterdam: John Benjamins. Gürel, A., & Yılmaz, G. (). Restructuring in the L Turkish grammar: Effects of L English and L Dutch. Language, Interaction and Acquisition, (), –.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Gutiérrez-Clellen, V. F., & Kreiter, J. (). Understanding child bilingual acquisition using parent and teacher reports. Applied Psycholinguistics, (), –. doi: ./ S Gutiérrez-Clellen, V., Simon-Cereijido, G., & Wagner, C. (). Bilingual children with language impairment: A comparison with monolinguals and second language learners. Applied Psycholinguistics, (), –. Hack, E. (). Multilingualism and the risk of Alzheimer disease and dementia. PhD thesis, University of Ontario. Hahne, A., Mueller, J. L., & Clahsen, H. (). Morphological processing in a second language: Behavioral and event-related brain potential evidence for storage and decomposition. Journal of Cognitive Neuroscience, , –. Håkansson, G. (). Syntax and morphology in language attrition: A study of five bilingual expatriate Swedes. International Journal of Applied Linguistics, , –. Håkansson, G. (). Tense morphology and verb-second in Swedish L children, L children and children with SLI. Bilingualism: Language and Cognition, (), –. Håkansson, G., & Nettelbladt, U. (). Similarities between SLI and L children. Evidence from the acquisition of Swedish word order. In J. Gilbert & C. Johnson (Eds.), Children’s Language (pp. –). Mahwah, NJ: Lawrence Erlbaum & Associates. Hakuta, K. (). A case study of a Japanese child learning English as a second language. Language Learning, (), –. Hakuta, K., & D’Andrea, D. (). Some properties of bilingual maintenance and loss in Mexican background high-school students. Applied Linguistics, , –. Hakuta, K., Bialystok, E., & Wiley, E. (). Critical evidence: A test of the Critical-Period Hypothesis for second language acquisition. Psychological Science, (), –. doi: ./ -. Halle, M., & Marantz, A. (). Distributed morphology and the pieces of inflection. In K. Hale & S. J. Keyse (Eds.), The View from Building, Vol. (pp. –). Cambridge: MIT Press. Halmari, H. (). ‘I’m forgetting both’: L maintenance and codeswitching in Finnish-English language contact. International Journal of Bilingualism, (–), –. Hameau, S., & Köpke, B. (). Cross-language transfer for cognates in aphasia therapy with multilingual patients: A case study. Aphasie et domaines associés, , –. Hammer, C. S., Davison, M. D., Lawrence, F. R., & Miccio, A. W. (). The effect of maternal language on bilingual children’s vocabulary and emergent literacy development during head start and kindergarten. Scientific Studies of Reading, (), –. doi: ./ Hammer, K., & Dewaele, J. M. (). Acculturation as the key to the ultimate attainment? The case of Poles in the UK. In F. F. Lundell & I. Bartning (Eds.), Cultural Migrants and Optimal Language Acquisition (pp. –). Bristol: Multilingual Matters. Hammill, D., & Larsen, S. (). The Test of Written Language (TOWL-). Austin, TX: Pro-ed. Hansen, L. (a). Not a total loss: The attrition of Japanese negation over three decades. In L. Hansen (Ed.), Second Language Attrition in Japanese Contexts. (pp. –). New York: Oxford University Press. Hansen, L. (Ed.) (b). Second Language Attrition in Japanese Contexts. New York: Oxford University Press. Hansen, L. (). Language attrition: The fate of the start. Annual Review of Applied Linguistics, , –. Hansen, L. (). The acquisition, attrition and relearning of mission vocabulary. In M. S. Schmid & W. Lowie (Eds.), Modelling Bilingualism: From Structure to Chaos (pp. –). Amsterdam: John Benjamins. Hansen, L., & Chen, Y. L. (). What counts in the acquisition and attrition of numeral classifiers? JALT Journal, (), –.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Hansen, L., & Chen, Y.-L. (). What counts in the retention of numeral classifiers in Japanese and Chinese. In L. Hansen (Ed.), Language Learning Abroad: The LDS Missionary Experience (pp. –). Amsterdam: John Benjamins. Hansen, L., & Newbold, J. (). Literacy as an anchor of the spoken language: Evidence from the acquisition and attrition of Japanese. Paper presented at the AAAL Annual Conference, Orlando, FL. Hansen, L., & Reetz-Kurashige, A. (). Investigating second language attrition: An introduction. In L. Hansen (Ed.), Second Language Attrition in Japanese Contexts (pp. –). New York: Oxford University Press. Hansen, L., Umeda, Y., & McKinney, M. (). Savings in the relearning of second language vocabulary: The effects of time and proficiency. Language Learning, (), –. Hansen, L., Colver, A., Chong, W., et al. (). The lost word: Vocabulary attrition in six mission languages. In L. Hansen (Ed.), Second Language Acquisition Abroad: The LDS Missionary Experience (pp. –). Amsterdam: John Benjamins. Hansen, L., Lam, K., Orikasa, L., Rama, P., Schwaller, G., & Miller, R. M. (). In the beginning was the word: Vocabulary learning in six mission languages. In L. Hansen (Ed.), Second Language Acquisition Abroad: The LDS Missionary Experience (pp. –). Amsterdam: John Benjamins. Hansen-Strain, L. (). The attrition of Japanese by English-speaking children: An interim report. Language Sciences, (), –. Hanslmayr, S., Staresina, B. P., & Bowman, H. (). Oscillations and episodic memory: Addressing the synchronization/desynchronization conundrum. Trends in Neuroscience, (), –. doi: ./j.tins... Harada, T. (). L influence on L speech in the production of VOT. In M.-J. Sole, D. Recasens, & J. Romero (Eds.), Proceedings of the th International Congress of Phonetic Sciences (pp. –). Barcelona: Causal Productions. Harley, B. (). Maintaining French as a second language in adulthood. Paper presented at the th AILA world congress, Amsterdam. Harrington, J. (). An acoustic analysis of ‘happy-tensing’ in the Queen’s Christmas broadcasts. Journal of Phonetics, (), –. doi: ./j.wocn... Harrington, J., Palethorpe, S., & Watson, C. (a). Monophthongal vowel changes in Received Pronunciation: an acoustic analysis of the Queen’s Christmas broadcasts. Journal of the International Phonetic Association, (–), –. doi: ./S Harrington, J., Palethorpe, S., & Watson, C. I. (b). Does the Queen speak the Queen’s English? Nature, (), –. doi: ./ Harris, J. D. (). Remarks on the determination of a differential threshold by the so-called ABX technique. Journal of the Acoustical Society of America, , . doi: ./. Hart, B., & Risley, T. (). Meaningful Differences in the Everyday Experience of Young American Children. Baltimore: Brookes. Hartsuiker, R. J., Pickering, M. J., & Veltkamp, E. (). Is syntax separate or shared between languages? Cross-linguistic syntactic priming in Spanish-English bilinguals. Psychological Science, (), –. Hartsuiker, R. J., Bernolet, S., Schoonbaert, S., Speybroeck, S., & Vanderelst, D. (). Syntactic priming persists while the lexical boost decays: Evidence from written and spoken dialogue. Journal of Memory and Language, (), –. Haugen, E. (). Language and immigration. Norwegian-American Studies, (), –. Haugen, E. (). Dialect, language, nation. American Anthropologist, (), –. doi: ./ aa....a Hauser, M., Chomsky, N., & Fitch, W. T. (). The faculty of language: What is it, who has it, and how did it evolve? Science, , –. Hawkins, R., & Liszka, S. (). Locating the source of defective past tense marking in advanced L English speakers. In R. van Hout, A. Hulk, F. Kuiken, & R. Towell (Eds.), The Lexicon-syntax Interface in Second Language Acquisition (pp. –). Amsterdam: John Benjamins.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Hayashi, B. (). Testing the regression hypothesis: The remains of the Japanese negation system in Micronesia. In Hansen, L. (Ed.) Second Language Attrition in Japanese Contexts (pp. –). Oxford: Oxford University Press. Hayashi, G. (). Language attrition of Japanese returnee students. Accents Asia, (), –. He, A. W. (). Toward an identity theory of the development of Chinese as a heritage language. The Heritage Language Journal, (), –. Hedgcock, J. (). Foreign language retention and attrition: A study of regression models. Foreign Language Annals, , –. Heine, B., & Kuteva, T. (). Language Contact and Grammatical Change. Cambridge: Cambridge University Press. Heine, B., Kaltenböck, G., Kuteva, T., & Long, H. (). An outline of discourse grammar. In S. Bischoff & C. Jany (Eds.), Functional Approaches to Language (pp. –). Berlin: Mouton de Gruyter. Henry, L. A., & Botting, N. (). Working memory and developmental language impairments. Child Language Teaching and Therapy, (), –. Herd, W. J., Walden, R. L., Knight, W. L., & Alexander, S. N. (). Phonetic drift in a first language dominant environment. Proceedings of Meetings on Acoustics, , . Herdina, P., & Jessner, U. (). A Dynamic Model of Multilingualism: Changing the Psycholinguistic Perspective. Clevedon: Multilingual Matters. Herdina, P., & Jessner, U. (). The implications of language attrition for Dynamic Systems Theory: Next steps and consequences. International Journal of Bilingualism. Special Issue on Dynamics of First Language Attrition across the Lifespan, (), –. Hermans, D., Bongaerts, T., de Bot, K., & Schreuder, R. (). Producing words in a foreign language: Can speakers prevent interference from their first language? Bilingualism: Language and Cognition, (), –. Herring, J., Deuchar, M., Parafita Couto, M. C., & Moro Quintanilla, M. (). When I went to Canada, I saw the madre: Evaluating two theories’ predictions about codeswitching between determiners and nouns using Spanish-English and Welsh-English bilingual corpora. International Journal of Bilingual Education and Bilingualism, , –. Hickey, R. (). Contact and language shift. In R. Hickey (Ed.), The Handbook of Language Contact (pp. –). Chichester: Wiley-Blackwell. Hickok, G. (). Towards an integrated psycholinguistic, neurolinguistic, sensorimotor framework for speech production. Language, Cognition and Neuroscience, (), –. Hickok, G., & Poeppel, D. (). Dorsal and ventral streams: A framework for understanding aspects of the functional anatomy of language. Cognition, (–), –. Higby, E., Kim, J., & Obler, L. K. (). Multilingualism and the brain. Annual Review of Applied Linguistics, , –. Higby, E., & Obler, L. K. (). Losing a first language to a second language. In J. Schwieter (Ed.), Cambridge Handbook of Bilingual Processing (pp. –). Cambridge: Cambridge University Press. Higby, E., Cahana-Amitay, D., Vogel, A., Spiro III, A., Albert, M. L., & Obler, L. K. (in press). The role of executive functions in naming among older adults. Experimental Aging Research. Hill, J. H. (). Review of B. Weltens, K. de Bot, & T. van Els (Eds.), Language Attrition in Progress: Studies on Language Acquisition. Dordrecht, NL: Foris Publications. Language in Society, (), –. Hilles, S. (). Interlanguage and the pro-drop parameter. Second Language Research, , –. Hinaut, X., & Dominey, P. F. (). Real-time parallel processing of grammatical structure in the fronto-striatal system: A recurrent network simulation study using reservoir computing. PloS One, (), e. Hipfner-Boucher, K., Milburn, T., Weitzman, E., Greenberg, J., Pelletier, J., & Girolametto, L. (). Narrative abilities in subgroups of English language learners and monolingual peers. International Journal of Bilingualism, (), –.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Hirai, S. (). Longitudinal L attrition: A case study of a Japanese-English bilingual child. Japan Journal of Multilingualism and Multiculturalism, , –. Hirakawa, M. (). A study of the L acquisition of English reflexives. Second Language Research, , –. Hiver, P. (). Attractor states. In Z. Dörnyei, P. D. MacIntyre, & A. Henry (Eds.), Motivational Dynamics in Language Learning (pp. –). Bristol: Multilingual Matter. Hlavac, J. (). Croatian in Melbourne: Lexicon, switching and morphosyntactic features in the speech of second-generation bilinguals. PhD thesis, Monash University. Hlavac, J. (). Second-generation Speech: Lexicon, Code-switching and Morpho-syntax of CroatianEnglish Bilinguals. Bern: Peter Lang. Hoff, E. (). How social contexts support and shape language development. Developmental Review, (), –. doi: ./j.dr... Hoff, E., & Tian, C. (). Socioeconomic status and cultural influences on language. Journal of Communication Disorders, , –. Hoff, E., Core, C., Place, S., Rumiche, R., Senor, M., & Parra, M. (). Dual language exposure and early bilingual development. Journal of Child Language, (), –. doi: ./S Hoff, E., Rumiche, R., Burridge, A., Ribot, K., M., & Welsh, S. (). Expressive vocabulary development in children from bilingual and monolingual homes: A longitudinal study from two to four years. Early Childhood Research Quarterly, (), –. doi: ./j.ecresq... Holland, J. H. (). Language acquisition as a complex adaptive system. In J. W. Minett & W. S.-Y. Wang (Eds.), Language Acquisition, Change and Emergence: Essays in Evolutionary Linguistics (pp. –). Hong Kong: City University of Hong Kong Press. Holmberg, A. (). Is there a little pro? Evidence from Finnish. Linguistic Inquiry, (), –. Hopp. H. (). Ultimate attainment in L inflection: Performance similarities between nonnative and native speakers. Lingua, , –. Hopp, H. (). Working memory effects on the L processing of ambiguous relative clauses. Language Acquisition, , –. Hopp, H. (a). Learning (not) to predict: Grammatical gender processing in second language acquisition. Second Language Research, (), –. Hopp, H. (b). The timing of lexical and syntactic processes in second language sentence comprehension. Applied Psycholinguistics, , –. Hopp, H., & Putnam, M. T. (). Syntactic restructuring in heritage grammars: Word order variation in Moundridge Schweitzer German. Linguistic Approaches to Bilingualism, , –. Hopp, H., & Schmid, M. S. (). Perceived foreign accent in first language attrition and second language acquisition: The impact of age of acquisition and bilingualism. Applied Psycholinguistics, (), –. doi: ./S Hoshino, N., & Kroll, J. F. (). Cognate effects in picture naming: Does cross-language activation survive a change of script? Cognition, (), –. Hosoda, C., Tanaka, K., Nariai, T., Honda, M., & Hanakawa, T. (). Dynamic neural network reorganization associated with second language vocabulary acquisition: A multimodal imaging study. Journal of Neuroscience, (), –. Houston, D. M., Stewart, J., Moberly, A., Hollich, G., & Miyamoto, R. T. (). Word learning in deaf children with cochlear implants: Effects of early auditory experience. Developmental Science, (), –. doi: ./j.-...x Huang, Y. (). On null subjects and null objects in generative grammar. Linguistics, , –. Hubel, D. H., & Wiesel, T. N. (). Binocular interaction in striate cortex of kittens reared with artificial squint. Journal of Neurophysiology, (), –. Huettel, S. A. (). Event-related fMRI in cognition. Neuroimage, (), –. Huettel, S. A., Song, A. W., & McCarthy, G. (). Functional Magnetic Resonance Imaging, rd edn. New York: Sinauer. Huettig, F., Rommers, J., & Meyer, A. S. (). Using the visual world paradigm to study language processing: A review and critical evaluation. Acta Psychologica, (), –.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Huffines, M. L. (). Pennsylvania German: Convergence and change as strategies of discourse. In H. W. Seliger & R. M. Vago (Eds.), First Language Attrition (pp. –). Cambridge: Cambridge University Press. Huffman, M. K., & Schuhmann, K. [S.] (a). Individual differences in phonetic drift by Englishspeaking learners of Spanish. Journal of the Acoustical Society of America, (), . Huffman, M. K., & Schuhmann, K. S. (b). Effect of early L learning on L stop voicing. Proceedings of Meetings on Acoustics, , . Hulk, A. (). L’acquisition des pronoms clitiques français par un enfant bilingue françaisnéerlandais [The acquisition of French clitic pronouns by a French-Dutch bilingual child.]. The Canadian Journal of Linguistics, (–), –. doi: ./SX Hulk, A., & Müller, N. (). Bilingual first language acquisition at the interface between syntax and pragmatics. Bilingualism: Language and Cognition, (), –. doi:./S Hulk, A., & Van der Linden, E. (). How vulnerable is gender? In P. Guijarro-Fuentes, & L. Domínguez (Eds.), New Directions in Language Acquisition: Romance Languages in the Generative Perspective (pp. –). Newcastle upon Tyne: Cambridge Scholars. Hulsen, M. (). Language loss and language processing: Three generations of Dutch migrants in New Zealand. PhD thesis, Katholieke Universiteit Nijmegen. Hulsen, M., de Bot, K., & Weltens, B. (). Between two worlds: Social networks, language shift, and language processing in three generations of Dutch migrants in New Zealand. International Journal of the Sociology of Language, , –. Hulstijn, J. H. (). Measuring second language proficiency. In E. Blom & S. Unsworth (Eds.), Experimental Methods in Language Research (pp. –). Amsterdam: John Benjamins. Hupbach, A., Gomez, R., Hardt, O., & Nadel, L. (). Reconsolidation of episodic memories: A subtle reminder triggers integration of new information. Learning & Memory, (–), –. doi:./lm. Hupet, M., & Nef, F. (). Vieillissement cognitif [Cognitive ageing]. In M. van der Linden & M. Hupet (Eds.), Le vieillissement cognitif (pp. –). Paris: PUF. Huttenlocher, J., & Smiley, P. (). Early word meanings: The case of object names. Cognitive Psychology, , –. Huttenlocher, J., Waterfall, H., Vasilyeva, M., Vevea, J., & Hedges, L. V. (). Sources of variability in children’s language growth. Cognitive Psychology, , –. Hutz, M. (). Is there a natural process of decay? A longitudinal study of language attrition. In M. S. Schmid, M. Keijzer, & L. Weilemar (Eds.), First Language Attrition: Interdisciplinary Perspectives on Methodological Issues (pp. –). Amsterdam: John Benjamins. Hyams, N. (). Language Acquisition and the Theory of Parameters. Dordrecht: Reidel. Hyams, N., & Wexler, K. (). On the grammatical basis of null subjects in child language. Linguistic Inquiry, , –. Hyltenstam, K. (). The code switching behaviour of adults with language disorders—with special reference to aphasia and dementia. In L. Milroy & P. Musken (Eds.), One Speaker, Two Languages: Cross-disciplinary Perspectives on Code Switching (pp. –). Cambridge: Cambridge University Press. Hyltenstam, K., & Abrahamsson, N. (). Maturational constraints in second language acquisition. In C. J. Doughty, & M. H. Long (Eds.), Handbook of Second Language Acquisition (pp. –). Oxford: Blackwell. Hyltenstam, K., & Obler, L. K. (). Bilingualism across the Lifespan: Aspects of Acquisition, Maturity and Loss. Cambridge: Cambridge University Press. Hyltenstam, K., & Stroud, C. (). Bilingualism in Alzheimer’s dementia: Two case studies. In K. Hyltenstam & L. K. Obler (Eds.), Bilingualism across the Lifespan: Aspects of Acquisition, Maturity and Loss (pp. –). Cambridge: Cambridge University Press. Hyltenstam, K., & Stroud, C. (). Second language regression in Alzheimer dementia. In K. Hyltenstam & A. Viberg (Eds.), Progression and Regression in Language: Sociocultural Neuropsychological and Linguistic Perspectives (pp. –). Cambridge: Cambridge University Press.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Hyltenstam, K., & Viberg, A. (). Linguistic progression and regression: an introduction. In K. Hyltenstam & A. Viberg (Eds.), Progression & Regression in Language: Sociocultural, Neuropsychological, & Linguistic Perspectives (pp. –). Cambridge: Cambridge University Press. doi:./cbo Hyltenstam, K., Bylund, E., Abrahamsson, N., Park, H. S. (). Dominant-language replacement: The case of international adoptees. Bilingualism: Language and Cognition, (), –. Indefrey, P. (). The spatial and temporal signatures of word production components: A critical update. Frontiers in Psychology, , . Ingram, D. (). Some characteristics of delayed phonological systems. In H. Winitz (Ed.), Native Language and Foreign Language Acquisition (pp. –). New York: New York Academy of Science. Ingram, D. (). The measurement of whole-word productions. Journal of Child Language, (), –. doi: ./S Ioup, G. (). Exploring age and loss using the savings paradigm. In K. de Bot & L. Hansen (Chairs), Reactivating a ‘forgotten’ language: The savings-paradigm applied. Symposium conducted at AAAL , St. Louis, MO. Isurin, L. (). ‘Deserted island’ or a child’s first language forgetting. Bilingualism: Language and Cognition, (), –. Isurin, L. (). Teachers’ language: L attrition in Russian-English bilinguals. Modern Language Journal, (), –. Isurin, L. (). Memory and first language forgetting. In J. Altarriba & L. Isurin (Eds.), Memory, Language, and Bilingualism (pp. –). Cambridge: Cambridge University Press. Isurin, L., & Ivanova-Sullivan, T. (). Lost in between: The case of Russian heritage speakers. Heritage Language Journal, (), –. Isurin, L., & Seidel, C. (). Traces of memory for a lost childhood language: The Savings Paradigm expanded. Language Learning, (), –. Ivanova, I., & Costa, A. (). Does bilingualism hamper lexical access in speech production? Acta Psychologica, , –. Ivanova, I., Salmon, D. P., & Gollan, T. H. (). The multilingual naming test in Alzheimer’s disease: Clues to the origin of naming impairments. Journal of the International Neuropsychological Society, (), –. Ivanova, I., Salmon, D. P., & Gollan, T. H. (). Which language declines more? Longitudinal versus cross-sectional decline of picture naming in bilinguals with Alzheimer’s disease. Journal of the International Neuropsychological Society, (), –. Iverson, M. (). Advanced language attrition of Spanish in contact with Brazilian Portuguese. PhD thesis, University of Iowa. Iverson, M., Kempchinsky, P., & Rothman, J. (). Interface vulnerability and knowledge of the subjunctive/indicative distinction with negated epistemic predicates in L Spanish, EUROSLA Yearbook, , –. Iverson, P., Kuhl, P. K., Akahane-Yamadac, R., et al. (). A perceptual interference account of acquisition difficulties for non-native phonemes. Cognition, , B–B. Jackendoff, R. (). Consciousness and the Computational Mind. Cambridge, MA: MIT Press. Jackendoff, R. (). The Architecture of the Language Faculty. Cambridge, MA: MIT Press. Jackendoff, R. (). Foundations of Language: Brain, Meaning, Grammar, Evolution. Oxford: Oxford University Press. Jackendoff, R. (). Linguistics in cognitive science: The state of the art. The Linguistic Review, , –. Jackendoff, R. (). What is the human language faculty? Two views. Language, (), –. Jacobs, A., Fricke, M., & Kroll, J. F. (). Cross-language activation begins during speech planning and extends into second language speech. Language Learning, (), –. Jaeggli, O., & Safir, K. J. (). The null subject parameter and parametric theory. In O. Jaeggli & K. J. Safir (Eds.), The Null Subject Parameter (pp. –). Dordrecht: Kluwer Academic Publishers.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Jaespart, K., & Kroon, S. (). Social determinants of language loss. ITL—Review of Applied Linguistics, /, –. Jake, J. L., Myers-Scotton, C., & Gross, S. (). Making a minimalist approach to codeswitching work: Adding the Matrix Language. Bilingualism: Language and Cognition, , –. Jakobson, R. (). The structure of the Russian verb. Reprinted in L. R. Waugh & M. Halle (Eds.), Russian and Slavic Grammar Studies – (pp. –). Berlin: Mouton. Jakobson, R. (). Kindersprache, Aphasie und allgemeine Lautgesetze. Uppsala: Almqvist. Jakobson, R. (). Child Language, Aphasia and Phonological Universals, Vol. . Berlin: Walter de Gruyter GmbH & Co KG. Jameson, K. A., & Alvarado, N. (). The relational correspondence between category exemplars and names. Philosophical Psychology, (), –. Jared, D., & Kroll, J. F. (). Do bilinguals activate phonological representations in one or both of their languages when naming words? Journal of Memory and Language, (), –. Jarvis, S. (). Probing the effects of the L on the L: A case study. In V. Cook (Ed.), Effects of the Second Language on the First (pp. –). Bristol: Multilingual Matters. Jarvis, S. (). Conceptual transfer: Crosslinguistic effects in categorization and construal. Bilingualism: Language and Cognition, , –. Jarvis, S. (). Lexical challenges in the intersection of applied linguistics and ANLP. In C. Boonthum-Denecke, P. M. McCarthy, & T. Lamkin (Eds.), Cross-disciplinary Advances in Applied Natural Language Processing: Issues and Approaches (pp. –). Hershey, PA: IGI Global. Jarvis, S. (a). Capturing the diversity in lexical diversity. Language Learning, (Suppl. ), –. Jarvis, S. (b). Defining and measuring lexical diversity. In S. Jarvis & M. Daller (Eds.), Vocabulary Knowledge: Human Ratings and Automated Measures (pp. –). Amsterdam: John Benjamins. Jarvis, S., & Pavlenko, A. (). Crosslinguistic Influence in Language and Cognition. London: Routledge. Jaspaert, K., & Kroon, S. (). Social determinants of language loss. I.T.L. Review of Applied Linguistics, , –. Jaspaert, K., & Kroon, S. (). Social determinants of language shift by Italians in the Netherlands and Flanders. International Journal of the Sociology of Language, , –. Jaspaert, K., & Kroon, S. (). From the typewriter of A.L.: A case study in language loss. In W. Fase, K. Jaspaert, & S. Kroon (Eds.), Maintenance and Loss of Minority Languages (pp. –). Amsterdam: John Benjamins. Jaspers, M., de Meer, G., Verhulst, F. C., Ormel, J., & Reijneveld, S. A. (). Limited validity of parental recall on pregnancy, birth and early childhood at age years. Journal of Clinical Epidemiology, , –. Jessner, U. (). A dynamic approach to language attrition in multilingual systems. In V. Cook (Ed.), Effects of the Second Language on the First (pp. –). Clevedon, Buffalo: Multilingual Matters. Jia, G. (). Heritage language development, maintenance, and attrition among recent Chinese immigrants in New York City. In A. W. He & Y. Xiao (Eds.), Chinese as a Heritage Language (pp. –). Honolulu, HI: National Foreign Language Resource Center / University of Hawaii Press. Jia, R., & Paradis, J. (). The use of referring expressions in narratives by Mandarin heritage language children and the roles of language environment factors in predicting individual differences. Bilingualism: Language and Cognition, , –. doi: ./S Jiménez Jiménez, A. F. (). A sociocultural approach to language attrition. In M. Schmid, B. Köpke, M. Keijzer, & L. Weilemar (Eds.), First Language Attrition (pp. –). Amsterdam: John Benjamins. Joh, J., Ko, H., Lim, I., & Lee, S. (). The impact of English language learning on the acquisition of native language among Korean children. Eungyong Eoneohag [Korean Journal of Applied Linguistics], (), –. Johannessen, J. B. (). Attrition in an American Norwegian heritage language speaker. In J. B. Johannssen & J. C. Salmons (Eds.), Germanic Heritage Languages in North America: Acquisition, Attrition, and Change (pp. –). Amsterdam: John Benjamins.