189 22 2MB
English Pages 255 [256] Year 2015
Barbara Hofer On the Dynamics of Early Multilingualism
Trends in Applied Linguistics
Edited by Ulrike Jessner Claire Kramsch
Volume 13
Barbara Hofer
On the Dynamics of Early Multilingualism
A Psycholinguistic Study
ISBN 978-1-61451-874-7 e-ISBN (PDF) 978-1-61451-867-9 e-ISBN (EPUB) 978-1-5015-0054-1 ISSN 1868-6362 Library of Congress Cataloging-in-Publication Data A CIP catalog record for this book has been applied for at the Library of Congress. Bibliographic information published by the Deutsche Nationalbibliothek The Deutsche Nationalbibliothek lists this publication in the Deutsche Nationalbibliografie; detailed bibliographic data are available on the Internet at http://dnb.dnb.de. 6 2015 Walter de Gruyter, Inc., Berlin/Boston Cover image: Roswitha Schacht/morguefile.com Typesetting: RoyalStandard, Hong Kong Printing and binding: CPI books GmbH, Leck ♾ Printed on acid-free paper Printed in Germany www.degruyter.com
Table of contents Acknowledgments List of figures Preface
I
xiii xiv
xvi Structure of the book
xviii
Theoretical background Theoretical underpinnings in early multilingual instruction
1 1.1 1.2 1.3 1.4 1.5 1.6
2 2.1
2.2 2.3 2.4 2.5 2.6 2.7
An Introduction to multilingualism 3 3 On the benefits of an early start 5 Cognitive consequences of being bi/multilingual 9 Recent trends in multilingualism research The dynamics and complexities of multilingual learning and 12 development Third language learning and the role of meta- and crosslinguistic 14 awareness Third language learning and the effects of prior linguistic 15 knowledge The representation of language in the mind 18 Models of language representation: How are words and concepts represented and interconnected in the mind of bilingual and 18 multilingual speakers? 20 The Word-Association and Concept Mediation Model 21 The Revised Hierarchical Model 21 The Distributed Feature Model The Integrated Model: A model for bilingual language representation 22 and processing Lexical Access in bilingual and multilingual speakers: The language 23 specific and language non-specific perspective 28 A dynamic view of the multilingual lexicon
vi 3 3.1 3.2 3.2.1 3.2.2 3.3 3.3.1 3.3.2 3.3.3 3.3.4 3.3.5 3.3.6 3.4 3.4.1 3.4.2 3.5
Table of contents
Language processing 30 Is multilingual processing any different from monolingual or 31 bilingual processing? 32 Language production 33 Levelt: A Monolingual Speech Model 35 Dell: Spreading Activation Model 36 Multilingual language production 37 Poulisse and Bongaerts’ Production Model 39 Myers-Scotton: A Matrix Language Frame 40 Grosjean: The Language Mode Hypothesis 43 Green: The Activation/Inhibition model 47 Clyne: The Plurilingual Model 50 De Bot: The Bilingual/Multilingual Production Model 52 Language comprehension Visual language comprehension in bilingual and multilingual 53 speakers: The BIA, the BIA+, and the MIA model Auditory language processing in bilingual and multilingual speakers: 57 The BIMOLA model for auditory speech comprehension Language representation and processing from a dynamic systems 59 perspective
4 4.1 4.2 4.3 4.4
Developmental models of second and third language learning 61 61 William & Hammarberg’s Model of L3 Production 62 The Dynamic Model of Multilingualism 66 The Biotic Model of Multilinguality A dynamic model of expert-novice co-adaptation during language 67 learning and acquisition
5 5.1 5.2 5.3 5.4
Memory, monitoring and fluency 68 70 Monitoring 70 Automaticity in L2/L3 acquisition 72 L2/L3 fluency Psycholinguistic factors contributing to fluency in L2/L3
6 6.1 6.2 6.3
Multiple language acquisition in children 77 77 Cognitive development in children 78 Theories of child language acquisition Early bilingual acquisition: One system or two?
79
74
vii
Table of contents
6.4 6.5 6.6
The acquisition of morphosyntax in bi- and multilingual 82 children 86 Code-mixing in bilingual and trilingual children Phonological development in bilingual/multilingual children
7
Multilingual proficiency
8 8.1
Multilingual learning in the classroom 101 Is there anything to be gained from multilingual learning?
9 9.1 9.2 9.3
Multilingual awareness 111 111 Learning without awareness? 111 Language awareness and metalinguistic abilities The emergence of metalinguistic awareness in monolingual 117 children Metalinguistic awareness in bilingual and multilingual 119 children 123 Measuring components of metalinguistic awareness
9.4 9.5
94
II A South Tyrol case study Experimental findings in a multilingual setting 10
The aim of the study
11
Historical background
12
The sociolinguistic context: Language education in a bi/trilingual community 136
13 13.1 13.2 13.3
Multilingual educational programmes In South Tyrol 139 140 Subject matter instruction in L2 German 143 Multilingual education and L1 and L2 development 144 Riflessione lingua or reflection on language
14
Research questions and hypotheses
15
Participants
148
91
131 133
146
107
viii
Table of contents
16 Theoretical and methodological framework and study design 151 16.1 The MAT-2 153 16.2 Metalinguistic Ability Test 2: A modified version 153 16.2.1 Acceptability 153 16.2.2 Ambiguity 155 16.2.3 Grammatical Function 155 16.2.4 Phonemic Segmentation 156 16.2.5 Syllable Scansion 156 16.2.6 Identification of Repeated Phonemes 157 16.2.7 Formation of Words 17
Data collection procedure and test administration
18
Scoring
158
159
19 19.1 19.1.1 19.1.2 19.2 19.3 19.4
Analysis and results 161 161 Results MAT-2: L questions and ML questions 162 L questions 163 ML questions 174 Summary and discussion of MAT-2 results 176 Results German and English test Summary and discussion of the German and English test 178 results 178 19.5 Discussion of overall results 182 19.5.1 Educational background mother 183 19.5.2 Educational background father 185 19.5.3 Languages spoken in the home 185 19.5.4 Language learning history
20 20.1 20.2 20.3 20.4 20.5 20.6 20.7 20.8 20.9
Some of the things multilingual children do 189 189 Qualitative data on cross-linguistic awareness 191 Qualitative data on multilingual productions 192 Mixing/Borrowing 193 Dialectal insertions and lexical inventions 194 L2 influences during L3 discourse 194 L3 influences during L2 discourse 195 Trilingual mixing 196 Learner language 196 Multilingual interaction
150
Table of contents
21
Reconceptualising language learning and teaching in South Tyrol. 198 Final recommendations.
22
Conclusion
Bibliography 229 Index
205
203
ix
To Heinz for his loving support, and to my parents who have always been there for me
Acknowledgments I am indebted to many people who have assisted me in various ways during the completion of this volume. I owe my deepest gratitude to Prof. Dr. Ulrike JessnerSchmid for her valuable suggestions and patient support, to Mag. Dagmar Morandell for her help with the statistical evaluation, to Herta Puff and Iris Geier for their kind assistance with the layout, to Dott.ssa Passarella and Dott.ssa Degasperi for granting me permission to conduct my research study at their schools, and to all the teachers at the Manzoni and Longon primary schools for being so immensely helpful in every way.
List of figures Figure Figure Figure Figure Figure Figure Figure
1 2 3 4 5 6 7
The Revised Hierarchical Model 21 34 Levelt’s Blueprint of the Speaker 42 The Trilingual Model 46 Green IC Model 48 Clyne Plurilingual Model The Trilingual or Multilingual Interactive Activation Model 73 The Fluency Model
56
Trilingualism is likely to gain in prominence due to the exigencies of globalization which will encourage people to have their first language, another language of significance in the region, and an international language. Clyne (1997 [2010]:242)
Preface An estimated 6,000 different languages are spoken on our planet (Dewaele, Housen & Wei, 2003:1). For reasons that may be social, political, economic or historical an ever growing number of individuals today speak more than one language. It is in fact estimated that 70% of the world’s population are bilingual or multilingual (see Keshavarz & Astaneh, 2004:295). Dewaele et al. (2003) point out that “[i]ndividual and societal bilingualism are neither recent nor temporary phenomena. Bilingualism has always been with us.” (1), and Edwards (2007) states that “Multilingualism is a powerful fact of life around the world, a circumstance arising, at the simplest level, from the need to communicate across speech communities” (447). Unfortunately, educators and politicians have only just started to acknowledge the prevalence and significance of bi- and multilingualism on both, the individual and the societal level. This notwithstanding, research interest in bilingual and multilingual phenomena has been growing rapidly over the last ten to twenty years. An increasing number of pertinent scientific publications and international conferences and symposiums testify to the growing interest in multilingualism and multiple language learning and development. Scholars are suggesting that the study of multilingual acquisition will “expand our understanding of the human language faculty and by extension the human mind because such research permits us to examine the capacity of the mind to acquire and use more than one language” (Genesee & Cenoz, 2001:2). In addition, it is to be expected that research into multilingualism will have profound effects on linguistic theory (Herdina & Jessner, 2002:1). The study of language and the mind is a central concern of psycholinguistic research. It may be of interest to the reader that systematic research into monolingual acquisition did not commence in earnest until the 1950s when Roger Brown produced his seminal work “Words and Things” (1958) (Genesee, 2003:204)1. By contrast, the study of bilingual acquisition dates back to the early 20th century when linguists such as Ronjat (1913) and Leopold (1939–1949) undertook their first investigations of bilingual phenomena (Barnes, 2006:1; Genesee, 2003:204). As regards the study of trilingual and multilingual phenomena, it has to be noted that TLA and multilingualism research have only very recently gained prominence in applied linguistics and in related areas of research. As pointed out by Jessner (2006), the field of Third Language Acquisi1 We do, however, have reports of early experiments carried out on infants who were deprived of language contact in their first months and years (Barnes, 2006:1).
Structure of the book
xvii
tion is a very young field of research and TLA research has only just begun to manifest itself as a subject in its own right (20). In spite of significant advancements in this young scientific discipline, it is probably true to say that there are still many unknowns about how multilingual learners acquire their languages, about how these languages are represented and how they interact in the mind (cf. Hoffmann, 2001:1). A general lack of interest in multilinguality (due in part to a widespread monolingual prejudice enforcing the application of monolingual norms with regards to learners’ language competence or proficiency) and the relatively small number of research findings produced thus far have been identified as the main causes of this unfortunate state of affairs (cf. Longxing Wei, 2003:60; Müller-Lance, 2003:132; De Bot, 2004:17; Jessner & Herdina, 2002:2). At the same time, the demand for multilingualism and multilingual educational models is increasing not only in Europe but worldwide. In response to this growing demand several countries including Spain (Balearic islands, Calaluña, Galicia, Navarre, the Basque Country, Valencia), France (Brittany), The Netherlands (Friesland), Finland (Vaasa, Turku, Helsinki), Luxembourg and Denmark, have implemented various bilingual or multilingual education programmes (Lasagabaster, 2001a:421). In addition, there is a clear tendency to introduce foreign languages at an earlier age than has been customary so far. These developments can, to some extent at least, be related to very recent findings in TLA research which have provided compelling evidence for the linguistic and cognitive benefits and intellectual gains of (early) multilingual learning (e.g. Galambos & Hakuta, 1988; Cenoz & Valencia, 1994; Lasagabaster, 1997, 1998; Bialystok, 2001, 2007; Cenoz, 2003, 2003a; Keshavearz & Astaneh, 2004; Oberhofer, 2008; De Angelis & Jessner, 2012). In the Autonomous Province of Bolzano/Bozen (Italy) the German language coexists with the Italian and with the Ladin language. During the Fascist years and during World War II the use of the German language was forbidden in schools, the media and public offices. After the war, a number of legal measures were introduced to protect and promote the German language. Today, mother tongue education, seen as a fundamental and irrevocable right, is considered of central importance for strengthening the German language and culture. Multilingual instruction is widely regarded with scepticism. As a consequence there are no German schools in the region that offer multilingual educational programmes. In contrast, a number of Italian schools have recently implemented various multilingual models. In South Tyrol, these programmes are widely referred to as ‘modelli bilingui’ or ‘classi bilingui2’ even though the denomina2 This is due to the fact that these ‘modelli bilingui’ utilise two main languages of instruction (Italian and German), while English L3 is taught as a foreign language (for 3 hours per week). In the present volume we will be referring to these models as multilingual models.
xviii
Preface
tion tri- or multilingual might seem more appropriate since most of them provide for instruction in three languages from first grade onwards. Unfortunately, the socio-political dimension of multilingual schooling in South Tyrol can only be discussed very briefly here (see part II of this book) since it is beyond the scope of the present volume. As will be outlined below, the principal aim of this research is to throw light on psycholinguistic and cognitive dimensions of tri- or multilingual learning and development. The author intends to show that multilingual schooling carries linguistic and cognitive advantages for young learners. The premise of the research study (presented in part II) is that learners in multilingual programmes demonstrate superiority in linguistic and metalinguistic tasks over learners in traditional educational programmes (cf. Hakuta & Diaz, 1985; Lee, 1996). It will be argued that multilingual instruction does not impede the child’s intellectual growth or interfere with the child’s development of the L1. Rather, it is posited here that (early) multilingual learning can lead to a high degree of proficiency in the L1 and L2s (if there is adequate support) and result in heightened levels of metalinguistic and crosslinguistic awareness.
Structure of the book This book is about early multilingual acquisition. It discusses some of the key issues in multilingualism and third language acquisition research and aims to contribute to the current debate of early multilingual development and metalinguistic awareness in young multilingual users by focusing on cognitive aspects of language acquisition and development. The book is divided into two major parts: a theoretical part and the presentation of the research study. The theoretical part provides important background information on a number of issues pertaining to multiple language acquisition and development. Great care has been taken to include a wide range of topics related to multilingual phenomena. There is detailed discussion of multiple language representation and processing, multilingual development in children, multilingual proficiency, multilingual instruction and metalinguistic awareness. The main focus is on how multiple language learning affects young users’ cognitive and linguistic development and how it influences their metalinguistic abilities. The second part of the book reports on a research project which was conducted in the trilingual province of South Tyrol in spring 2011. The study examines the role of multilingual education and the impact of bilingualism on additional language learning and on learners’ awareness of language. The focus is on how children in high-immersion multilingual educational programmes compare to children in
Structure of the book
xix
traditional low-immersion programmes in terms of their linguistic and metalinguistic development. It will be argued that early multilingualism carries positive implications for young learners’ linguistic performance and metalinguistic awareness, and that early contact with an L2/L3, as provided by high-immersion multilingual education, promotes learners’ (meta-)linguistic sensitivity and facilitates the acquisition of additional languages. The author of this book subscribes to a ‘multilingualist’ view (Cook, 2002:22) of language acquisition and positions herself within the framework of the Dynamic Model of Multilingualism, henceforth DMM (Herdina & Jessner, 2002). The DMM postulates a holistic approach to multilingualism. It conceives of multiple language systems as complex, dynamic and interrelated systems (Herdina & Jessner, 2002; Jessner, 2003a; Jessner, 2008a). The DMM proposes to apply Dynamic Systems Theory to multilingual acquisition and development. According to Jessner (2008a), a “Dynamic Systems Theory approach, which uses dynamic modeling to investigate properties of the dynamic adaptation to contexts in change, is able to take all the relevant characteristics of multilingual learning and use into account” (273). Following scholars like Cook (1997, 2002), Grosjean (2001), Clyne (2003), De Bot (2004), Herdina & Jessner (2002) and Jessner (1999, 2006, 2008, 2008a, 2008b) the present author advocates a holistic perspective of multiple language acquisition which conceives of multilingual competence as a complex and constantly changing construct and of the multilingual user as a “complete linguistic entity, an integrated whole” (Baker, 2006:12). Such an approach carries important implications for how multilingual learners and multilingual performance3 are perceived and characterised. Crucially, a holistic approach does not restrict itself to the investigation of what learners will acquire in terms of linguistic structures and grammatical rules in each of their languages. Rather, it entails a conception of multilingual systems which comprises the speaker’s entire linguistic repertoire and a range of language-specific and non-lingustic skills and strategies, including speakers’ socio-linguistic and pragmatic competence and, perhaps most importantly, their metalinguistic and cross-linguistic awareness or sensitivity. In addition it is posited here that a distinction needs to be made between SLA and TLA, for “it is known that learning a second language differs in many respects from learning a third language” (Jessner, 2006:13; see also Jessner, 1999:201). Robust evidence indicating that bilingual learners show superiority over monolingual learners when learning an additional language (Cenoz & Valencia, 1994; Sanz, 2000; Keshavarz & Astaneh, 2004) supports this notion.