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Johnson, J. S., & Newport, E. L. (). Critical period effects in second language learning: The influence of maturational state on the acquisition of English as a second language. Cognitive Psychology, , –. Jonker, T. R., Seli, P., & MacLeod, C. M. (). Putting retrieval-induced forgetting in context: An inhibition-free, context-based account. Psychological Review, (), –. Jonker, T. R., Seli, P., & MacLeod, C. M. (). Retrieval-induced forgetting and context. Current Directions in Psychological Science, (), –. Jordens, P., de Bot, K., van Os, C., & Schumans, J. (). Regression in German case marking. In B. Weltens, K. de Bot, & T. van Els (Eds.), Language Attrition in Progress: Studies on Language Acquisition (pp. –). Dordrecht, NL: Foris Publications. Jordens, P., de Bot, K., & Trapman, H. (). Linguistic aspects of regression in German case marking. Studies in Second Language Acquisition, , –. Ju, M., & Luce, P. A. (). Falling on sensitive ears: Constraints on bilingual lexical activation. Psychological Science, (), –. Judy, T., & Rothman, J. (). The processing of pronominal subjects by near-native L Persian-L Spanish adult bilinguals. In W. Orman & M. J. Valleau (Eds.), Proceedings of Boston University Conference on Language Development, Vol. (pp. –). Somerville, MA: Cascadilla Press. Juffs, A. (). Representation, processing, and working memory in a second language. Transactions of the Philological Society, , –. Jun, S.-A. (). Phonological development of Korean: A case study. In S. Kuno, I.-H. Lee, J. Whitman, et al. (Eds.), Harvard Studies in Korean Linguistics, Vol. XI (pp. –). Korea: Hanshin Publishing Company. Jurafsky, D., Bell, A., Gregory, M., & Raymond, W. (), Probabilistic relations between words: Evidence from reduction in lexical production. In J. Bybee & P. Hopper (Eds.), Frequency and the Emergence of Linguistic Structure (pp. –). Amsterdam: John Benjamins. Just, M. A., & Carpenter, P. A. (). A theory of reading: From eye fixations to comprehension. Psychological Review, , –. Juvonen, P. (). Språkkontakt och språkförändring: Om användningen av demonstrativa pronomen i en sverigefinsk kontaktsituation [Language contact and language change: On the use of demonstrative pronouns in a Swedish-Finnish contact situation.]. In L. Huss (Ed.), Många vägar till tvåspråkighet [Many ways to bilingualism] (pp. –). Uppsala: Uppsala Universitet. Kaan, E. (). Event-related potentials and language processing: A brief overview. Language and Linguistics Compass, (), –. Kaiser, E., & Trueswell, J. C. (). The role of discourse context in the processing of a flexible wordorder language. Cognition, (), –. Kaltenböck, G., Heine, B., & Kuteva, T. (). On thetical grammar. Studies in Language, (), –. Kaltsa, M., Tsimpli, I. M., & Rothman, J. (). Exploring the source of differences and similarities in L attrition and heritage speaker competence: Evidence from pronominal resolution. Lingua, , –. Kamide, Y., Altmann, G. T., & Haywood, S. L. (). The time-course of prediction in incremental sentence processing: Evidence from anticipatory eye movements. Journal of Memory and Language, (), –. Kan, P. F., & Windsor, J. (). Word learning in children with primary language impairment: A meta-analysis. Journal of Speech, Language, and Hearing Research, (), –. Kan, R., & Schmid, M. S. (). Development of tonal discrimination in young heritage speakers of Cantonese. Journal of Phonetics, , –. Kang, K.-H., & Guion, S. G. (). Phonological systems in bilinguals: Age of learning effects on the stop consonant systems of Korean-English bilinguals. Journal of the Acoustical Society of America, (), –. Kang, S.-G. (). English attrition in Korean-English bilingual children. PhD thesis, University of Hawaii.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Kang, S.-G. (). The role of language dominance in cross-linguistic syntactic influence: A Korean child’s use of null subjects in attriting English. Bilingualism: Language and Cognition, , –. doi: ./S Kaplan, E., Goodglass, H., & Weintraub, S. (). Boston Naming Test (BNT). Manual, nd edn. Philadelphia: Lea & Fabiger. Karayayla, T. (). Turkish as an immigrant and heritage language in the UK: Effects of exposure and age at onset of bilingualism on grammatical and lexical development of the first language. PhD thesis, University of Essex. Karayayla, T. (). Effects of first language attrition on heritage language input and ultimate attainment: two generations of Turkish immigrants in the UK. In B. Brehmer, J. Treffers-Daller, & D. Berndt (Eds.), Lost in Transmission: The Role of Attrition and Input in Heritage Language Development. Amsterdam: John Benjamins. Karayayla, T., & Schmid, M. S. (). L attrition as a function of age at onset of bilingualism: L attainment of Turkish-English bilinguals in the UK. Language Learning (), –. Kargar, H., & Rezai, M. J. (). The contribution of lexicon to the language attrition of Iranian immigrants to Canada. Procedia—Social and Behavioral Sciences, , –. Karni, A., & Sagi, D. (). A memory system in the adult visual cortex. In B. Julesz & I. Kovacs (Eds.), Maturational Windows and Adult Cortical Plasticity (pp. –). New York: AddisonWesley. Kartushina, N. (). Second language phonological acquisition in adults. PhD thesis, University of Geneva. Kartushina, N., Hervais-Adelman, A., Frauenfelder, U. H., & Golestani, N. (). Mutual influences between native and non-native vowels in production: Evidence from short-term visual articulatory feedback training. Journal of Phonetics, , –. Kasparian, K. (). The case of the non-native-like first-language: Neurophysiological investigations of first-language attrition and second-language processing. PhD Dissertation, McGill University. Kasparian, K., & Steinhauer, K. (). Confusing similar words: ERP correlates of lexical-semantic processing in first language attrition and late second language acquisition. Neuropsychologia, , –. Kasparian, K., & Steinhauer, K. (a). When the second language takes the lead: Neurocognitive processing changes in the first language of adult attriters. Frontiers in Psychology, , . doi: ./fpsyg... Kasparian, K., & Steinhauer, K. (b). On missed opportunities and convenient “truths”. Linguistic Approaches to Bilingualism, (), –. Kasparian, K., Vespignani, F., & Steinhauer, K. (). The case of a non-native-like first language: ERP evidence of first language (L) attrition in lexical and morphosyntactic processing. International Journal of Psychosterphysiology, (), –. Kasparian, K., Vespignani, F., & Steinhauer, K. (). First language attrition induces changes in online morphosyntactic processing and re-analysis: An ERP study of number agreement in complex Italian sentences. Cognitive Science, (), –. Kasparian, K., Postiglione, F., & Steinhauer, K. (under revision). In my other language: An ERP study of Italian-English cognates and homographs in L-attrition and L-acquisition. PLoS-ONE. Kaufman, D., & Aronoff, M. (). Morphological disintegration and reconstruction in first language attrition. In H. W. Seliger & R. M. Vago (Eds.), First Language Attrition (pp. –). Cambridge: Cambridge University Press. Kavé, G., & Goral, M. (). Do age-related word retrieval difficulties appear (or disappear) in connected speech? Aging, Neuropsychology, and Cognition. doi: ./... Kavé, G., Samuel-Enoch, K., & Adiv, S. (). The association between age and the frequency of nouns selected for production. Psychology and Aging, (), –.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Kavé, G., Knafo, A., & Gilboa, A. (). The rise and fall of word retrieval across the lifespan. Psychology and Aging, (), –. Keating, G. D. (). Eye-tracking with text. In J. Jegerski & B. VanPatten (Eds.), Research Methods in Second Language Psycholinguistics (pp. –). London: Taylor & Francis. Keating, G. D., VanPatten, B., & Jegerski, J. (). Who was walking on the beach? Anaphora resolution in Spanish heritage speakers and adult second language learners. Studies in Second Language Acquisition, (), –. Kecskes, I., & Papp, T. (). Foreign Language and Mother Tongue. Mahwah, NJ: Lawrence Erlbaum Associates. Kehoe, M. (). Developing vowel systems as a window to bilingual phonology. International Journal of Bilingualism, (), –. doi: ./ Kehoe, M. (). Cross-linguistic interaction: A retrospective and prospective view. In E. Babatsouli, & D. Ingram (Eds.), Proceedings of the International Symposium on Monolingual and Bilingual Speech (pp. –). Chania: Institute of Monolingual and Bilingual Speech. Kehoe, M. (). The development of rhotics: A comparison of monolingual and bilingual children. Bilingualism: Language and Cognition, (): –. doi: .. Kehoe, M., Lleó, C., & Rakow, M. (). Voice onset time in bilingual German-Spanish children. Bilingualism: Language and Cognition, (), –. doi: ./S Keijzer, M. (). First language attrition: A cross-linguistic investigation of Jakobson’s regression hypothesis. International Journal of Bilingualism, (), –. Keijzer, M. (). Last in first out? An investigation of the Regression Hypothesis in Dutch emigrants in anglophone Canada. PhD thesis, Vrije Universiteit, Amsterdam. Keijzer, M. (a). The regression hypothesis as a framework for first language attrition. Bilingualism: Language and Cognition, (), –. doi:./S Keijzer, M. (b). Inside the attriter’s mind. A comparative exploration of the cognitive constraints in Dutch L attrition in an L English environment and advanced Dutch L acquisition. In M. Pütz & L. Sicola (Eds.), Cognitive Processing and Second Language Acquisition (pp. –). Amsterdam: John Benjamins. Keijzer, M. (c). First Language Acquisition and First Language Attrition: Parallels and Divergences. Saarbrücken: Lambert Publishing. Keijzer, M. (). Language reversion versus general cognitive decline: Towards a new taxonomy of language change in elderly bilingual immigrants. In M. S. Schmid & W. Lowie (Eds.), Modeling Bilingualism: From Structure to Chaos (pp. –). Amsterdam: John Benjamins. Keijzer, M. (). Changes in neural activation patterns and brain anatomy as a function of nonpathological first language attrition. Journal of Neurological Disorders, (), . doi:./. Keijzer, M., & de Bot, K. (). Contexts associated with the unlearning of languages across the lifespan. In Ortega, L. & De Houwer, A. (Eds.), Handbook of Bilingualism (pp. –). Cambridge: Cambridge University Press. Kempchinsky, P. (). What can the subjunctive disjoint reference effect tell us about the subjunctive? Lingua, (), –. Kemper, S., & Sumner, A. (). The structure of verbal abilities in young and older adults. Psychology and Aging, (), –. Kennedy, L. R. (). The retention of certain Latin syntactical principles by first and second year Latin students after various time intervals. Journal of Educational Psychology, (), –. Kerswill, P. (). Children, adolescents and language change. Language Variation and Change, (), –. https://doi.org/./S Kim, A., & Osterhout, L. (). The independence of combinatory semantic processing: Evidence from event-related potentials. Journal of Memory and Language, , –.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Kim, J.-H., Montrul, S., & Yoon, J. (). Binding interpretation of anaphors in Korean heritage speakers. Language Acquisition, (), –. Kim, J.-H., Montrul, S., & Yoon, J. (). Dominant language influence in acquisition and attrition of binding: Interpretation of the Korean relfexive caki. Bilingualism: Language and Cognition, (): –. Kim, K., Relkin, N., Lee, K., & Hirsch, J. (). Distinct cortical areas associated with native and second languages. Nature, , –. Kimura, Y. (). Assessing communicative competence of high school returnees: The implications for English classes in Japan. Paper presented at JALT Kansai, Kyoto, Japan. King, J. (Ed.). (). The Dynamic Interplay between Context and the Language Learner. Basingstoke: Palgrave Macmillan. Kleim, J. A., & Jones, T. A. (). Principles of experience-dependent neural plasticity: Implications for rehabilitation after brain damage. Journal of Speech, Language, and Hearing Research, (), S–S. Klein, D., Zatorre, R. J., Milner, B., & Zhao, V. (). A cross-linguistic PET study of tone perception in Mandarin Chinese and English speakers. NeuroImage, (), –. doi: ./nimg.. Klimesch, W. (). Alpha-band oscillations, attention, and controlled access to stored information. Trends in Cognitive Sciences, (), –. doi:./j.tics... Knightly, L. M., Jun, S.-A., Oh, J. S., & Au, T. K. (). Production benefits of childhood overhearing. Journal of the Acoustical Society of America, (), –. doi: ./. Knoeferle, P., Crocker, M. W., Scheepers, C., & Pickering, M. J. (). The influence of the immediate visual context on incremental thematic role-assignment: Evidence from eyemovements in depicted events. Cognition, (), –. Knudsen, E. (). Sensitive periods in the development of the brain and behavior. Journal of Cognitive Neuroscience, (), –. doi: ./ Koelkebeck, K., Miyata, J., Kubota, M., et al. (). The contribution of cortical thickness and surface area to gray matter asymmetries in the healthy human brain. Human Brain Mapping, (), –. Kohnert, K. (). Bilingual children with Primary Language Impairment: Issues, evidence and implications for clinical actions. Journal of Communication Disorders, , –. Kohnert, K., Windsor, J., & Yim, D. (). Do language-based processing tasks separate children with language impairment from typical bilinguals? Learning Disabilities Research & Practice, (), –. Koike, I. (). Descriptive analysis of loss of English as a second language. Paper presented at the th JACET Conference, Fukuoka, Japan. Kolk, H. (). Does agrammatic speech constitute a regression to child language? A three-way comparison between agrammatic, child, and normal ellipsis. Brain and Language, (), –. doi: ./brln.. Konopka, A. E., & Bock, J. K. (). Helping syntax out: How much do words do? Talk presented at the th CUNY Human Sentence Processing Conference. Köpke, B. (). L’attrition de la première language chez le bilingue tardif: implications pour l’étude psycholinguistique du bilinguisme [First language attrition in late bilinguals: Implications for the psycholinguistic research of bilingualism.]. PhD thesis, Université de Toulouse-Le Mirail. Köpke, B. (). Effet du pays d’accueil sur le maintien de la langue: Le cas des immigrés d’origine allemande [Effect of the host country on language maintenance: The case of immigrants with German origin.]. Education et Societes Plurilingues, , –. Köpke, B. (). Quels changements linguistiques dans l’attrition de la L chez le bilingue tardif? [What are the linguistic changes in L attrition in late bilinguals?] Traveaux neuchâtelois de linguistique, /, –. Köpke, B. (). Activation thresholds and non-pathological first language attrition. In F. Fabbro (Ed.), Advances in the Neurolinguistics of Bilingualism (pp. –). Udine: Forum. Köpke, B. (a). Attrition. Special issue of Journal of Neurolinguistics, (), –. Köpke, B. (b). Neurolinguistic aspects of attrition. Journal of Neurolinguistics, (), –.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Köpke, B. (c). Attrition is not a unitary phenomenon: On different possible outcomes of language contact situations. In A. M. L. Suarez, F. Ramallo, & X. P. Rodiguez-Yanez (Eds.), Bilingual Socialization and Bilingual Language Acquisition (pp. –). Vigo: Universidade de Vigo. Köpke, B. (). Language attrition at the crossroads of brain, mind, and society. In B. Köpke, M. S. Schmid, M. Keijzer, & S. Dostert (Eds.), Language Attrition: Theoretical Perspectives (pp. –). Amsterdam: John Benjamins. Köpke, B. (). First language attrition: From bilingual to monolingual proficiency? In A. De Houwer & L. Ortega (Eds.), The Cambridge Handbook of Bilingualism (–). Cambridge: Cambridge University Press. Köpke, B., & Nespoulous, J.-L. (). First language attrition in production skills and metalinguistic abilities in German-English and German-French bilinguals. In T. Ammerlaan, M. Hulsen, H. Strating, & K. Yağmur (Eds.), Sociolinguistic and Psycholinguistic Perspectives on Maintenance and Loss of Minority Languages (pp. –). Münster: Waxmann. Köpke, B., & Schmid, M. S. (). Language attrition: The next phase. In M. Schmid, B. Köpke, M. Keijzer, & L. Weilemar (Eds.), First Language Attrition: Interdisciplinary Perspectives on Methodological Issues (pp. –). Amsterdam: John Benjamins. Köpke, B., & Schmid, M. S. (). L’attrition de la première langue en tant que phénomène psycholinguistique [Attrition of the first language as a psycholinguistic phenomenon.]. Language, Interaction and Acquisition, (), –. Köpke, B., Schmid, M. S., Keijzer, M. C., & Dostert, S. (Eds.) (). Language Attrition: Theoretical Perspectives. Amsterdam: John Benjamins. doi: ./sibil. Kostakopoulpu, D. (). The anatomy of civic integration. The Modern Law Review, (), –. Kotz, S. A. (). A critical review of ERP and fMRI evidence on L syntactic processing. Brain and Language, (), –. Kovács, Á. M., & Mehler, J. (). Cognitive gains in -month-old bilingual infants. Proceedings of the National Academy of Sciences, (), –. doi: ./pnas. Kövari, E., Gold, G., Herrmann, F. R., et al. (). Cortical microinfarcts and demyelination significantly affect cognition in brain aging. Stroke, (), –. Kowoll, M. A., Degen, C., Gladis, S., & Schroder, J. (). Neuropsychological profiles and verbal abilities in lifelong bilinguals with mild cognitive impairment and Alzheimer’s disease. Journal of Alzheimer’s Disease, , –. Kral, A. (). Auditory critical periods: A review from system’s perspective. Neuroscience, , –. doi: ./j.neuroscience... Krefeld, T. (). Einführung in die Migrationslinguistik [Introduction to migration linguistics]. Tübingen: Narr. Kritchevsky, M., Zouzounis, J., & Squire, L. R. (). Transient global amnesia and functional retrograde amnesia: Contrasting examples of episodic memory loss. Philosophical Transactions of the Royal Society B: Biological Sciences, (), –. Kroll, J. F., & Bialystok, E. (). Understanding the consequences of bilingualism for language processing and control. Journal of Cognitive Psychology, (), –. Kroll, J. F., & Gollan, T. H. (). Speech planning in two languages: What bilinguals tell us about language production. In V. Ferreira, M. Goldrick, & M. Miozzo (Eds.), The Oxford Handbook of Language Production (pp. –). Oxford: Oxford University Press. Kroll, J. F., & Stewart, E. (). Category interference in translation and picture naming: Evidence for asymmetric connections between bilingual memory representations. Journal of Memory and Language, , –. Kroll, J. F., & Sunderman, G. (). Cognitive processes in second language learners and bilinguals: The development of lexical and conceptual representations. In C. Doughty & M. Long (Eds.), The Handbook of Second Language Acquisition (pp. –). Oxford: Blackwell. Kroll, J. F., Bobb, S. C., & Wodniecka, Z. (). Language selectivity is the exception, not the rule: Arguments against a fixed locus of language selection in bilingual speech. Bilingualism: Language and Cognition, , –.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Kroll, J. F., Bobb, S. C., Misra, M., & Guo, T. (). Language selection in bilingual speech: Evidence for inhibitory processes. Acta Psychologica, , –. Kroll, J. F., Van Hell, J. G., Tokowicz, N., & Green, D. W. (). The Revised Hierarchical Model: A critical review and assessment. Bilingualism: Language and Cognition, (), –. Kroll, J. F., Dussias, P. E., Bice, K., & Perrotti, L. (). Bilingualism, mind, and brain. Annual Review of Linguistics, , –. Kroll, J. F., Gullifer, J. W., McClain, R., Rossi, E., & Martín, M. C. (). Selection and control in bilingual comprehension and production. In J. Schwieter (Ed.), Cambridge Handbook of Bilingual Processing (pp. –). Cambridge: Cambridge University Press. Kuhberg, H. (). Longitudinal L-attrition versus L-acquisition in three Turkish children: Empirical findings. Second Language Research, (), –. Kuhl, P. K., Williams, K. A., Lacerda, F., Stevens, K. N., & Lindblom, B. (). Linguistic experience alters phonetic perception in infants by months of age. Science, (), –. Kuhl, P. K., Conboy, B. T., Coffey-Corina, S., Padden, D., Rivera-Gaxiola, M., & Nelson, T. (). Phonetic learning as a pathway to language: New data and native language magnet theory expanded (NLM-e). Philosophical Transactions of the Royal Society B: Biological Sciences, (), –. doi: ./rstb.. Kuperberg, G. (). Neural mechanisms of language comprehension: Challenges to syntax. Brain Research, , –. Kupisch, T. (). Determiners in bilingual German-Italian children: What they tell us about the relation between language influence and language dominance. Bilingualism: Language and Cognition, (), –. doi: ./S Kupisch, T. (). Generic subjects in the Italian of early German-Italian bilinguals and German learners of Italian as a second language. Bilingualism: Language and Cognition, (), –. doi: ./S Kupisch, T. (). A new term for a better distinction? A view from the higher end of the proficiency scale. Theoretical Linguistics, (–), –. doi: ./tl-- Kupisch, T. (). Adjective placement in simultaneous bilinguals (German-Italian) and the concept of cross-linguistic overcorrection. Bilingualism: Language and Cognition, (), –. doi: ./S Kupisch, T., & Barton, D. (). How bilinguals deal with variation: Generic reference in adult German bilinguals. Studia Linguistica, (), –. doi: ./stul. Kupisch, T., & Lleó, C. (). Voice Onset Time in German-Italian simultaneous bilinguals: Evidence on crosslanguage influence and markedness. In M. Yavaş, M. Kehoe, & W. Cardoso (Eds.), RomanceGermanic Bilingual Phonology: Studies in Phonetics and Phonology (pp. –). London: Equinox. Kupisch, T., & Rothman, J. (). Interfaces in the acquisition of syntax. In S. Fischer, & C. Gabriel (Eds.), Manual of Grammatical Interfaces in Romance (pp. –). Berlin/Boston: De Gruyter. doi: ./- Kupisch, T., & Rothman, J. (). Terminology matters! Why difference is not incompleteness and how early child bilinguals are heritage speakers. International Journal of Bilingualism, (). pp. –. doi: ./ Kupisch, T., & Van de Weijer. J. (). The role of the childhood environment for language dominance: A study of adult simultaneous bilingual speakers of German and French. In C. Silva-Corvalán, & J. Treffers-Daller (Eds.), Language Dominance in Bilinguals: Issues of Measurement and Operationalization (pp. –). Cambridge: Cambridge University Press. Kupisch, T., Müller, N., & Cantone, K. F. (). Gender in monolingual and bilingual first language acquisition: Comparing Italian and French. Lingue e Linguaggio, (), –. doi: ./ Kupisch, T., Akpinar, D., & Stöhr, A. (). Gender assignment and gender agreement in adult bilingual and second language speakers of French. Linguistic Approaches to Bilingualism, (), –. doi: ./lab...kup Kupisch, T., Barton, D., Klaschik, E., Lein, T., Stangen, I., & Van de Weijer, J. (). Foreign accent in adult simultaneous bilinguals. Heritage Language Journal, (), –.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Kupisch, T., Barton, D., Lein, T., Schröder, D. J., Stangen, I., & Stöhr, A. (). Acquisition outcomes across domain in adult heritage speakers of French. Journal of French Language Studies, (), –. doi: ./S Kupisch, T., Belikova, A., Özcelik, Ö., Stangen, I., & White, L. (). Restrictions on definiteness in the grammars of German-Turkish heritage speakers. Linguistic Approaches to Bilingualism, (), –. doi: ./lab..kup Kutas, M., & Federmeier, K. D. (). Electrophysiology reveals semantic memory use in language comprehension. Trends in Cognitive Sciences, (), –. Kutas, M., & Federmeier, K. D. (). Thirty years and counting: Finding meaning in the N component of the event related brain potential (ERP). Annual Review of Psychology, , –. Kutas, M., & Hillyard, S. A. (). Brain potentials during reading reflect word expectancy and semantic association. Nature, , –. Kutas, M., & Van Petten, C. (). Psycholinguistics electrified. Event-related brain potential investigations. In M. A. Gernsbacher (Ed.), Handbook of Psycholinguistics (pp. –). San Diego, CA: Academic Press. Kutas, M., Lindamood, T. E., & Hillyard, S. A. (). Word expectancy and event-related brain potentials during sentence processing. In S. Kornblum & J. Requin (Eds.), Preparatory States and Processes (pp. –). Hillsdale, NJ: Lawrence Erlbaum. La Morgia, F. (). Bilingual first language acquisition: The nature of the weak language and the role of the input. PhD thesis, Dublin City University. Labov, W. (). The sociophonetic orientation of the language learner. In C. Celata & S. Calamai (Eds.), Advances in Sociophonetics (pp. –). Amsterdam: John Benjamins. Ladefoged, P., & Maddieson, I. (). The Sounds of the World’s Languages. Oxford: Blackwell. Laghzaoui, M. (). Emergent academic language at home and at school: A longitudinal study of -to -year-old Moroccan Berber children in the Netherlands. PhD thesis, Tilburg University. Laks, B., Calderone, B., & Celata, C. (). French liaison in the light of corpus phonology: From lexical information to patterns of usage variation. In D. K. Brentari & J. L. Lee (Eds.), Shaping Phonology: A Festschrift for John Goldsmith (pp. –). Chicago: University of Chicago Press. Laleko, O. (). A study of language attrition: American Russian in Minnesota. In K. Shaw, S. Wagner, and E. Yasui (Eds.), Symposium about Language and Society—Austin (SALSA), Vol. XV (pp. –). Austin, TX: University of Texas. Laleko, O. (). Restructuring of verbal aspect in Heritage Russian: Beyond lexicalization. International Journal of Language Studies, (), –. Lam, N. H., Schoffelen, J.-M., Uddén, J., Hultén, A., & Hagoort, P. (). Neural activity during sentence processing as reflected in theta, alpha, beta, and gamma oscillations. Neuroimage, , –. Lambert, R. D., & Freed, B. (). The Loss of Language Skills. Rowley, MA: Newbury House. Lambert, R. D., & Moore, S. J. (). Problem areas in the study of language attrition. In B. Weltens, K. de Bot, & T. van Els (Eds.), Language Attrition in Progress: Studies on Language Acquisition (pp. –). Dordrecht, NL: Foris Publications. Landau, I. (). The Locative Syntax of Experiencers. Cambridge, MA: MIT Press. Lang, B., & Davidson, L. (). Effects of exposure and vowel space distribution on phonetic drift: Evidence from American English learners of French. Language and Speech (), –. Lardiere, D. (). Case and tense in the ‘fossilized’ steady state. Second Language Research, , –. Lardiere, D. (). Dissociating syntax from morphology in a divergent L end-state grammar. Second Language Research, (), –. Lardiere, D. (). Mapping features to forms in second language acquisition. In J. Archibald (Ed.), Second Language Acquisition and Linguistic Theory (pp. –). Malden, MA: Blackwell. Lardiere, D. (). On morphological competence. In L. Dekydtspotter, R. Sprouse, & A. Liljestrand (Eds.), Proceedings of the th Generative Approaches to Second Language Acquisition Conference (GASLA ) (pp. –). Somerville, MA: Cascadilla Press. Lardiere, D. (). Ultimate Attainment in Second Language Acquisition. Mahweh, NJ: Lawrence Erlbaum.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Lardiere, D. (). Feature-assembly in second language acquisition. In J. Liceras, H. Zobl, & H. Goodluck (Eds.), The Role of Features in Second Language Acquisition (pp. –). Mahwah, NJ: Lawrence Erlbaum Associates. Lardiere, D. (). Some thoughts on the contrastive analysis of features in second language acquisition. Second Language Research, (), –. Larsen-Freeman, D. (). Chaos/complexity science and second language acquisition. Applied Linguistics, (), –. Larsen-Freeman, D. (). The emergence of complexity, fluency, and accuracy in the oral and written production of five Chinese learners of English. Applied Linguistics, (), –. Larsen-Freeman, D. (). Complex, dynamic systems: A new transdisciplinary theme for applied linguistics? Language Teaching, (), –. Larsen-Freeman, D. (). Introduction. Complexity Theory: A new way to think. Revista Brasileira de Linguística Aplicada [en linea], (). http://revele.com.veywww.redalyc.org/articulo.oa? id= (accessed August ). Larsen-Freeman, D. (). Ten ‘lessons’ from Complex Dynamic Systems Theory: What is on offer. In Z. Dörnyei, P. D. MacIntyre, & A. Henry (Eds.), Motivational Dynamics in Language Learning (pp. –). Bristol: Multilingual Matters. Larsen-Freeman, D. (). Foreword. In J. King (Ed.), The Dynamic Interplay between Context and the Language Learner (pp. xi–xiii). Basingstoke: Palgrave Macmillan. Larsen-Freeman, D. (). Complexity Theory: The lessons continue. In L. Ortega & Z. Han (Eds.), Complexity Theory and Language Development: In Celebration of Diane Larsen-Freeman (pp. –). Amsterdam: John Benjamins. Larsen-Freeman, D., & Cameron, L. (). Research methodology on language development from a Complex Systems perspective. The Modern Language Journal, (), –. Larsen-Freeman, D., Schmid, M. S., & Lowie, W. (). From structure to chaos. Twenty years of modeling bilingualism. In M. S. Schmid & W. Lowie (Eds.), Modeling Bilingualism: From Structure to Chaos. In Honor of Kees de Bot (pp. –). Amsterdam: John Benjamins. Larson-Hall, J. (). Our statistical intuitions may be misleading us: Why we need robust statistics. Language Teaching, (), –. doi: ./S Lattey, E., & Tracy, R. (). Well, I tell you, das war’n Zeiten [Well, I tell you, those were the days.]. In V. Hinnenkamp & K. Meng (Eds.), Sprachgrenzen überspringen: Sprachliche Hybridität und polykulturelles Selbstverständnis [Overcoming language borders: linguistic hybridity and polycultural identity.] (pp. –). Tübingen: Narr. Laufer, B. (). The influence of L on L collocational knowledge and on L lexical diversity in free written expression. In V. Cook (Ed.), Effects of the Second Language on the First (pp. –). Bristol: Multilingual Matters. Laufer, B., & Nation, P. (). Vocabulary size and use: Lexical richness in L written production. Applied Linguistics, , –. Lawton, D. M., Gasquoine, P. G., & Weimer, A. A. (). Age of dementia diagnosis in community dwelling bilingual and monolingual Hispanic Americans. Cortex, , –. Leal, T., Rothman, J., & Slabakova, R. (). A rare structure at the syntax-discourse interface: Heritage and Spanish-dominant native speakers weigh in. Language Acquisition, (), –. Leal Méndez, T., Rothman, J., & Slabakova, R. (). Discourse-sensitive clitic-doubled dislocations in heritage Spanish. Lingua, , –. Lee, S. W. S., Oyserman, D., & Bond, M. H. (). Am I doing better than you? That depends on whether you ask me in English or Chinese: Self-enhancement effects of language as a cultural mindset prime. Journal of Experimental Social Psychology, , –. Lee-Ellis, S. (). The elicited production of Korean relative clauses by heritage speakers. Studies in Second Language Acquisition, , –. Lefebvre, L., & Rinaldi, R. (). Langage et mémoire de travail dans la maladie d’Alzheimer: Quelles interrelations? [Language and working memory in Alzheimer’s Disease: How are they related?] In P. Gatignol & N. Joyeux (Eds.), Mémoire de travail [Working memory](pp. –). Isbergues: Ortho.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Leisiö, L. (). Genitive subjects and objects in the speech of Finland Russians. Journal of Slavic Linguistics, , –. Lemhöfer, K., Schriefers, H., & Indefrey, P. (). Idiosyncratic grammars: Syntactic processing in second language comprehension uses subjective feature representations. Journal of Cognitive Neuroscience, (), –. Lenneberg, E. H. (). Biological Foundations of Language. New York: Wiley. Lenzing, A. (). Exploring regularities and Dynamic Systems in L development. Language Learning, (), –. Leonard, L. (). Children with Specific Language Impairment. Cambridge, MA: MIT Press. Leonard, L. (a). Children with Specific Language Impairment, nd edn. Cambridge, MA: MIT Press. Leonard, L. (b). Children with specific language impairment and their contribution to the study of language development. Journal of Child Language, , –. Leonard, L. (). The primacy of priming in grammatical learning and intervention: A tutorial. Journal of Speech, Language, and Hearing Research, (), –. Leonard, L., Ellis Weismer, S., Miller, C., Francis, D. J., Tomblin, J. B., & Kail, R. V. (). Speed of processing, working memory and language impairment in children. Journal of Speech, Language and Hearing Research, , –. Leopold, W. (). A child’s learning of two languages. In E. Hatch (Ed.), Second Language Acquisition: A Book of Readings (pp. –). Rowley, MA: Newbury House. Letts, C., Edwards, S., Sinka, I., Schaefer, B., & Gibbons, W. (). Socio-economic status and language acquisition: Children’s performance on the new Reynell Developmental Language Scales. International Journal of Language & Communication Disorders, , –. Lev-Ari, S., & Peperkamp, S. (). Low inhibitory skill leads to non-native perception and production in bilinguals’ native language. Journal of Phonetics, (), –. Levelt, W. (). Speaking, from Intention to Articulation. Cambridge, MA: MIT Press. Levelt, W. (). A Blueprint of the speaker. In C. Brown & P. Hagoort (Eds.), The Neurocognition of Language (pp. –). Oxford: Oxford University Press. Levy, J. (). Policy implications/complications arising from native language attrition in U.S. ethnolinguistic minority groups. In R. D. Lambert & B. Freed (Eds.), The Loss of Language Skills (pp. –). Rowley, MA: Newbury House. Levy, B. J., & Anderson, M. C. (). Inhibitory processes and the control of memory retrieval. Trends in Cognitive Sciences, (), –. Levy, B. J., McVeigh, N. D., Marful, A., & Anderson, M. C. (). Inhibiting your native language: The role of retrieval-induced forgetting during second-language acquisition. Psychological Science, (), –. Lew-Williams, C., & Fernald, A. (). Young children learning Spanish make rapid use of grammatical gender in spoken word recognition. Psychological Science, , –. Li, C. N., & Thompson, S. A. (). The acquisition of tone in Mandarin speaking children. Journal of Child Language, , –. doi: ./S Li, P., & Farkas, I. (). A self-organizing connectionist model of bilingual processing. In R. Heredia & J. Altarriba (Eds.), Bilingual Sentence Processing (pp. –). Amsterdam: Elsevier. Li, P., & Grant, A. (). Second language learning success revealed by brain networks. Bilingualism: Language and Cognition, (), –. Li, P., Zhao, X., & MacWhinney, B. (). Dynamic self-organization and early lexical development in children. Cognitive Science, , –. Li, P., Legault, J., & Litcofsky, K. A. (). Neuroplasticity as a function of second language learning: anatomical changes in the human brain. Cortex, , –. doi: ./j. cortex... Li, W. (). The Bilingualism Reader. London: Routledge. Li, W. (). Interlanguage: The Abstract Level in Language Acquisition. Lewiston, NY: Edwin Mellen Press.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Libben, M. R., & Titone, D. A. (). Bilingual lexical access in context: Evidence from eye movements during reading. Journal of Experimental Psychology: Learning, Memory, and Cognition, (), –. Liberman, A. M., Harris, K. S., Hoffman, H. S., & Griffith, B. C. (). The discrimination of speech sounds within and across phoneme boundaries. Journal of Experimental Psychology, , –. doi: ./h Liceras, J. (). Syntax and stylistics: More on the pro-drop parameter. In S. Flynn & W. O’Neil (Eds.), Learnability and Second Languages: A Book of Readings (pp. –). Dordrecht: Foris. Liceras, J. (). On some properties of the pro-drop parameter: Looking for missing subjects in nonnative Spanish. In S. M. Gass & J. Schacter (Eds.), Perspectives on Second Language Acquisition (pp. –). Cambridge: Cambridge University Press. Liceras, J. M., Fernández Fuertes, R. F., Perales, S., Pérez-Tattam, R., & Spradlin, K. T. (). Gender and gender agreement in bilingual native and non-native grammars: A view from child and adult functional-lexical mixings. Lingua, (), –. Liceras, J. M., Zobl, H., & Goodluck, H. (Eds.) (). The Role of Formal Features in Second Language Acquisition. London: Routledge. Lightfoot, D. (). How New Languages Emerge. Cambridge: Cambridge University Press. Liljencrants, J., & Lindblom, B. (). Numerical simulation of vowel quality systems: The role of perceptual contrast. Language, , –. Linck, J. A., Kroll, J. F., & Sunderman, G. (). Losing access to the native language while immersed in a second language: Evidence for the role of inhibition in second-language learning. Psychological Science, , –. Lindblom, B. (). Phonetic universals in vowel systems. In J. J. Ohala & J. J. Jaeger (eds.), Experimental Phonology (pp. –). Orlando: Academic Press. Lindsay, S., & Gaskell, M. (). A complementary systems account of word learning in L and L. Language Learning, , –. Linnarud, M. (). Lexis in Composition: A Performance Analysis of Swedish Leaners’ Written English. Malmö: Liber Förlag. Liu, H., Bates, E., & Li, P. (). Sentence interpretation in bilingual speakers of English and Chinese. Applied Psycholinguistics, , –. Lleó, C. (). Bilingualism and child phonology. Oxford Handbooks Online. http://www.oxfordhandbooks.com/view/./oxfordhb/../oxfordhb--e-? rskey=HvYQiy&result= Lleó, C. (). Aspects of the phonology of Spanish as a heritage language. Bilingualism: Language and Cognition, (): –. doi: ./S Lleó, C., Kuchenbrandt, I., Kehoe, M., & Trujillo, C. (). Syllable final consonants in Spanish and German monolingual and bilingual acquisition. In N. Müller (Ed.), (Non)Vulnerable Domains in Bilingualism (pp. –). Amsterdam: John Benjamins. Lloyd-Smith, A., Gyllstad, H., & Kupisch, T. (). Global accent in German-dominant heritage speakers of Turkish. Linguistic Approaches to Bilingualism, (), –. doi: ./lab..llo Long, M. (). Maturational constraints on language development. Studies in Second Language Acquisition, (), –. https://doi.org/./S Longtin, C., & Meunier, F. (). Morphological decomposition in early visual word processing. Journal of Memory and Language, , –. Lopez Pousa, S., & de Pinedo, R. (). Enfermedad de Alzheimer. In S. Lopez Pousa, J. Vilalta Franch, & J. Linas Regia (Eds.), Manual de Demencias (pp. –). Barcelona: Prous Science. Lord, G. (). Second language acquisition and first language phonological modification. In J. B. de Garavito & E. Valenzuela (Eds.), Selected Proceedings of the th Hispanic Linguistics Symposium (pp. –). Somerville, MA: Cascadilla Proceedings Project. Lövdén, M., Wenger, E., Mårtensson, J., Lindenberger, U., & Bäckman, L. (). Structural brain plasticity in adult learning and development. Neuroscience & Biobehavioral Reviews, (), –.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Low, E. L., Grabe, E., & Nolan, F. (). Quantitative characterizations of speech rhythm: Syllabletiming in Singapore English. Language and Speech, (), –. Lowe, J. (). U.K. PM David Cameron: Migrants must learn to speak English, January. Newsweek. http://www.newsweek.com/uk-prime-minister-david-cameron-migrants-learn-speak-english- (accessed August ). Lowe, P., Jr., (). The U.S. Government’s foreign language attrition and maintenance experience. In R. D. Lambert & B. Freed (Eds.), The Loss of Language Skills (pp. –). Rowley, MA: Newbury House. Lowie, W., & Verspoor, M. (). Variability and variation in second language acquisition orders: A dynamic reevaluation. Language Learning, (), –. Lozano, C. (). The development of the syntax–discourse interface: Greek learners of Spanish. In V. Torrens & L. Escobar (Eds.), The Acquisition of Syntax in Romance Languages (pp. –). Amsterdam: John Benjamins. Lozano, C. (). Selective deficits at the syntax–discourse interface: Evidence from the CEDEL corpus. In N. Snape, Y. I. Leung, & M. Sharwood-Smith (Eds.), Representational Deficits in SLA: Studies in Honor of Roger Hawkins (pp. –). Amsterdam: John Benjamins Publishing Co. Lubińska, D. (). Förstaspråksattrition hos vuxna: Exemplet polsktalande i Sverige [Adult first language attrition: The case of Polish speakers in Sweden]. PhD thesis, Stockholm University. Luck, S. (). An Introduction to the Event-related Potential Technique. Cambridge, MA: MIT Press. Luderus, L. (). Language Choice and Language Separation in Bilingual Alzheimer Patients. Amsterdam: IFOTT. Luk, G., Anderson, J. A. E., Craik, F. I. M., Grady, C., & Bialystok, E. (). Distinct neural correlates for two types of inhibition in bilinguals: Response inhibition versus interference suppression. Brain and Cognition, (), –. doi: ./j.bandc... Luk, G., Bialystok, E., Craik, F. I., & Grady, C. L. (). Lifelong bilingualism maintains white matter integrity in older adults. The Journal of Neuroscience, (), –. Lybeck, K. (). Cultural identification and second language pronunciation of Americans in Norway. The Modern Language Journal, , –. Łyskawa, P., Maddeaux, R., Melara, E., & Nagy, N. (). Heritage speakers follow all the rules: Language contact and convergence in Polish devoicing. Heritage Language Journal, (), –. Mack, M. (). Phonetic transfer in a French-English bilingual child. In P. H. Nelde (Ed.), Language Attitudes and Language Conflict (pp. –). Bonn: Dümmler. MacKay, A. J., Connor, L. T., Albert, M. L., & Obler, L. K. (). Noun and verb retrieval in healthy aging. Journal of the International Neuropsychological Society, (), –. MacLeod, C. (). Bilingual episodic memory: Acquisition and forgetting. Journal of Verbal Learning and Verbal Behavior, (), –. MacLeod, C. M. (). Forgotten but not gone: Savings for pictures and words in long-term memory. Journal of Experimental Psychology: Learning, Memory, and Cognition, (), –. MacWhinney, B. (). A unified model of language acquisition. In J. Kroll & A. de Groot (Eds.), Handbook of Bilingualism: Psycholinguistic Approaches (pp. –). Oxford: Oxford University Press. MacWhinney, B. (). Item-based patterns in early syntactic development. In T. Herbst, H.-J. Schmid, & S. Faulhaber (Eds.), Constructions Collocations Patterns (pp. –). Berlin: de Gruyter Mouton. MacWhinney, B. (a). Introduction: Language emergence. In B. MacWhinney & W. O’Grady (Eds.), Handbook of Language Emergence (pp. –). New York: Wiley. MacWhinney, B. (b). Multidimensional SLA. In S. Eskilde & T. Cadierno (Eds.), Usage-based Perspectives on Second Language Learning (pp. –). New York: Oxford University Press.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
MacWhinney, B. (). A unified model of first and second language learning. In M. Hickmann, E. Veneziano, & H. Jisa (Eds.), Sources of Variation in First Language Acquisition. Languages, contexts, and learners (pp. –). Amsterdam: John Benjamins. MacWhinney, B., & Bates, E. (Eds.) (). The Crosslinguistic Study of Sentence Processing. New York: Cambridge University Press. MacWhinney, B., Malchukov, A., & Moravcsik, E. (Eds.) (). Competing Motivations in Grammar and Usage. New York: Oxford University Press. Madden, D. J., Bennett, I. J., & Song, A. W. (). Cerebral white matter integrity and cognitive aging: Contributions from diffusion tensor imaging. Neuropsychology Review, (), –. Maher, J. (). A crosslinguistic study of language contact and language attrition. In H. W. Seliger & R. M. Vago (Eds.), First Language Attrition (pp. –). Cambridge: Cambridge University Press. Major, R. C. (). Losing English as a first language. The Modern Language Journal, (), –. Major, R. C. (). Sociolinguistic factors in loss and acquisition of phonology. In K. Hyltenstam & A. Viberg (Eds.), Progression and Regression in Language: Sociocultural, Neuropsychological, and Linguistic Perspectives (pp. –). Cambridge: Cambridge University Press. Major, R. C. (). L acquisition, L loss, and the critical period hypothesis. In A. James & J. Leather (Eds.), Second-Language Speech: Structure and Process (pp. –). Berlin: Mouton de Gruyter. Major, R. C. (). The phonology of the L user. In V. Cook (Ed.), Portraits of the L User (pp. –). Bristol: Multilingual Matters. Major, R. C. (). First language attrition in foreign accent perception. International Journal of Bilingualism, , –. Malt, B. C., & Sloman, S. A. (). Linguistic diversity and object naming by non-native speakers of English. Bilingualism: Language and Cognition, , –. Malt, B. C., Li, P., Pavlenko, A., Zhu, H., & Ameel, E. (). Bidirectional lexical interaction in late immersed Mandarin-English bilinguals. Journal of Memory and Language, , –. Manzini, R. M., & Wexler, K. (). Parameters, binding theory, and learnability. Linguistic Inquiry, (), –. Marchman, V., Martinez-Sussman, C., & Dale, P. S. (). The language-specific nature of grammatical development: Evidence from bilingual language learners. Developmental Science, (), –. doi: ./j.-...x Margaza, P., & Bel, A. (). Null subjects at the syntax–pragmatics interface: Evidence from Spanish interlanguage of Greek speakers. In M. G. O’Brien, C. Shea, & J. Archibald (Eds.), Proceedings of GASLA (pp. –). Somerville, MA: Cascadilla Press. Marian, V., & Neisser, U. (). Language-dependent recall of autobiographical memories. Journal of Experimental Psychology: General, , –. Marian, V., & Spivey, M. (). Competing activation in bilingual language processing: Within-and between-language competition. Bilingualism: Language and Cognition, , –. Marslen-Wilson, W. D., Tyler, L. K., Waksler, R., & Older, L. (). Morphology and meaning in the English mental lexicon. Psychological Review, , –. Martensson, J., Eriksson, J., Bodammer, N. C., et al. (). Growth of language-related brain areas after foreign language learning. Neuroimage, (), –. doi: ./j.neuroimage... Matras, Y. (). Utterance modifiers and universals of grammatical borrowing. Linguistics, , –. Matras, Y. (). Language Contact. Cambridge: Cambridge University Press. Matras, Y. (). Contact, convergence and typology. In R. Hickey (Ed.), The Handbook of Language Contact (pp. –). Chichester: Wiley-Blackwell. Mattock, K., Molnar, M., Polka, L., & Burnham, D. (). The developmental course of lexical tone perception in the first year of life. Cognition, (), –. doi: ./j.cognition... Maus, B., Van Breukelen, G. J., Goebel, R., & Berger, M. P. (). Optimal design of multi-subject blocked fMRI experiments. Neuroimage, (), –.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Mayberry, R. I. (). First-language acquisition after childhood differs from second-language acquisition: The case of American Sign Language. Journal of Speech, Language, and Hearing Research: JSLHR, (), –. doi: ./jshr.. Mayberry, R. I. (). When timing is everything: Age of first-language acquisition effects on secondlanguage learning. Applied Psycholinguistics, (), –. doi: ./S Mayberry, R. I. (). Early language acquisition and adult language ability: What sign language reveals about the critical period for language. In M. Marschark & P. Spencer (Eds.), Oxford Handbook of Deaf Studies, Language, and Education, Vol. (pp. –). Oxford: Oxford University Press. Mayberry, R. I., Chen, J.-K., Witcher, P., & Klein, D. (). Age of acquisition effects on the functional organization of language in the adult brain. Brain and Language, (), –. doi: ./j.bandl... Maye, J., Werker, J. F., & Gerken, L. (). Infant sensitivity to distributional information can affect phonetic discrimination. Cognition, (), B–B. doi: ./S-()- Mayer, E., & Sánchez, L. (). Feature variability in the bilingual-monolingual continuum: Clitics in Bilingual Quechua-Spanish, Bilingual Shipibo-Spanish and in Monolingual Limeño Spanish contact varieties. International Journal of Bilingual Education and Bilingualism. doi: ./ .. Mayer, M. (). Frog, Where Are You? New York: Penguin Books. Mayer, M. (). A Boy, a Dog and a Frog. New York: Dial Books. Mayerthaler, W. (). System-independent morphological naturalness. In W. Dressier (Ed.), Leitmotifs in Natural Morphology (pp. –). Amsterdam: John Benjamins. Mayo, A. Y., & Leseman, P. P. M. (). Dagelijkse Informele Educatie: Gezinsvragenlijst [Daily informal education: Family questionnaire.]. Utrecht: Langeveld Instituut. Mayr, R., Price, S., & Mennen, I. (). First language attrition in the speech of Dutch-English bilinguals: The case of monozygotic twin sisters. Bilingualism: Language and Cognition, (), –. Mayr, R., Morris, J., Mennen, I., & Williams, D. (). Disentangling the effects of long-term language contact and individual bilingualism: The case of monophthongs in Welsh and English. International Journal of Bilingualism, . doi: ./ Mayr, R., López-Bueno, L., & Tomé Lourido, G. (). The role of early experience and continued language use in bilingual speech production: A study of Galician and Spanish mid vowels by Galician-Spanish bilinguals. Journal of Phonetics, , –. McClosky, M., & Cohen, N. J. (). Catastrophic interference in connectionist networks: The sequential learning problem. Psychology of Learning and Motivation, , –. McCormack, B. (). The maintenance and loss of reflexive anaphors in L English. PhD thesis, University of Durham. McDonald, H. G., & Balgopal, P. R. (). Conflicts of American immigrants: Assimilate or retain ethnic identity. Migration World Magazine, (), –. McDonald, J. (). The acquisition of cue-category mappings. In B. MacWhinney & E. Bates (Eds.), The Crosslinguistic Study of Language Processing (pp. –). New York: Cambridge University Press. McDonald, J., & Heilenman, K. (). Determinants of cue strength in adult first and second language speakers of French. Applied Psycholinguistics, , –. McGaugh, J. L. (). Memory—A century of consolidation. Science, (), –. doi: ./ science... McKee, G., Malvern, D., & Richards, B. (). Measuring vocabulary diversity using dedicated software. Literary and Linguistic Computing, (), –. McKhann, G. M., Knopman, D. S., Chertkow, H., et al. (). The diagnosis of dementia due to Alzheimer’s disease: Recommendations from the National Institute on Aging and the Alzheimer’s Association workgroup. Alzheimer’s & Dementia, (), –.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