3 No distinction is made here in the Chomskyan sense between competence and performance.
xx
Preface
This book purports to demonstrate that early acquisition of a second and third language (as provided by high-immersion multilingual education schemes) fosters learners’ awareness of language and benefits their knowledge of the L1, L2 and L3. The first part of the volume outlines some of the latest developments in bilingualism and multilingualism research and reviews some of the most relevant theoretical perspectives relating to the study of bilingual and multilingual acquisition. A range of theoretical approaches and viewpoints are presented in an attempt to bring together various strands of research. It will be argued that a holistic and dynamic perspective, such as proposed by DMM (Herdina & Jessner, 2002) and by the continuity-of-mind framework (Lowie & Verspoor, 2011), promises to be a most fruitful approach to discussing multilingual phenomena. The second part of the book presents the research study conducted in two Italian primary schools in the city of Bolzano/Bozen. Both schools offer high-immersion multilingual education, i.e. both Italian and German are languges of instruction for approximately 50% of the school time and in addition pupils are taught L3 English (for 3 hours per week) from first grade onwards. The benefits (or dangers, as critics imply) of these and similar multilingual programmes are hotly debated in South Tyrol. It is hoped that this research will contribute to identify and make evident the cognitive and linguistic advantages of (early) multilingual learning. As indicated, the study aims to ascertain the role of multilingual education and the effects of multilingualism on subjects’ linguistic and metalinguistic performance. Following Jessner (2006) it is argued that metalinguistic awareness is “an influential cognitive component which shows its effect on the way multilinguals use and learn their languages” (117) and that raising meta- and cross-linguistic awareness in the classroom should be made a central aim in multilingual schooling (121). The present author concurs with Clyne (2003) that one of the “best way[s] to acquire metalinguistic awareness is through exposure to a second [or third; comment added by this author] language” (49). It will therefore be suggested that multilingual education is a particularly effective way of exposing children to second or third languages from an early age. Most available publications in the field have focused and reported on the bilingual speaker and on bilingual phenomena4. Therefore the terminus bilingualism will be employed throughout this volume whenever it was utilised in the original literature. Apart from these citations or references to the extant literature, the terms trilingualism and multilingualism will be preferred whenever more than two languages are involved. 4 It ought to be noted, however, that the notion of bilingualism is generally used in a wider sense, relating to any additional language, not just the L2.
Structure of the book
xxi
Given that an exhaustive analysis and discussion of all issues relating to early multilingual learning is hardly accomplishable, a selection had to be made as to which thematic fields to include in this volume. Since the main focus of the book is on multilingual acquisition at the primary level, it was estimated that issues of multiple child acquisition and multilingual proficiency and education needed to be addressed in some depth. Further thematic choices concern aspects of multilingual storage and language processing since it was felt that multilingual acquisition and development can be much better understood if one has knowledge of how languages are represented and processed in the mind. As the central questions driving this research focus on multilingual competence and metalinguistic awareness, the impact of multilingual schooling on children’s linguistic and metalinguistic abilities is of major interest here. The starting point of this book (Chapter 1) is a discussion of the positive and negative effects of being bi- and multilingual. Differential perceptions of the value and purpose of bi- and multilingualism are considered. The chapter enlarges upon recent trends in multilingualism research and upon the role of metalinguistic awareness and previous linguistic knowledge in L3 learning. Chapter 2 elaborates on issues of language representation and storage and on the multilingual lexicon. In Chapter 3 and 4, aspects pertaining to speech processing are discussed and some of the most influential models of speech production and comprehension are presented. The main focus is on multilingual language processing and on the mechanisms and computational routines underpinning processing operations in multilingual users. Special attention is given to the Dynamic Model of Multilingualism proposed by Herdina & Jessner (2002) and to Lowie & Verspoor’s (2011) recent proposition of a continuous model of language processing. Chapter 5 explores aspects of memory, fluency and automaticity. This is followed by a detailed account (in Chapter 6) of how children acquire and cope with multiple languages. The chapter also investigates code-switching and code-mixing. It is argued that code-switching and mixing are typical in bi- and multilingual (adults and) children and that switches serve a variety of useful communicative purposes. Multilingual behaviour in general and code-switching in particular are discussed in terms of multilinguals’ enhanced sensitivity to cross-linguistic requirements of well-formedness and linguistic regularity. Chapter 7 focuses on aspects of multilingual proficiency, while Chapter 8 explores issues related to multilingual education and learning. Chapter 9 deals with metalinguistic and cross-linguistic awareness as emergent properties of multilingual development. It is posited that the verbal flexibility and creativity deriving from the constant negotiation of multiple language systems result in higher levels of metalinguistic awareness and cross-linguistic abilites and that contact with two or more linguistic and/or cultural systems carries positive implications for the individual
xxii
Preface
learner/user even if the languages are not mastered equally well. This said, it is to be expected, that a high level of competence in all languages will result in greater overall benefits. The second part of the book centres on the research study. First, an introduction to the history and the sociocultural context in which the study was carried out is given. This is followed by a brief discussion of the effects of multilingual education on the development of pupils’ L1 and L2. Next, the research questions and working hypotheses are laid out and the methodological approach is delineated. In the final part the results of the study are presented and discussed within the broader context of subjects’ bi/multilingual and sociocultural background. The chapter closes with the recommendation to promote early multilingual learning and foster cross-linguistic approaches to language teaching.
I Theoretical Background . . . the multilingual child experiences the world through two or more languages which are used in alternation. For the multilingual, linguistic experience is spread over several languages . . . Adapted from Malakoff & Hakuta (1991:142)
Theoretical underpinnings in early multilingual instruction 1 An Introduction to Multilingualism The past decade has seen an unprecedented interest in the study of multiple language acquisition and multilingual education. As intimated above, this trend has been paralleled and possibly promoted by the rapid growth in psycholinguistic research not only in SLA but increasingly also in the field of TLA. The marked interest in all matters multilingual can also be linked to worldwide societal changes brought about by globalisation and increased migration. In addition, it has become clear that languages have a market value (Clyne, 2003:68; cf. Olshtain & Nissim-Amitai, 2004:34) and that cultural and linguistic diversity constitute an asset to a country’s or region’s economy. Moreover, the perception appears to be winning ground that multilingualism is “not an aberration – as many, in particular monolingual speakers, may still think – but a normal necessity for the world’s majority” (Jessner, 2006:1), and that cultural and linguistic diversity constitute a rich linguistic and intellectual capital and resource for the whole of society (Cummins, 2000, 2001; Garcia, 2012).
1.1 On the benefits of an early start Parents but also teachers and educational authorities have frequently voiced concern as to the cognitive and developmental repercussions of early second or foreign language teaching on children. We know today, that this concern is quite unfounded (cf. Galambos & Hakuta, 1988; Lasagabaster, 2001b:313; Bialystok, 2007; De Angelis & Jessner, 2012). Early research into bilingualism and second language acquisition mostly identified detrimental effects of bilinguality on the speaker’s cognition. Today, our understanding of bilingualism has greatly advanced. There is strong empirical evidence indicating that the benefits of early foreign language acquisition are very real (as we shall also see in Part II of this book) and that any potential deficits are by far outweighed by the advantages of mastering two or more languages. This is illustrated by the recommendations laid down by the European Commission. The Commission’s White Paper (1995) calls for the introduction of second or foreign language teaching at the preschool level and recommends that every European citizen should, in addition to one’s first language, acquire two further community languages. Although
4
An Introduction to Multilingualism
there is no conclusive evidence attesting to the benefits of very early programmes (cf. Etxeberrìa, 2004; Cenoz, 2004, Munoz, 2006, 2011), it is probably true to say that even in the absence of any immediately measurable benefit, longterm benefits are to be expected. This claim is here made on the grounds of the following observations and findings. Firstly, extended exposure to an L2 or L3 has widely been found to benefit learners (cf. Hakuta & Diaz, 1985:323 and Lee, 1996). Secondly, studies on L2 users’ phonological competence show that there is a correlation between age of onset and phonological performance and/ or phonological awareness (Bruck & Genesee, 1995; Price et al., 1999; Andreou, 2007), and thirdly, research findings on the impact of motivation in foreign language learning indicate that young children show positive attitudes and high motivation towards second or foreign language learning. This positive disposition has been found to decrease as children get older (Cenoz, 2004:205). Robinson (1998) suggests that early (second) language learning greatly benefits the child in that it “enhances students’ basic skills in such areas as English language arts, reading, and math. This holds for academically talented children as well as for those of below-average ability, those with handicaps that affect learning, and those at risk” (43; see also Da Fontoura & Siegel, 1995 in Thompson, 2008:135). Robinson’s assertion is echoed by Haas (1998) who notes that “[i]n any area of study, students benefit from starting early and continuing through a long sequence of learning that grows and deepens as they mature” (43). Similarly, Cook (2002) posits that children who acquire a second language at an early age develop better communicative skills, learn to read more rapidly and are more aware of grammatical structures (7). Bilingual or multilingual instruction provides pupils with a vast amount of L2/L3 input which allows young learners to extrapolate the regularities of phonology, morphology, and syntax. The headway that young learners thereby gain in terms of receptive and productive competence constitutes one of the major advantages of early foreign language education. As indicated, this headway is by no means restricted to vocabulary learning but extends to the acquisition and development of morpho-syntactical structures and phonology. Genesee (2003), for instance, intimates that, in contrast to monolingual children, bilingual children possess additional processing capacities which permit them to coordinate the grammars of their two languages on-line (223). Similarly, Price et al. (1999) argue that “individuals who acquire both their languages early are better able to programme phonological responses and are less likely to have a foreign accent” (2233). This obtains because early acquisition of an L2 or L3 causes morphosyntactic structures and phonological features to become deeply entrenched or anchored in the child’s cognitive system. Basing herself on a substantial body of research on the cognitive and academic benefits of (early)
Cognitive consequences of being bi/multilingual
5
language learning, Robinson (1998:41) concludes that “all students can profit academically from participation in an early language program. Time spent in studying a foreign language is time well-spent.” This does, however, not imply that ‘earlier is always better’ (cf. Munoz, 2005:83) or that older students are necessarily disadvantaged with regards to second and third language learning (cf. Munoz, 2006). In point of fact, one ought not to forget that in certain aspects of L3 learning older learners may be more advantaged since they are cognitively more mature, possess a more profound knowledge of the world and can build on a more developed L1 and possibly L2 (particularly as far as literacy skills are concerned).
1.2 Cognitive consequences of being bi/multilingual Until 1962, the widespread assumption held amongst linguists and lay persons was that bilingualism and multilingualism may inhibit the acquisition and full mastery of the speaker’s first language and that it may cause mental confusion (Baker, 2006:143; see also Lee, 1996 and Hakuta & Diaz, 1985). Genesee (2002:170) explicates that the simultaneous acquisition of two languages was or is oftentimes “thought to exceed the language learning capacity of the young child and thus to incur potential costs, such as delayed or incomplete language development or even deviant development”. The early nineteenth century, in particular, saw scholars propound the detrimental effects of bilingualism on the individual’s identity and cognition. It was claimed that bilingualism places an immense strain on the speaker’s mental capacity and that it could even trigger schizophrenia (ibid.; cf. Wandruska, 1979:52). This extreme position was seen as corroborated by the results of early studies on the relationship between bilingualism and intelligence. The findings generated by these studies seemed to imply that bilinguals have lower IQs than monolinguals. Saer (1923) in particular condemned bilingualism for bringing forth mentally confused and retarded individuals. Bilingualism, it was maintained, would burden the brain and cause mental confusion and identity conflicts (Baker, 2006:143). Lüdi (2007) intimates that negative connotations associated with bilingualism and multilingualism might stem from the fact that in the bible people were portrayed as originally having and using only one language. Multilingualism, he suggests, calls to mind Sprachverwirrung, the language confusion which was seen as God’s punishment to his people (39). At the end of the nineteenth century, Laurie, a professor of English at Cambridge University, famously proclaimed that
6
An Introduction to Multilingualism
If it were possible for a child to live in two languages at once equally well, so much the worse. His intellectual and spiritual growth would not thereby be doubled, but halved. Unity of mind and character would have great difficulty in asserting itself in such circumstances (Laurie, 1890:15 cited in Lüdi, 2007: 39).