McLaughlin, J., Osterhout, L., & Kim, A. (). Neural correlates of second-language word learning: Minimal instruction produces rapid change. Nature Neuroscience, , –. McLaughlin, J., Tanner, D., Pitkänen, I., et al. (). Brain potentials reveal discrete stages of L grammatical learning. Language Learning, (s), –. McMurray, B., Horst, J. S., & Samuelson, L. K. (). Word learning emerges from the interaction of online referent selection and slow associative learning. Psychological Review, (), –. Meara, P. (). Matrix models of vocabulary acquisition. AILA Review, , –. http://www.aila. info/download/publications/review/AILA.pdf Meara, P. (). Modelling vocabulary loss. Applied Linguistics, , –. Meara, P. (). Reactivating a dormant vocabulary. In S. H. Foster-Cohen, M. P. Garcia Mayo, & J. Cenoz (Eds.), EUROSLA Yearbook, Vol. (pp. –). Amsterdam: John Benjamins. Mechelli, A., Crinion, J. T., Noppeney, U., et al. (). Neurolinguistics: Structural plasticity in the bilingual brain. Nature, (), . doi: ./a Mecklinger, A., Schriefers, H., Steinhauer, K., & Friederici, A. D. (). Processing relative clauses varying on syntactic and semantic dimensions: An analysis with event-related potentials. Memory and Cognition, (), –. Mehotcheva, T. H. (). -month non-use of a language: A case study. A longitudinal study on the effects of non-use of one of the languages (Spanish) of a German L multilingual. Master’s thesis, Pompeu Fabra University. Mehotcheva, T. H. (). After the fiesta is over: Foreign language attrition of Spanish in Dutch and German Erasmus students. PhD thesis, University of Groningen. Meisel, J. (). Word order and case marking in early child language. Evidence from simultaneous acquisition of two first languages: French and German. Linguistics, (), –. doi: ./ ling.... Meisel, J. (Ed.) (). Two First Languages. Dordrecht: Foris. Meisel, J. (Ed.) (). Bilingual First Language Acquisition: French and German Grammatical Development. Amsterdam: John Benjamins. Meisel, J. (). The bilingual child. In T. K. Bhatia, & W. C. Ritchie (Eds.), The Handbook of Bilingualism (pp. –). Oxford: Blackwell. Meisel, J. (a). On autonomous syntactic development in multiple first language acquisition. In H. Caunt-Nulton, S. Kulatilake, & I. Woo (Eds.), st BUCLD Proceedings (pp. –). Somerville: Cascadilla Press. Meisel, J. (b). The weaker language in early child bilingualism: Acquiring a first language as a second language? Applied Psycholinguistics, (), –. doi: ./S Meisel, J. (). Second language acquisition in early childhood. Zeitschrift für Sprachwissenschaft, , –. Meisel, J. (). First and Second Language Acquisition: Parallels and Difference. Cambridge: Cambridge University Press. Meisel, J. (). Heritage language learners: Unprecedented opportunities for the study of grammars and their development? Theoretical Linguistics, (–), –. https://doi.org/./tl-- Meisel, J. (). Heritage language learners: Incomplete acquisition of grammar in early childhood. In A. Enrique-Arias, M. J. Gutiérrez, A. Landa, & F. Ocampo (Eds.), Perspectives in the Study of Spanish Language Variation (pp. –). Santiago de Compostela: Universidade de Santiago de Compostela. Meisel, J. (). On first language attrition in second language learners. Linguistic Approaches to Bilingualism, (), –. Meisel, J. (). Early child second language acquisition: French gender in German. Bilingualism: Language and Cognition, (), –. doi: ./S Melinger, A., Branigan, H. P., & Pickering, M. J. (). Parallel processing in language production. Language, Cognition, and Neuroscience, , –. Mendes, C. G., & Iribarren, I. C. (). Fixação do parâmetro do sujeito nulo na aquisição do português europeu por hispanofalantes [Fixation of the null subject parameter in the Portuguese of Hispanophones.]. In M. Lobo & M. A. Coutinho (Eds.), XXII Encontro Nacional da Associação
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Portuguesa de Linguística: Textos selecionados [Meeting of the Portuguese National Association of Linguistics: Selected texts.] (pp. –). Lisbon: Associação Portuguesa de Linguística. Mendez, M. F., Perymann, K. M., Ponton, M. O., & Cummings, J. (). Bilingualism and dementia. Journal of Neuropsychiatry and Clinical Neurosciences, (), –. Mennen, I. (). Bi-directional interference in the intonation of Dutch speakers of Greek. Journal of Phonetics, , –. Mennen, I., & Stansfield, J. (). Speech and language therapy service delivery for bilingual children: A survey of three cities in Great Britain. International Journal of Language & Communication Disorders, , –. Mennen, I., Scobbie, J. M., Leeuw, E. de, Schaeffler, S., & Schaeffler, F. (). Measuring languagespecific phonetic settings. Second Language Research, (), –. doi: ./ Menyuk, P., & Brisk, M. E. (). Language Development and Education: Children with Varying Language Experience. New York: Palgrave. Merino, B. (). Language loss in bilingual Chicano children. Journal of Applied Developmental Psychology, , –. doi.org/./-()- Meulman, N., Stowe, L. A., Sprenger, S. A., Bresser, M., & Schmid, M. S. (). An ERP study on L syntax processing: When do learners fail? Frontiers in Psychology, (), –. Meulman, N., Wieling, M., Sprenger, S. A., Stowe, L. A., & Schmid, M. S. (). Age effects in L grammar processing as revealed by ERPs and how (not) to study them. PloS one, (), e. Meuter, R., & Allport, A. (). Bilingual language switching in naming: Asymmetrical costs of language selection. Journal of Memory and Language, , –. Miccio, A. W., Hammer, C. S., & Rodríguez, B. (). Code-switching and language disorders in bilingual children. In J. Toribio & B. Bullock (Eds.), The Cambridge Handbook of Linguistic Codeswitching (pp. –). Cambridge: Cambridge University Press. Mikhaylova, A. (). Aspectual knowledge of high proficiency L and heritage speakers of Russian. Heritage Language Journal, (), –. Miličević, M., & Kraš, T. (). Translation between L acquisition and L attrition: Anaphora resolution in Italian by English-Italian trainee translators. Applied Linguistics, (), –. Miller, J. (). The sampling distribution of d’. Perception and Psychophysics, , –. Misra, M., Guo, T., Bobb, S. C., & Kroll, J. F. (). When bilinguals choose a single word to speak: Electrophysiological evidence for inhibition of the native language. Journal of Memory and Language, , –. Mohades, S. G., Struys, E., Van Schuerbeek, P., Mondt, K., Van De Craen, P., & Luypaert, R. (). DTI reveals structural differences in white matter tracts between bilingual and monolingual children. Brain Research, , –. doi: ./j.brainres... Mohades, S. G., Van Schuerbeek, P., Rosseel, Y., Van De Craen, P., Luypaert, R., & Baeken, C. (). White-matter development is different in bilingual and monolingual children: A longitudinal DTI study. PLoS One, (), e. doi: ./journal.pone. Molinaro, N., Barber, H. A., & Carreiras, M. (). Grammatical agreement processing in reading: ERP findings and future directions. Cortex, (), –. Molinaro, N., Vespignani, F., Zamparelli, R., & Job, R. (). Why brother and sister are not just siblings: Repair processes in agreement computation. Journal of Memory and Language, (), –. Molnar, M., Polka, L., Baum, S., & Steinhauer, K. (). Learning two languages from birth shapes preattentive processing of vowel categories: Electrophysiological correlates of vowel discrimination in monolinguals and simultaneous bilinguals. Bilingualism: Language and Cognition, (), –. Montalbetti, M. (). After binding. PhD thesis, Massachusetts Institute of Technology. Montoya-Abat, B. (). Phonological features of attrition. The shift from Catalan to Spanish in Alicante. In J. N. Stanford & D. R. Preston (Eds.), Variation in Indigenous Minority Languages (pp. –). Amsterdam: John Benjamins. Montrul, S. (). Clitic-doubled dative subjects in Spanish. In Zagona, K. (Ed.), Current Issues in Linguistic Theory (pp. –). Amsterdam: John Benjamins.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Montrul, S. (). Incomplete acquisition and attrition of Spanish Tense/Aspect distinctions in adult bilinguals. Bilingualism: Language and Cognition, (), –. Montrul, S. (a). Subject and object expression in Spanish heritage speakers: A case of morphosyntactic convergence. Bilingualism: Language and Cognition, (), –. Montrul, S. (b). Convergent outcomes in L acquisition and L loss. In M. S. Schmid, B. Köpke, M. Keijzer, & L. Weilemar (Eds.), First Language Attrition: Interdisciplinary Perspectives on Methodological Issues (pp. –). Amsterdam: John Benjamins. Montrul, S. (). On the bilingual competence of Spanish heritage speakers: Syntax, lexical semantics and processing. International Journal of Bilingualism, , –. Montrul, S. (). Incomplete Acquisition in Bilingualism: Re-examining the Age Factor. Amsterdam: Benjamins. Montrul, S. (). Incomplete acquisition of tense-aspect and mood in Spanish heritage speakers. The International Journal of Bilingualism, (), –. Montrul, S. (a). Multiple interfaces and incomplete acquisition. Lingua, (), –. Montrul, S. (b). First language retention and attrition in an adult Guatemalan adoptee. Language, Interaction and Acquisition, (), –. doi: ./lia...mon Montrul, S. (). El bilingüismo en el mundo hispanohablante [Bilingualism in the Spanish-speaking world.]. Malden, MA: Wiley-Blackwell. Montrul, S. (a). The Acquisition of Heritage Languages. Cambridge: Cambridge University Press. Montrul, S. (b). Age of onset of bilingualism effects and availability of input in first language attrition. In E. Nicoladis & S. Montanari (Eds.), Bilingualism Across the Lifespan: Factors Moderating Language Proficiency (pp. –). Berlin: Mouton de Gruyter. Montrul, S., Bhatt, R., & Girju, R. (). Differential object marking in Spanish, Hindi and Romanian as heritage languages. Language, (), –. Montrul, S., & Bowles, M. (). Back to basics: Differential object marking under incomplete acquisition in Spanish heritage speakers. Bilingualism: Language and Cognition, , –. Montrul, S., & Sánchez-Walker, N. (). Differential object marking in child and adult Spanish heritage speakers. Language Acquisition, (), –. doi: ./.. Montrul, S., & Sánchez-Walker, N. (). Subject expression in bilingual school-age children in the United States. In A. M. Carvalho, R. Orozco, & N. L. Shin (Eds.), Subject Pronoun Expression in Spanish: A Cross-dialectal Perspective (pp. –). Washington, DC, Georgetown University Press. Montrul, S., & Slabakova, R. (). Competence similarities between native and near-native speakers: An investigation of the Preterite/Imperfect contrast in Spanish. Studies in Second Language Acquisition, , –. Montrul, S., Foote, R., & Perpiñán, S. (a). Gender agreement in adult second language learners and Spanish heritage speakers: The effects of age and context of acquisition. Language Learning, , –. Montrul, S., Foote, R., & Perpiñán, S. (b). Knowledge of wh-movement in Spanish L learners and heritage Speakers. In J. Bruhn de Garavito & E. Valenzuela (Eds.), Selected Proceedings of the th Hispanic Linguistics Symposium (pp. –). Somerville, MA: Cascadilla Press. Moorcroft, R., & Gardner, R. C. (). Linguistic factors in second language loss. Language Learning, (), –. Morales, L., Paolieri, D., & Bajo, T. (). Grammatical gender inhibition in bilinguals. Frontiers in Psychology, , . doi: ./fpsyg.. Morford, J. P., Wilkinson, E., Villwock, A., Piñar, P., & Kroll, J. F. (). When deaf signers read English: Do written words activate their sign translations? Cognition, (), –. Morgan-Short, K. (). Electrophysiological approaches to understanding second language acquisition: A field reaching its potential. Annual Review of Applied Linguistics, , –. Morgan-Short, K., Steinhauer, K., Sanz, C., & Ullman, M. T. (). Explicit and implicit second language training differentially affect the achievement of native-like brain activation patterns. Journal of Cognitive Neuroscience, (), –.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Morrison, J. H., & Baxter, M. G. (). The ageing cortical synapse: Hallmarks and implications for cognitive decline. Nature Reviews Neuroscience, (), –. Müller, N. (). Transfer in bilingual first language acquisition. Bilingualism: Language and Cognition, (), –. doi: ./S Müller, N., & Hulk, A. (). Crosslinguistic influence in bilingual language acquisition: Italian and French as recipient languages. Bilingualism: Language and Cognition, (), –. doi: ./ S Muñoz, C. (). Symmetries and asymmetries of age effects in naturalistic and instructed L learning. Applied Linguistics, (), –. doi: ./applin/amm Muñoz, C., & Singleton, D. (). A critical review of age-related research on L ultimate attainment. Language Teaching, (), –. doi: ./S Munro, M. J., Derwing, T. M., & Flege, J. E. (). Canadians in Alabama: A perceptual study of dialect acquisition in adults. Journal of Phonetics, (), –. doi: ./jpho.. Murayama, K., Miyatsu, T., Buchli, D., & Storm, B. C. (). Forgetting as a consequence of retrieval: A meta-analytic review of retrieval-induced forgetting. Psychological Bulletin, (), –. Murphy, V. A., & Pine, K. J. (). L influence on L linguistic representations. In V. Cook (Ed.), Effects of the Second Language on the First (pp. –). Bristol: Multilingual Matters. Murre, J. M. J., & Dros, J. (). Replication and analysis of Ebbinghaus’ forgetting curve. Plos One, (), e. Murtagh, L. (). Retention and attrition of Irish as a second language: A longitudinal study of general and communicative proficiency in Irish among second level school leavers and the influence of instructional background, language use and attitude/motivation variables. PhD thesis, University of Groningen. Murtagh, L. (). Second language attrition: Theory, research and challenges. In M. S. Schmid, & W. Lowie (Eds.), Modelling Bilingualism—From Structure to Chaos. Amsterdam: John Benjamins. Murtagh, L., & Van der Slik, F. (). Retention of Irish skills: A longitudinal study of a schoolacquired L. International Journal of Bilingualism, (), –. Muysken, P. (). Creole tense/mood/aspect systems: The unmarked case? in P. Muysken (Ed.), Generative Studies on Creole Languages. Dordrecht: Foris, –. Myers-Scotton, C. (). Contact Linguistics: Bilingual Encounters and Grammatical Outcomes. Cambridge: Cambridge University Press. Myers-Scotton, C. (). Supporting a differential access hypothesis: Code switching and other contact data. In J. F. Kroll & A. M. B. DeGroot (Eds.), Handbook of Bilingualism: Psycholinguistic Approaches (pp. –). New York: Oxford University Press. Myers-Scotton, C., & Jake, J. (). Four types of morphemes: Evidence from aphasia, codeswitching, and second language acquisition. Linguistics, (), –. Myers-Scotton, C., & Jake, J. (). Revisiting the -M model: Codeswitching and morpheme election at the abstract level. International Journal of Bilingualism, (), –. Nadel, L., Samsonovich, A., Ryan, L., & Moscovitch, M. (). Multiple trace theory of human memory: Computational, neuroimaging, and neuropsychological results. Hippocampus, (), –. Nagasawa, S. (). The effects of initial achievement, learning experience and classroom instruction on adult attrition/retention of L Japanese. In P. Robinson (Ed.), Representation and Process: Proceedings of the rd Pacific Second Language Research Forum (PacSLRF’), Vol. (pp. –). Tokyo: Aoyama University. Nagy, N. (). A multilingual corpus to explore geographic variation. Rassegna Italiana di Linguistica Applicata, (–), –. Nagy, N., & Kochetov, A. (). Voice onset time across the generations. A cross-linguistic study of contact-induced change. In P. Siemund, I. Gogolin, M. E. Schulz, & J. Davydova (Eds.), Multilingualism and Language Diversity in Urban Areas: Acquisition, Identities, Space, Education (pp.–). Amsterdam: John Benjamins Publishing Company.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Nagy, N., Aghdasi, N., Denis, D., & Motut, A. (). Null subjects in heritage languages: Contact effects in a crosslinguistic context. University of Pennsylvania Working Papers in Linguistics, (), –. Namjoshi, J., Tremblay, A., Spinelli, E., et al. (). Speech segmentation is adaptive even in adulthood: Role of the linguistic environment. In The Scottish Consortium for ICPhS (Ed.), Proceedings of the th International Congress of Phonetic Sciences (paper number ). Glasgow: University of Glasgow. Negrisanu, R. M. (). Aspects of first language attrition: A case study of German immigrants in East Tennessee. PhD thesis, University of Tennessee. Neisser, U. (). Interpreting Harry Bahrick’s discovery: What confers immunity against forgetting? Journal of Experimental Psychology: General, (), –. Nelson, R. (). Expanding the role of connectionism in SLA theory. Language Learning, (), –. Nelson, T. (). Detecting small amounts of information in memory: Savings for nonrecognized items. Journal of Experimental Psychology: Human Learning and Memory, , –. Nesdale, D. (). Acculturation attitudes and the ethnic and host-country identification of immigrants. Journal of Applied Social Psychology, (), –. Newman, R. S., & German, D. J. (). Life span effects of lexical factors on oral naming. Language and Speech, (), –. Newport, E. L. (). Maturational constraints on language learning. Cognitive Science, , –. doi: ./scog_ Newport, E. L., Bavelier, D., & Neville, H. J. (). Critical thinking about critical periods: Perspectives on a critical period for language acquisition. In E. Dupoux (Ed.), Language, Brain and Cognitive Development (pp. –). Cambridge, MA: MIT Press. Nicholas, M., Obler, L., Albert, M., & Goodglass, H. (). Lexical retrieval in healthy aging. Cortex, (), –. Nickels, S., & Steinhauer, K. (). Semantic blocking revisited: Investigating individual variability in the elicitation of N and P components using Generalized Additive Mixed Models (GAMMs). Poster presented at the Annual Meeting of the Society for the Neurobiology of Language, Chicago, – October. Nickels, S., & Steinhauer, K. (in preparation). ‘Semantic blocking’ solved at last: Investigating individual variability in the elicitation of N and P components using Generalized Additive Mixed Models (GAMMs). Nicoladis, E., & Grabois, H. (). Learning English and losing Chinese: A case study of a child adopted from China. International Journal of Bilingualism, (), –. Niparko, J., Tobey, E., Thal, D., et al. (). Spoken language development in children following cochlear implantation. JAMA, (), –. doi: ./jama.. Nippold, M. (). Later Language Development: The school-age and adolescent years. Austin, TX: Pro-ed. Nippold, M. (). Research on later language development: International perspectives. In R. Berman (Ed.), Language Development Across Childhood and Adolescence (pp. –). Amsterdam: John Benjamins. Nittrouer, S., & Burton, L. T. (). The role of early language experience in the development of speech perception and phonological processing abilities: Evidence from -year-olds with histories of otitis media with effusion and low socioeconomic status. Journal of Communication Disorders, (), –. doi: ./j.jcomdis... Nodari, R., Celata, C., & Nagy, N. (). Immigrants’ speech. Is phonetic attrition a necessary precondition for phonological attrition to occur? Paper presented at rd International Conference on Language Attrition (ICLA), Colchester, July. Nodari, R., Celata, C., & Nagy, N. (). Socio-indexical phonetic features in the heritage language context: Voiceless stop aspiration in the Calabrian community in Toronto. Journal of Phonetics, , –.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Norman, G. (). Likert scales, levels of measurement and the ‘laws’ of statistics. Advances in Health Sciences Education, (), –. Norrman, G., & Bylund, E. (). The irreversibility of sensitive period effects in language development: Evidence from second language acquisition in international adoptees. Developmental Science, (), –. doi: ./desc. Norrman, G., Hyltenstam, K., & Bylund, E. (). Long-term language development in international adoptees. In F. Genesee & A. Delcenserie (Eds.), Starting Over—The Language Development of Internationally-adopted Children (pp. –). Amsterdam: John Benjamins. Nosofsky, R. (). Attention, similarity, and the identification-categorization relationship. Journal of Experimental Psychology: General, , –. Nutley, S., Powell, A., & Davies, H. (for Alliance for Useful Evidence) (). What counts as good evidence? https://www.allianceusefulevidence.org/assets/What-Counts-as-Good-EvidenceWEB.pdf Nuwer, M. R., Comi, G., Emerson, R., et al. (). IFCN standards for digital recording of clinical EEG. Electroencephalography and Clinical Neurophysiology, (), –. O’Regan, J. K., & Lévy-Schoen, A. (). Eye-movement strategy and tactics in word recognition and reading. In M. Coltheart (Ed.), Attention and Performance, Vol. (pp. –). Hillsdale, NJ: Lawrence Erlbaum and Associates Inc. O’Grady, W. (a). Syntactic Carpentry: An Emergentist Approach to Syntax. Mahwah, NJ: Erlbaum. O’Grady, W. (b). Emergentism. In P. Hogan (Ed.), The Cambridge Encyclopedia of the Language Sciences (pp. –). Cambridge: Cambridge University Press. O’Grady, W., Lee, M., & Choo, M. (). The acquisition of relative clauses by heritage and nonheritage learners of Korean as a second language: A comparative study. Journal of Korean Language Education, , –. O’Grady, W., Kwak, H.-Y., Lee, M., & Lee, O.-S. (). An emergentist perspective on partial language acquisition. Studies in Second Language Acquisition, , –. Obler, L. K. (). Neurolinguistic aspects of language loss as they pertain to second language attrition. In R. D. Lambert, & B. Freed (Eds.), The Loss of Language Skills (pp. –). Rowley, MA: Newbury House. Obler, L. K. (). Language and brain dysfunction in dementia. In S. Segolowitz (Ed.), Language Functions and Brain Organization (pp. –). New York: Academic Press. Obler, L. K., & Albert, M. L. (). Language skills across adulthood. In J. E. Birren & K. W. Schaie (Eds.), Handbook of the Psychology of Aging (pp. –). New York: Van Nostrand Reinhold Co. Obler, L. K., & Mahecha, N. R. (). First language loss in bilingual and polyglot aphasics. In H. W. Seliger & R. M. Vago (Eds.), First Language Attrition (pp. –). Cambridge: Cambridge University Press. Obler, L. K., De Santi, S., & Goldberger, J. (). Bilingual dementia: Pragmatic breakdown. In R. Lubinski (Ed.), Dementia and Communication (pp. –). San Diego, CA: Singular Publishing Group. Obler, L. K., Rykhlevskaia, E., Schnyer, D., et al. (). Bilateral brain regions associated with naming in older adults. Brain and Language, (), –. Oh, G. E., Guion-Anderson, S., Aoyama, K., Flege, J. E., Akahane-Yamada, R., & Yamada, T. (). A one-year longitudinal study of English and Japanese vowel production by Japanese adults and children in an English-speaking setting. Journal of Phonetics, (), –. Oh, J. S., Au, T. K., & Jun, S.-A. (). Benefits of childhood language experience for adult L learners’ phonology. In B. Skarabela, S. Fish, & A. H.-J. Do (Eds.), Proceedings of the th Annual Boston University Conference on Language Development, Vol. (pp. –). Somerville, MA: Cascadilla Press. Oh, J. S., Jun, S.-A., Knightly, L. M., & Au, T. K. (). Holding on to childhood language memory. Cognition, (), B–B. doi: ./s-()- Oh, J., Au, T. K., & Jun, S.-A. (). The nature of childhood language memory: Korean adoptees learning Korean as adults. In J. Chandlee, M. Franchini, S. Lord, & G.-M. Rheiner (Eds.),