Early studies of bilingualism characterised bilinguals, and bilingual children in particular, as having “lower IQ scores than monolinguals”, as being “socially malajusted”, and as trailing monolinguals in academic performance (Reynolds, 1991:145). Diaz (1983) reports on studies conducted prior to 1962 which relate to bilingualism as some sort of social plague and as a language handicap found predominantly in the lower social strata (25). Today, as pointed out by Baker (2006) and numerous other scholars, it is generally acknowledged that those early studies were frought with weaknesses and did not comply with scientific measures of reliability and validity (144). The 1960s were to witness radically different developments and attitudes in terms of scholars’ and the general public’s perception of bilingualism. In 1962 two scientific publications brought about a rather unexpected change of perception regarding the consequences of bilingualism and the impact of bilinguality on the speaker’s cognition. One was Peal & Lambert’s seminal article on “The relation of bilingualism to intelligence”, the other was the English edition of Lev Vygotsky’s Thought and Language (Reynolds, 1991:146–147). Vygotsky suggested that learning a second language in childhood “facilitates mastering the higher forms of the native language” (in Swain & Lapkin, 1991:206). This, according to Vygotsky, causes bilingual children to perceive their language(s) as a system among many and invariably makes them more aware of their linguistic operations (ibid.; see also Reynolds, 1991:147). Similarly, Peal & Lambert (1962) posited that bilingualism carries positive implications for the individual learner. The authors found that bilinguals possess a more diversified set of cognitive abilities than monolinguals. The bilingual children participating in their study had obtained better results than the monolingual control group on most cognitive tests and subtests. In particular, the bilingual group performed better on tests measuring verbal and non-verbal abilities. It emerged that bilingual children’s ability to mentally manipulate and reorganise visually presented stimuli was superior to that of the monolingual controls. Peal & Lambert thus concluded that bilingual children are advantaged, compared to monolingual children, particularly in tasks which require mental flexibility and concept formation (Reynolds, 1991:146). Penfield even went as far as intimating that “[t]he bilingual brain is a better brain” (in Lambert, 1991:238). In contradistinction to the large body of research literature on bilingualism and second language acquisition, studies on TLA and multilingualism are some-
Cognitive consequences of being bi/multilingual
7
what sparse. This is partly due to the fact that the field of tri- and multilingualism research is a relatively young one. Encouragingly, though, many recent studies suggest that children who learn an L2 or L3 may be “more intellectually acute” (Andreou & Anastassiou, 2011:111) and that “bilingual learners will acquire an additional language faster and more efficiently” (Safont Jorda`, 2005:46). In contrast to the formerly held belief that bilingualism may have negative implications on the speaker, there are strong indications that bilingualism furthers speakers’ cognitive and linguistic development and that it promotes third language learning. Keshavarz & Astaneh (2004), for instance, evidenced positive effects of bilingualism on TLA. The authors intimate that bilinguals are better and more efficient at foreign language vocabulary learning than monolinguals (300). Keshavarz & Astaneh’s findings are consistent with Rivers (1996) who compared (adult) bilingual and monolingual groups enrolled on various intensive Slavic and non-Slavic language training programmes. Rivers hypothesised that respondents who had previously learned a second language would display different learning behaviours than monolingual subjects who had no prior foreign language learning experience. As expected, their test results yielded “dramatic evidence that experienced language learners do indeed learn faster than novice language learners” (4). Rivers found that third language learners exhibited a high degree of learner autonomy and self-assessment (3) and demonstrated self-directed learning behaviours (5) which enabled them to control and modify learning processes so as to optimise their learning success. Rivers’ study investigated adult learners of a foreign language. However, nothing seems to speak against assuming similarly positive effects of bilingualism for young learners of a third language. Rahimpour (2002) intimates that bilinguals who learn an L3 “derive more benefit from their special learning experience of L2” than L2 learners could ever derive from their L1 (31). Rahimpour claims that bilingual speakers profit immensely from their knowledge of a previously learned language and that they show greater metalinguistic awareness and abilities. According to Rahimpour, bilinguals are better equipped to learn a third language (42; cf. Bono, 2011:25; Ringbom, 2011:20). In a similar vein, Jessner (2006) notes that under specific favourable conditions “being bilingual can have tremendous advantages, not only in terms of linguistic competence but also in terms of cognitive and social development.” (27; see also Jessner, 2008:29). Herdina & Jessner (2002) posit that “[t]he use of two or more languages not only influences the linguistic and cognitive skills of multilinguals, but also their social skills” (64) implying that multilinguals may be more sensitive than monolinguals in interpersonal communication (65). The authors further note that young multilinguals “show definite advantages in cognitive flexibility, creativity,
8
An Introduction to Multilingualism
divergent thought or problem solving” (64) which lends support to the assumption that multilinguals and monolinguals “differ in thinking styles” (64). Various propositions have been advanced in an attempt to account for this bi- or multilingual superiority in certain cognitive and verbal tasks. One such proposal holds that very early on the bi- or multilingual child learns to differentiate linguistic form from meaning thereby developing an increased awareness of language (Reynolds, 1991:169; see also Bruck & Genesee, 1995:308). Reynolds (1991) suggests that the “necessity of monitoring and controlling two (potentially interfering) symbol systems leads to increased metacomponential abilities” (169). Having to control two language systems instead of just one, the bilingual and in particular the bilingual child is seen as developing a heightened awareness of linguistic form and function. Children growing up with two languages from a very early age are predicted to “apprehend the arbitrariness of language earlier than monolinguals” (Reynolds, 1991:168). In addition, they are believed to develop an early understanding and judgement of the nature of certain problems they are faced with, and they are said to possess a greater sensitivity with regard to making decisions and choices not only in terms of language but of a more general nature (169; cf. Bialystok, 2001:179). Bialystok (2002) argues that bilingual5 children are not only advantaged with regard to language but also when faced with general problem solving, since they seem to adapt more readily to novel rules or task instructions and are therefore faster at producing a viable solution to a given problem (158). Bialystok explicates that this is due to the bilingual child’s enhanced capacity to revise, re-interpret and, if necessary, re-code task descriptions (159). According to Bialystok, these latter processing operations require a fair amount of control and inhibition as interfering and misleading information needs to be ignored and new information has to be attended to (ibid.). As specified by Bialystok, the bilingual experience of constantly managing two diverse language systems “may carry over to processing in disparate cognitive domains” (160). In particular, Bialystok (2007) proposes that . . . bilingualism as a normal lifelong experience creates a context in which the executive processes for attentional control are employed routinely, and this use for language management has the felicitous benefit of boosting their function across other cognitive domains, even those that apparently have little connection to linguistic performance. (220)
In bilingual children this control over executive functions has been found to develop earlier than in monolingual children, and in bilingual adults decline of 5 The reader will recall that our reference to the bilingual child does not preclude multilingual children but merely reflects the respective author’s word choice in the original literature.
Recent trends in multilingualism research
9
executive control appears to be delayed and less abrupt (Bialystok, 2007:220). According to Bialystok, this bilingual superiority obtains regardless of the subject’s degree of bilingualism (Ricciardelli, 1992:314). This is contrasted by Ricciardelli’s findings (1992) which imply that a minimum of language competency is required if positive effects on cognition are to be expected (313). Ricciardelli (1992) reports a distinct bilingual advantage for individuals who achieve high levels of proficiency in both their languages, but not for bilinguals who perform poorly in their languages (301). Following Cummins (1976), Ricciardelli suggests that bilinguals need to attain a threshold level of language proficiency in order to gain cognitively from their bilingualism. If no such level is reached, the bilingual speaker’s cognitive development may suffer (302). In line with Bialystok (2007:220), Ricciardelli concludes that the bilingual advantage does not extend to all aspects of human cognition but that it is restricted to some select areas. Ricciardelli conceives of divergent thinking, imagination, grammatical awareness, perceptual organization, and (word) reading as those aspects of cognition which are positively affected by the speaker’s bilinguality (1992:313–14). The underlying assumption is that in order to acquire a second or third language learners need to engage in a number of complex processing operations as they are required to take in, store and recall novel information. It has been suggested that these operations may well “make one think harder” (Robinson, 1998:37). Diaz (1983) points out that “[w]hen compared to monolinguals, balanced bilingual children show definite advantages on measures of metalinguistic abilities, concept formation, field independence, and divergent thinking skills” (48). From this perspective, bilingual competence can be said to differ from monolingual competence not only along linguistic dimensions but also in terms of thinking, in terms of perceiving the world, and in terms of approaching nonlinguistic problems or tasks (44, 46; cf. Gabrys-Barker, 2012). In a similar way, Pavlenko (2010) asserts that multilingualism has beneficial effects on cognition and language (383) inasfar as it enriches speakers’ linguistic repertoire and offers them alternative conceptualizations which may be crucial for flexible and critical thinking (383–384).
1.3 Recent trends in multilingualism research The second part of the 20th century saw many researchers engaged in the construction of psycholinguistic models of bilingual representation and processing, proposing “neat, straightforward, linear and monodimensional solutions” (Schmid & Lowie, 2011:1). These models largely consisted of hierarchically structured, static elements or components. The underlying assumption was that one
10
An Introduction to Multilingualism
event in the processing chain would cause another and that procedures would always occur in the same fashion and order. In recent years, it has, however, become increasingly apparent that modular, monodimensional models can not account for “non-static scenarios, such as linguistic change, language development and multilingualism” (2). In response to the weaknesses of these sequential models, a number of linguists have therefore taken new and radically different directions. Conceiving of languages as complex and dynamic systems and criticising the widespread monolingual or fractional view of bi- and multilingualism, these scholars argue that being bi- or multilingual entails having differential knowledge of one’s languages which is necessarily distinct from monolingual competence or knowledge of only one language. These researchers investigate and conceptualise multilingual behaviour and competence in terms of speakers’ total language repertoire (Grosjean, 2001; Herdina & Jessner, 2002; Clyne, 2003; De Bot, 2004; Larsen-Freeman & Cameron, 2008). Coined by Cook (1992, 2002), the concept of ‘holistic multicompetence’ emerged, designating a unique form of language competence which differs in significant ways from that of monolinguals. In line with Cook’s holistic view of multilingualism, Herdina & Jessner (2002) posit that multilingual systems contain components which monolingual systems lack and even those components that are found in both must be expected to play differential roles within the respective systems (130). Herdina & Jessner focalise on the complexity and dynamics of interacting language systems in multilingual speakers. In their Dynamic Model of Multilingualism, multilingual proficiency is the manifestation of complex and dynamic interactions of the various psycholinguistic systems involved (Jessner, 2008a:275). In DMM, a characteristic feature of multilingual proficiency is the so-called M(ultilingualism) factor which comprises all those qualities and skills that develop in multilinguals as a function of their prior language learning experience and due to an increase in language contact(s) (ibid.). According to the DMM, multilingual systems are dynamic (i.e. characterised by continuous change and nonlinear growth), and adaptive (because they adapt to changes in the environment and develop new properties in response to modified conditions). As specified by Herdina & Jessner (2002), multilingual systems are open systems because they are influenced by psychological and social factors (see also Hoffmann, 2001:1). It is argued that the stability of multilingual systems depends on the resources available to the individual speaker. If resources are not sufficiently available, language maintenance will not be successful. Maintenance, it is posited, depends on the speaker’s perceived communicative needs and on a range of individual cognitive and psychological factors such as aptitude, motivation or anxiety (Jessner, 2008a:273–4; see also van Geert et al., 2011). As explicated by Jessner (2008a:274), a holistic approach such as
Recent trends in multilingualism research
11
adopted in DMM, presupposes a dynamic view of multilingualism which in turn “assumes that the presence of one or more language systems influences the development not only of the L2, but also the development of the overall multilingual system.” Herdina & Jessner’s approach can be said to accord with Hoffmann’s (2001) perspective on multiple language learning and with her description of trilingual competence. Hoffmann advocates a holistic view of trilingual competence arguing that it encompasses knowledge of three language systems plus knowledge of how to move and function in the respective sociolinguistic contexts. This, so the author, results in trilinguals creating “their own linguistic means in order to master particular communicative situations” (14; cf. Garcia’s notion of translanguaging, 2009:51). In a similar way, Hoffmann & Stavans (2007) point to the linguistic resourcefulness of multilinguals suggesting that speakers’ linguistic versatility is enhanced when three languages are in constant contact and use (55). The authors argue that trilingual competence differs profoundly from monolingual and bilingual competence (cf. Hoffmann, 2001:1). Trilingual output, they note, is typically characterised by codeswitching and mixing which allows speakers to “compensate for linguistic gaps in one or several of their languages or to capitalize on the typological efficiency of one linguistic system versus the other so as to achieve successful communication” (56). From this perspective, typical multilingual behaviours such as switching and mixing languages are no longer seen as linguistic inadequacies but rather as strategies that trilingual and multilingual speakers have at their disposal (cf. Garcia, 2009:53–54). Summarising we can state that most of the early (but also many of the more recent) accounts of language representation and processing have proved inadequate to accommodate the dynamics and complexities of linguistic processing and development. Beginning with linguists like Herdina & Jessner (1994, 2002), Larsen-Freeman (1997), or de Bot (2004), to name but a few, it was increasingly felt that traditional monolingual and structuralist-oriented conceptualisations of language could not satisfactorily account for many aspects of language learning and usage. In consequence, a non-negligible number of scholars in applied linguistics and particularly in TLA research have abandoned conventional approaches to language learning and development turning instead to Dynamic Systems or Complexity Theory for answers. De Bot has recently described these changes in perspective as the onset of a new division in applied linguistics into an ‘old school’ and a ‘new school’ (Schmid & Lowie, 2011:5). The following section provides a first introduction to the latter (for more on dynamic approaches to language see sections 2.7, 3.5 and chapter 4). Particular attention will be given to the Dynamic Model of Multilingualism (Herdina & Jessner, 2002).
12
An Introduction to Multilingualism
1.4 The dynamics and complexities of multilingual learning and development Following the Dynamic Model of Multilingualism, multilingual learning is here envisaged as a highly complex and dynamic process. As outlined above, the DMM takes a holistic approach to multilingual acquisition and development. The model investigates and describes multilingual phenomena from a Dynamic Systems Theory perspective, that is, it rests on the assumption that individuals are complex biological systems who are best studied within the framework of Dynamic Systems Theory (Herdina & Jessner, 2002:154). Dynamic Systems and Complexity Theory have been employed in other scientific disciplines for several decades (Weltens, 2011:295; see also Larsen-Freeman & Cameron, 2008). They aim to account “for how the interacting parts of a complex system give rise to the system’s collective behaviour and how such a system simultaneously interacts with its environment” (Larsen-Freeman & Cameron, 2008:1). A defining feature of a complex system is “that it emerges from the interactions of its components” (2). Hence, in a DST approach the emphasis is on dynamism, interconnectedness, complexity and variability (cf. Ellis & Larsen-Freeman, 2006:560). Herdina & Jessner (2002) have been among the first to apply DST and Complexity Theory to applied linguistics and multilingualism research. The authors posit that multilingual learning is determined by a range of unpredictable and dynamic factors which constantly interact with one another in complex ways and which vary from individual to individual. Crucially, from a dynamic systems approach, mental activity is seen as emerging from the activities of everyday life (Larsen-Freeman & Cameron, 2008:5). It is important to note that this contradicts the assumption of static internal representations of language as postulated by traditional models of language representation and processing (elaborated upon in chapters 2 and 3). This said, it is also important to acknowledge that, even though traditional psycholinguistic models are clearly unable to capture the complexity and dynamism inherent in multilingual phenomena, they are still significant insofar as they have allowed researchers to arrive at an initial understanding of the organization and processing of language in the mind. Contrary to traditional theories of language learning and development, the DMM states that contact between two or more language systems does not result in a mere overlap of systems but that it causes a complete metamorphosis of all the systems implicated (Herdina & Jessner, 2002:116). In contrast to previous models, the DMM holds that language acquisition is dynamic, non-linear and not necessarily always associated with positive growth. In other words, fast progress in the early stages of learning can be followed by fossilisation or a flattening of the growth curve and even by language loss. Positive and negative
The dynamics and complexities of multilingual learning and development
13
growth patterns are seen as a function of the individual speaker’s perceived communicative needs (Herdina & Jessner, 2002:111). In DMM, multilingual “impediments”, such as partial achievement, are counterbalanced by several advantages which multilinguals have over monolinguals. One such advantage is the availability of enhanced monitor functions in multilingual speakers (114). Inextricably linked to this increased monitor function are multilinguals’ enhanced metalinguistic awareness and abilities (63). As indicated by Herdina & Jessner (2002), it is not entirely clear as yet how (multi)language aptitude relates to metalinguistic awareness (116). In DMM, the acronym MLA is therefore chosen to include both, multilingual awareness and aptitude (116). Aptitude, it should be noted, is considered an emergent property of multilingual systems (117). As outlined above, MLA is posited to exert strong catalytic effects on third language acquisition (116). The assumption is that MLA is acquired, and not genetically given, as nativist theorists suggest (117). This said, DMM does allow for the possibility of “a number of dispositions” which facilitate the development of skills and abilities (117). Traditional theories of SLA have typically investigated second language learning in terms of how learners’ linguistic productions relate to the target language forms. In the DMM the endpoint, that is, native-like attainment, is of minor importance as the focus is on the process of learning. In addition, in the DMM, language competence in multilinguals is “considered as consisting of dynamically interacting language subsystems which [. . .] do not necessarily represent any kind of constant but are subject to variation” (Jessner, 1997:27; Herdina & Jessner, 2002:126). According to the DMM, multilingual proficiency is therefore variable, rather than stable, and dependent on the so-called LME or language maintenance effort. LME is seen as a function of the speaker’s perceived communicative needs (Herdina & Jessner, 2002:126). As expounded by Herdina & Jessner (2002), multilingual speakers develop skills which monolingual speakers do not possess (129). These include language learning skills, language management skills and language maintenance skills. The emergence of these skills is related to qualitative changes in the (multiple) language learning process and to multilingual speakers’ heightened linguistic awareness. In the DMM an enhanced multilingual monitor (which is dependent on general cognitive functions) is assumed. The monitor allows for more effective use of the available language systems and resources (Herdina & Jessner, 2002:114). Crucially, the development of the EMM correlates with the number of language systems available to the multilingual speaker (63) and with the frequency of use of each of these languages (129). The EMM forms part of the M-factor which is described as “a dispositional effect which will have a priming or catalytic effect in TLA” (129) and as an emergent property of multilingual
14
An Introduction to Multilingualism
systems (131). The M-factor as a function of “the interaction between more than one language system” (130) is thus said to refer to proficiency skills which develop only in multilingual speakers (131). In particular, the M-factor is linked to changes in language awareness and in language strategy development which obtain due to the speaker’s increased language learning experience.