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Proceedings of the rd Annual Boston University Conference on Language Development, Vol. (pp. –). Somerville, MA: Cascadilla Press. Oh, J. S., Au, T. K., & Jun, S.-A. (). Early childhood language memory in the speech perception of international adoptees. Journal of Child Language, (), –. Oller, D. K., Pearson, B. Z., & Cobo-Lewis, A. B. (). Profile effects in early bilingual language and literacy. Applied Psycholinguistics, , –. Olshtain, E. (). The attrition of English as a second language with speakers of Hebrew. In B. Weltens, K. de Bot, & T. J. M. van Els (Eds.), Language Attrition in Progress (pp. –). Dordrecht: Foris. Olshtain, E. (). Is second language attrition the reversal of second language acquisition? Studies in Second Language Acquisition, , –. Olshtain, E., & Barzilay, M. (). Lexical retrieval difficulties in adult language attrition. In H. W. Seliger & R. M. Vago (Eds.), First Language Attrition (pp. –). Cambridge: Cambridge University Press. Opitz, C. (). L attrition in adult bilinguals: The impact of L proficiency, language use and linguistic aptitude. Paper presented at AILA World Congress, Essen, – August. Opitz, C. (). The role of linguistic aptitude and language use for adult L acquisition and L attrition. Paper presented at ISB, Utrecht, – July. Opitz, C. (a). First language attrition and second language acquisition in a second language environment. PhD thesis, Trinity College Dublin. Opitz, C. (b). The impact of long-term residence abroad on the L and L proficiency of adult bilinguals. Paper presented at the Modern Languages Symposium, ‘Language, Migration and Diaspora’, Dublin, – December. Opitz, C. (). A dynamic perspective on late bilinguals’ linguistic development in an L environment. International Journal of Bilingualism, (), –. Opitz, C. (a). Little to lose and everything to gain: L maintenance and L attainment in longterm migrants. CALL: Irish Journal for Culture, Arts, Literature and Language, (), Article . doi: ./DBCR. http://arrow.dit.ie/priamls/vol/iss/ Opitz, C. (b). Attrition profiles. What CDST can and can’t tell us about attrition. Paper presented at the rd International Conference on Language Attrition, Colchester, – July. Opitz, C. (). Language destabilization and (re-)learning from a Complexity Theory perspective: Timescales and patterns across four studies. In L. Ortega & Z. Han (Eds.), Complexity Theory and Language Development: In Celebration of Diane Larsen-Freeman (pp. –). Amsterdam: John Benjamins. Opitz, C., & Smyth, S. (). ‘Dear diary’—L proficiency development through diary writing. Paper presented at the AILA World Congress, Brisbane, – August. Opitz, C., Lubińska, D., & de Leeuw, E. (Eds.) (). Dynamics of first language attrition across the lifespan. Special Issue of the International Journal of Bilingualism, (), pp. –. Ossher, L., Flegal, K. E., & Lustig, C. (). Everyday memory errors in older adults. Aging, Neuropsychology, and Cognition, (), –. Osterhout, L., & Hagoort, P. (). A superficial resemblance does not necessarily mean you are part of the family: Counterarguments to Coulson, King, and Kutas () in the P/SPS-P debate. Language and Cognitive Processes, , –. Osterhout, L., & Holcomb, P. J. (). Event-related brain potentials elicited by syntactic anomaly. Journal of Memory and Language, , –. Osterhout, L., & Nicol, J. (). The neural architecture of syntactic processing. In C. M. Brown & P. Hagoort (Eds.), Neurocognition of Language (pp. –). Oxford: Oxford University Press. Osterhout, L., McKinnon, R., Bersick, M., & Corey, V. (). On the language-specificity of the brain response to syntactic anomalies: Is the syntactic positive shift a member of the P family? Journal of Cognitive Neuroscience, , –. Osterhout, L., McLaughlin, J., Kim, A., Greenwald, R., & Inoue, K. (). Sentences in the brain: Event-related potentials as real-time reflections of sentence comprehension and language learning.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
In M. Carreiras & C. Clifton, Jr. (Eds.), The On-line Study of Sentence Comprehension: Eyetracking, ERP, and Beyond (pp. –). New York: Psychology Press. Osterhout, L., McLaughlin, J., Pitkänen, I., Frenck-Mestre, C., & Molinaro, N. (). Novice learners, longitudinal designs, and event-related potentials: A means for exploring the neurocognition of second language processing. Language Learning, , –. Osterhout, L., Poliakov, A., Inoue, K., et al. (). Second-language learning and changes in the brain. Journal of Neurolinguistics, (), –. doi: ./j.jneuroling... Osterhout, L., Kim, A., & Kuperberg, G. R. (). The neurobiology of sentence comprehension. In M. Spivey, M. Joannisse, & K. McCrae (Eds.), The Cambridge Handbook of Psycholinguistics (pp. –). Cambridge: Cambridge University Press. Otheguy, R. (). The linguistic competence of second-generation bilinguals: A critique of ‘incomplete acquisition’. In C. Tortora, M. den Dikken, I. L. Montoya, & T. O’Neill (Eds.), Romance Linguistics : Selected Papers from the rd Linguistic Symposium on Romance Languages (LSRL) (pp. –). Amsterdam: Benjamins. Otheguy, R., & Lapidus, N. (). An adaptive approach to noun gender in New York contact Spanish. In R. Cameron, L. Ortíz-López, & R. Núñez-Cedeño (Eds.), A Romance Perspective on Language Knowledge and Use (pp. –). Amsterdam: John Benjamins. Otheguy, R., & Zentella, A. C. (). Spanish in New York: Language Contact, Dialectal Leveling, and Structural Continuity. Oxford: Oxford University Press. Otheguy, R., Zentella, A. C., & Livert, D. (). Language and dialect contact in Spanish in New York: Towards the formation of a speech community. Language, (), –. Oxford, R. L. (). Technical issues in designing and conduction research on language skill attrition. In R. D. Lambert, & B. Freed (Eds.), The Loss of Language Skills (pp. –). Rowley, MA: Newbury House. Oyama, S. (). A sensitive period for the acquisition of a nonnative phonological system. Journal of Psycholinguistic Research, , –. doi: ./BF Oyserman, D., & Lee, S. W. S. (). Does culture influence what and how we think? Effects of priming individualism and collectivism. Psychological Bulletin, , –. Pallier, C. (). Critical periods in language acquisition and language attrition. In B. Köpke, M. S. Schmid, M. Keijzer, & S. Dostert (Eds.), Language Attrition: Theoretical Perspectives (pp. –). Amsterdam: John Benjamins. Pallier, C., Dehaene, S., Poline, J. B., et al. (). Brain imaging of language plasticity in adopted adults: Can a second language replace the first? Cerebal Cortex, (), –. Paradis, J. (). Do bilingual two-year-olds have separate phonological systems? The International Journal of Bilingualism, (), –. Paradis, J. (). Grammatical morphology in children learning English as a second language: Implications of similarities with Specific Language Impairment. Language, Speech, and Hearing Services in Schools, (), –. Paradis, J. (). Bilingual children with specific language impairment: Theoretical and applied issues. Applied Psycholinguistics, (), –. Paradis, J. (). The interface between bilingual development and specific language impairment. Keynote article for special issue with peer commentaries. Applied Psycholinguistics, , –. Paradis, J. (). Individual differences in child English second language acquisition: Comparing child-internal and child-external factors. Linguistic Approaches to Bilingualism, (), –. doi:./lab...par Paradis, J., & Crago, M. (). Tense and temporality: A comparison between children learning a second language and children with SLI. Journal of Speech, Language, and Hearing Research, (), –. Paradis, J., & Genesee, F. (). Syntactic acquisition in bilingual children: Autonomous or interdependent? Studies in Second Language Acquisition, (), –. doi: ./S
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Paradis, J., & Navarro, S. (). Subject realization and crosslinguistic interference in the bilingual acquisition of Spanish and English: What is the role of the input? Journal of Child Language, (), –. Paradis, J., Crago, M., Genesee, F., & Rice, M. (). French-English bilingual children with SLI: How do they compare with their monolingual peers? Journal of Speech, Language, and Hearing Research, , –. Paradis, M. (). Linguistic, psycholinguistic, and neurolinguistic aspects of ‘interference’ in bilingual speakers: The Activation Threshold hypothesis. International Journal of Psycholinguistics, (), –. Paradis, M. (). Neurolinguistic aspects of implicit and explicit memory: Implications for bilingualism and SLA. In N. Ellis (Ed.), Implicit and Explicit Learning of Languages (pp. –). London: Academic Press. Paradis, M. (). A Neurolinguistic Theory of Bilingualism. Amsterdam: John Benjamins. Paradis, M. (). L attrition features predicted by a neurolinguistic theory of bilingualism. In B. Köpke, M. S. Schmid, M. Keijzer, & S. Dostert (Eds.), Language Attrition: A Theoretical Perspective (pp. –). Amsterdam: John Benjamins. Paradis, M. (). Declarative and Procedural Determinants of Second Languages. Amsterdam: John Benjamins. Paradis, M., Elias, J., & Gomez, I. (). Test d’afàsia en els bilingües (Catalan version) [Bilingual aphasia test]. Hillsdale, NJ: Lawrence Erlbaum. Park, H. (). A minimalist approach to null subjects and objects in second language acquisition. Second Language Research, (), –. Park, H. S. (). Korean adoptees in Sweden: Have they lost their first language completely? Applied Psycholinguistics, , –. Parker Jones, O., Green, D. W., Grogan, A., et al. (). Where, when and why brain activation differs for bilinguals and monolinguals during picture naming and reading aloud. Cerebral Cortex, (), –. Parlato-Oliveira, E., Cristophe, A., Hirose, Y., & Dupoux, E. (). Plasticity of illusory vowel perception in Brazilian-Japanese bilinguals. Journal of the Acoustical Society of America, (), –. Parodi, C. (). Aspect in the syntax of Spanish psych-verbs. PhD thesis, University of California, Los Angeles. Pascual y Cabo, D. (). Agreement reflexes of emerging optionality in heritage speaker Spanish. PhD thesis, University of Florida. Pascual y Cabo, D. (). Examining the role of cross-generational attrition in the development of Spanish as a heritage language. Linguistic Approaches to Bilingualism. Advance online publication. doi: ./lab..pas Pascual y Cabo, D., & Rothman, J. (). The (il)logical problem of heritage speaker bilingualism and incomplete acquisition. Applied Linguistics, , –. doi: ./applin/ams Pashler, H., Rohrer, D., Cepeda, N. J., & Carpenter, S. K. (). Enhancing learning and retarding forgetting: Choices and consequences. Psychonomic Bulletin and Review, , –. Passoni, E., Mehrabi, A., Levon, E., & de Leeuw, E. (). Bilingualism, pitch range and social factors: Preliminary results from sequential Japanese-English bilinguals. In Speech Prosody Conference Proceedings (pp. –). Poznań: Adam Mickiewicz University. Patel, A. D., Gibson, E., Ratner, J., Besson, M., & Holcomb, P. J. (). Processing syntactic relations in language and music: An event related potential study. Journal of Cognitive Neuroscience, (), –. Pater, J. (). Weighted constraints in generative linguistics. Cognitive Science, , –. Pavlenko, A. (). I feel clumsy speaking Russian: L influence on L in narratives of Russian L users of English. In V. Cook (Ed.), Effects of the Second Language on the First (pp. –). Bristol: Multilingual Matters.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Pavlenko, A. (). L Influence and L attrition in adult bilingualism. In M. S. Schmid, B. Köpke, M. Keijzer, & L. Weilemar (Eds.), First Language Attrition: Interdisciplinary Perspectives on Methodological Issues (pp. –). Amsterdam: Benjamins. Pavlenko, A. (). Verbs of motion in L Russian of Russian-English bilinguals. Bilingualism: Language and Cognition, (), –. Pavlenko, A., & Jarvis, S. (). Bidirectional transfer. Applied Linguistics, (),–. Pavlenko, A., & Malt, B. C. (). Kitchen Russian: Cross-linguistic differences and first-language object naming by Russian-English bilinguals. Bilingualism: Language and Cognition, (), –. Pavlik, P., & Anderson, J. (). Using a model to compute the optimal schedule of practice. Journal of Experimental Psychology: Applied, , –. Pearson, B. Z. (). Narrative competence among monolingual and bilingual school children in Miami. In D. K. Oller, & R. E. Eilers (Eds.). Language and Lliteracy in Bilingual Children (pp. –). Clevedon: Multilingual Matters. Peelle, J. E., Troiani, V., Grossman, M., & Wingfield, A. (). Hearing loss in older adults affects neural systems supporting speech comprehension. Journal of Neuroscience, (), –. Pelc, L. (). L lexical, morphological and morphosyntactic attrition in Greek-English bilinguals. PhD thesis, City University of New York. Penfield, W. (). A consideration of the neurophysiological mechanisms of speech and some educational consequences. Proceedings of the American Academy of Arts and Sciences, , –. Penfield, W., & Roberts, L. (). Speech and Brain-Mechanisms. Princeton: Princeton University Press. Peng, S. (). Cross-language influence on the production of Mandarin /f/ and /x/ and Taiwanese / h/ by native speakers of Taiwanese Amoy. Phonetica, (), –. Pereltsvaig, A. (). Aspect lost, aspect regained: Restructuring of aspectual marking in American Russian. In P. Kempchinsky & R. Slabakova (Eds.), Aspectual Inquiries (pp. –). Dordrecht: Kluwer. Pereltsvaig, A. (). Aspect in Russian as grammatical rather than lexical notion: Evidence from heritage Russian. Russian Linguistics, (), –. Perez-Cortes, S. (). Acquiring obligatory and variable mood selection: Spanish heritage speakers’ and L learners’ performance in desideratives and reported speech contexts. PhD thesis, Rutgers University. Pérez-Leroux, A. T., & Glass, W. R. (). OPC effects on the L acquisition of Spanish. In A. T. Pérez-Leroux & W. R. Glass (Eds.), Contemporary Perspectives on the Acquisition of Spanish (pp. –). Somerville, MA: Cascadilla Press. Pérez-Leroux, A., Roberge, Y., & Pirvulescu, M. (). Bilingualism as a window into the language faculty: The acquisition of objects in French-speaking children in bilingual and monolingual contexts. Bilingualism: Language and Cognition, (), –. doi: ./SX Perquin, M., Vaillant, M., Schuller, A.-M., Pastore, J., Dartigues, J.-F., & Lair, M.-L., on behalf of the MemoVie Group (). Lifelong exposure to multilingualism: New evidence to support cognitive reserve hypothesis. PLOS One, (), e. Perry, R. J., & Hodges, J. R. (). Attention and executive deficits in Alzheimer’s disease. Brain, (), –. Peters, A., Sethares, C., & Luebke, J. I. (). Synapses are lost during aging in the primate prefrontal cortex. Neuroscience, (), –. Phinney, M. (). The pro-drop parameter in second language acquisition. In T. Roeper & E. Williams (Eds.), Parameter Setting (pp. –). Dordrecht: Reidel. Pica, T. (). Adult acquisition of English as a second language under different conditions of exposure. Language Learning, (), –. Pickering, M. J., & Branigan, H. P. (). The representation of verbs: Evidence from syntactic priming in language production. Journal of Memory and Language, (), –. Pierce, L. J., Genesee, F., & Paradis, J. (). Acquisition of English grammatical morphology by internationally adopted children from China. Journal of Child Language, (), –. doi: ./S
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Pierce, L. J., Klein, D., Chen, J. K., Delcenserie, A., & Genesee, F. (). Mapping the unconscious maintenance of a lost first language. Proceedings of the National Academy of Sciences, (), –. doi: ./pnas. Pierce, L. J., Chen, J. K., Delcenserie, A., Genesee, F., & Klein, D. (). Past experience shapes ongoing neural patterns for language. Nature Communications, , . doi: ./ ncomms Pierce, L. J., Genesee, F., & Klein, D. (). Language loss or retention in internationally-adopted children: Neurocognitive implications for language learning. In F. Genesee, & A. Delcenserie (Eds.), Starting Over—The Language Development in Internationally-Adopted Children (pp. –). Trends in Language Acquisition Research series (TiLAR). Amsterdam: John Benjamins Publishing. Pierce, L. J., Genesee, F., Delcenserie, A., & Morgan, G. (). Variations in phonological working memory: Linking early language experiences and language learning outcomes. Applied Psycholinguistics, (), –. doi: ./S Pinker, S. (). Language Learnability and Language Development. Cambridge, MA: Harvard University Press. Pinker, S. (). The Language Instinct. New York: William Morrow. Pires, A. (). The Minimalist Syntax of Defective Domains: Gerunds and Infinitives. Amsterdam: John Benjamins. Pires, A., & Rothman, J. (). Disentangling sources of incomplete acquisition: An explanation for competence divergence across heritage grammars. International Journal of Bilingualism, (), –. doi: ./ Pitkänen, I. (). Electrophysiological investigations of second language word learning, attrition and bilingual processing. PhD thesis, University of Washington. Pitres, A. (). Etude sur l’aphasie chez les polyglottes [Study of aphasia among polyglots]. Revue de Medicine, , –. (Translated in Paradis, , pp. –.) Place, S., & Hoff, E. (). Properties of dual language exposure that influence two-year-olds’ bilingual proficiency. Child Development, , –. Place, S., & Hoff, E. (). Effects and noneffects of input in bilingual environments on dual language skills in ½-year-olds. Bilingualism: Language and Cognition, (), –. doi: ./ S Plag, I., Homann, J., & Kunter, G. (). Homophony and morphology: The acoustics of word-final S in English. Journal of Linguistics, (), –. Platzack, C. (). The initial hypothesis of syntax: A minimalist perspective on language acquisition and attrition. In H. Clahsen (Ed.), Generative Perspectives on Language Acquisition: Empirical Findings, Theoretical Considerations, and Crosslinguistic Comparisons (pp. –). Amsterdam: John Benjamins Publishing. Pliatsikas, C., Moschopoulou, E., & Saddy, J. D. (). The effects of bilingualism on the white matter structure of the brain. Proceedings of the National Academy of Sciences, (), –. doi: ./pnas. Polat, B., & Kim, Y. (). Dynamics of complexity and accuracy: A longitudinal case study of advanced untutored development. Applied Linguistics, (), –. Polinsky, M. (). Structural dimensions of first language loss. Chicago Linguistic Society, , –. Polinsky, M. (). American Russian: Language loss meets language acquisition. In W. Brown, E. Dornisch, N. Kondrashova, & D. Zec (Eds.), Formal Approaches to Slavic Linguistics (pp. –). Ann Arbor: Michigan Slavic Publishers. Polinsky, M. (). Incomplete acquisition: American Russian. Journal of Slavic Linguistics, , –. Polinsky, M. (a). Russian gender under incomplete acquisition: Heritage speakers’ knowledge of noun categorization. Heritage Language Journal, (), –.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Polinsky, M. (b). Without aspect. In G. G. Corbett & M. Noonan (Eds.), Case and Grammatical Relations: Studies in Honor of Bernard Comrie (pp. –). Amsterdam: John Benjamins. Polinsky, M. (). Reanalysis in adult heritage language: New evidence in support of attrition. Studies in Second Language Acquisition, , –. doi: ./SX Polinsky, M. (). Looking ahead. In D. Pascual y Cabo (Ed.), Advances in Spanish as a Heritage Language (pp. –). Philadelphia, PA: John Benjamins. Polinsky, M. (). Heritage Languages and Their Speakers. Cambridge: Cambridge University Press. Polinsky, M., & Kagan, O. (). Heritage languages: In the ‘wild’ and in the classroom. Language and Linguistics Compass, (), –. doi: ./j.-x...x Polka, L., & Werker, J. F. (). Developmental changes in perception of nonnative vowel contrasts. Journal of Experimental Psychology: Human Perception and Performance, (), –. doi: ./-... Poplack, S. (). Sometimes I start a sentence in Spanish y termino en Español. Linguistics, , –. Porte, G. (). English from a distance: Code-mixing and blending in the L output of long-term resident overseas EFL teachers. In V. Cook (Ed.), Effects of the Second Language on the First (pp. –). Bristol: Multilingual Matters. Portes, A., & Rumbaut, R. (). Immigrant America: A Portrait. Berkeley: University of California Press. Potowski, K., & Rothman, J. (Eds.) (). Bilingual Youth: Spanish in English-speaking Societies. Amsterdam: John Benjamins. Presson, N., Sagarra, N., MacWhinney, B., & Kowalski, J. (). Compositional production in Spanish second language conjugation. Bilingualism: Language and Cognition, , –. Preston, D. R. (). How to lose a language. Interlanguage Studies Bulletin, (), –. Prevoo, M. J., Malda, M., Mesman, J., et al. (). Predicting ethnic minority children’s vocabulary from socioeconomic status, maternal language and home reading input: Different pathways for host and ethnic language. Journal of Child Language, (), –. Price, C. J. (). A review and synthesis of the first years of PET and fMRI studies of heard speech, spoken language and reading. NeuroImage, (), –. Price, J. R., Pollock, K. E., & Oller, D. K. (). Speech and language development in six infants adopted from China. Journal of Multilingual Communication Disorders, (), –. doi: ./ Pufahl, I., & Rhodes, N. (). Foreign language instruction in U.S. schools: Results of a national survey of elementary and secondary schools. Foreign Language Annals, (), –. Pulvermüller, F., & Schumann, J. (). Affect in second language learning. Language Learning, , –. Putnam, M. (). Feature reassembly as constraint satisfaction. The Linguistic Review, (), –. Putnam, M., & Klosinski, R. (). The good, the bad, and the gradient: The role of ‘losers’ in codeswitching. Linguistic Approaches to Bilingualism. https://doi.org/./lab..put Putnam, M., & Sánchez, L. (). What’s so incomplete about incomplete acquisition? A prolegomenon to modeling heritage grammars. Linguistic Approaches to Bilingualism, (), –. Putnam, M., & Schwarz, L. (). How interrogative pronouns can become relative pronouns: The case of was in Misionero German. STuF, (), –. Quay, S. (). The bilingual lexicon: Implications for studies of language choice. Journal of Child Language, (), –. doi: ./S Quesada, M. L. (). *Yo gusto pasteles de chocolate: De la transitividad hacia la intransitividad en la adquisición de verbos psicológicos en español [I like chocolate cakes. On transitivity and intransitivity in the acquisition of Spanish psychological verbs.]. In R. M. Ortiz-Ciscomani (Ed.), Memoria del IX EILN (pp. –). Hermosillo: Editorial Universidad de Sonora. Radford, A. (). Small Children’s Small Clauses. Transactions of the Philological Society, (), –. doi: ./j.-X..tb.x
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Rajagopal, A., Holland, S. K., Walz, N. C., Staat, M. A., Altaye, M., & Wade, S. (). A functional magnetic resonance imaging study of language function in international adoptees. The Journal of Pediatrics, (), –. doi: ./j.jpeds... Raposo, E. (). Case theory and Infl-to-Comp: The inflected infinitive in European Portuguese. Linguistic Inquiry, , –. Ratcliff, R. (). A theory of memory retrieval. Psychological Review, , –. Ravid, D. (). Derivational morphology revisited. Later lexical development in Hebrew. In R. Berman (Ed.), Language Development across Childhood and Adolescence (pp. –). Amsterdam: John Benjamins. Ravid, D., & Tolschinsky, L. (). Developing linguistic literacy: A comprehensive model. Journal of Child Language, (), –. doi: ./S Rayner, K. (). Eye movements in reading and information processing. Psychological Bulletin, (), –. Rayner, K. (). Eye movements in reading and information processing: years of reading. Psychological Bulletin, , –. Rayner, K., & Pollatsek, A. (). Eye movements in reading: A tutorial review. In M. Coltheart (Ed.), Attention and Performance, Vol. (pp. –). Hillsdale, NJ: Lawrence Erlbaum and Associates Inc. Reber, P. (, May ). What is the memory capacity of the human brain? Scientific American Mind. https://www.scientificamerican.com/article/what-is-the-memory-capacity/ Reetz-Kurashige, A. (). Japanese returnees’ retention of English-speaking skills: Changes in verb usage over time. In L. Hansen (Ed.), Second Language Attrition in Japanese Contexts (pp. –). Oxford: Oxford University Press. Reilly, J., Zamora, A., & McGivern, R. (). Acquiring perspective in English: The development of stance. Journal of Pragmatics, (), –. https://doi.org/./j.pragma... Reilly, S., Tomblin, B., Law, J., et al. (). Specific language impairment: A convenient label for whom? International Journal of Language & Communication Disorders, (), –. Reinhart, T., & Reuland, E. (). Anaphors and logophors: An argument structure perspective. In J. Koster & E. Reuland (Eds.), Long-distance Anaphora (pp. –). Cambridge: Cambridge University Press. Rescorla, R. A. (). Retraining of extinguished Pavlovian stimuli. Journal of Experimental Psychology: Animal Behavior Processes, (), . doi: ./-... Restrepo, M. A., & Kruth, K. (). Grammatical characteristics of a Spanish-English bilingual child with specific language impairment. Communication Disorders Quarterly, , –. Restrepo, M. A., & Silverman, S. (). Validity of the Spanish Preschool Language Scale- for use with bilingual children. American Journal of Speech Language Pathology, , –. Restrepo, M. A., Morgan, G. P., & Smyk, E. (). Bilingual children with language impairment. In J. Guendozi, F. Loncke, & M. J. Williams (Eds.), The Handbook of Psycholinguistic & Cognitive Processes: Perspectives in Communication Disorders (pp. –). Abingdon: Taylor & Francis. Ribbert, A., & Kuiken, F. (). L-induced changes in the L of Germans living in the Netherlands. Bilingualism: Language and Cognition, (), –. Ribot, T. (). Diseases of the Memory: An Essay in the Positive Psychology. New York: D. Appleton and Company. Internet Archive. https://archive.org/stream/diseasesofmemoryribouoft#page (accessed April ). Rice, J., Jackson, A., Mahoney, E. E., & Tan, T. X. (). Pre-adoption stress, adversity and later development in IA children. In F. Genesee & A. Delcenserie (Eds.), Starting Over: The Language Development of Internationally-adopted Children (pp. –). Amsterdam: John Benjamins. Rice, M. L. (). A unified model of specific and general language delay: Grammatical tense as a clinical marker of unexpected variation. In Y. Levy & J. Schaeffer (Eds.), Language Competence Across Populations: Toward a Definition of Specific Language Impairment (pp. –). Mahwah, NJ: Erlbaum. Rice, M. L. (). Toward epigenetic and gene regulation models of specific language impairment: Looking for links among growth, genes, and impairments. Journal of Neurodevelopmental Disorders, , .