1.5 Third language learning and the role of meta- and crosslinguistic awareness Third language acquisition differs from the acquisition of a second language in a number of significant ways (Jessner, 2008a:270; 1999:204; see also Hoffmann & Stavans, 2007:55). Profound qualitative changes have been found to obtain in TLA compared to SLA. According to Jessner (1999) the major difference between SLA and TLA consists in tri/multilingual speakers’ heightened levels of metaand crosslinguistic awareness (203) and in their differential metacognitive skills (205). In other words, learners of a third or fourth language develop higher levels of meta- and crosslinguistic skill and metacognitive awareness than learners of a second language. Meta- and crosslinguistic ability and metacognitive awareness have been identified as central cognitive components involved in language learning (203). In the DMM, meta- and crosslinguistic ability and metacognitive knowledge are seen as contributing to the catalytic or accelerating effects that bilingualism can have on the acquisition of a third language (Jessner, 2008a:270). The interaction between multiple language systems is said to result in the development of so-called emergent properties. The M-factor or Multilingualism factor is one such emergent property and contributes to the catalytic or accelerating effects typical of TLA. The M-factor “is made up of a set of skills and abilities that the multilingual user develops owing to her/his prior linguistic and metacognitive knowledge” (Jessner, 2008a:275). Its core components are meta- and crosslinguistic awareness. Meta- and crosslinguistic awareness are seen as important advantages that learners gain from contact with multiple languages. Importantly, linguistic awareness “becomes more crucial in TLA than in SLA, as a speeding up of the language-learning process can be expected with increased learning experience” (Jessner, 1999:203; see also Hoffmann & Stavans, 2007:55). To summarise what has been said thus far, it appears that multilingual users develop special (linguistic and nonlinguistic) skills and abilities which are not found in monolingual or even bilingual speakers (Jessner, 2008a:275; see also chapters 19 and 20 in Part II of this book). These skills play an important role
Third language learning and the effects of prior linguistic knowledge
15
in language learning and maintenance. They form part of a larger metasystem which develops in trilingual and multilingual learners and which is not available to the monolingual speaker or to learners of a second language. As specified by Herdina & Jessner (2002), trilingual learners can approach a new language from a bilingual norm rather than from a monolingual one because they have two language systems to help them with the acquisition of the third language. Thanks to their enhanced meta- and crosslinguistic sensitivity they tend to be more aware of linguistic forms and functions than monolingual learners of a second language. They also tend to be more skilled at drawing comparisons between their various languages and at exploiting their linguistic resources and language learning strategies (cf. Jessner, 2006). It is important to note that metalinguistic and crosslinguistic development can be fostered through teaching (Jessner, 2006:122; Allgäuer-Hackl & Jessner, 2013). In the classroom, this entails sensitising pupils to the forms and functions of their first and second languages. Reflecting on linguistic structures, engaging in cross-linguistic comparisons (i.e. searching for similarities and/or differences), and targeted strategy training are good ways of creating synergies and fostering pupils’ awareness of language(s). Importantly, prior language learning experience has been found to facilitate and accelerate the language learning process (Jessner, 2006; Valencia & Cenoz, 1992; see also Hobbs, 2012:204). With regard to the language learning classroom, it is therefore strongly recommended to build on pupils’ knowledge of previously learned languages (cf. Hobbs, 2012:223–225; Allgäuer-Hackl & Jessner, 2013; Krumm & Reich, 2013). Clearly, this requires a re-examination of the roles of the L1 and L2s in the foreign language classroom. More particularly, it requires multilingual and cross-language approaches to teaching which replace reductionist practices that “treat each curricular language as an isolated unit” (Jessner, 2006:130; see also Schöpp, 2013:169–170).
1.6 Third language learning and the effects of prior linguistic knowledge In TLA the learner can, as highlighted in the previous section, draw on two languages or two different forms of prior linguistic knowledge for support (Ringbom, 2011:20). On account of their “extended linguistic repertoire”, learners of a third or additional language “do not have one but several source languages at their disposal” (Bono, 2011:25). It follows therefore that they have more resources available to them (cf. Jessner, 2006; De Angelis, 2007; De Angelis & Jessner, 2012) than monolingual speakers who are learning a second language. They know how language(s) function(s) because they have already acquired an L2 and
16
An Introduction to Multilingualism
language learning skills. They know how to learn a new language and they can apply this knowledge and transfer the acquired skills and strategies to the new language. Hall & Ecke (2003) found that L3 learners routinely draw on their prior linguistic knowledge thus mobilising all the resources and strategies available to them (72; see Gibson & Hufeisen, 2003 and Jessner, 2006 for a similar view). According to these authors, ‘parasitic learning’ constitutes the default cognitive mechanism in L3 acquisition at least as far as vocabulary learning is concerned (71). In Hall & Ecke’s Parasitic Model, acquisition is construed as involving complex and dynamic processes. The model predicts that “new lexical representations will be integrated, where possible, into the rest of the network via connections with pre-existing representations (‘hosts’), at points of similarity or overlap between them” (71–72). The authors suggest that low levels of proficiency will, in the early stages of learning, result in incomplete non-target representations and deviant access routes. In other words, ‘interlanguage’ representations and ‘interlanguage’ connections and access routes are held to lead to non-target productions or errors (2003:77). This notwithstanding, Hall & Ecke predict positive effects of cross-linguistic interaction on L3 vocabulary learning because parasitic processes or strategies allow the L3 learner to draw on prior linguistic knowledge by activating and summoning already existing representations and/ or links (78; see also Ringbom, 1983:168). Previously learned languages have been found to have differential roles and functions in the language learning process (see for example Hammarberg, 2001). Hall & Ecke (2003) intimate that the speaker’s L2 typically functions as the most important source language (80, 83; cf. Williams & Hammarberg, 1998 for a similar position). Other researchers have suggested that factors such as typology, psychotypology, proficiency or frequency and recency of use may determine which language(s) the L3 learner uses as a support language in TLA (see Williams & Hammarberg, 1998; Hammarberg, 2001). In a recent study of trilingual students of L1 Italian, L2 German and L3 English, De Angelis & Jessner (2012) evidenced a number of beneficial effects of bilinguals’ previous linguistic knowledge on TLA. The authors concluded that prior knowledge of two languages facilitates the acquisition of a third language (cf. Gibson & Hufeisen, 2011). This appears to be in line with Ringbom (1983) who claims that learners typically try to facilitate their learning by making use of any relevant prior linguistic knowledge (166). De Angelis & Jessner (2012) also found that adolescent trilingual learners’ expanding linguistic skills are closely linked with their cognitive development and thinking processes. This is congruent with Cummins (2001:2) who suggests that speakers of more languages may develop more flexibility in their thinking because they habitually process
Third language learning and the effects of prior linguistic knowledge
17
information through two or more languages (see also McLaughlin, 1990:170). De Angelis & Jessner (2012) posit that multilingual awareness and language learning awareness, which emerge in multilingual users as a function of the interaction between the various language systems, guide the learning process. Again this is consistent with Cummins (2001), who states that children who develop their abilities in two or more languages gain a deeper understanding of language and language usage (2). Finally, De Angelis & Jessner suggest that the increased amount of crosslinguistic interaction in multilingual users promotes the acquisition of a third language. Once more, this is in line with Cummins (2001:2) who intimates that the multilingual speaker’s languages influence and nurture each other (3), implying that the skills or abilities in the speaker’s two or more languages are significantly related and interdependent. In particular, Cummins (ibid.) proposes that bilingual children are not merely learning two or more languages but that they also acquire intellectual skills which they can transfer across languages thanks to a common processing system or common underlying proficiency (CUP) (see also Kecskes & Papp’s proposition of a Common Underlying Conceptual Base or CUCB in bilinguals, 2000:47, 81). In the subsequent chapter we consider differential accounts of how language is represented and organised in the speaker’s mind. Many of the models discussed in chapter 2 have recently been put into question. An alternative account to these representation models is therefore presented at the end of the chapter. It has its theoretical foundations in Dynamic Systems Theory. As indicated in the Preface, Part I of this book serves as an introduction to some of the central conceptions and notions in bi- and multilingualism research. It also lays the groundwork for the second part of the volume where the experimental findings of the study are presented and discussed.
2 The representation of language in the mind A distinction has traditionally been made between organisational and operational aspects of language. The former relate to the storage, that is, the representation and organisation of language in the brain, the latter account for the computations implicated in linguistic processing. The ensuing two chapters present and discuss various psycholinguistic accounts of language representation and processing. In so doing the following fundamental questions are examined: – How is language represented in the mind? – How do humans access linguistic knowledge? – What path does speech production follow from the original intention and the conception of an idea to the generation of an utterance, and what are the computational routines underpinning speech comprehension? – How are multiple languages stored in the brain and how do they interact? – What happens in the brain when multilingual speakers seek to convey a verbal message to an interlocutor? Do the languages they speak influence the processes of production or comprehension and if so, in what way? This is followed by a discussion of language representation and processing from a dynamic systems point of view. The DST approach to language representation and processing is premised on the assumption that language is not a static but a highly complex and dynamic system. The components of complex systems constantly interact with each other, and the behaviour of complex systems emerges from these complex and dynamic interactions and from the self-organisation of the various components implicated (Lowie & Verspoor, 2011:268).
2.1 Models of language representation: How are words and concepts represented and interconnected in the mind of bilingual and multilingual speakers? As Levelt notes “the systematic study of how we speak did not begin before the end of the 19th century” (2001:13464) when Meringer & Mayer (1896) and Wilhelm Wundt (1900) published their theories on the processing of language. While Wundt proposed a theory of how sentences “emerge in the speaker’s mind” (ibid.), Meringer & Mayer arrived at the conclusion that there must be “a distinction between meaning- and form-based operations in word generation” (ibid). In the first decades of the 20th century, scholars, among them Ronjat (1913) and Leopold (1939), first began to record their own (bilingual) children’s
Models of language representation
19
linguistic development (Marini & Fabbro, 2007:13). However, their observations were mostly subjective impressions since they failed to comply with current scientific research criteria and paradigms. A more rigorous and systematic study of bilingual phenomena did not take off until much later, i.e. until the beginning of the 1950s (French & Jacquet, 2004:87). Research in the field of tri- and multilingualism by contrast has only recently come into its own and it is still negotiating its position and status. One of the oldest questions in bilingualism research hinges on whether concepts in the bilingual speaker’s mind are stored separately or together (De Bot & Broersma, 2006:5). The early stages of bilingual memory research saw scholars mostly engaged in the task of investigating issues related to the storage and representation of language in the mind. The aim was to establish whether linguistic information from the bilingual’s two languages is stored in a single shared lexicon, or whether words from the two languages are represented in separate compartments (De Angelis, 2007:89). One of the first attempts to formulate a model of language representation was undertaken by Weinreich in 1953 (Marini & Fabbro, 2007:13). Drawing on Saussure’s distinction between signifier and signified, i.e. the concept and the expression of a word (De Angelis, 2007:88), Weinreich proposed that word form and word meaning can be represented in three possible configurations, namely in a coordinate, a compound or a subordinate system. In the coordinate system the bilingual speaker is assumed to have two distinct word forms each of which is linked to its own distinct conceptual representation. Configuration number one thus consists of two separate stores for the bilingual’s words and two separate stores for their meanings. The compound configuration, by contrast, is characterised by two separate word forms which share the same conceptual system while in the subordinate system the L2 word accesses meaning through the L1 word (italics added). Weinreich hypothesised that these different types of associations need not be seen as mutually exclusive but that they could be conceived of as co-existing in the bilingual’s mind. He suggested that while for some words compound relationships are formed, others may be in a subordinate or coordinate relation with one another (De Angelis, 2007:88; De Bot & Broersma, 2006:5). This might explain why even proficient bilinguals are sometimes found to revert to lexical mediation in certain language tasks instead of relying on conceptual interconnections (as is commonly expected of fluent speakers) (Schwartz & Kroll, 2006:968). It has further been suggested that certain characteristics of words and concepts (such as concrete versus abstract or high frequency versus low frequency) may affect processing and speakers’ reliance on conceptual or lexical links (ibid.). In consequence, it is conceivable that even low proficiency
20
The representation of language in the mind
users of an L2 may occasionally access a concept directly via an L2 word if that word is encountered or used frequently by the speaker. As mentioned above, the reverse may hold for proficient L2 speakers. In other words, highly proficient speakers of an L2 may access concepts indirectly via the L1 translation when faced with abstract and low frequency words. Ervin & Osgood (1954) elaborated Weinreich’s Coordinate/Compound Model and suggested that language learning history may play a critical role in how the two languages of a bilingual are represented in the mind. Later, Kolers (1963) argued that bilingual memory consists of two separate and functionally independent storage and retrieval systems. Interaction, so Kolers, occurs only through translation, otherwise the two language systems are completely “insulated from each other” (Paivio, 1991:115). More recently, this position has been contested. Cummins (1980, 1981), for instance, posits that the bilingual’s languages do not function separately, but that they are fused and operate through the same central processing system (Baker, 2006:169). Similarly, Brysbaert & Dijkstra (2006) have challenged the assumption of functionally independent lexicons and a language-selective input switch (see further down), arguing that “evidence is rapidly acculmulating that both ideas are wrong.” (11).
2.2 The Word-Association and Concept Mediation Model In the 1980s researchers revised and extended Weinreich’s Coordinate/Compound/ Subordinate Model of bilingual language representation. The newly devised models are generally referred to as hierarchical models or three-component models for they consist of three components or nodes. Hierarchical models presuppose a separation between the form and the meaning of a word and posit separate lexicons for the speaker’s languages (French & Jacquet, 2004:88). The twolexicon view of language representation or storage postulates that words from the bilingual’s two languages are stored separately. Three-component models differ in terms of the number and type of connections between the single nodes. Potter et al. (1984) proposed three diverse models of bilingual memory representation, the Word Association Model, the Concept Mediation Model, and an Intermediate Model which is generally known as Developmental Model or Mixed Model. The Word Association Model largely corresponds to Weinreich’s subordinative model structure while the Concept Mediation Model matches Weinreich’s Compound Model (De Groot, 2002:37; see also Kecskes & Papp, 2000:65). The Intermediate Model, by contrast, differs from the above mentioned models in that it assumes a shift from word association to concept mediation as proficiency in the L2 increases (De Angelis, 2007:90).
The Revised Hierarchical Model
21
2.3 The Revised Hierarchical Model Kroll & Stewart (1994) reformulated the Mixed or Developmental Model and proposed a revised version which has since become known as the Revised Hierarchical Model (see Fig. 1 below). The model adopts the basic structure found in the Developmental Model, including a common concept system and two distinct lexical stores. In addition, it assumes that the connections between the two lexical nodes and the conceptual node differ in strength depending on how proficient the speaker is in the two languages. Significantly, links between the various nodes are held to be bidirectional. A further difference regards the size of the two lexicons. Kroll & Stewart (1994) assume a larger lexicon for L1 than for L2, also in very advanced speakers of L2. As regards the links between the lexicons and the concept store, the model posits stronger conceptual links for the L1 than for L2. The model also implies that learners initially exploit the existing word-to-concept connections in their L1 to access meaning for words in L2 (Kroll & Tokowicz, 2005:546). Progress in L2 achievement is believed to redress this since the existing connections between the L2 lexicon and the conceptual store get strengthened and thus allow for faster performance.