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Rice, M. L. (). Language growth and genetics of specific language impairment. International Journal of Speech-Language Pathology, (), –. Rice, M. L., Oetting, J. B., Marquis, J., Bode, J., & Pae, S. (). Frequency of input effects on word comprehension of children with specific language impairment. Journal of Speech, Language, and Hearing Research, (), –. Riches, N. G., Tomasello, M., & Conti-Ramsden, G. (). Verb learning in children with SLI. Frequency and spacing effects. Journal of Speech, Language, and Hearing Research, (), –. Riehl, C. M. (). Diskursmarkierung im mehrsprachigen Dialog [Discourse markers in multilingual communication.]. In M. Dannerer, P. Mauser, H. Scheutz, & A. E. Weiss (Eds.), Gesprochen— geschrieben—gedichtet: Variation und Transformation von Sprache [Spoken language—written language—poetric language. Variation and transformation of language.] (pp. –). Berlin: Erich Schmidt. Riehl, C. M. (). Norm and variation in language minority settings. In A. Lenz & A. Plewnia (Eds.), Grammar between Norm and Variation (pp. –). Amsterdam: John Benjamins. Riehl, C. M. (). Mehrsprachigkeit und Sprachkontakt [Multilingualism and language contact.]. In P. Auer (Ed.), Einführung in die Sprachwissenschaft: Grammatik—Interaktion—Kognition [Introduction to Linguistics: grammar—interaction—cognition.] (pp. –). Stuttgart: Metzler. Riehl, C. M. (). Sprachkontaktforschung: Eine Einführung. [Language contact research: an introduction.]. Tübingen: Narr. Riehl, C. M. (). Language contact and language attrition: The case of Barossa German. International Journal of the Sociology of Language, , –. Riksem, B. R., Grimstad, M. B., Lohndal, T., & Åfarli, T. (to appear). Language mixing with verbs and nouns in American Norwegian. Journal of Comparative Germanic Linguistics. Rizio, A. A., Moyer, K. J., & Diaz, M. T. (). Neural evidence for phonologically based language production deficits in older adults: An fMRI investigation of age-related differences in pictureword interference. Brain and Behavior, (). doi: ./brb. Rizzi, L. (). Issues in Italian Syntax. Dordrecht: Foris. Rizzi, L. (). Early null Subjects and root null subjects. In T. Hoekstra & B. Schwartz (Eds.), Language Acquisition Studies in Generative Grammar (pp. –). Amsterdam: John Benjamins. Roberts, I. (). Diachronic Syntax. Oxford: Oxford University Press. Rodina, Y., & Westergaard, M. (). Grammatical gender in bilingual Norwegian–Russian acquisition: The role of input and transparency. Bilingualism: Language and Cognition, (), –. doi: ./S Rodriguez, E. T., Tamis-LeMonda, C. S., et al. (). The formative role of home literacy experiences across the first three years of life in children from low-income families. Journal of Applied Development Psychology, (), –. doi: ./j.appdev... Romano, F. B. (). Remarks on research of anaphora resolution in situations of language contact: Cross-linguistic influence and the PAS. International Journal of Bilingualism, (), –. https:// doi.org/./ Rosenberg, P. (). Dialect convergence in the German language islands (Sprachinseln). In P. Auer, F. Hinskens, & P. Kerswill (Eds.), Dialect Change: Convergence and Divergence of Dialects in Contemporary Europe (pp. –). Cambridge: Cambridge University Press. Rossi, E., Cheng, H., Kroll, J. F., Diaz, M. T., & Newman, S. D. (). Bilingualism impacts whitematter connectivity: Evidence from DTI. Manuscript submitted for publication. Rossi, E., Newman, S., Kroll, J. F., & Diaz, M. T. (). Neural signatures of inhibitory control in bilingual spoken production. Cortex, , –. Rothman, J. (). Heritage speaker competence differences, language change and input type: Inflected infinitives in heritage Brazilian Portuguese. International Journal of Bilingualism, (), –. doi: ./ Rothman, J. (a). Pragmatic deficits with syntactic consequences? L pronominal subjects and the syntax-pragmatics interface. Journal of Pragmatics, , –.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Rothman, J. (b). Understanding the nature and outcomes of early bilingualism: Romance languages as heritage languages. International Journal of Bilingualism, (), –. doi: ./ Rothman, J., & Iverson, M. (). On L clustering and resetting the null subject parameter in L Spanish: Implications and observations. Hispania, , –. Rothman, J., & Iverson, M. (). Poverty-of-the-stimulus and SLA epistemology: Considering L knowledge of aspectual phrasal semantics evidence in L Portuguese. Language Acquisition, (), –. Rothman, J., & Treffers-Daller, J. (). A prolegomenon to the construct of the native speaker: Heritage speaker bilinguals are natives too! Applied Linguistics, (), –. doi: ./applin/ amt Rothman, J., Tsimpli, I., & Pascual y Cabo, D. (). Formal linguistic approaches to heritage language acquisition: Bridges for pedagogically oriented research. In D. Pascual y Cabo (Ed.), Advances in Spanish as a Heritage Language (pp. –). Amsterdam: John Benjamins. Rothweiler, M., Chilla, S., & Clahsen, H. (). Subject–verb agreement in specific language impairment: A study of monolingual and bilingual German-speaking children. Bilingualism: Language and Cognition, (), –. Rottschy, C., Langner, R., Dogan, I., et al. (). Modelling neural correlates of working memory: A coordinate-based meta-analysis. NeuroImage, (), –. doi: ./j.neuroimage... Roy, P., & Chiat, S. (). Teasing apart disadvantage from disorder: The case of poor language. In C. R. Marshall (Ed.), Current Issues in Developmental Disorders (pp. –). London: Psychology Press. Runnqvist, E., & Costa, A. (). Is retrieval-induced forgetting behind the bilingual disadvantage in word production? Bilingualism: Language and Cognition, (), –. Runnqvist, E., Gollan, T. H., Costa, A., & Ferreira, V. S. (). A disadvantage in bilingual sentence production modulated by syntactic frequency and similarity across languages. Cognition, , –. Rüschemeyer, S., Nojack, A., & Limbach, M. (). A mouse with a roof? Effects of phonological neighbors on processing of words in sentences in a non-native language. Brain and Language, , –. Russell, R. A. (a). Lexical maintenance and attrition in Japanese as a second language. In L. Hansen (Ed.), Second Language Attrition in Japanese Contexts (pp. –). Oxford: Oxford University Press. Russell, R. A. (b). Measuring attrition in L Japanese syntactic competence. In P. Robinson (Ed.), Representation and Process: Proceedings of the rd Pacific Second Language Research Forum (PacSLRF’), Vol. (pp. –). Tokyo: Aoyama University. Russell, R. A. (). Syntactic attrition in L Japanese missionary language. In L. Hansen (Ed.), Second Language Acquisition Abroad: The LDS Missionary Experience (pp. –). Amsterdam: John Benjamins. Sabourin, L., Stowe, L. A., & De Haan, G. J. (). Transfer effects in learning a second language grammatical gender system. Second Language Research, (), –. Saito, K. (). Education and adjustment of children returned from overseas: An approach from cross-cultural psychology. Japanese Journal of Social Psychology, (), –. Sale, A., Maya Vetencourt, J. F., Medini, P., et al. (). Environmental enrichment in adulthood promotes amblyopia recovery through a reduction of intracortical inhibition. Nature Neuroscience, (), –. Salthouse, T. A. (). Selective review of cognitive aging. Journal of the International Neuropsychological Society, (), –. Salvatierra, J., Rosselli, M., Acevedo, A., & Duara, R. (). Verbal fluency in bilingual Spanish/ English Alzheimer’s disease patients. American Journal of Alzheimer’s Disease and Other Dementias, (), –. Sam, D. L., & Berry, J. W. (). Acculturation: When individuals and groups of cultural backgrounds meet. Perspectives on Psychological Science, (). –.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Sancier, M. L., & Fowler, C. A. (). Gestural drift in a bilingual speaker of Brazilian Portuguese and English. Journal of Phonetics, (), –. Sanders, A. E., Hall, C. B., Katz, M. J., & Lipton, R. B. (). Non-native language use and risk of incident dementia in the elderly. Journal of Alzheimer’s Disease, (), –. Sanoudaki, E., & Thierry, G. (). Language non-selective syntactic activation in early bilinguals: The role of verbal fluency. International Journal of Bilingual Education and Bilingualism, (), –. Santos, A. L., & Flores, C. (). Comparing heritage speakers and late L-learners of European Portuguese: verb movement, VP ellipsis and adverb placement. Linguistic Approaches to Bilingualism, (), –. doi: ./lab..san Sapir, E. (/). Language: An Introduction to the Study of Speech. Mineola, NY: Dover Publications. Sardinha, J. (). ‘Returned’ second-generation Portuguese-Canadians and Portuguese-French: Return motivations and sense of belonging. Journal of Mediterranean Studies, (), –. Savage, C., Lieven, E., Theakston, A., & Tomasello, M. (). Structural priming as implicit learning in language acquisition: The persistence of lexical and structural priming in -year-olds. Language Learning and Development, (), –. Scheele, A. F., Leseman, P. M., & Mayo, A. Y. (). The home language environment of monolingual and bilingual children and their language proficiency. Applied Psycholinguistics, , –. Scherag, A., Demuth, L., Rösler, F., Neville, H. J., & Röder, B. (). The effects of late acquisition of L and the consequences of immigration on L for semantic and morpho-syntactic language aspects. Cognition, (), B–. doi: ./j.cognition... Scherer, G. A. C. (). The forgetting rate of learning German. German Quarterly, , –. Schiffrin, D. (). Discourse Markers. Cambridge: Cambridge University Press. Schjerve, R. R., & Vetter, E. (Eds.) (). European Multilingualism: Current Perspectives and Challenges. Bristol: Multilingual Matters. Schlegel, A. A., Rudelson, J. J., & Peter, U. T. (). White matter structure changes as adults learn a second language. Journal of Cognitive Neuroscience, (), –. Schlichting, M. L., Zeithamova, D., & Preston, A. R. (). CA subfield contributions to memory integration and inference. Hippocampus, (), –. Schlyter, S. (). The weaker language in bilingual Swedish–French children. In K. Hyltenstam, & A. Viberg (Eds.), Progression and Regression in Language: Sociocultural, Neuropsychological and Linguistic Perspectives (pp. –). Cambridge: Cambridge University Press. Schlyter, S. (). Early morphology in Swedish as the weaker language in French-Swedish bilingual children. Scandinavian Working Papers on Bilingualism, , –. Schlyter, S., & Håkansson, G. (). Word order in Swedish as the first language, second language and weaker language in bilinguals. Scandinavian Working Papers on Bilingualism, , –. Schmid, M. S. (). First Language Attrition, Use and Maintenance: The Case of German Jews in Anglophone Countries. Amsterdam: John Benjamins. Schmid, M. S. (a). Language attrition research: An annotated bibliography. In M. S. Schmid, B. Köpke, M. Keijzer, & L. Weilemar (Eds.), First Language Attrition: Interdisciplinary Perspectives on Methodological Issues (pp. –). Amsterdam: John Benjamins. Schmid, M. S. (b). A new blueprint for language attrition research. In M. S. Schmid, B. Köpke, M. Keijzer, & L. Weilemar (Eds.), First Language Attrition: Interdisciplinary Perspectives on Methodological Issues (pp. –). Amsterdam: John Benjamins. Schmid, M. S. (c). First language attrition: The methodology revised. International Journal of Bilingualism, (), –. Schmid, M. S. (). The Language Attrition Test Battery: A research manual. Unpublished manuscript, Amsterdam. Schmid, M. S. (). The role of L use for L attrition. In B. Köpke, M. S. Schmid, M. Keijzer, & S. Dostert (Eds.), Language Attrition: Theoretical Perspectives (pp. –). Amsterdam: John Benjamins. Schmid, M. S. (a). L attrition across the lifespan. In K. de Bot & R. W. Schrauf (Eds.), Language Development Over the Lifespan (pp. –). New York: Routledge.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Schmid, M. S. (b). Languages at play: The relevance of L attrition to the study of bilingualism. Bilingualism: Language and Cognition, (), –. Schmid, M. S. (a). Language Attrition. Cambridge: Cambridge University Press. Schmid, M. S. (b). Contact x time: External factors and variability in L attrition. In M. S. Schmid & W. Lowie (Eds.), Modeling Bilingualism: From Structure to Chaos. In Honor of Kees de Bot (pp. –). Amsterdam: John Benjamins. Schmid, M. S. (). The impact of age and exposure on bilingual development in international adoptees and family migrants: A perspective from Holocaust survivors. Linguistic Approaches to Bilingualism, (), –. doi: ./lab...sch Schmid, M. S. (a). First language attrition. WIRE’s Cognitive Science, (), –. Schmid, M. S. (b). First language attrition. Linguistic Approaches to Bilingualism, (), –. Schmid, M. S. (c). First language attrition as a window to constraints on bilingual development. International Symposium on Bilingualism (ISB). Singapore, June. Schmid, M. S. (). The debate on maturational constraints in bilingual development: A perspective from first-language attrition. Language Acquisition, (), –. Schmid, M. S., & Beers Fägersten, K. (). Disfluency markers in L attrition. Language Learning, (), –. Schmid, M. S., & de Bot, K. (). Language attrition. In A. Davies & C. Elder (Eds.), The Handbook of Applied Linguistics (pp. –). Oxford: Blackwell Publishing. Schmid, M. S., & Dusseldorp, E. (). Quantitative analyses in a multivariate study of language attrition: The impact of extralinguistic factors. Second Language Research, (), –. Schmid, M. S., & Jarvis, S. (). Lexical access and lexical diversity in first language attrition. Bilingualism: Language and Cognition, (), –. Schmid, M. S., & Karayayla, T. (). Language development in L attrition and heritage speakers. Paper presented at EuroSLA at the University of Reading, August – September. https:// reading.ac.uk/celm-eurosla/ Schmid, M. S., & Keijzer, M. (). First language attrition and reversion among older migrants. International Journal of the Sociology of Language, , –. Schmid, M. S., & Köpke, B. (). Bilingualism and attrition. In B. Köpke, M. S. Schmid, M. Keijzer, & S. Dostert (Eds.), Language Attrition: Theoretical Perspectives, Vol. (pp. –). Amsterdam: John Benjamins. Schmid, M. S., & Köpke, B. (). L attrition and the mental lexicon. In A. Pavlenko (Ed.), The Bilingual Mental Lexicon: Interdisciplinary Approaches (pp. –). Bristol: Multilingual Matters. Schmid, M. S., & Köpke, B. (). Second language acquisition and attrition. Language, Interaction and Acquisition, (), –. Schmid, M. S., & Köpke, B. (a). The relevance of first language attrition to theories of bilingual development. Linguistic Approaches to Bilingualism, (), –. Schmid, M. S., & Köpke, B. (b). When is a bilingual an attriter? Response to the commentaries. Linguistic Approaches to Bilingualism, (), –. Schmid, M. S., & Lowie, W. (Eds.). (). Modeling Bilingualism: From Structure to Chaos. In Honor of Kees de Bot. Amsterdam: John Benjamins. Schmid, M. S., & Mehotcheva, T. (). Foreign language attrition. Dutch Journal of Applied Linguistics. In H. W. Seliger & R. M. Vago (Eds.), First Language Attrition (pp. –). Cambridge: Cambridge University Press. Schmid, M. S., Köpke, B., Keijzer, M., & Weilemar, L. (Eds.) (). First Language Attrition: Interdisciplinary Perspectives on Methodological Issues. Amsterdam: John Benjamins. Schmid, M. S., Köpke, B., and de Bot, K. (). Language attrition as a complex, non-linear development. International Journal of Bilingualism, (), –. Schmid, M. S., Berends, S. M., Bergmann, C., et al. (). German gender in second language acquisition and first language attrition: An EEG study. Third International Conference on Attrition (ICLA). University of Essex, July.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Schmid, M. S., & Yılmaz, G. (). Predictors of language dominance: an integrated analysis of first language attrition and second language acquisition in late bilinguals. Frontiers in psychology, . Schmitt, E. (). Overt and covert codeswitching in immigrant children from Russia. International Journal of Bilingualism, , –. Schmitt, E. (). Beneath the surface: Signs of language Attrition in immigrant Children from Russia. PhD dissertation, University of South Carolina. Schmitt, E. (). No more reductions—To the problem of evaluation of language attrition data. In M. S. Schmid, B. Köpke, M. Keijzer, & L. Weilemar (Eds.), First Language Attrition: Interdisciplinary Perspectives on Methodological Issues (pp. –). Amsterdam: John Benjamins. Schmitt, E. (). Early bilinguals—Incomplete acquirers or language forgetters? Issues of Multiculturalism and Multilingualism in Modern Education System: Studia humaniora et paedagogica collegii narovensis, III(), –. Schmitt, E. (). When boundaries are crossed: Evaluating language attrition data from two perspectives. Bilingualism: Language and Cognition, (), –. Schmitter-Edgecombe, M., Vesneski, M., & Jones, D. W. R. (). Aging and word-finding: A comparison of spontaneous and constrained naming tests. Archives of Clinical Neuropsychology, (), –. Schneider, V. I., Healey, A. F., & Bourne, L. E., Jr. (). What is learned under difficult conditions is hard to forget: Contextual interference effects in foreign vocabulary acquisition, retention, and transfer. Journal of Memory and Language, , –. Scholten, P. W. A., Entzinger, H., Kofman, E., Vacchelli, E., Kraler, A., & Brechner, C. (). Integration from abroad? Perception and impacts of pre-entry tests for third country nationals. Final report of PROSINT: Work package . Schoofs, P. (). The mutual influence of the first and second languages in German and English L speakers in second language environments. PhD dissertation, University of Newcastle Upon Tyne. Schuhmann, K. S., & Huffman, M. K. (). L drift and L category formation in second language learning. In The Scottish Consortium for ICPhS (Ed.), Proceedings of the th International Congress of Phonetic Sciences (paper number ). Glasgow: University of Glasgow. Schumann, J. H. (). Research on the acculturation model for second language acquisition. Journal of Multilingual and Multicultural Development, (), –. Schumann, J. H. (). Where is cognition? Emotion and cognition in second language acquisition. Studies in Second Language Acquisition, , –. Schwartz, A. I., & Kroll, J. F. (). Bilingual lexical activation in sentence context. Journal of Memory and Language, (), –. Schwartz, G., Balas, A., & Rojczyk, A. (). Phonological factors affecting L phonetic realization of proficient Polish users of English. Research in Language, (), –. Schweizer, T. A., Ware, J., Fischer, C. E., Craik, F. I., & Bialystok, E. (). Bilingualism as a contributor to cognitive reserve: Evidence from brain atrophy in Alzheimer’s disease. Cortex, (), –. doi: ./j.cortex... Schwieter, J. W. (Ed.) (). The Cambridge Handbook of Bilingual Processing. New York: Cambridge University Press. Scobbie, J. (). Interface and overlap in phonetics and phonology. In G. Ramchand & C. Reiss (Eds.), The Oxford Handbook of Linguistic Interfaces (pp. –). Oxford: Oxford University Press. Scontras, G., Fuchs, S., & Polinsky, M. (). Heritage language and linguistic theory. Frontiers in Psychology, , . https://doi.org/./fpsyg.. Scott, C. (). Syntactic contribution to literacy. In A. Stone, E. Silliman, B. Ehren, & K. Apel (Eds.), Handbook of Language and Literacy (pp. –). New York: Guilford. Scott, K. A., & Roberts, J. A. (). Language development during the preschool years. In F. Genesee & A. Delcenserie (Eds.), Starting Over—The Language Development of Internationally-adopted Children (pp. –). Amsterdam: John Benjamins. Sebastián-Gallés, N. (). Bilingual language acquisition: Where does the difference lie? Human Development, , –. https://doi.org/./
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Segalowitz, N. (). Does advanced skill in a second language reduce automaticity in the first language? Language Learning, (), –. Seidenberg, M. S. (). Language acquisition and use: Learning and applying probabilistic constraints. Science, , –. Seliger, H. (). Implications of a multiple critical periods hypothesis for second language learning. In W. C. Ritchie (Ed.), Second Language Acquisition Research: Issues and Implications (pp. –). New York: Academic Press. Seliger, H. W. (). Deterioration and creativity in childhood bilingualism. In K. Hylstenstam & L. K. Obler (Eds.), Bilingualism Across the Lifespan (pp. –). Cambridge: Cambridge University Press. Seliger, H. W. (). Language attrition, reduced redundancy, and creativity. In H. W. Seliger & R. M. Vago (Eds.), First Language Attrition (pp. –). Cambridge: Cambridge University Press. Seliger, H. W. (). Primary language attrition in the context of bilingualism. In T. J. Bhatia & W. C. Ritchie (Eds.), Handbook of Second Language Acquisition (pp. –). New York: Academic Press. Seliger, H. W., & Vago, R. M. (a). The study of first language attrition: An overview. In H. W. Seliger & R. M. Vago (Eds.), First Language Attrition (pp. –). Cambridge: Cambridge University Press. Seliger, H. W., & Vago, R. M. (Eds.) (b). First Language Attrition. Cambridge: Cambridge University Press. Selinker L. (). Interlanguage. International Review of Applied Linguistics, , –. Semel, E., Wiig, E. H., & Secord, W. A. (). Clinical Evaluation of Language Fundamentals, th edn. San Antonio, TX: Harcourt Assessment. Semyonov, M., Raijman, R., & Gorodzeisky, A. (). The rise of antiforeigner sentiment in European societies –. American Sociological Review, (), –. Serratrice, L. (). Crosslinguistic influence in bilingual development. Linguistic Approaches to Bilingualism, (), –. doi: ./lab...ser Serratrice, L., Sorace, A., & Paoli, S. (). Crosslinguistic influence at the syntax–pragmatics interface: Subjects and objects in English-Italian bilingual and monolingual acquisition. Bilingualism: Language and Cognition, (), –. Serratrice, L., Sorace, A., Filiaci, F., & Baldo, M. (). Bilingual children’s sensitivity to specificity and genericity: Evidence from metalinguistic awareness. Bilingualism: Language and Cognition, (), –. doi: ./S Serratrice, L., Sorace, A., Filiaci, F., & Baldo, M. (). Pronominal objects in English-Italian and Spanish-Italian bilingual children. Applied Psycholinguistics, (), –. Seton, B., & Schmid, M. S. (). Multi-competence and first language attrition. In V. Cook & W. Li (Eds.), The Cambridge Handbook of Linguistic Multi-competence (pp. –). Cambridge: Cambridge University Press. Shafto, M. A., Burke, D. M., Stamatakis, E. A., Tam, P. P., & Tyler, L. K. (). On the tip-of-thetongue: Neural correlates of increased word-finding failures in normal aging. Journal of Cognitive Neuroscience, (), –. Shao, Z., Roelofs, A., & Meyer, A. S. (). Sources of individual differences in the speed of naming objects and actions: The contribution of executive control. The Quarterly Journal of Experimental Psychology, (), –. Share, D. L. (). Phonological recoding and self-teaching: Sine qua non of reading acquisition. Cognition, , –. Sharma, A., Gilley, P. M., Dorman, M. F., & Baldwin, R. (). Deprivation-induced cortical reorganization in children with cochlear implants. International Journal of Audiology, (), –. doi: ./ Sharwood Smith, M. (). Cross-linguistic aspects of second language acquisition. Applied Linguistics, (), –.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Sharwood Smith, M. (a). On first language loss in the second language acquirer. In S. Gass & L. Selinker (Eds.), Language Transfer in Language Learning (pp. –). Rowley: Newbury House. Sharwood Smith, M. (b). Crosslinguistic aspects of second language acquisition. Applied Linguistics, () –. Sharwood Smith, M. (). Second Language Learning: Theoretical Foundations. New York: Routledge. Sharwood Smith, M. (). Attrition by processing: A MOGUL account. A promissory note? Paper presented at the Second International Conference on First Language Attrition, Amsterdam, August. Sharwood Smith, M. (). Understanding Attrition within a MOGUL framework. In B. Köpke, M. Schmid, M. Keijzer, & S. Dostert (Eds.), Language Attrition: Theoretical Perspectives (pp. –). Amsterdam: John Benjamins. Sharwood Smith, M. (). In search of conceptual frameworks for relating brain activity to language function. Frontiers in Psychology, , . Sharwood Smith, M. (). Introducing Language and Cognition: A Map of the Mind. Cambridge: Cambridge University Press. Sharwood Smith, M., & Truscott, J. (). Stages or continua in second language acquisition: A MOGUL solution. Applied Linguistics, (), –. Sharwood Smith, M., & Truscott, J. (). Consciousness and language: A processing perspective. In E. Perry, D. Collerton, F. LeBeau, & H. Ashton (Eds.), New Horizons in the Neuroscience of Consciousness (pp. –). Amsterdam: John Benjamins. Sharwood Smith, M., & Truscott, J. (). The Multilingual Mind: A Modular Processing Perspective. Cambridge: Cambridge University Press. Sharwood Smith, M., & Van Buren, P. (). First language attrition and the parameter setting model. In H. W. Seliger & R. M. Vago (Eds.), First Language Attrition (pp. –). Cambridge: Cambridge University Press. Shatz, M. (). On the development of the field of language development. In E. Hoff & M. Shatz (Eds.), Blackwell Handbook of Language Development (pp. –). Malden, MA: WileyBlackwell. Sheldon, A. (). The role of parallel function in the acquisition of relative clauses in English. Journal of Verbal Learning and Verbal Behavior, (), –. http://dx.doi.org/./S()- Sheng, L., Lu, Y., & Kan, P. F. (). Lexical development in Mandarin-English bilingual children. Bilingualism: Language and Cognition, (), –. doi: ./S Sherkina-Lieber, M. (). Tense, aspect, and agreement in heritage Labrador Inuttitut: Do receptive bilinguals understand functional morphology? Linguistic Approaches to Bilingualism, (), –. Sherkina-Lieber, M., Pérez-Leroux, A. T., & Johns, A. (). Grammar without speech production: The case of Labrador Inuttitut heritage receptive bilinguals. Bilingualism: Language and Cognition, (), –. https://doi.org/./S Shibuya, M. (). The potential of ‘returnees’ for criticism and bring about change: A focus on school regulations. The Japan Society of Educational Sociology, , –. Shin, S. (). Developing in Two Languages: Korean Children in America. Clevedon: Multilingual Matters. Shockey, L. (). All in a flap: Long-term accommodation in phonology. International Journal of the Sociology of Language, (), –. doi: ./ijsl... Shohamy, E. (). Language Policy, Hidden Agenda and New Approaches. London: Routledge. Silva-Corvalán, C. (). Language Contact and Change: Spanish in Los Angeles. New York: Oxford University Press. Silva-Corvalán, C., & Treffers-Daller, J. (). Language Dominance in Bilinguals: Issues of Measurement and Operationalization. Cambridge: Cambridge University Press. doi: ./ CBO
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Singer, T., Verhaeghen, P., Ghisletta, P., Lindenberger, U., & Baltes, P. B. (). The fate of cognition in very old age: Six-year longitudinal findings in the Berlin Aging Study (BASE). Psychology and Aging, (), –. Singh, L., Liederman, J., Mierzejewski, R., & Barnes, J. (). Rapid reacquisition of native phoneme contrasts after disuse: You do not always lose what you do not use. Developmental Science, , –. doi: ./j.-...x Slabakova, R. (). What is so difficult about telicity marking in L Russian? Bilingualism: Language and Cognition, , –. Slabakova, R. (). Meaning in the Second Language. Berlin: Mouton de Gruyter. Slabakova, R. (). Features or parameters: Which one makes SLA easier and more interesting to study? Second Language Research, (), –. Slabakova, R. (). What is easy and what is hard in second language acquisition: A generative perspective. In M. d. P. García Mayo, M. Junkal Gutierrez-Mangado, & M. Martínez Adrián (Eds.), Contemporary Approaches to Second Language Acquisition (pp. –). Amsterdam: John Benjamins. Slabakova, R. (). Acquiring temporal meanings without tense morphology: The case of L Mandarin Chinese. Modern Language Journal, (), –. Slabakova, R. (). Second Language Acquisition. Oxford: Oxford University Press. Slabakova, R., & Montrul, S. (). Genericity and aspect in L acquisition. Language Acquisition, (), –. Smith, C. (). The Parameter of Aspect. Dordrecht: Kluwer. Smolensky, P. (). Computational levels and integrated connectionist/symbolic explanation. In P. Smolensky & G. Legendre (Eds.), The Harmonic Mind: From Neural Computation to Optimality-Theoretic Grammar. Vol. : Linguistic and Philosophical Implications (pp. –). Cambridge, MA: MIT Press. Smolensky, P., Goldrick, M., & Mathis, D. (). Optimization and quantization in gradient symbol systems: A framework for integrating the continuous and the discrete in cognition. Cognitive Science, , –. Smythe, P. C., Jutras, G. C., Bramwell, J. R., & Gardner, R. C. (). Second language retention over varying intervals. Modern Language Journal, (), –. Snape, N., Hirakawa, M., Hirakawa, Y., Hosoi, H., & Matthews, J. (). L English generics: Japanese child returnees’ incomplete acquisition or attrition? In R. T. Miller et al. (Ed.), Selected Proceedings of the Second Language Research Forum (pp. –). Somerville, MA: Cascadilla Proceedings Project. Snape, N., Matthews, J., Hirakawa, M., Hirakawa, Y., & Hosoi, H. (). Aspect in L English: A longitudinal study of four Japanese child returnees. EUROSLA Yearbook, , –. Snedeker, J., Geren, J., & Shafto, C. L. (). Disentangling the effects of cognitive development and linguistic expertise: A longitudinal study of the acquisition of English in internationally-adopted children. Cognitive Psychology, (), –. doi: ./j.cogpsych... Sniderman, P. M., Hagendoorn, L., & Markus, P. (). Predisposing factors and situational triggers. Exclusionary reactions to immigrant minorities. American Political Science Review, (), –. Soesman, A. (). An experimental study on native language attrition in Dutch adult immigrants in Israel. In J. Klatter-Folmer & S. Kroon (Eds.), Dutch Overseas: Studies in Maintenance and Loss of Dutch as an Immigrant Language (pp. –). Tilburg: Tilburg University Press. Solé, M.-J., & Recasens, D. (). Editors’ introduction. In M.-J. Solé & D. Recasens (Eds.), The Initiation of Sound Change: Perception, Production, and Social Factors (pp. –). Amsterdam: John Benjamins. Sommers, M. S., & Danielson, S. M. (). Inhibitory processes and spoken word recognition in young and older adults: The interaction of lexical competition and semantic context. Psychology and Aging, , –. Sndergaard, B. (). Language maintenance, code mixing, and language attrition: Some observations. Nowele: North-Western European Language Evolution, –, –.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Sopata, A. (). Placement of infinitives in successive child language acquisition. In E. Rinke & T. Kupisch (Eds.), The Development of Grammar: Language Acquisition and Diachronic Change (pp. –). Amsterdam: John Benjamins. Sorace, A. (a). Differential effects of attrition in the L syntax of L near-native speakers. In C. Howell, S. Fish, & T. Keith-Lucas (Eds.), Proceedings of the Boston University Conference on Language Development (pp. –). Somerville, MA: Cascadilla Press. Sorace, A. (b). Syntactic optionality in L acquisition. Second Language Research, , –. Sorace, A. (). Near-nativeness. In M. Long & C. Doughty (Eds.), Handbook of Second Language Acquisition (pp. –). Oxford: Blackwell. Sorace, A. (). Native language attrition and developmental instability at the syntax-discourse interface: Data, interpretations and methods. Bilingualism: Language and Cognition, , –. doi: ./S Sorace, A. (). Selective optionality in language development. In L. Cornips & K. Corrigan (Eds.), Syntax and Variation: Reconciling the Biological and the Social (pp. –). Amsterdam: John Benjamins. Sorace, A. (). The more, the merrier: Facts and beliefs about the bilingual mind. In S. Della Sala (Ed.), Tall Tales about the Mind and the Brain: Separating Fact from Fiction (pp. –). Oxford: Oxford University Press. Sorace, A. (). Pinning down the concept of ‘interface’ in bilingualism. Linguistic Approaches to Bilingualism, (), –. Sorace, A. (). Referring expressions and executive functions in bilingualism. In I. A. Sekerina & L. Spradlin (Eds.), Bilingualism and Executive Function: An Interdisciplinary Approach (pp. –). Amsterdam: John Benjamins. Sorace, A., & Filiaci, F. (). Anaphora resolution in near-native speakers of Italian. Second Language Research, , –. Sorace, A., & Serratrice, L. (). Internal and external interfaces in bilingual development: Beyond structural overlap. International Journal of Bilingualism, (), –. Sorace, A., Serratrice, L., Filiaci, F., & Baldo, M. (). Discourse conditions on subject pronoun realization: Testing the linguistic intuitions of bilingual children. Lingua, , –. Spaulding, T. J., Plante, E., & Vance, R. (). Sustained selective attention skills of preschool children with specific language impairment: Evidence for separate attentional capacities. Journal of Speech, Language, and Hearing Research, (), –. ‘Speak German at home’ row as conservatives target migrants (). BBC News, December. http:// www.bbc.com/news/world-europe- (accessed August ). Spinner, P. (). The second language acquisition of number and gender in Swahili: A feature reassembly approach. Second Language Research, (), –. Spivey, M. J., & Marian, V. (). Cross talk between native and second languages: Partial activation of an irrelevant lexicon. Psychological Science, (), –. Spivey, M. J., & Marian, V. (). Competing activation in bilingual language processing: Withinand between-language competition. Bilingualism: Language and Cognition, (), –. Sprouse, R. A. (). Full Transfer and Relexification: Second Language Acquisition and Creole Genesis. In C. Lefebvre, C. Jourdan, & L. White (Eds.), L Acquisition and Creole Genesis: Dialogues (pp. –). Philadelphia, PA: John Benjamins. Stager, C. L., & Werker, J. F. (). Infants listen for more phonetic detail in speech perception than in word-learning tasks. Nature, (), –. doi: ./ Stangen, I., Kupisch, T., Proietti Ergün, A. L., & Zielke, M. (). Foreign accent in heritage speakers of Turkish in Germany. In H. Peukert (Ed.), Transfer Effects in Multilingual Language Development (pp. –). Amsterdam: John Benjamins. Starren, M. (). Phenomenes de reduction linguistique en situation d’acquisition et d’attrition d’une langue second—A la recherche des elements linguistiques les plus susceptibles de regression. CALAP, –, –.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Stein, M., Federspiel, A., Koenig, T., et al. (). Structural plasticity in the language system related to increased second language proficiency. Cortex, (), –. doi:./j.cortex... Stein, M., Winkler, C., Kaiser, A., & Dierks, T. (). Structural brain changes related to bilingualism: Does immersion make a difference? Frontiers in Psychology, , . doi: ./fpsyg.. Stein, N. L., & Glenn, C. G. (). An analysis of story comprehension in elementary school children. In R. Freedle (Ed.), New Directions in Discourse processing II (pp. –). Norwood, NJ: Ablex. Steinhauer, K. (). Event-related potentials (ERPs) in second language research: A brief introduction to the technique, a selected review, and an invitation to reconsider critical periods in L. Applied Linguistics, (), –. doi: ./applin/amu Steinhauer, K., & Connolly, J. F. (). Event-related potentials in the study of language. In B. Stemmer & H. Whitaker (Eds.), Handbook of the Cognitive Neuroscience of Language (pp. –). New York: Elsevier. Steinhauer, K., White, E. J., & Drury, J. E. (). Temporal dynamics of late second language acquisition: Evidence from event-related brain potentials. Second Language Research, (), –. Steinhauer, K., Drury, J. E., Portner, P., Walenski, M., & Ullman, M. T. (). Syntax, concepts, and logic in the temporal dynamics of language comprehension: Evidence from event-related potentials. Neuropsychologia, (), –. Steinhauer, K., Drury, J. E., Royle, P., & Fromont, L. A. (). The priming of priming: Evidence that the N reflects context-dependent post-retrieval word integration in working memory. Neuroscience Letters, , –. doi: ./j.neulet... Steinkrauss, R., & Schmid, M. S. (). Entrenchment and language attrition. In H.-J. Schmid (Ed.), Entrenchment and the Psychology of Language Learning: How We Reorganize and Adapt Linguistic Knowledge (pp. –). Berlin: De Gruyter Mouton. Stilwell, B. L., Dow, R. M., Lamers, C., & Woods, R. T. (). Language changes in bilingual individuals with Alzheimer’s disease. International Journal of Language & Communication Disorders, (), –. Stocco, A., Lebiere, C., & Anderson, J. R. (). Conditional routing of information to the cortex: A model of the basal ganglia’s role in cognitive coordination. Psychological Review, (), –. Stöhr, A., Akpinar, D., Bianchi, G., & Kupisch, T. (). Gender marking in Italian-German heritage speakers and L-learners of German. In K. Braunmüller & C. Gabriel (Eds.), Multilingual Individuals Multilingual Societies (MIMS) (pp. –). Amsterdam: Benjamins. Stolberg, D., & Münch, A. (). ‘Die Muttersprache vergisst man nicht’—or do you? A case study in L attrition and its (partial) reversal. Bilingualism: Language and Cognition, (), –. Strik, T. (). Integration tests: helping or hindering integration? Report no. to Parliamentary Assembly, Council of Europe, Strasbourg. Stringer, D. (). The gloss trap. In Z.-H. Han and T. Cadierno (Eds.), Linguistic Relativity in SLA: Thinking for Speaking (pp. –). Clevedon: Multilingual Matters. Sundara, M., Polka, L., & Baum, S. (). Production of coronal stops by simultaneous bilingual adults. Bilingualism: Language and Cognition, (), –. doi: ./S Sunderman, G., & Kroll, J. F. (). An investigation of lexical form, meaning, and grammatical class. Studies in Second Language Acquisition, , –. Suomi, K., McQueen, M. J., & Cutler, A. (). Vowel harmony and speech segmentation in Finnish. Journal of Memory and Language, , –. Swanenberg, J. (). Leerlingen van Poolse scholen in Nederland. Taal, cultuur en identiteit [Students of Polish schools in the Netherlands. Language, culture and identity]. In N. de Jong, K. Juffermans, M. Keijzer, & L. Rasier (Eds.), Papers of the Anéla, Applied Linguistics Conference (pp. –). Delft: Eburon. Sypiańska, J. (). L vowels of multilinguals: The applicability of SLM in multilingualism. Research in Language, (), –. Szupica-Pyrzanowska, M. (). Language attrition: Implications for second/foreign language acquisition. Lingwistyka Stosowana/Applied Linguistics/Angewandte Linguistik, , –.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Takesian, A. E., & Hensch, T. K. (). Balancing plasticity/stability across brain development. Progress in Brain Research, , –. Takeuchi, H., Sekiguchi, A., Taki, Y., et al. (). Training of working memory impacts structural connectivity. Journal of Neuroscience, (): –. Tamargo, R. E. G., Kroff, J. R. V., & Dussias, P. E. (). Examining the relationship between comprehension and production processes in code-switched language. Journal of Memory and Language, , –. Tanenhaus, M. K., & Trueswell, J. C. (). Eye movements and spoken language comprehension. In M. J. Traxler and M. A. Gernsbacher (Eds.), Handbook of Psycholinguistics (pp. –). Amsterdam: Elsevier. Tanenhaus, M. K., Spivey-Knowlton, M. J., Eberhard, K. M., & Sedivy, J. C. (). Integration of visual and linguistic information in spoken language comprehension. Science, (), –. Taniguchi, J. (). Literacy retention and development in Japanese returnee children. The Japanese Journal of Language in Society, (), –. Tanner, D. (). On the left anterior negativity (LAN) in electrophysiological studies of morphosyntactic agreement: A commentary on ‘Grammatical agreement processing in reading: ERP findings and future directions’ by Molinaro et al., . Cortex, , –. Tanner, D., & Van Hell, J. G. (). ERPs reveal individual differences in morphosyntactic processing. Neuropsychologia, , –. Tanner, D., McLaughlin, J., Herschensohn, J., & Osterhout, L. (). Individual differences reveal stages of L grammatical acquisition: ERP evidence. Bilingualism: Language and Cognition, (), –. Tanner, D., Inoue, K., & Osterhout, L. (). Brain-based individual differences in on-line L sentence comprehension. Bilingualism: Language and Cognition, (), –. Taura, H. (). Language Attrition and Retention in Japanese Returnee Students. Tokyo: Akashi Shoten. Taura, H. (). Brain activation, connectivity and the L onset age effect. Paper presented at Bilingualism vs Monolingualism: A new perspective on limitations onto L acquisition (BiMo ), Toulouse, France. Taura, H., & Nakanan, M. (). Bilingual first language attrition from linguistic and neuroimaging perspectives: A functional near-infrared spectroscopy (fNIRS) study. Studies in Language Science, , –. Taura, H., & Taura, A. (). Bilingual first language development of Japanese-English bilingual siblings: Literature review on possible dependent variables. Gengo Bunka Gaku Kenkyuu, Gengo Jouhouhen [Research in Language and Cultural Studies]. hdl.handle.net// Tavakolian, S. (). Structural principles in the acquisition of complex sentences. PhD dissertation, University of Massachusetts. Tees, R. C., & Werker, J. F. (). Perceptual flexibility: Maintenance or recovery of the ability to discriminate non-native speech sounds. Canadian Journal of Psychology/Revue canadienne de psychologie, (), –. Terpstra, J., & Magel, R. (). A new nonparametric test for the ordered alternative problem. Journal of Nonparametric Statistics, (), –. Thelen, E., & Smith, L. B. (). A Dynamic Systems Approach to the Development of Cognition and Action. Cambridge: MAMIT Press. Thelen, E., & Smith, L. B. (). Dynamic systems theories. In W. Damon and R. M. Lerner (Eds.), Handbook of Child Psychology, Vol. I. New York: John Wiley and Sons. Thierry, G., & Wu, Y. J. (). Brain potentials reveal unconscious translation during foreignlanguage comprehension. Proceedings of the National Academy of Sciences, , –. doi: ./pnas. Thiessen, E., & Erickson, L. (). Perceptual development and statistical learning. In B. MacWhinney & W. O’Grady (Eds.), The Handbook of Language Emergence (pp. –). New York: Wiley.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Thomas, E. (). Sociophonetics: An Introduction. Basingstoke: Palgrave Macmillan. Thomas, M. (). Knowledge of Reflexives in a Second Language. Amsterdam: John Benjamins. Thomason, S. G. (). Language Contact: An Introduction. Washington, DC: Georgetown University Press. Thomason, S. G., & Kaufman, T. (). Language Contact, Creolization, and Genetic Linguistics. Berkeley: University of California Press. Thordardottir, E. (). The relationship between bilingual exposure and vocabulary development. International Journal of Bilingualism, (), –. doi: ./ Thordardottir, E. (). The effect of amount of input on simultaneous and sequential bilingual acquisition. Paper presented at the International Conference on Multilingualism at McGill University, Montreal, – October. Thordardottir, E. (). The typical development of simultaneous bilinguals: Vocabulary, morphosyntax and language processing in two age groups of Montreal preschoolers. In T. Grüter & J. Paradis (Eds.), Input and Experience in Bilingual Development (pp. –). Amsterdam: John Benjamins. Tice, M., & Woodley, M. (). Paguettes and bastries: Novice French learners show shifts in native phoneme boundaries. UC Berkeley Phonology Lab Annual Report, , –. Tobin, S. (). A dynamic approach to phonetic change. In The Scottish Consortium for ICPhS (Ed.), Proceedings of the th International Congress of Phonetic Sciences (paper number ). Glasgow: University of Glasgow. Tokowicz, N., & MacWhinney, B. (). Implicit and explicit measures of sensitivity to violations in second language grammar: An event-related potential investigation. Studies in Second Language Acquisition, , –. Tolsma, J., Lubbers, M., & Gijsberts, M. (). Education and cultural integration among ethnic minorities and natives in The Netherlands: A test of the integration paradox. Journal of Ethnic and Migration Studies, , –. Tomasello, M. (). The item-based nature of children’s early syntactic development. In M. Tomasello & E. Bates (Eds.), Language Development: The Essential Readings (pp. –). Malden, MA: Blackwell. Tomblin, J. B., Records, N., Buckwalter, P., Zhang, X., Smith, E., & O’Brien, M. (). Prevalence of specific language impairment in kindergarten children. Journal of Speech, Language, and Hearing Research, , –. Tomiyama, M. (a). The first stage of second language attrition: A case study of a Japanese returnee. In L. Hansen (Ed.), Second Language Attrition in Japanese Context (pp. –). Oxford: Oxford University Press. Tomiyama, M. (b). The later stages of natural L attrition. In P. Robinson (Ed.), Representation and Processes: Proceedings of the rd Pacific Second Language Forum, Vol. (pp. –). Tokyo: PacSLRF. Tomiyama, M. (). Child second language attrition: A longitudinal case study. Applied Linguistics, (), –. Tomiyama, M. (). Age and proficiency in L attrition: Data from two siblings. Applied Linguistics, (), –. Tomoda, M. (). Nihonjin kikokushijo ni mirareru dainigengo soushitsukatei [L attrition process in Japanese returnees]. Language, Information, Text, , –. Toribio, A. J., & Nye, C. (). Restructuring of reverse psychological predicate. Amsterdam Studies in the Theory and History of Linguistic Science Series, , –. doi: ./cilt..tor Torrens, V., Escobar, L., & Wexler, K. (). The acquisition of experiencers in Spanish L and the external argument requirement hypothesis. Language Acquisition and Language Disorders, , –. doi: ./lald..tor Tóth, G. (). Linguistic Interference and First Language Attrition: German and Hungarian in the San Francisco Bay Area. New York: Peter Lang.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Tournier, J. D., Mori, S., & Leemans, A. (). Diffusion tensor imaging and beyond: Magnetic Resonance in Medicine, (), –. doi: ./mrm. Toyoizumi, T., Miyamoto, H., Yazaki-Sugiyama, Y., Atapour, N., Hensch, T. K., & Miller, K. D. (). A theory of the transition to critical period plasticity: Inhibition selectively suppresses spontaneous activity. Neuron, (), –. Traxler, M. J., & Tooley, K. M. (). Priming in sentence comprehension: Strategic or syntactic? Language and Cognitive Processes, (), –. Treffers-Daller, J., Ozsoy, A. S., & Van Hout, R. (). (In)Complete acquisition of Turkish among Turkish German bilinguals in Germany and Turkey: An analysis of complex embeddings in narratives. International Journal of Bilingual Education and Bilingualism, (), –. doi: ./beb. Treffers-Daller, J., Daller, M., Furman, R., & Rothman, J. (). Ultimate attainment in the use of collocations among heritage speakers of Turkish in Germany and Turkish–German returnees. Bilingualism: Language and Cognition, (), –. doi: ./S Truscott, J. (a). Consciousness and Second Language Learning. Bristol: Multilingual Matters. Truscott, J. (b). Consciousness in SLA: A modular perspective. Second Language Research, (), –. Truscott, J. (). Modularity, working memory, and second language acquisition: a research program. Second Language Research, (), –. Truscott, J., & Sharwood Smith, M. (). Representation, Processing and Code-switching. Bilingualism: Language & Cognition, (), –. Tse, L. (). Affecting affect: The impact of heritage language programs on student attitudes. In S. Krashen, L. Tse, & J. Mcquillan (Eds.), Heritage Language Development (pp. –). Culver City: CA: Language Education Associates. Tse, L. (a). Why Don’t They Learn English? Separating Fact from Fallacy in the U.S. Language Debate. New York: Teachers College Press. Tse, L. (b). Resisting and reversing language shift: Heritage-language resilience among U.S. native biliterates. Harvard Educational Review, (), –. https://doi.org/./haer... kumj Tsimpli, I. M. (). First language attrition from a minimalist perspective: Interface vulnerability and processing effects. In B. Köpke, M. S. Schmid, M. Keijzer, & S. Dostert (Eds.), Language Attrition: A Theoretical Perspective (pp. –). Amsterdam: John Benjamins. Tsimpli, I. M. (). Early, late or very late? Timing acquisition and bilingualism. Linguistic Approaches to Bilingualism, (), –. doi: ./lab...tsi Tsimpli, I. M., & Dimitrakopoulou, M. (). The interpretability hypothesis: Evidence from whinterrogatives in second language acquisition. Second Language Research, (), –. Tsimpli, I. M., & Sorace, A. (). Differentiating interfaces: L performance in syntax-semantics and syntax-discourse phenomena. In D. Bamman, T. Magnitskaia, & C. Zaller (Eds.), Proceedings of the th Annual Boston University Conference on Language Development (pp. –). Somerville, MA: Cascadilla Press. Tsimpli, I. M., Sorace, A., Heycock, C., & Filiaci, F. (). First language attrition and syntactic subjects: A study of Greek and Italian near-native speakers of English. International Journal of Bilingualism, (), –. Tuller, L. (). Clinical use of parental questionnaires in multilingual contexts. In S. Armon-Lotem, J. de Jong & N. Meir (Eds.), Assessing Multilingual Children: Disentangling Bilingualism from Language Impairment (pp. –). Bristol: Multilingual Matters. Turian, D., & Altenberg, E. P. (). Compensatory strategies of child first language attrition. In H. W. Seliger & R. M. Vago (Eds.), First Language Attrition (pp. –). Cambridge: Cambridge University Press. doi: ./CBO. Ulbrich, C., & Ordin, M. (). Can L-English influence L-German? The case of post-vocalic /r/. Journal of Phonetics, , –.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Ullman, M. T. (a). The declarative/procedural model of lexicon and grammar. Journal of Psycholinguistic Research, , –. Ullman, M. T. (b). The neural basis of lexicon and grammar in first and second language: The Declarative/Procedural Model. Bilingualism: Language and Cognition, (), –. Ullman, M. T. (c). A neurocognitive perspective on language: The declarative/procedural model. Nature Reviews: Neuroscience, , –. Ullman, M. T. (). Contributions of memory circuits to language: The declarative/procedural model. Cognition, , –. Ullman, M. T. (). A cognitive neuroscience perspective on second language acquisition: the declarative/procedural model. In C. Sanz (Ed.), Mind and Context in Adult Second Language Acquisition: Methods, Theory, and Practice (pp. –). Washington, DC: Georgetown University Press. Unsworth, S. (). Assessing the role of current and cumulative exposure in simultaneous bilingual acquisition: The case of Dutch gender. Bilingualism: Language and Cognition, (), –. doi: ./S Unsworth, S. (). Comparing the role of input in bilingual acquisition across domains. In T. Grüter & J. Paradis (Eds.), Input and Experience in Bilingual Development (pp. –). Amsterdam: John Benjamins. Unsworth, S. (). Amount of exposure as a proxy for dominance in bilingual language acquisition. In C. Silva-Corvalan & J. Treffers-Daller (Eds.), Language Dominance in Bilinguals: Issues of Measurement and Operationalization (pp. –). Cambridge: Cambridge University Press. Unsworth, S. (a). Quantity and quality of language input in bilingual language development. In E. Nicoladis & S. Montanari (Eds.), Lifespan Perspectives on Bilingualism (pp. –). Berlin: Mouton de Gruyter/APA. Unsworth, S. (b). Early child L acquisition: Age or input effects? Neither, or both? Journal of Child Language, (), –. doi: ./SX Unsworth, S., & Hulk, A. (). L acquisition of neuter gender in Dutch: Production and judgement. Proceedings of GALA (pp. –). Cambridge: Cambridge Scholars Publishing. US Census Bureau (). American Community Survey. http://www.census.gov/programs-surveys/acs Vago, R. (). Paradigmatic regularity in first language attrition. In H. Seliger & R. Vago (Eds.), First Language Attrition (pp. –). Cambridge: Cambridge University Press. Valdés Kroff, J. R., Dussias, P. E., Gerfen, C., Perrotti, L., & Bajo. M. T. (). Experience with code-switching modulates the use of grammatical gender during sentence processing. Linguistic Approaches to Bilingualism, (), –. doi: ./lab..val Valdés, G. (). The world outside and inside schools: Language and immigrant children. Educational Researcher, (), –. Valdés, G. (). The teaching of heritage languages: an introduction for Slavic-teaching professionals. In O. Kagan & B. Rifkin (Eds.), The Learning and Teaching of Slavic Languages and Cultures: Toward the st Century (pp. –). Bloomington, IN: Slavica. Valdman, A. (). Language attrition and the administration of secondary school and college foreign language instruction. In R. D. Lambert & B. Freed (Eds.), The Loss of Language Skills (pp. –). Rowley, MA: Newbury House. Van Assche, E., Duyck, W., & Hartsuiker, R. J. (). Bilingual word recognition in a sentence context. Frontiers in Psychology, , . http://doi.org/./fpsyg.. Van Assche, E., Duyck, W., & Gollan, T. H. (). Whole-language and item-specific control in bilingual language production. Journal of Experimental Psychology: Learning, Memory, and Cognition, (), –. Van Coetsem, F. (). A General and Unified Theory of the Transmission Process in Language Contact. Heidelberg: Winter. Van de Meerendonk, N., Kolk, H. H., Chwilla, D. J., & Vissers, C. T. W. (). Monitoring in language perception. Language and Linguistics Compass, (), –. Van der Hoeven, N., & de Bot, K. (). Relearning in the elderly: Age-related effects on the size of savings. Language Learning, (), –.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Van der Hulst, H., & Van de Weijer, J. (). Vowel harmony. In J. A. Goldsmith (Ed.), The Handbook of Phonological Theory (pp. –). Oxford: Blackwell Publishers. Van Dijk, M., Verspoor, M., & Lowie, W. (). Variability and DST. In M. Verspoor, K. de Bot, & W. Lowie (Eds.), A Dynamic Approach to Second Language Development: Methods and Techniques, Vol. (pp. –). Amsterdam: John Benjamins. Van Geert, P. (). A dynamic systems model of cognitive and language growth. Psychological Review, (), –. Van Geert, P. (). Dynamic systems in second language learning: Some general methodological reflections. Bilingualism: Language and Cognition, (), –. Van Geert, P. (). The Dynamic Systems approach in the study of L and L acquisition: An introduction. The Modern Language Journal, (), –. Van Geert, P. (). A comprehensive Dynamic Systems Theory of language development. In K. de Bot & R. W. Schrauf (Eds.), Language Development Over the Lifespan (pp. –). New York: Routledge. Van Geert, P., & Van Dijk, M. (). Ambiguity in child language, the problem of interobserver reliability in ambiguous observation data. First Language, (), –. Van Geert, P., & Verspoor, M. (). Dynamic systems and language development. In B. MacWhinney & W. O’Grady (Eds.). Handbook of Language Emergence (pp. –). New York: Wiley. Van Hell, J. G., & De Groot, A. M. (). Sentence context modulates visual word recognition and translation in bilinguals. Acta psychologica, (), –. Van Hell, J. G., & Dijkstra, T. (). Foreign language knowledge can influence native language performance in exclusively native contexts. Psychonomic Bulletin & Review, (), –. Van Hell, J. G., & Tokowicz, N. (). Event-related brain potentials and second language learning: Syntactic processing in late L learners at different L proficiency levels. Second Language Research, (), –. Van Heuven, W. J., Dijkstra, T., & Grainger, J. (). Orthographic neighborhood effects in bilingual word recognition. Journal of Memory and Language, (), –. Van Osch, B., Hulk, A., Sleeman, P., & Van Suchtelen, P. I. (). Gender agreement in interface contexts in the oral production of heritage speakers of Spanish in the Netherlands. Linguistics in the Netherlands, (), –. Van Versendaal, D., Rajendran, R., Saiepour, M. H., et al. (). Elimination of inhibitory synapses is a major component of adult ocular dominance plasticity. Neuron, (), –. Varga, Z. (). First language attrition and maintenance among Hungarian speakers in Denmark. PhD thesis, Aarhus University. Ventureyra, V. A. G., & Pallier, C. (). In search of the lost language: The case of adopted Koreans in France. In M. S. Schmid, B. Köpke, M. Keijzer, & L. Weilemar (Eds.), First Language Attrition: Interdisciplinary Perspectives on Methodological Issues (pp. –). Amsterdam: John Benjamins. Ventureyra, V. A. G., Pallier, C., & Yoo, H. (). The loss of first language phonetic perception in adopted Koreans. Journal of Neurolinguistics, , –. Verhaegen, C., & Poncelet, M. (). Changes in naming and semantic abilities with aging from to years. Journal of the International Neuropsychological Society, (), –. Verhaeghen, P. (). Aging and vocabulary score: A meta-analysis. Psychology and Aging, (), –. Verhoeven, L., Steenge, J., & Van Balkom, H. (). Verb morphology as clinical marker of specific language impairment: Evidence from first and second language learners. Research in Developmental Disabilities, , –. Verhoeven, L., Steenge, J., Van Weerdenburg, M., & Van Balkom, H. (). Assessment of second language proficiency in billingual children with specific language impairment: A clinical perspective. Research in Developmental Disabilities, , –. Verspoor, M., & Van Dijk, M. (). Variability in a Dynamic Systems Theory approach to second language acquisition. In C. A. Chapelle (Ed.), The Encyclopedia of Applied Linguistics (pp. –). Oxford: Wiley-Blackwell.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Verspoor, M., Lowie, W., & Van Dijk, M. (). Variability in second language development from a Dynamic Systems perspective. The Modern Language Journal, (), –. Verspoor, M., de Bot, K., & Lowie, W. (Eds.). (). A Dynamic Approach to Second-language Development: Methods and Techniques. Amsterdam: John Benjamins. Verspoor, M., Lowie, W., Van Geert, P., Van Dijk, M., & Schmid, M. S. (). How to sections. In M. Verspoor, K. de Bot, & W. Lowie (Eds.), A Dynamic Approach to Second-language Development: Methods and Techniques (pp. –). Amsterdam: John Benjamins. Vihman, M. (). Getting started without a system: From phonetics to phonology in bilingual development. International Journal of Bilingualism, (), –. doi: ./ Vogel, I. (). One system or two? An analysis of a two-year-old Romanian-English bilingual’s phonology. Papers and Reports on Child Language Development, , –. Vogel-Eyny, A., Obler, L. K., Hyun, J., Paplikar, A., & Higby, E. (in prep.). Lexical retrieval abilities in discourse in healthy older adults. Volterra, V., & Taeschner, T. (). The acquisition and development of language by bilingual children. Journal of Child Language, (), –. doi: ./S Vugs, B., Cuperus, J., Hendriks, M., & Verhoeven, L. (). Visuospatial working memory in specific language impairment: A meta-analysis. Research in Developmental Disabilities, (), –. Waas, M. (). Language Attrition Downunder. Frankfurt: Peter Lang. Wakabayashi, S. (). The nature of interlanguage: SLA of reflexives. Second Language Research, , –. Walters, J. (). Bilingualism: The Sociopragmatic-Psycholinguistic Interface. Mahwah, NJ: Erlbaum. Walton, K. A., Murray, L. J., Gallagher, A. M., Cran, G. W., Savage, M. J., & Boreham, C. (). Parental recall of birthweight: A good proxy of recorded birthweight? European Journal of Epidemiology, (), –. doi: ./A: Wang, X. (). Patterns and causes of attrition of English as a foreign language. PhD thesis, Shandong University. Ward, L. M. (). Synchronous neural oscillations and cognitive processes. Trends in Cognitive Sciences, (), –. Webb, S. (). Receptive and productive vocabulary sizes of L learners. Studies in Second Language Acquisition, , –. Weber-Fox, C. M., & Neville, H. J. (). Maturational constraints on functional specializations for language processing: ERP and behavioral evidence in bilingual speakers. Journal of Cognitive Neuroscience, (), –. Wei, J. (). Selectivity of second language attrition. Theory and Practice in Language Studies, (), –. doi: ./tpls...- Weinreich, U. (). Languages in Contact: Findings and Problems. The Hague: Mouton. Weiss, S., & Mueller, H. M. (). The contribution of EEG coherence to the investigation of language. Brain and Language, (), –. Wells, J. C. (). Accents of English. Cambridge: Cambridge University Press. Weltens, B. (). The attrition of foreign-language skills: A literature review. Applied Linguistics, , –. Weltens, B. (). The Attrition of French as a Foreign Language. Dordrecht: Foris Publications. Weltens, B., & Cohen, A. D. (). Language attrition research: An introduction. Studies in Second Language Acquisition, , –. Weltens, B., & Grendel, M. (). Attrition of vocabulary knowledge. In R. Schreuder & B. Weltens (Eds.), The Bilingual Lexicon (pp. –). Amsterdam: John Benjamins. Weltens, B., & van Els, T. (). The attrition of French as a foreign language: Interim results. In B. Weltens, K. de Bot, & T. van Els (Eds.), Language Attrition in Progress (pp. –). Berlin: Walter de Gruyter. Weltens, B., de Bot, K., & van Els, T. (). Language Attrition in Progress. Berlin: Walter de Gruyter.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Weltens, B., van Els, T., & Schils, E. (). The long-term retention of French by Dutch students. Studies in Second Language Acquisition, (), –. Werker, J. F., & Hensch, T. K. (). Critical periods in speech perception: New directions. Annual Review of Psychology, , –. doi: ./annurev-psych-- Werker, J. F., & Lalonde, C. E. (). Cross-language speech perception: Initial capabilities and developmental change. Developmental Psychology, (), –. doi: ./-... Werker, J. F., & Tees, R. C. (). Cross-language speech perception: Evidence for perceptual reorganization during the first year of life. Infant Behavior and Development, (), –. doi: ./S-()- Werker, J. F., & Tees, R. C. (). Speech perception as a window for understanding plasticity and commitment in language systems of the brain. Developmental Psychobiology, , –. Westergaard, M. (). The acquisition of linguistic variation: Parameters vs. micro-cues. In T. Lohndal (Ed.), In Search of Universal Grammar: From Old Norse to Zoque (pp. –). Amsterdam: John Benjamins. White, E. J., Genesee, F., & Steinhauer, K. (). Brain responses before and after intensive second language learning: Proficiency based changes and first language background effects in adult learners. PloS One, (), e. White, E. J., Hutka, S. A., Williams, L. J., & Moreno, S. (). Learning, neural plasticity and sensitive periods: Implications for language acquisition, music training and transfer across the lifespan. Frontiers in Systems Neuroscience, , . White, L. (). The pro-drop parameter in adult second language acquisition. Language Learning, , –. White, L. (). Universal Grammar and Second Language Acquisition. Amsterdam: John Benjamins. White, L. (). Second language acquisition and Binding Principle B: Child/adult differences. Second Language Research, (), –. White, L. (). Second Language Acquisition and Universal Grammar. Cambridge: Cambridge University Press. White, L. (). Second language acquisition at the interfaces. Lingua, , –. White, L., & Genesee, F. (). How native is near-native? The issue of ultimate attainment in adult second language acquisition. Second Language Research, , –. Whitford, V., & Titone, D. (). The effects of reading comprehension and launch site on frequency—Predictability interactions during paragraph reading. The Quarterly Journal of Experimental Psychology, (), –. Whong-Barr, M., & Schwartz, B. (). Morphological and syntactic transfer in child L acquisition of the English dative alternation. Studies in Second Language Acquisition, , –. Wilson, F. (). Processing at the syntax–discourse interface in second language acquisition. PhD thesis, University of Edinburgh. Wilson, F., Keller, F., & Sorace, A. (). Antecedent preferences for anaphoric demonstratives in L German. In J. Chandlee, M. Franchini, S. Lord, & G. Rheiner (Eds.), Proceedings of Boston University Conference on Language Development, Vol. (pp. –). Somerville, MA: Cascadilla Press. Winford, D. (). An Introduction to Contact Linguistics. Oxford: Blackwell. Winford, D. (). Contact and borrowing. In R. Hickey (Ed.), The Handbook of Language Contact (pp. –). Chichester: Wiley-Blackwell. Wingfield, A., Tun, P. A., & McCoy, S. L. (). Hearing loss in older adulthood what it is and how it interacts with cognitive performance. Current Directions in Psychological Science, (), –. Wittenberg, G., Sullivan, M., & Tsien, J. (). Synaptic reentry reinforcement based network model for long-term memory consolidation. Hippocampus, , –. Wixted, J. T. (). On common ground: Jost’s () law of forgetting and Ribot’s () law of retrograde amnesia. Psychological Review, (), –. Wong-Fillmore, L. (). When learning a second language means losing the first. Early Childhood Research Quarterly, , –.