Figure 1: The Revised Hierarchical Model (based on De Groot & Kroll, 1997:178)
2.4 The Distributed Feature Model Recent studies have suggested that interlanguage connections are influenced not only by the speaker’s level of proficiency but also by the very nature of
22
The representation of language in the mind
words. De Groot and colleagues (1992, 1993, 1995), for instance, have noted that words in one language do not always have a precise translation equivalent in another language and that sometimes translations can capture only an approximate meaning (Kroll & Dussias, 2004:174). De Groot (1992) therefore proposed the Distributed Feature Model postulating that lexical units consist of semantic features that overlap to different degrees in different languages (cf. Lowie & Verspoor, 2011:280–81). The implicit assumption is that certain words such as concrete words or cognates share more or even the same features across languages while abstract words and non-cognates require more time to translate because they overlap only partially in meaning. The amount or degree of overlap is thus held to be determined (at least in part) by word category. While concrete nouns and cognates are said to be represented similarly across languages, abstract nouns and noncognates are held to share fewer features. The model predicts that the larger the featural overlap between two translation equivalents, the faster the speaker is able to provide the translation of a word. The Distributed Feature Model assumes that the semantic system is shared across languages and that the semantic primitives contained in a given lexical unit are available to all the speaker’s languages (Kroll & Tokowicz, 2005:537). De Bot & Schreuder’s (1993) proposition that conceptual structure can be described in terms of primitives and principles of combinations (193) bears some resemblance to the Distributed Feature Model. De Bot & Schreuder posited that the conceptual structure realised by the conceptualiser consists of sets of primitives and principles of combinations (193). In contrast, Kormos (2006) maintains that as far as “conceptual activation is concerned, we can reject the assumption that concepts are made up of lists of semantic features” (35). Drawing on Levelt (1999) and Roelofs (1992), Kormos intimates that in L1 speech production research “the most widely held view is that concepts are undivided wholes, which cannot be broken down into semantic components” (ibid.).
2.5 The Integrated Model: A model for bilingual language representation and processing An interesting proposal of bilingual language representation comes from Hartsuiker, Pickering & Veltkamp (2004). Their model accounts not only for language representation but also for processing in bilinguals and is particularly concerned with “the interface between the mental lexicon and syntactic encoding in bilingualism” (Hartsuiker et al. 2008:480). The model posits a common lexicon
Lexical Access in bilingual and multilingual speakers
23
and syntax for both languages suggesting that lemma nodes are linked to language nodes and to combinatorial nodes. These latter specify the word’s syntactic properties (Hartsuiker, 2008:481) and indicate whether an item is a verb or a noun or whether it is active or passive. Combinatorial nodes are shared between lemmas, i.e., all the lexical items (from both L1 and L2) which share the same properties are connected to the same combinatorial node (Hartsuiker et al. 2004:412; Hartsuiker et al. 2008:481). Word meaning and word syntax can thus be seen as points of contact between languages (Hartsuiker, 2004:413). In this respect the integrated model may be said to be in line with a dynamic view of language which assumes that there is no clear border between lexicon and grammar and that words and grammatic structures form a kind of continuum (Langacker, 2008:67 in Lowie & Verspoor, 2011:281). The integrated view of lemma representation as postulated by the integrated model can easily account for the occurance of code-switching.
2.6 Lexical Access in bilingual and multilingual speakers: The language specific and language non-specific perspective The study of language and the mind, as undertaken in the field of psycholinguistics, and increasingly also in neurolinguistics, has thrown up a range of important questions. Among the major issues raised by SLA research are those relating to the organisation of language(s) in the mind and the retrieval of lexical knowledge in speakers of two (or more) languages. Two major positions can be identified with regards to the representation and retrieval of languages, the language-specific or selective view and the non-specific or non-selective view. The former predicts that “the activation flow from the conceptual system to the lexical system is channeled in such a way that only those words belonging to the response language receive activation” (Costa, 2005:312). In other words, non-target items are not seen as competing for selection. This entails the assumption that bilingual lexical selection is comparable to lexical selection in monolinguals since lexical items in the non-target language are not considered in the selection process. Today, there is overwhelming support for the non-selective position which predicts that the activation flow during multilingual processing is not restricted to the response language but obtains across languages6 (cf. Hell & Dijkstra, 6 An overview of studies and research findings pertaining to the selective/non-selective debate can be found in De Bot (2004).
24
The representation of language in the mind
2002). Support for the idea of completely separate lexical stores and separate language processors has dwindled in recent years as scholars have increasingly concurred on the assumption that language organisation in bi- and multilinguals is “at least partially overlapping” (French & Jacquet, 2004:89). As noted by De Bot (2004), there is now fairly conclusive evidence in support of the nonselective view while in the past the language selective view had been favoured (18; cf. Kroll & Stewart, 2005:531–2). Similarly, Müller-Lance (2003) intimates that “[t]he frequency of switching between various languages, including L1, makes it plausible that the languages of an individual are not separated in different ‘compartments of the mind’” (124). Other scholars have suggested that both theories, the language selective and the language non-selective, may be right to some extent (Brysbaert & Dijkstra, 2006:3; De Angelis, 2007:89; Singleton, 2003:176). An intriguing proposition comes from Kroll & Tokowicz (2005). The authors posit that there is no reason to assume that the way languages are represented (and, as is here suggested, accessed) should be the same for orthography, phonology, semantics, and syntax (531). According to these authors, different aspects of language may be organised in different ways, depending on a wealth of variables including typology, language learning history, learning environment, age, and even aptitude. This is in line with Cook (2002:12) who states that “neither total separation nor total integration can be completely true” suggesting that it is conceivable that different relationships exist for different areas of language such as vocabulary, syntax or pragmatic functions (see also Jessner, 2003 and Singleton, 2003). Individual differences and language typology are also predicted to affect the structural representation of words in the speakers’s mind. As Cook (2002) notes, some individuals keep the languages predominantly separate, others integrate them massively. Every mind does not necessarily relate two languages in the same way. . . These variations might be due entirely to the users’ minds, or might be the product of the language situations they find themselves in or of the teaching they have been subjected to or of the age at which they learnt the second language. (13)
Code-switching or -mixing phenomena are widely seen as challenging the language selective view (cf. Singleton, 2003; Green, 1886, 1998). Code-switching and -mixing are also said to provide evidence against the assumption of a switch mechanism in the brain which, as some researchers have contended, allows speakers of multiple languages to avoid interference. In the late 1950s Penfield & Roberts (1959) first advanced the proposition of an automatic language switch mechanism maintaining that the “functional separation of languages takes place by an automatic switch at the neurophysiological level” (Brysbaert
Lexical Access in bilingual and multilingual speakers
25
& Dijkstra, 2006:5). Some years later, MacNamara & Kushnir (1967, 1971) similarly argued in favour of an automatically operating input switch which supposedly channels the input to the appropriate language system (De Groot, 2002:50). The authors claimed that a switch mechanism would allow the bilingual speaker to turn their languages on or off according to the requirements of the communicative situation. On very much the same line, Lipski (1978) and Sridhar & Sridhar (1980) maintained that a ‘comparator’ in the brain permits the speaker to ascertain whether constituents from another language are compatible with the syntactic structure of the base language (Poulisse, 1997:208). Both, switch models and comparator models, have been widely criticised. Singleton (2003) carefully analysed the arguments for and against an integrated mental lexicon in speakers of multiple languages. He concluded that there is powerful evidence for a very high degree of integration and connectivity (e.g. in the form of cross-linguistic influence). However, according to Singleton, there are a number of arguments which speak against total integration (formal differences between languages, selective recovery of languages in aphasia patients, or the fact that multilingual speakers mostly keep their languages apart when communicating with others). According to Singleton, “a position which would claim that lexical knowledge is radically unitary” appears untenable, but “the fact that lexical stimuli in one language also automatically activate words in other languages known to an individual – at least in some circumstances – is a powerful argument for integration in the sense of a very high degree of crosslexical connectivity an interaction” (176). A fundamental question concerning the bilingual or multilingual lexicon relates to the precise locus of selection, i.e., the moment at which one language is selected over another. Most scholars today converge on the assumption that language choice occurs at the conceptual level (cf. Kormos, 2006:83). Clyne (2003) and Hammarberg (2001:37) have advanced alternative propositions. These authors intimate that language choice can occur locally in the speech sequence, and therefore also at a lower level of processing so as to allow for switches to other languages. More recently, Lowie & Verspoor (2011) have proposed a DST approach to the multilingual lexicon (see below). As indicated above, it is not entirely clear how, in speakers of multiple languages, conceptual material is mapped onto lexical items in the lexicon and how the appropriate target-language word is selected. It has been suggested that the conceptual system activates “lexical representations of the response language with more intensity than those of the nonresponse language” (see Costa, 2005:315). This is thought to ensure that target language lexical items receive higher levels of activation and are thus selected as the items corresponding best to the conceptual representation. The assumption is that the bi- or
26
The representation of language in the mind
multilingual lexicon encompasses lexical items or lemmas, which apart from meaning specifications also store language-specific information regarding the lexical item’s syntactic, morphological and phonological characteristics (cf. Wei, 2002:694). In addition, there appears to be general consensus that lemmas are tagged for language (Wei, 2002:692; De Bot, 2004; Poulisse & Bongaerts, 1994 but see La Heij, 2005 for a different view). Once a lemma (tagged for a particular language) is activated, morphosyntactic and phonological processes of that language are triggered (694). Some researchers, most notably Green (1998), posit additional inhibitory mechanisms which operate on the nontarget lexical items by suppressing all potential competitors, thereby allowing the system to select the most appropriate candidate in the intended language. La Heij (2005) advocates a lexical selection or retrieval process which he succinctly describes as “complex access, simple selection” (304). The underlying assumption is that the preverbal message consists of a complex set of information. On the basis of this comprehensive information a lemma is then selected. A special language cue ensures that the target-language word receives a higher level of activation than non-target items. In other words, the lexical item corresponding most closely to the information contained in the preverbal message reaches the highest activation and is selected (305). It is thus assumed that the preverbal message contains all the information necessary to activate a particular lexical item. This information comprises semantic, affective, pragmatic, and situational features and includes the language that is to be used in the communicative act (289). It follows that the actual selection process “does not evaluate the appropriateness of a word that is about to be selected or involve the selective activation or inhibition of words that belong to a particular language” (290). Instead, the target word is selected by way of direct activation on the basis of conceptual information (304). La Heij’s (2005) propositions that the to-be-used language is specified in the preverbal message in the form of a language cue (305) is consistent with Poulisse & Bongaert (1994) who propose that the preverbal message contains a language cue which cues the intended language (see De Bot & Schreuder, 1993 for a similar proposition). The language cue and the conceptual information together activate the lexical item that matches the conceptual message in both meaning and language (La Heij, 2005:302, Poulisse, 1997:216). Poulisse & Bongaerts (1994) predict that the speaker’s intention to use one language over another results in higher activation levels for words from that particular language (La Heij, 2005:302). As in La Heij’s model, no additional selective activation or inhibiton at the lexical level is required (304, 305). Unlike La Heij (2005), Poulisse & Bongaerts (1994) additionally assume language tags which indicate whether an item belongs to the L1 or L2.
Lexical Access in bilingual and multilingual speakers
27
Meuter (2005) argues that, during lexical retrieval processes, bilingual speakers can not ignore the non-response language because they “cannot help but process what is seen or heard” (358). Meuter (2005) suggests that upon reading or hearing words in one language, lexical candidates in the other language(s) are automatically activated (352). Meuter & Allport (1999) observed that switching from the weaker L2 to the more dominant L1 entailed larger time costs than the other way round (35). This, according to the authors, happens because active inhibition of the stronger L1 “persists involuntarily” into the processing of the next item rendering “active disentanglement” from, or suppression of, the dominant L1 necessary (1999:26,35; italics in the original) thus causing a processing delay and, in consequence, slower response latencies (25). Meuter (2005) intimates that the processing system exhibits strong inertia, i.e. a “tendency to continue responding in the same language” (354). When the dominant L1 is suppressed because the speaker is required to respond in the weaker L2, the so-called ‘language set inertia’ appears to be particularly strong and hard to overcome. Hence, switching into the suppressed L1 entails a greater time cost (355; cf. Green’s IC Model for a similar account). The observed asymmetry is hypothesised to diminish gradually as the speaker’s mastery of her/his languages becomes more balanced (Meuter & Allport, 1999:36). In contrast to the above authors, Costa, Miozzo & Caramazza (1999) postulate a language-selective access to the lexicon (302) arguing that “the two languages of a bilingual do not compete during lexical access” ([2010]:134). According to Costa et al. (1999), a language-specific selection mechanism “allows the two languages to have a certain degree of autonomy” (ibid.). The authors posit a selection procedure which ensures that only words in the intended language are selected (La Heji, 2005:302). The language specific view implies that the lexical selection mechanism operates in such a manner as to ignore all potential candidates from the nontarget language. Even if lexical items from the other language are available because they have received some activation, the system is “’blind’ to the activation levels of the lexical nodes belonging to the nonresponse language” (Costa, 2005:313). According to this latter position bilingual lexical selection does not really differ from monolingual lexical selection because the nontarget item does not interfere with or disrupt the retrieval of the target item (ibid.). In summary, we can state that scholars today largely converge on the assumption that the multilingual speaker’s languages are highly connected (cf. Lowie & Verspoor, 2011 for a dynamic systems view of linguistic representation and the lexicon) and that a parallel search is carried out in the multilingual lexicon whenever a given word or item needs to be retrieved. Phenomena such
28
The representation of language in the mind
as mixing and switching are seen as providing strong evidence for the interconnectedness of languages. The resulting implication is that whenever speakers access one language, the other language(s) are also to some degree activated, that is, when one language is activated, the others are not switched off (Grosjean, 2001; Williams & Hammarberg, 1998; Clyne, 2003; Dijkstra, 2003; De Bot, 2004).
2.7 A Dynamic View of the Multilingual Lexicon In contrast to linear, modular theories (outlined in chapter 3), dynamic approaches to linguistic representation and processing leave no room for separately operating modules (i.e., conceptualiser, lexicon, formulator, articulator) in the mind (Lowie & Verspoor, 2011:269). In the DST framework, there are no fixed representations and no sequential or hierarchical processing stages as in linear models. The DST approach does not assign distinct, sequential roles to a lexicon or a formulator (282). This is not to say that “no particularly focused processing components can be assumed” (269), as long as one conceives of such components as interconnected and dynamically interacting subsystems which mutually influence each other. As specified by Lowie & Verspoor (2011), the notion of lexical storage (and the storage of information in general) is incompatible with a DST approach because it implies that mental representations are static. In a dynamic systems framework, there are no fixed representations. All mental activity is dynamic and constructed over time in a multidimensional mental state space through which the mind travels continuously (277). Unlike modular processing models, the DST approach does not assume distinct and sequential processing stages for a preverbal plan, for lexical retrieval, for morphosyntactic and phonological encoding and for articulation. In other words, the conceptualiser, lexicon, formulator, and articulator are not seen as separate, impermeable modules which store and carry out specific types of operations. Rather, the DST approach assumes mental subsystems which co-operate and which can be understood as dynamic, fluctuating and overlapping entities with no clearly defined boundaries and with graded or vague category and language membership (279). It follows that the lexicon “is not a neatly defined and separately stored set of words” (280). Instead, multilingual speakers’ languages are construed as fuzzy subsets which change over time. Rather than being conceived as stored information, speakers’ languages are seen as “highly unstable and context dependent, cognitive routines that exist in a multilingual state space” (280). Importantly, from a DST perspective, words are not “one-to-one mappings of meaning and form, but [. . .] the simultaneous association of multiple aspects of meaning and form that
A Dynamic View of the Multilingual Lexicon
29
occur in a particular context” (ibid.). In this sense, words are not separate from other symbolic units such as chunks, morphological or syntactic constructions (282). The reader is referred to chapter 3 for more on the lexicon and on linguistic processing from a DST view. Chapter 3 presents some of the most widely cited language production and comprehension models. The final section of the chapter introduces an entirely novel perspective on (multiple) language processing within the continuity-of-mind framework.
3 Language Processing Any discussion of language and language learning would be incomplete without a thorough analysis of how the human mind processes language. A central question in psycholinguistic research therefore relates to the structures and mechanisms underlying speech production and comprehension, viz. the mental processes implicated in the production and comprehension of language. Unless suffering from some kind of impairment most humans are capable of producing and perceiveing language accurately and efficiently at the remarkably fast rate of 2 to 3 words per second on average (Levelt, 1989:22). When communicating their ideas, speakers do not limit themselves to retrieving single words from memory but, instead, combine a series of lexical items to generate coherent speech. This online construction entails the integration of semantic and syntactic relations between lexical units on the one hand and the phonological realisation of words and their final articulation on the other. It thus becomes clear that speech processing involves a succession of highly complex operations. Linguists have traditionally distinguished between the production and comprehension of language. Production involves top-down procedures triggered by conceptual activity arising from the speaker’s mental faculty. Production starts with the intention to communicate an idea or message and results in the overt output of speech. Comprehension, by contrast, proceeds the other way. It is triggered by external stimuli, by utterances or images from outside. Comprehension or the recognition of words follows a bottom-up route in the course of which acoustic signals are said to be mapped onto letters and words and at the end of which process the meaning of a word or utterance is retrieved. In spite of an increase in research activity over the last decades the operations implicated in (both, monolingual and multilingual) language production and comprehension are still very little understood. Many fundamental questions regarding speech processing and language development remain unanswered. In consequence, the majority of the psycholinguistic models devised thus far are at best partial accounts of very complex processes and further research is clearly needed to allow more light to be shed on the mechanisms and computational routines underpinning the processing of language. This said, we direct the reader’s attention to sections 3.3 and 3.5, and to chapter 4 where novel and very promising perspectives on linguistic processing are presented.
Is multilingual processing any different from monolingual or bilingual processing?
31
3.1 Is multilingual processing any different from monolingual or bilingual processing? The answer to this question is, without any doubt, in the affirmative (cf. Stavans & Swisher, 2006:193,215 and Gibson, Hufeisen & Libben, 2001:138). With each new language added to the system, the dynamics and complexity of language learning and linguistic processing increase. As a result, the linguistic and cognitive systems in the brain must be anticipated to undergo significant changes. As explicated by Jessner (2003a), the multilingual system is “a complex system with its own parameters, which are exclusive to the multilingual speaker, that is they are not to be found in the monolingual speaker or the bilingual speaker” (48). This is congruent with Hoffmann (2001) who points out that it suffices to consider the mechanisms and processes which seem to be involved in the selection of the appropriate language, and also language mixing, switching and borrowing -either consciously or unconsciously-, then it becomes apparent that a comparison between bilinguals and trilinguals may show up differences which go beyond quantitative ones. (15)
Jessner (2008a) posits that the presence of several language systems in the mind and the increase in language contact(s) influence the development of the overall multilingual system (274–5; see also De Angelis & Selinker, 2001:45). In the same way, the DMM predicts that contact between multiple psycholinguistic systems will change the entire system and will result in a complete metamorphosis of all the systems involved (Jessner, 2008b:95). According to Jessner (2008b), multilingual processing and “[c]hanges of quality between second and third language learning are based on the differences in norms that the language learners relate to, that is, a bilingual norm in third language learning as opposed to a monolingual norm in second language learning” (95). In addition, it is assumed that in contrast to monolingual and bilingual speakers, multilingual learners can draw on a unique “metasystem which develops in third language learning as the result of its relation to a bilingual norm” (Jessner, 2006:89) and which results in “an increased level of metalinguistic awareness” in multilingual learners (2008a:270). Grosjean (2001) famously used the analogy of the hurdler to exemplify the differential skills of monolingual and bilingual users. In a similar way, the analogy of the triathlete could here serve to illustrate that the comparison of monolingual and/or bilingual speakers with multilingual speakers is equally inappropriate, and that multilingual processing must be taken to differ not only from monolingual but also from bilingual processing.
32
Language Processing
The ensuing section explores some of the most influential models of language production and comprehension presented by linguists in the late 20th and early 21st century. It is important to recall that from today’s perspective many of these models are considered inadequate since they fail to capture significant aspects pertaining to language processing and development (cf. Lowie & Verspoor, 2011). This notwithstanding, they have been included here for the sake of completeness and, of course, because they have contributed significantly to researchers’ understanding of the nature of the human language faculty (in a generic, not in a Chomskyan sense). The last section of this chapter is of particular interest. It presents an entirely novel approach to linguistic processing within a continuity-of-mind framework and discusses language processing from a dynamic systems perspective.
3.2 Language Production Language production research typically investigates the processes and operations implicated in the production of language and the components that constitute the production system. Two major theories of L1 speech production have been proposed: spreading activation theories, as advanced by Dell (1986) and modular theories as put forward by Garret (1976), Levelt (1989) or Levelt, Roelofs & Meyer (1999). Proponents of both theories broadly agree on the basic components constituting the production system. These components include the conceptualiser, where message planning takes place, the formulator where grammatical and phonological encoding are attended to, and the articulator which generates overt speech. The single stages involved in language production are presumed to follow the above order, i.e. conceptualisation precedes formulation and formulation precedes articulation (Kormos, 2006:xviii). There are, however, a number of significant differences between the two theories. One central point of contention concerns the way activation occurs within the processing system. Modular theories maintain that the components implicated in the speech production process work autonomously and independently of each other. Activation is held to occur in a forward direction only, i.e. from the higher levels to the lower levels. This stands in contrast to Spreading Activation Theory where it is posited that activation can also flow backwards, i.e. from a subordinate to a superordinate level. A further theoretical difference regards the nature of syntactic and phonological encoding in the formulator. Spreading activation theories assume framedriven encoding procedures. According to Spreading Activation Theory, the processing system utilises insertion rules to build frames with slots. These slots
Language Production
33
are then filled with the appropriate linguistic elements. The assumption is that “speakers first construct frames for sentences and for phonetic representations and then select the appropriate words or phonetic features for the slots in the frame” (Kormos, 2006:xix). This contrasts with modular theories where the reverse is the case. Modular models are lexically driven, that is, words activate syntactic procedures and syntactic encoding can not commence before lexical encoding is completed (ibid.). It has been repeatedly criticised that neither theory considers the role of memorised chunks or formulaic utterances in the generation of speech (ibid.). An exhaustive discussion of the modular or minimalist and spreading activation or maximalist view can be found in Vigliocco & Hartsuiker (2002). As specified by these authors, the minimalist view assumes modularity and information encapsulation and posits that production processes at one level are encapsulated from processes at all other levels (semantic, syntactic, and phonological). Each level computes a specific kind of information or representation which is then passed on to the next level or module. It thus follows that each processing level receives only the minimal necessary information from the preceding module. Additionally, minimalist theory postulates that the flow of information is unidirectional. It does not allow for interaction or feedback between the various processing levels. Proponents of modular language processing systems maintain that unidirectionality and encapsulation of information allow the system to function accurately and efficiently with little interference from non-necessary information from other levels (443). The maximalist view by contrast assumes interactivity and postulates that information from any of the above outlined processing levels (semantic, syntactic, phonological) can influence processing at other levels. This view entails maximal input in the form of spreading activation and bidirectional flow of information. This is said to permit fast and efficient encoding by way of preactivation of items and feedback between the various levels.
3.2.1 Levelt: A Monolingual Speech Model Levelt’s model of monolingual speech production (1989) is widely recognised as one of the most influential psycholinguistic models in TLA research (cf. Lowie & Verspoor, 2011:269). It will therefore be discussed in some detail here. Originally proposed to account for monolingual production it was later extended to accommodate language production in bilingual and multilingual speakers. Levelt (1989) describes the speaker as “a highly complex information processor who can, in some still rather mysterious way, transform intentions, thoughts, feelings into fluently articulated speech” (1). According to Levelt’s Blueprint for the
34
Language Processing
Speaker (1989:9) language production comprises a number of processes including the conceptual preparation of the to-be-conveyed message, lexical selection, syntactic encoding, phonological and phonetic encoding and the final articulation of the utterance. Levelt proposes three major processing components or subsystems: the conceptualizer, the formulator and the articulator. In addition, Levelt proposes a speech comprehension system for decoding incoming speech and for self-monitoring. The processing components or subsystems are predicted to operate independently of each other since they “do their work in rather autonomous fashion” (2). The assumption is that they operate “in a highly automatic, reflex-like way” which allows them to work in parallel (ibid.). Levelt assumes incremental processing operations whereby each processing component can start operations as soon as it receives input from the previous module (De Angelis, 2007:66). Crucially, any fragment of characteristic input, even a minimal amount of input, suffices to trigger a given processing component into activity (Levelt, 1989:24–26). The ensuing figure depicts Levelt’s blueprint for the speaker and illustrates the flow of information during on-line processing.
Figure 2: A blueprint for the speaker (based on Levelt, 1989:9)
Language Production
35
The assumption is that language production begins at the conceptualisation level where the conceptual information originates and where the communicative intention leads to the generation of a pre-verbal, non-linguistic message. The conceptualiser has access to knowledge about the world, the communicative situation and the interlocutor (Levelt, 1989:10). Once an idea or concept is born – Levelt defines concepts as bundles of semantic or conceptual features – a preverbal message is passed on to the formulator. The formulator consists of two subcomponents for grammatical and phonological encoding. As postulated by Levelt (1993), grammatical encoding precedes phonological encoding (242). As soon as the formulator receives the pre-verbal message, lemmas in the mental lexicon are activated by matching the meaning components in the concepts with the appropriate lemmas (243). The mental lexicon stores a vast amount of information about the meaning of a word, its syntax, morphology and phonology. Levelt (1989) specifies that lexical access involves the retrieval of two different representations: lemmas and lexemes (231). Lemmas are defined as “abstract elements that underlie actual surface-level morphemes (and therefore words)” (Myers-Scotton, 2006:268). Lemmas contain the word’s syntactic and semantic information. The activation of a particular lemma “frees the syntactic information” needed to activate syntactic procedures (Poulisse, 1997:204). Lexemes, by contrast, represent a word’s morphological and phonological information and therefore constitute its form. The implicit assumption is that at the same time as syntactic procedures are activated, the lemma is also looking for its lexeme. Lexeme retrieval and syntactic processing occur in parallel as predicted by the principle of incrementality. Activating the lexeme starts a very complex process of word formation procedures (De Bot, 2004:26). As soon as operations at this stage are completed, the phonological encoder transforms the surface structure prepared by the grammatical encoding subcomponent into a phonetic plan. Once the phonetic plan or articulatory program (Levelt, 1993:242) is put together, it is fed into the articulator which then converts it into overt speech.
3.2.2 Dell: Spreading Activation Model Of all interactive activation spreading models, Dell’s model is probably the most famous and most frequently cited (Kormos, 2006:4). This said, Dell’s model of spreading activation (1986) has not been adopted as widely as Levelt’s speaking model (Kormos, 2006:3). Dell posits generative rules which trigger frame building operations and insertion rules which specify the linguistic elements to be
36
Language Processing
filled into the slots prepared in the course of frame-building (ibid.). As in Levelt’s model, processing proceeds from the highest to lowest level, from higher-level nodes to lower-level nodes. Dell’s model operates on the principle of spreading activation. A particular node gets selected if its activation level is higher than that of other nodes of the same class or category. The activated node subsequently spreads activation to the lower level where again the most appropriate node of that category is selected and activated. Dell proposes three levels of processing, one for syntactic encoding, one for morphological encoding and one for phonological encoding. At each level a specific set of rules determines and constrains the processing operations in terms of the combinatorial possibilities. As indicated above, generative rules produce frames with slots which are then filled with the appropriate items. These items are stored in the lexicon and are labelled according to the category they belong to. The nature of slots differs according to the level of encoding. At the syntactic encoding level, slots are filled with words. At the morphological and phonological level morphemes and phonemes are inserted respectively (Poulisse, 1997:202, De Angelis, 2007:64). As explicated, activation of nodes is achieved by virtue of spreading activation. This entails the simultaneous activation of several representations which does, however, not signify that all activated nodes get the same amount of activation (Dell cited in De Angelis, 2000:65). A selection mechanism is posited to mediate the selection of the preferred representation at each level. Dell proposes that item selection is achieved by raising the activation level for one particular node. Additionally, “a passive decay of activation over time” is presumed to keep levels of unnecessary or unwanted activation down (ibid., Levelt, 1989:19). Spreading activation is thus seen as encompassing three stages: spreading, summation and decay (De Angelis, 2007:65). In contrast to Levelt’s strictly modular architecture and feedforward operations, Dell’s spreading activation network allows for interaction between the various processing stages and components. Interaction is predicted to occur in both directions, from higher to lower levels and from lower to higher levels of processing.
3.3 Multilingual language production Existing models of bilingual and multilingual language production are mostly based on monolingual models. This is, at least in part, due to the fact that multilingual language development and acquisition is a very young field of research, with the consequence that our knowledge and understanding of multiple language processing is as yet rather limited. Although it makes sense to assume
Multilingual language production
37
that additional languages entail an increase in the complexity of speech processing operations (see Herdina & Jessner, 2002; Clyne, 2003; Aronin & O’Laoire, 2004; De Bot, 2004), it has to be conceded that many aspects pertaining to multilingual processing and development remain at present unresolved. Recent approaches to psycholinguistic modelling have been influenced by dynamic and holistic conceptualisations of language acquisition and development (Lowie & Verspoor, 2011; see also Herdina & Jessner, 2002; De Bot, 2004). From a holistic perspective, languages are viewed as dynamic, complex and constantly interacting and changing systems. Dynamic approaches to linguistic processing reject the strictly modular and hierarchical structure of earlier models (Levelt, 1989; De Bot, 1992). Instead of the sequential and static modules or processing components of early models, Lowie & Verspoor (2011:284) propose dynamic subsystems which interact in complex ways. Modularity is thus replaced with continuity (-of-mind) (Lowie & Verspoor, 2011:268). As specified by Lowie & Verspoor (2011:267), in “current dynamic approaches to language processing, it is assumed that the language system is not stable and static, but instable and dynamic. In these approaches, language production and comprehension are not seen as sequential processing in a series of separate modules, but as an integrated dynamic system that is characterised by a multidimensional state space with numerous attractors at different levels.” The reader is referred to van Geert et al. (2011) and Larsen-Freeman et al. (2011) for an overview of recent developments in psycholinguistic modelling. The following section provides a comprehensive review of the early and more recent literature on bilingual and multilingual language processing and discusses some of the most influential models in geater detail. Some of these models are processing models in the narrow sense, while others cover a wider scope and integrate aspects which are not strictly related to linguistic processing. The models are presented in chronological order with earlier models preceding more recent ones.