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
Wong-Fillmore, L. (). Loss of family languages: Should educators be concerned? Theory into Practice, (), –. Woumans, E., Santens, P., Sieben, A., Versijpt, J., Stevens, M., & Duyck, W. (). Bilingualism delays clinical manifestation of Alzheimer’s disease. Bilingualism: Language and Cognition, (), –. Wright, R. (). Factors of lexical competition in vowel articulation. In J. Local, R. Ogden, & R. Temple (Eds.), Papers in Laboratory Phonology, Vol. VI (pp. –). Cambridge: Cambridge University Press. Wu, Y. J., & Thierry, G. (). Chinese-English bilinguals reading English hear Chinese. The Journal of Neuroscience, (), –. Wurm, L. H., & Fisicaro, S. A. (). What residualizing predictors in regression analyses does (and what it does not do). Journal of Memory and Language, , –. doi: ./j.jml... Xu, X. (). English language attrition and retention in Chinese and Dutch university students. PhD thesis, University of Groningen. Yager, L. (). Morphosyntactic variation and change in Wisconsin Heritage German. PhD thesis, University of Wisconsin-Madison. Yager, L., Hellmold, N., Joo, H., et al. (). New structural patterns in moribund grammar: Case marking in Heritage German. Frontiers in Psychology, , . Yağmur, K. (). First language attrition among Turkish Speakers in Sydney. PhD thesis, Tilburg University Press. Yağmur, K. (). Language maintenance patterns of Turkish immigrant communities in Australia and Western Europe: The impact of majority attitudes on ethnolinguistic vitality perceptions. International Journal of the Sociology of Language, , –. Yağmur, K., & Van de Vijver, F. J. R. (). Acculturation and language orientations of Turkish immigrants in Australia, France, Germany, and the Netherlands. International Association for Cross-Cultural Psychology, (), –. Yağmur, K., de Bot, K., & Korzilius, H. (). Language attrition, language shift and ethnolinguistic vitality of Turkish in Australia. Journal of Multilingual and Multicultural Development, (), –. Yang, C. (). Knowledge and Learning in Natural Language. New York: Oxford University Press. Yao, Y., & Chang, C. B. (). On the cognitive basis of contact-induced sound change: Vowel merger reversal in Shanghainese. Language, (), –. Yeni-Komshian, G. H., Flege, J. E., & Liu, S. (). Pronunciation proficiency in the first and second languages of Korean-English bilinguals. Bilingualism: Language and Cognition, (), –. Yeung, H. H., Chen, K. H., & Werker, J. F. (). When does native language input affect phonetic perception? The precocious case of lexical tone. Journal of Memory and Language, , –. doi: ./j.jml... Yılmaz, G. (). Complex embeddings in free speech production among late Turkish-Dutch bilinguals. Language, Interaction and Acquisition, (), –. Yılmaz, G. (). First language attrition and second language acquisition among Turkish speakers in the Netherlands. PhD thesis, University of Groningen. Yılmaz, G., & Schmid, M. S. (). L accessibility among Turkish-Dutch bilinguals. Mental Lexicon, (), –. Yip, V., & Matthews, S. (). Syntactic transfer in a Cantonese–English bilingual child. Bilingualism: Language and Cognition, (), –. Yip, V., & Matthews, S. (). The Bilingual Child: Early Development and Language Contact. Cambridge: Cambridge University Press. Yoel, J. (). Evidence for first-language attrition of Russian Sign Language among immigrants to Israel. In D. Quinto-Pozos (Ed.), Sign Languages in Contact (p. –). Washington, DC: Gallaudet University Press. Yoon, S.-Y. (). Acoustic properties of Korean stops as L produced by L learners of the English language. Communication Sciences and Disorders, (), –. Yoshida, K., & Arai, K. (). Kikoku shijo no risuningu noryoku no hoji ni knsuru kosatsu [On the retention of listening skills of returnees]. In Kikoku shijo no gaikokugo hoji ni kansuru chosa
OUP CORRECTED PROOF – FINAL, 22/6/2019, SPi
kenkyu hokokusho [A survey on the foreign language retention of returnees], Vol. (pp. –). Tokyo: Kaigai shijo kyouikushinko zaidan [Japan overseas educational services]. Yoshida, K., Arai, K., Fujita, T., et al. (). Kikoku shijo no gaikokugo hoji ni kansurucihi kosatsu [A consideration on the retention of a foreign language by returnees]. In Kikoku shijo no gaikokugo hoji ni kansuru chousa kenkyu hokokusho [A survey on the foreign language retention of returnees], Vol. (pp. –). Tokyo: Kaigai shijo kyoiku shinko zaidan [Japan overseas educational services]. Yoshitomi, A. (). Towards a model of language attrition: Neurobiological and psychological contributions. Issues in Applied Linguistics, , –. Yoshitomi, A. (). On the loss of English as a L by Japanese returnee children. In L. Hansen (Ed.), Second Language Attrition in Japanese Contexts (pp. –). Oxford: Oxford University Press. Yuan, B. (). Interpretation of binding and orientation of the Chinese reflexive ziji by English and Japanese speakers. Second Language Research, (), –. Yukawa, E. (). L Japanese Attrition and Regaining: Three Case Studies of Two Early Bilingual Children. Tokyo: Kurosio Publishing. Yusa, N., Nasukawa, K., Koizumi, M., Kim, J., Kimura, N., & Emura, K. (). Unexpected effects of the second language on the first. In K. Dziubalska-Kołaczyk, M. Wrembel, & M. Kul (Eds.), New Sounds : Proceedings of the th International Symposium on the Acquisition of Second Language Speech (pp. –). Poznań: Adam Mickiewicz University. Zahodne, L. B., Schofield, P. W., Farrell, M. T., Stern, Y., & Manly, J. J. (). Bilingualism does not alter cognitive decline or dementia risk among Spanish-speaking immigrants. Neuropsychology, (), . Zappalà, G., Thiebaut de Schotten, M., & Eslinger, P. J. (). Traumatic brain injury and the frontal lobes: What can we gain with diffusion tensor imaging? Cortex, , –. Zdorenko, T., & Paradis, J. (). Articles in child L English: When L and L acquisition meet at the interface. First Language, (–), –. Zeithamova, D., Schlichting, M. L., & Preston, A. R. (). The hippocampus and inferential reasoning: Building memories to navigate future decisions. Frontiers in Human Neuroscience, , . Zevin, J. D. (). A sensitive period for shibboleths: The long tail and changing goals of speech perception over the course of development. Developmental Psychobiology, (), –. Zhou, L., Duff, F. J., & Hulme, C. (). Phonological and semantic knowledge are causal influences on learning to read words in Chinese. Scientific Studies of Reading, (), –. doi: ./ .. Zhou, W. (). Assessing birth language memory in young adoptees. PhD thesis, Radboud University. Zürrer, P. (). Sprachkontakt in Walser Dialekten: Gressoney und Issime im Aostatal (Italian). Stuttgart: F. Steiner.
OUP CORRECTED PROOF – FINAL, 25/6/2019, SPi
I .......................
-M model , , , –,
A
Activation Threshold Hypothesis , , , , , , , –, –, , –, –, , –, , –, , , Adaptive Control Hypothesis , , adoption , –, , , , –, , , –, , , , –, –, , –, , , , , , , –, , ageing , , –, , –, , , , , , , , –, – aphasia , , –, –, , , , , aptitude , , , , , –, , , , , , aspect –, , , , , , , , , , , attitude , , –, , , , –, , , –, –, –, , , , see also motivation
B
behavioural ecology of bilingualism , biculturalism , Bilingual Interactive Activation Model (BIA) binding , , , , , –, –, , , , , borrowing , , , , , , –, , , Bottleneck Hypothesis –, , brain imaging , , , , , , –,
C
case –, , –, , , –, –, , –, , , , , , , , , , catastrophic interference –, , , circumlocution , , co-activation , , , , –, , , , code-switching see switching competence , , –, , , –, , –, , –, , , –, , , –, , , , , , , –, , , –, , , –, –, –, , –, –, , –, , competition –, –, , , , –, –, , , , , , , , , , , , , , –, , , , , , ,
Competition Model , –, , , complementarity , complexity , , , –, , , , , , , –, –, –, , –, , , , , , , , , , , –, , , , , , , –, , , , , , comprehension , –, , , , –, , , –, , –, –, –, , , , , , , –, –, , –, , , –, , , , , , , , , , , , , , , connectionism , , , , , control mechanism , , , convergence , , , , , , , , , , , , –, –, , , , critical period , , , , –, –, , , , , , –, , , , , , –, , see also maturational constraints; sensitive period Critical Threshold Hypothesis , , , –, –, –, , –, C-Test , , , , , , ,
D
decay , , , , , , , , , , , determiner , , , , , , developmental language disorder (DLD) –, – see also specific language impairment differential object marking (DOM) –, , , –, dominance , , , , , , , , , , , , , , , , , , , , , , , –, , Dynamic Systems Theory (DST) –, –, , –, , , , –, , , , , , , , , ,
E
education –, , –, , , , , , , –, , –, , –, , , –, , , , , , –, , , , , , , , electroencephalogram (EEG) see event-related potentials
OUP CORRECTED PROOF – FINAL, 25/6/2019, SPi
emergentism –, , , , , –, , –, , , , emotions , , , , , , , , , –, , ethnolinguistics , , , –, – European Graduate Network on Language Attrition , , , , , , , , event-related potentials (ERPs) , , , –, , –, –, –, , , –, , , , , , , , –, , –, –, , , executive function , –, , –, , , , eye-tracking , –, , , , –, –,
F
fluency , , , , –, –, , , , –, –, , , , , , , –, , –, –, , , , –, , , –, –, , –, fMRI , , , –, –, , , , , , , , fossilization , free speech , , , , , , , –, , , , , –, , , –, –, –
G gender biological gender , , , grammatical gender , , , , , –, , , –, , , –, –, –, , , , –, , –, –, , –, , , grammaticality judgements , , , , , , , , –, , , , , , , ,
H
Heritage Bilingual Paradox , hypnosis , –,
I
identity , , , , , , , , , –, , , incomplete acquisition , , –, , –, , , , , , , , , , , , , –, –, , –, , , , –, , incomplete transmission , incubation –, –, , , –, , –, –, –, , –, –, inhibition , , , , –, , , , , , –, , , –, –, , , , , , , –, , , –, , , , see also suppression inhibitory control , , , , , inner speech –
interface , –, –, , , , –, –, , , –, , –, , –, , , , , , , , –, –, , interference , –, , , , , , , –, , –, , –, –, , , , , , , , , –, , , , , , , , –, , , , , , interlanguage , interpretability ,
K knowledge explicit knowledge , –, , , –, , , , , , implicit knowledge , , , , ,
L
language acquisition device (LAD) , language change , , , , , , , , –, , , –, , , –, , , , –, , , , , , language contact , , , , –, , , , , –, , –, , –, , –, , , , –, , , language mode , , , , , , language shift , , , , , , , , language switching see switching language use , –, –, , , , –, , , , , –, , , –, –, , , , , , , , –, , , , –, , , , , , , , , , , , , –, , , , , , , –, , , , , length of residence , –, –, , , , , , –, –, , –, , –, , , , , –, –, –, –, , , –, , , –, , , – lexical access –, , , , , , , , , –, , , , , , –, , , , , , , , , lexical network , , , , , lexical retrieval –, , –, –, –, , , , , , , –, –, , , , –, , , , , lifespan –, –, , , , , , , , , , , –, , , , , , , , , , literacy , , , , , , , –, , , , , –, –, , –, , , –, , , , , , Logical Form –, , , , , , ,
M
Markedness Theory , –, , maturation , , , , , , –, , , , –, , , , , – see also Critical Period
OUP CORRECTED PROOF – FINAL, 25/6/2019, SPi
MEG , memory declarative memory , –, episodic memory –, procedural memory , –, working memory , , –, , , , , –, , , mental grammar, mental lexicon , , , –, , –, –, , metalinguistics , , , , , , , , minimalism –, , , , , , , –, , minority languages , , , , –, , , , –, –, , –, –, , , modularity –, MOGUL , –, mood , , , , motivation –, , –, , , , , –, , –, , –, , , –, –, , , , –, , , , , , , see also attitude multicompetence ,
recall , , , –, , , , , , , , , , , , , , , reduction –, , , , –, –, , , , –, , , –, , , , , , , , , , , , –, , , , Regression Hypothesis , , , –, –, , , , , , –, –, , , –, , , , , , –, , –, –, relative clauses , , –, , , , , , , , , , , , , relative pronouns see pronouns relearning , –, , , –, , , , , , , –, , , –, –, , repetitions , –, , , , , , , repression , research design –, , , , –, –, , , , restructuring , –, –, , , , , , , , –, , , , , , –, , , –, , –, , , , , , ,
N
Savings Paradigm , , , –, , , selectivity –, , , , , –, , –, , , , , , , , –, , , , , , , , , , –, , , self-correction , , , , sensitive period , , , –, simplification , , , –, , –, , , , , Social Networks , , , , –, , social turn , socioculturalism , specific language impairment (SLI) , , , –, see also developmental language disorder subject, null see pro-drop subsystem , , , , , , , , –, , suppression , , , , , , , –, – see also inhibition switch cost – see also language switching switching code-switching –, , , , , , , –, , , , –, , , , , , , , , , , , , language switching –, , ,
naming tasks , –, –, –, , , –, –, , , , , , , , , , , , , , , , , neural activation –, –, , –, , –, Neurolinguistic Theory of Bilingualism
O
offline tasks –, , , , , , , , , , , , ,
P
plasticity –, , , , –, , , , –, , , –, –, , , , , , , , , , , see also maturation priming , –, , , pro-drop null pronouns –, , , , , , –, –, , , null subjects –, , , , , –, –, , , , , pronouns null see pro-drop reflexive , , , , –, , puberty , , , , , –, , , , , , , –, –, , , –, –, , , –, , see also maturation
R
reactivation , , , , , , real-time processing , , , , , ,
S
T
tense –, , , , , , , –, , , –, , , topicalization , , transfer –, , , –, –, , , , , , –, , , , , , , , , –, , , , –, , –, –, , , –, , –, , ,
OUP CORRECTED PROOF – FINAL, 25/6/2019, SPi
translation equivalent –, , , , , trauma , , –, ,
U
ultimate attainment , , , , , , –, , , ungrammaticality , –, , , , , , , , , Universal grammar (UG) , , , –, , – Usage-based (UB) , , , – see also emergentism
V
variability , , , , –, –, , , –, , , –, , , –, , –, , , , , –, , , , , , , , verbal fluency , , –, , , , , , , , –, –, –, , , –, , , VOCD ,
W
weaker links , , word recognition , , , , , , –
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
OXFORD HANDBOOKS IN LINGUISTICS THE OXFORD HANDBOOK OF AFRICAN AMERICAN LANGUAGE Edited by Sonja Lanehart
THE OXFORD HANDBOOK OF APPLIED LINGUISTICS Second edition Edited by Robert B. Kaplan
THE OXFORD HANDBOOK OF ARABIC LINGUISTICS Edited by Jonathan Owens
THE OXFORD HANDBOOK OF CASE Edited by Andrej Malchukov and Andrew Spencer
THE OXFORD HANDBOOK OF CHINESE LINGUISTICS Edited by William S-Y. Wang and Chaofen Sun
THE OXFORD HANDBOOK OF COGNITIVE LINGUISTICS Edited by Dirk Geeraerts and Hubert Cuyckens
THE OXFORD HANDBOOK OF COMPARATIVE SYNTAX Edited by Gugliemo Cinque and Richard S. Kayne
THE OXFORD HANDBOOK OF COMPOSITIONALITY Edited by Markus Werning, Wolfram Hinzen, and Edouard Machery
THE OXFORD HANDBOOK OF COMPOUNDING Edited by Rochelle Lieber and Pavol Štekauer
THE OXFORD HANDBOOK OF COMPUTATIONAL LINGUISTICS Edited by Ruslan Mitkov
THE OXFORD HANDBOOK OF CONSTRUCTION GRAMMAR Edited by Thomas Hoffman and Graeme Trousdale
THE OXFORD HANDBOOK OF CORPUS PHONOLOGY Edited by Jacques Durand, Ulrike Gut, and Gjert Kristoffersen
THE OXFORD HANDBOOK OF DERIVATIONAL MORPHOLOGY Rochelle Lieber and Pavol Štekauer
THE OXFORD HANDBOOK OF DEVELOPMENTAL LINGUISTICS Edited by Jeffrey Lidz, William Snyder, and Joe Pater
THE OXFORD HANDBOOK OF ENDANGERED LANGUAGES Edited by Kenneth L. Rehg and Lyle Campbell
THE OXFORD HANDBOOK OF ELLIPSIS
Edited by Jeroen van Craenenbroeck and Tanja Temmerman
THE OXFORD HANDBOOK OF ERGATIVITY
Edited by Jessica Coon, Diane Massam, and Lisa deMena Travis
THE OXFORD HANDBOOK OF EVENT STRUCTURE Edited by Robert Truswell
THE OXFORD HANDBOOK OF EVIDENTIALITY Edited by Alexandra Y. Aikhenvald
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
THE OXFORD HANDBOOK OF EXPERIMENTAL SEMANTICS AND PRAGMATICS Edited by Chris Cummins and Napoleon Katsos
THE OXFORD HANDBOOK OF GRAMMATICALIZATION Edited by Heiko Narrog and Bernd Heine
THE OXFORD HANDBOOK OF HISTORICAL PHONOLOGY Edited by Patrick Honeybone and Joseph Salmons
THE OXFORD HANDBOOK OF THE HISTORY OF ENGLISH Edited by Terttu Nevalainen and Elizabeth Closs Traugott
THE OXFORD HANDBOOK OF THE HISTORY OF LINGUISTICS Edited by Keith Allan
THE OXFORD HANDBOOK OF INFLECTION Edited by Matthew Baerman
THE OXFORD HANDBOOK OF INFORMATION STRUCTURE Edited by Caroline Féry and Shinichiro Ishihara
THE OXFORD HANDBOOK OF JAPANESE LINGUISTICS Edited by Shigeru Miyagawa and Mamoru Saito
THE OXFORD HANDBOOK OF LABORATORY PHONOLOGY Edited by Abigail C. Cohn, Cécile Fougeron, and Marie Hoffman
THE OXFORD HANDBOOK OF LANGUAGE AND LAW Edited by Peter Tiersma and Lawrence M. Solan
THE OXFORD HANDBOOK OF LANGUAGE AND SOCIETY Edited by Ofelia García, Nelson Flores, and Massimiliano Spotti
THE OXFORD HANDBOOK OF LANGUAGE ATTRITION Edited by Monika S. Schmid and Barbara Köpke
THE OXFORD HANDBOOK OF LANGUAGE EVOLUTION Edited by Maggie Tallerman and Kathleen Gibson
THE OXFORD HANDBOOK OF LANGUAGE POLICY AND PLANNING Edited by James W. Tollefson and Miguel Pérez-Milans
THE OXFORD HANDBOOK OF LEXICOGRAPHY Edited by Philip Durkin
THE OXFORD HANDBOOK OF LINGUISTIC ANALYSIS Second edition Edited by Bernd Heine and Heiko Narrog
THE OXFORD HANDBOOK OF LINGUISTIC FIELDWORK Edited by Nicholas Thieberger
THE OXFORD HANDBOOK OF LINGUISTIC INTERFACES Edited by Gillian Ramchand and Charles Reiss
THE OXFORD HANDBOOK OF LINGUISTIC MINIMALISM Edited by Cedric Boeckx
OUP CORRECTED PROOF – FINAL, 21/6/2019, SPi
THE OXFORD HANDBOOK OF LINGUISTIC TYPOLOGY Edited by Jae Jung Song
THE OXFORD HANDBOOK OF LYING Edited by Jörg Meibauer
THE OXFORD HANDBOOK OF MODALITY AND MOOD Edited by Jan Nuyts and Johan van der Auwera
THE OXFORD HANDBOOK OF MORPHOLOGICAL THEORY Edited by Jenny Audring and Francesca Masini
THE OXFORD HANDBOOK OF NAMES AND NAMING Edited by Carole Hough
THE OXFORD HANDBOOK OF PERSIAN LINGUISTICS Edited by Anousha Sedighi and Pouneh Shabani-Jadidi
THE OXFORD HANDBOOK OF POLYSYNTHESIS Edited by Michael Fortescue, Marianne Mithun, and Nicholas Evans
THE OXFORD HANDBOOK OF PRAGMATICS Edited by Yan Huang
THE OXFORD HANDBOOK OF REFERENCE Edited by Jeanette Gundel and Barbara Abbott
THE OXFORD HANDBOOK OF SOCIOLINGUISTICS Second Edition Edited by Robert Bayley, Richard Cameron, and Ceil Lucas
THE OXFORD HANDBOOK OF TABOO WORDS AND LANGUAGE Edited by Keith Allan
THE OXFORD HANDBOOK OF TENSE AND ASPECT Edited by Robert I. Binnick
THE OXFORD HANDBOOK OF THE WORD Edited by John R. Taylor
THE OXFORD HANDBOOK OF TRANSLATION STUDIES Edited by Kirsten Malmkjaer and Kevin Windle
THE OXFORD HANDBOOK OF UNIVERSAL GRAMMAR Edited by Ian Roberts
THE OXFORD HANDBOOK OF WORLD ENGLISHES Edited by Markku Filppula, Juhani Klemola, and Devyani Sharma