3.3.1 Poulisse and Bongaerts’ Production Model The Production Model designed by Poulisse & Bongaerts is concerned with the mental processes implicated in translating an idea which was formed in the speaker’s mind into overt speech. Based on Levelt’s model of monolingual speech production (1989), Poulisse & Bongaerts (1994) proposed the so-called model of Spoken Language Production. It consists of three levels or stages of processing. At the highest level, the conceptual level, concepts are stored. The conceptual level is endowed with a language cue component which provides
38
Language Processing
information as to what language is to be spoken. As in De Bot (1992), language choice is contained and specified in the preverbal message at the conceptual level (Poulisse, 1997:216, Müller-Lance, 2003:121). The conceptual level has a distributed form. In other words, each concept consists of differential meaning elements which are shared by semantically related items and translation equivalents (Poulisse, 1999:62). Poulisse & Bongaerts suggest that at the conceptual level the “conceptual information and the language cue work together in activating lemmas of the appropriate meaning and language” (Poulisse, 1997:216). Lemmas are stored in a shared multilingual network on the subsequent level, the so-called lemma level. In order to allow the system to identify which words belong to which language, the authors posit language tags which (in addition to the semantic and syntactic tags attached to each lemma) specify each lemma’s language membership. This is in line with Green’s proposition that lemmas are tagged with a language label (Green, 1986 [2007]:376). In the Production Model preverbal messages are thought to contain the specification [+1] or [+2] (Poulisse, 1997:217). This ensures that an L2 word receives more activation than an L1 word if the speaker’s intention is to use the L2. Conversely, an L1 lemma is activated if an L1 specification is given in the preverbal message. Unlike De Bot (1992), Poulisse & Bongaerts (1994) do not assume parallel speech plans to explain code-switching (Poulisse, 1999:63). Instead, they propose that L2 lemmas receive more activation than the corresponding L1 item if the preverbal message contains the feature (+L2). Conversely, an L1 specification in the preverbal message will cause an L1 lemma to be activated (Poulisse, 1999:62). The authors suggest that switching is made possible by activating single lexical items rather than by activating complete subsets which contain all the items from a particular language (63). Activation occurs by way of activation spreading (Poulisse, 1999:62; De Angelis, 2007:83, Müller-Lance, 2003:121). Similar to De Bot (1992), who proposed that sounds and articulatory patterns are contained in one single store (Poulisse, 1999:63; De Angelis, 2007:84), Poulisse & Bongaerts posit that the phonological level, where “the phonology of the spoken word is specified” (Kroll & Tokowicz, 2005:539), contains all the sounds and pitch patterns of a speaker’s languages, which explains why interference is also observed in pronunciation (Poulisse, 1997:218). The assumption thus is that bi/multilingual speakers have a shared phonological system. Like lemmas, phonemes are tagged for the language they belong to. This does however not preclude the possibility that an L2 word may accidentally be encoded with an L1 phoneme (Poulisse, 1997:218) thereby resulting in the unintentional production of a non-target phonetic item mapped onto a target lexical item.
Multilingual language production
39
3.3.2 Myers-Scotton: A Matrix Language Frame Like the Production Model outlined above, Myers-Scotton’s account of speech production (1993) is modelled on Levelt’s speaking model (1989). The so-called Matrix Language Frame (MLF) model predicts that the bilingual speaker, when engaging in mixed speech, selects a Matrix Language which supplies the morphosyntactic structure or frame for the utterance to be produced (Myers-Scotton, 2005:327). The frame designates “the set of abstract well-formedness constraints” (Myers-Scotton, 2005:334) which determines the structure of a sentence. Once the speaker has decided on what message to convey, semantic/pragmatic feature bundles in the conceptualizer are accessed. These feature bundles are languagespecific but they are not linguistic units in the narrow sense (Myers-Scotton, 2005:327). The feature bundles then activate lemmas. According to Myers-Scotton, lemmas consist of a three-layered abstract grammatical information store. They contain all the information necessary to initiate the building of a surface-level morphosyntactic structure (Myers-Scotton, 2005:327). More particularly, lemmas contain specifications for semantic and pragmatic features, specifications for thematic roles (such as ‘agent’ and ‘patient’), specifications for the subcategorisation of verbs (such as transitive and intransitive) and specifications regarding surface-level arrangement of morphemes (Myers-Scotton, 2005:328; cf. Longxing Wei, 2002). The assumption is that even though the bilingual formulator does, most of the time, carry out language-specific operations, it is not itself languagespecific since it contains and implements specifications for both languages (Myers-Scotton, 2005:332). Myers-Scotton posits that morphological elements are activated or accessed at different stages during language production (Myers-Scotton, 2005:332). In addition, not all words are accessed in the same way (337). Content words and regular inflections are stored as separate lexical items in the mental lexicon and are built on-line (333). In contrast, irregular forms and non-finite forms are stored as holistic units and are retrieved as such during language production (337). It is suggested that some types of morphemes7 (such as content morphemes) are more easily accessed than others because they become available earlier on in the production process and because they are more salient (337). The Matrix Language Frame model is premised on the notion that one of the bilingual’s languages is usually more dominant than the other. Even balanced bilinguals are held to be stronger in one language than in the other. MyersScotton refers to the stronger language as the Matrix Language. In the case of 7 Myers-Scotton (2006) defines morphemes as the “basic building block in words”. According to the author, the frequently employed definition of morphemes as “the minimal meaningful units in language” does not take into consideration that certain grammatical morphemes do not really carry much/any meaning (236).
40
Language Processing
code-switching, the Matrix Language provides the basic grammatical frame (Matrix Frame) into which elements from the less dominant or Embedded Language are inserted (Myers-Scotton, 2006:243; Kormos, 2006:84,108). While in the MLF model the focus is clearly on the structural features of the languages implicated, other researchers, notably Clyne (2003), Grosjean (2001), Cook (2002), and Herdina & Jessner (2002) have called for a more complete approach to multilingual processing which also takes into account cognitive and psycholinguistic aspects of multilingual development. In Myers-Scotton these aspects are not considered.
3.3.3 Grosjean: The Language Mode Hypothesis Accounting for bilingual production and perception, Grosjean’s Language Mode Hypothesis (1998, 2001) is one of the most important models to date (Grosjean, 1998 [2010]:9–10). The model is not a processing model in the traditional sense, but it nevertheless makes important predictions about the procedures that mediate language choice and selection in bilingual and multilingual subjects. As specified by Grosjean (2008), bilinguals are not simply “the sum of two complete or incomplete monolinguals” but have instead “a unique and specific linguistic configuration” (13/14). Grosjean characterises bilinguals as “speakers-hearers in their own right who will often not give exactly the same kind of results as monolinguals” (1998 [2010]:5). The author therefore advocates “a newer – wholistic – view of bilingualism which states that bilinguals are fully competent speakers-hearers” (Grosjean, 2008:2). Such a view carries weighty implications since it apprehends the bilingual as a fully competent communicator and not lacking or deficient. This is particularly relevant in light of the criticism and doubts that are frequently voiced with regards to the learning and teaching of multiple languages in school. The reader is referred to the second part of the book where the author will be elaborating on the controversies surrounding the discussion of multilingual educational programmes in South Tyrol. Grosjean (1998) describes bilinguals as “people who use two (or more) languages (or dialects) in their everyday lives” (1998 [2010]:3). As predicted by the complementarity principle, bilinguals tend to use their languages in different situations, for different purposes, with different people and in different domains of life (ibid.). This constellation entails that fluency and proficiency in each of the bilingual’s languages are determined by domain specificity. In addition, it has been found that “the bilingual’s two grammars can be quite different from the corresponding monolingual grammars or that language competence (and especially first language competence) can actually change when it comes into contact with another language” (Grosjean, 2008:13). As noted by
Multilingual language production
41
Grosjean, dynamic interferences, or deviations from the norm due to interference from the non-target language, are characteristic features of multilingual systems (2001:1). This nonwithstanding, the bilingual speaker is regarded as having a complete, albeit different language system (Grosjean, 2008:13–14; see also De Angelis & Selinker, 2001:45). In this sense, bilingual speakers’ communicative competence is viewed in terms of their “total language repertoire” as it is used in their everyday life (Grosjean, 2008:14). This is in line with the Dynamic Model of Multilingualism which posits the need for a holistic approach to multilingual acquisition and development. Conversely, the generally prevalent monolingual or fractional view of bilingualism implies that bilingual speakers are two monolinguals in one person with separate competencies in their two languages which entails that the bilingual speaker’s communicative competencies are frequently perceived as imperfect and unbalanced. Derogative qualifiers such as ‘semilingual’, ‘alingual’, ‘dominant’ etc. are widely employed in the literature to refer to bilingual speakers’ language competence or fluency when it is anything less but perfect. Because bilinguals will never be an “ideal speaker-hearer” in the way that monolinguals are, there is generally little acceptance, even among linguists, for “bilingual specificities” (Grosjean, 1998 [2010]:22). Grosjean (2001) postulates that bilingual language behaviour is profoundly affected by language mode (13). He defines language mode as “the state of activation of the bilingual’s languages and language processing mechanisms at a certain point in time” (2001:2). The assumption is that speakers consciously select a base language, which is the language with the highest level of activation. At the same time the speaker also chooses if and how many other languages are to be activated, that is, “[i]n the monolingual speech mode, the bilingual speaker deactivates one language (but never totally) and in the bilingual mode, the bilingual speaker chooses a base language, activates the other language and calls on it from time to time in the form of code-switches and borrowings” (Grosjean, 2001:2). An ‘intermediate mode’ allows for the possibility of various intermediary positions along the language mode continuum. The base language is the most active language in a particular speech situation. The base language dominates the production or perception process insofar as it plays the most prominent role in the communicative act. A second and a third language can be co-activated, albeit to a lesser degree (3–4). Grosjean (2001) states that [a]ny number of factors can help position a bilingual speaker or listener at a particular point on the language mode continuum, that is, set the activation level of the bilingual’s languages and language processing mechanism. Among these we find the participant(s), that is the person(s) being spoken or listened to (this includes such factors as language proficiency, language mixing habits and attitudes, usual mode of interaction, kinship relation, socioeconomic status, etc.), the situation (physical location, presence of mono-
42
Language Processing
linguals, degree of formality and of intimacy), the form and content of the message being uttered or listened to (language used, topic, type of vocabulary needed, amount of mixed language), the function of the language act (to communicate information, to request something, to create a social distance between the speakers, to exclude someone, to take part in an experiment, etc.) and specific research factors (the aims of the study taking place [. . .], the type and organization of the stimuli, the task used, etc.). (5)
In the same way as bilingual speakers move along a continuum when interacting with others, trilinguals, too, can “be seen in a monolingual, bilingual or trilingual mode with various levels of activation depending on her or his position on the language mode continuum” (Grosjean, 2001:3–4). The ensuing figure shows the visual representation of a trilingual speaker in a monolingual, bilingual and trilingual mode. The level of activation of each language is indicated by the degree of darkness of the squares: black stands for active, white for inactive. Language A is the base language in each case (Grosjean, 2001:18).
Figure 3: The Trilingual mode (based on Grosjean, 2001:18)
Multilingual language production
43
Crucially, language mode is predicted to impact the speaker’s perception and speed of lexical access. At the same time, the language mode itself can be affected by the multilingual speaker’s expectations and by a range of nonlinguistic context effects (cf. Dijkstra, 2003:22 for a different view). Grosjean’s notion of language mode also features in Clyne’s (2003) Plurilingual Processing Model (see below). Clyne suggests that “[a] language, usually in the form of a mode in Grosjean’s sense, is chosen for the purpose” of conveying a message (210), adding that “[o]nce a speaker has decided to speak in the (‘monolingual mode’ of a) particular language, this raises the activation threshold of one system, but does not totally inhibit the other(s)” (211).
3.3.4 Green: The Activation/Inhibition model Different psycholinguistic models have focused on different aspects of language processing. In Green this focus is on language control and activation, viz. on “the nature of the control requirements when two . . . [language] systems are involved” (Green, 1986 [2007]:374). Green’s IC (Inhibitory Control) model aims to explicate how bilingual and multilingual speakers accomplish the complex task of generating an utterance in the response language without too much disruptive interference from competing non-target items. The IC model specifies the locus, the means (reactive inhibition of lemma tags) and the control mechanisms (language task schemas and SAS) involved in the selection process (Green, 1998 [2010]:47,49). The following section outlines the basic tenets of the model. A central assumption underlying the Inhibitory Control model is that control in language processing and control of action are mediated by analogous mechanisms (Green, 1998 [2010]:32). As propounded by Green, “speech production can be understood in the same way as skilled action in general” (Green, 1986 [2007]:371; cf. Bialystok, 2007:212 for a similar proposition). The IC model posits that language control is achieved by way of alternating the activation level of language networks, or elements within these networks (Green, 1986 [2007]:371). Language selection is partly seen as a matter of increasing the activation of one language and partly as a matter of suppressing the activation of the nonintended language so that words from that system do not get produced (376). Green first presented the so-called Activation/Inhibition Model (1986) to account for control processes in language production. Later he revised and extended the model. Its successor model, the so-called Inhibitory Control Model (IC) (1998) offers a more detailed account of control operations. Like the earlier model, the IC is premised on the notion that resources are utilised to effect
44
Language Processing
control and to excite or inhibit the language subsystems implicated (Green, 1986 [2007]:375). The IC Model specifies the mechanisms of activation and inhibition during speech production. These mechanisms are held to operate concurrently. Green’s model builds on the assumption that different mental schemas guide or regulate different human actions. Schemas are defined as mental devices or networks which the speaker relies on to carry out a specific task. Schemas may be constructed on the spot or can be retrieved from memory (if a given task has already been performed previously) and can be adapted to meet the requirements of the task at hand. If a person is confronted with a new task the Supervisory Attentional System (SAS) steps in and effectuates the construction of a new schema. The SAS also monitors the performance of these schemas as they regulate task activities (Green, 1998 [2010]: 33–34). In other words, the SAS controls the activation and/or inhibition of “task schemas” which guide language selection. More specifically, the SAS informs the task schemas about the language to be selected and the task schemas then send directions to the “lexico-semantic system” to prepare for the appropriate lemma to be accessed (Myers-Scotton, 2006:308). Green posits that bilinguals do not switch their languages on or off. Instead, he suggests that the speaker’s languages have differential levels of activation, that is, the languages are continuously activated but they are so to different degrees. When a language is selected, the highest level of activation occurs. When a language is not used, its level of activation is predicted to decrease (Green, 1986 [2007]:375). Languages can be selected as the language to speak, they can be active (and take part in the speech process), or they can be dormant (i.e. stored in long-term memory without interfering in the speech production process) (Green, 1986[2007]:375). Frequency and recency of use are held to play a major role in determining whether a given language is selected, active or dormant (De Angelis, 2007:74). When the bilingual speaker chooses to speak a particular language, activation of that language must exceed that of all other languages. Words belonging to the response language must reach a required threshold level in order for activation to occur. Importantly, the Inhibitory Control Model (1998) predicts that the conceptual system activates lexical items from both languages. However, at the lexical or lemma level, non-target items are inhibited. Inhibition is assumed to take place through special reactive inhibition mechanisms which ensure that lexical items that have been activated by the conceptual system but do not belong to the target language are suppressed. At the same time target language items receive more activation. To allow the system to distinguish between target language items and non-target items Green posits the existence of language tags attached to the lemmas at the lexical level.
Multilingual language production
45
Lemmas carrying the wrong tag are identified by the system and are inhibited (Costa, 2005:320; Green, 1998 [2010]:42). Green postulates a so-called specifier which “sets the activation going” and guides the processing operations whenever translation or switching is called for (Poulisse, 1997:209–10). The specifier specifies “the type of control needed in order to execute a given task” (De Angelis, 2007:74). As noted by Green, the inhibitory mechanisms can not entirely prevent nonresponse lemmas from interfering in the lexical selection process (Costa, 2005:320). Green adduces code-switching phenomena to sustain his claim that languages are simultaneously activated. Like De Bot, he considers code-switching to be compelling evidence in support of simultaneous activation of languages and against the widely held view that one of the speaker’s languages is switched off during speech processing. Like De Bot, Green postulates that language choice, i.e. the intention to use a specific language, originates in the conceptualizer. The author notes that “[i]n order to produce a word in a specific language the intention to do so must be part of the conceptual representation and this conceptual representation must contact the relevant lemmas” (Green, 1998 [2010]:37). This, according to Green, does not preclude the simultaneous activation of the other languages spoken by the bi/multilingual subject. On the contrary, if the speaker lacks sufficient knowledge in the selected language, he may draw on words from one of the other languages. Other researchers, notably Costa & Caramazza (1999), dismiss the idea of inhibitory mechanisms which actively suppress lexical items (or lexical nodes, to use their terminology). Instead, these authors posit that lexical selection processes only consider “the lexical nodes of the language-in-use and therefore there is no need to postulate the existence of a mechanism that actively inhibits the lexical nodes of the language-not-in-use” ([2010]:135). Hence, no active inhibition of the non-response lexicon is deemed necessary (121). In line with Levelt (1989) and De Bot (1992, 2004), Green assumes that the production process starts at the conceptual level where, driven by the goal to communicate an idea, conceptual representations are generated (Green, 1998 [2010]:34). The conceptualiser then activates the bilingual lexico-semantic system and, at the same time, also the supervisory attentional system. As outlined above, the SAS is mainly concerned with the activation of the appropriate language task schema for a particular task in a given communicative situation. Naming a picture in the L1, for instance, is assumed to require a different task schema from naming a picture in the L2, or translating from one language to another (Poulisse, 1997:209). The task schema specifies and controls the single steps or sequences involved in language processing (Dijkstra, 2007:22). Language task schemas are formed on the basis of cues, such as linking a given colour or figure
46
Language Processing
cue to the production of a particular language as in “Name a picture in L1, not L2, upon hearing a high tone” (Meuter, 2005:358). Task schemas activate lemmas in the intended language and inhibit lemmas in the non-target language. As indicated, inhibition occurs at the lemma level by virtue of language tags8. A language task schema can alter the level of activation of representations in the system and can inhibit inappropriate items from the non-target language (Green, 1998 [2010]:35; Dijkstra, 2007:22).
Figure 4: The IC Model (Source: Green, 1998:69; printed with persmission from the copyright holder)
An important aspect of Green’s model is that of attentional resources. Resources are supplied at a certain rate by the “resource generator” (Green, 1986 [2007]:375). In Green’s IC model resources are a vital requisite and the major supplier of energy. They allow the system to regulate activation levels and select the appropriate words. Resources are by nature limited or “capacity constrained” (Green, 1998 [2010]:48). Hence, when resources are not sufficiently available, control processes are impeded and production is flawed. This is particularly evident in beginning learners of an L2 who have not yet developed automatised procedures and therefore require more resources and need to make greater efforts. Their low proficiency causes them to allocate all their resources on L2 processing operations with the result that insufficient capacity remains to suppress the dominant L1. This explains why beginning learners’ L1 frequently interferes during L2 production. 8 Green’s suggestion that lemmas might possess two tags, one for comprehension and one for production seems rather unparsimonious (Green, 1998 [2010]:48).
Multilingual language production
47
3.3.5 Clyne: The Plurilingual Model Like most of the here presented speech processing models, the Plurilingual Model proposed by Clyne (2003b) is based on Levelt’s monolingual model for speech production (1989). Adopting a dynamic and holistic conception of language, the Plurilingual Model (see below) seeks to account for speech processing and development in multilingual speakers. The model differs from previous models in that it considers the impact of various sociolinguistc and psycholinguistic factors on multilingual processing, including the speaker’s multiple identity, language mode, contextual and situational variables, and phenomena such as convergence, transference and transversion (Clyne, 2003b:211). According to Clyne, dynamic models of speech processing are better suited to accomodate plurilingual speech and language contact phenomena than traditional linear models (103). Clyne conceives of plurilingual speech as the product of multiple factors, which explains why plurilingual speech will “often defy hard and fast linear models” (241). Clyne’s holistic approach invites a comparison with the Dynamic Model of Multilingualism developed by Herdina & Jessner (2002) which will be outlined further down. An important aspect in Clyne’s model relates to the issue of language choice. Language choice is achieved by virtue of language tags attached to lemmas. Lemmas can be tagged for one, two or more languages. Multiple tagging is expected to facilitate or induce transversion (i.e. code-switching) from one language to another. As specified by Clyne, tags guide the formulator in choosing the appropriate grammatical encoding operations and the articulator in selecting the right sound patterns (Clyne, 2003b:211). According to Clyne, language choice is conditioned by social and motivational factors (Clyne, 200b3:210; Jessner, 2004:30). Clyne explicates that language “choice is taken on the basis of context (e.g. setting, participants and genre), identity and other attitudinal issues” (Clyne, 2003b:211). The assumption is that sociolinguistic and psycholinguistic factors, in particular affective and attitudinal factors, impact the speaker’s language mode (212). In line with De Bot (1992) and Green (1998), Clyne posits “a processing model with joint storage of material from the two or more languages of the plurilingual” (Clyne, 2003b:241) arguing that “storage and access of the two or more languages of a plurilingual cannot be separate” (Clyne, 2003b:214). The speaker’s languages are conceived of as interconnected networks in which shared items or items that are perceived to be similar form a link thereby allowing for cross-language activation (Clyne, 2003b:211; see also Herdina & Jessner, 2002). Connections between the language systems can exist at all levels of processing (Clyne, 2003b:211–212). Cross-language activation is said to be facilitated by convergence, viz. the “relatedness of two or
48
Language Processing
Figure 5: The Plurilingual Model (Source: Clyne, 2003b:213; printed with permission from the copyright holder)
more languages” (Clyne, 2003b:211). It is assumed that transference or “the use of forms, features or structures from another language” (234) “is the result of unclear or multiple language tagging of lemmas” while transversion, i.e. the crossing over to another language “is further down the processing schema because it is often the product of transference and/or convergence” (Clyne, 2003:214). Clyne differentiates between two types of convergence, one type relating to the adoption, in the L3, of a pattern shared by L1 and L2 and the other type consisting of conversion formulae (Clyne, 2003b:105; Jessner, 2006:80). In contrast to the first type of convergence, where two languages share the same pattern which is then adopted in the L3, conversion formulae or rules “transform an item from one language to one in another language” (Clyne, 2003b:109, italics added). The occurance of convergence phenomena is seen as evidence
Multilingual language production
49
for common storage and non-selective access to languages in the brain (214; see also Singleton, 2003:176). There is common agreement that trilinguals typically transfer features from one Ls to another Ls. Transfer has been found to take place at all levels of language production (Clyne, 2003b:76–79; Jessner, 2006:80). It is not always clear what prompts speakers to transfer or switch. According to Clyne’s triggering hypothesis, words with similar form and meaning in two languages can trigger or facilitate transversion, i.e. a switch to the other language (De Bot & Broersma, 2006:2). Lexical transfers (items from one language which have also become part of the speaker’s other language lexicon), proper nouns and bilingual homophones (items that sound the same or almost the same in the speaker’s two or more languages) can all function as trigger words and cause or facilitate code-switching. Conversely, there are indications that “a high degree of phonic/ phonological integration tends to inhibit transversion” (Clyne, 2003b:211; Jessner, 2006:105). Transversion, which denotes a switch from one language to another (Clyne, 2003b:162) is seen as a possible indication that parallel speech plans are prepared for (all) the languages known to the speaker (214). Clyne (1997) suggests that transversion between three languages operates on similar principles as transversion between two languages, with the difference that the links between each “additional language complicates the operations of the processes” involved ([2010]:262). According to Clyne, the notion of transversion facilitation requires adjustments to be made to traditional modular models of speech processing, since strictly feedforward modular models can not accommodate the occurance of transversion facilitation. Unlike unidirectional models of processing, the Plurilingual Model therefore assumes perceptual feedback from the phonological level to the lemma level (Clyne, 2003:242). Clyne observed that trilingual speakers frequently use only one of their languages as support to help them with the target language. The language used as support language is often the speaker’s L2, rather than the speaker’s L1, as one might expect. This coincides with Williams & Hammarberg (1998) who found that there is a tendency in L3 learners to suppress their native language in favour of a related second language, even when native and target language structures are similar (see also Hammarberg, 2001:21–41). Williams & Hammarberg (1998) explain the observed behaviour with “an automatic activation of other second languages in the context of second language learning [. . .] and the desire to suppress the first language in the belief that an inherently ‘foreign’ language would serve better as a point of reference for another ‘foreign’ language” (Herwig, 2001:132). The implication thus appears to be that multilinguals engaged in L3 discourse show greater preference for another ‘foreign language’
50
Language Processing
to support them in their communicative efforts. This is consistent with Cenoz (2003b) who reports studies in which only “the second language and not the first is activated while the third language is selected” indicating that learners tend to transfer more from their L2 than from their L1 in spontaneous L3 production (106). Meisel (1983) relates to this phenomenon as “foreign language effect”, while Hammarberg (2001) speaks of the “L2 status” (Cenoz, 2003b:106; 2001:9).
3.3.6 De Bot: The bilingual/multilingual production model De Bot’s first major contribution to the field was the development of the Bilingual Production Model (1992) which is based on Levelt’s model for monolingual production. The Bilingual Production model has since been revised and extended to accomodate multilingual production. The new Multilingual Model proposes a more comprehensive theory of linguistic processing in multilingual speakers. De Bot assumes no distinctive difference between bi-, tri- and multilingualism. According to De Bot, the same principles underlie linguistic processing in bilingual, trilingual and multilingual speakers (2004:30; cf. Clyne, 2003). In line with Levelt (1993), De Bot postulates three levels for linguistic representation and processing, one for conceptual features, one for syntactic procedures and one comprising form elements including sounds, syllables and gestures (De Bot, 2004:28). Each level contains language-specific subsets which can overlap if the speaker’s languages are typologically related, i.e. if elements are shared across languages. Shared elements function as a bridge within subsets and between languages. Shared syntactic features, for instance, result in the activation of all the subsets that share these procedures (26). In the same way, shared forms at the phonological level can activate items from various languages (De Bot, 2004:24). Access to words in the mental lexicon is held to be non-selective. Non-selective access of words entails a “parallel search through all languages” (De Bot, 2004:18). The resulting implication is that items belonging to different languages compete for activation. Activation levels and cross-language competition are held to be dependent on factors such as language proficiency, frequency, recency, or age of acquisition. A ‘default level of activation’ must be attained for words from a given language if these words are to be considered in the selection process (De Bot, 2004:22). According to De Bot (2004), the intention to use a particular language originates at the concept or intention level. At the level of the conceptualizer communicative intentions are turned into pre-verbal messages which are then fed into the formulator. The formulator has access to the lexicon which contains lexical items from all the languages (De Bot, 1992 [2007]:402). The lexicon com-
Multilingual language production
51
prises lemmas containing the word’s semantic and syntactic information and lexemes which specify the possible forms of a word. Two subcomponents in the formulator carry out grammatical and phonological encoding respectively. The subcomponent for grammatical encoding produces a surface structure. The subcomponent for phonological encoding receives the surface structure and produces a phonetic plan. This plan is then passed on to the articulator where all the possible sounds and prosodic patterns of the speaker’s languages are stored. The articulator then generates the utterance (Clyne, 2003:196). In accordance with Levelt (1989), De Bot (2004) assumes one single articulator which contains sounds and syllables from all the speaker’s languages. Depending on frequency, recency of use and linguistic proficiency, L2 learners are hypothesised to develop their own norms for L2 sounds. In low-proficiency learners these norms can be very close to L1 norms. This explains why beginning learners tend to display more phonological interference (Poulisse, 1997:212). In addition, it appears that “individual sounds are encapsulated in syllables and therefore less easily adapted” (De Bot, 2004:26; cf. Levelt, 1989). Earlier, De Bot (1992) had predicted that speakers produce parallel speech plans for their different languages. This allowed the author to account for code-switching and cross-linguistic interference. However, as noted by Poulisse (1997), the assumption of two parallel speech plans (for the selected and the active language) is problematic. In particular, so Poulisse, it is impossibile to reconcile the proposition that language choice occurs in the conceptualizer (from which it follows that only one speech plan - that for the intended language – is prepared) with the notion of two parallel speech plans. Speech production, she concludes, would be too effortful and unparsimonious if several speech plans had to be prepared every time an utterance is generated. This line of reasoning appears particularly relevant in the case of multilingual speech production since there would be no limit to “the number of alternative plans” if one were to endorse the parallel plan hypothesis (Poulisse, 1997:213). The recognition that the simultaneous generation of speech plans is incompatible with the idea of language specifications instantiated at the conceptualiser level eventually led De Bot (and other proponents of this theory) to discard the idea of parallel plans (Poulisse, 1997:212). Like Green (1998), De Bot maintains that languages are activated to varying degrees. In particular, De Bot states that “[i]n selection the elements from a given language/subset are preferred, but the elements from other languages are not completely blocked out” (De Bot, 2004:30). As regards the organisation of the mental lexicon, De Bot (2004) proposes a shared lexicon for all the speaker’s languages. The lexicon contains language-specific subsets that can be activated or deactivated independently, that is, languages can be activated and inhibited as sets (24).
52
Language Processing
More recently (inter alia at the DST and Multilingualism Workshop held at Innsbruck University in 2010), De Bot has challenged existing modular processing models, including his own (1992, 2004), on the grounds that they fail to accomodate the complex and dynamic nature of language and language acquisition. The author of this book subscribes to De Bot’s criticism of extant speech processing models since many of these models contrast with more recent holistic views of language(s) as complex and dynamic systems and because they fail to acknowledge the correlation between linguistic development and more general cognitive processes (cf. Herdina & Jessner, 2002; Aronin & O’Laoire, 2004; De Angelis, 2007; Lowie & Verspoor, 2011; Bialystok, 2001,2005,2007).
3.4 Language Comprehension A fundamental difference between models of speech production and models of speech perception/comprehension is said to relate to the directionality of processing operations. Language production is generally thought to implicate top-down procedures. Comprehension, by contrast, is held to entail bottom-up mechanisms. A further difference between production and comprehension models is believed to concern the issue of language selection. In speech production some kind of selective mechanism or language cue is widely thought to come into play very early on in the production process (see Clyne, 2003 and Hammarberg, 2001 for a somewhat different proposition9) so as to allow multilingual speakers to select the intended language for a particular communicative situation. Conversely, in speech comprehension the processor is not presumed to activate the target language node until much later in the processing sequence (Kroll & Tokowicz, 2005:540). Dijkstra (2007) provides a good insight into the complexities of language perception in particular and language processing in general: . . . after entering the ear, an input speech signal is represented as a phonetic/phonological representation of sounds. Next, in a number of steps this phonetic/phonological representation of the utterance is recoded into a meaning representation. In order to do so, knowledge about language (e.g., syntactic rules) and the world needs to be retrieved from longterm memory, temporary products of processing need to be stored in a working memory, and the whole process must be monitored to detect possible derailments in an early stage. This requires cognitive control and the investment of attentional capacity. In all, language processing requires a number of complex components (214).
9 These authors suggest that language choice can occur locally in the speech sequence (cf. Hammarberg, 2001:37).
Language Comprehension
53
In what follows we examine the mechanisms and computational routines that are said to allow multilingual speakers to perceive and identify the sounds and words of their languages. We look at different accounts of how speakers of multiple languages recognise visually and orally presented language and how they assign language membership. We begin with a discussion of visual speech comprehension. One of the most influential computational models of bilingual visual word recognition, the Bilingual Interactive Activation model, is presented. First the architecture and the assumptions underlying the model are outlined. Then two recent extensions to the BIA model are discussed. The second part of this section deals with aspects of auditory speech perception. This is followed by an overview of the BIMOLA model for auditory speech comprehension. The chapter closes with a dynamic systems perspective of multilingual representation and processing.
3.4.1 Visual language comprehension in bilingual and multilingual speakers: The BIA, the BIA+, and the MIA model The Bilingual Interactive Activation model was developed by Grainger & Dijkstra (1992) and Dijkstra & Van Heuven (1998) (Kroll & Tokowicz, 2005:534). The BIA model posits an integrated lexicon with non-selective lexical access in the initial processing phase but cross-language and lateral inhibition at subsequent processing stages (Dijkstra & van Heuven, 1998[2010]:63, 65). Initially, the processing system considers all potentially relevant candidates in parallel. The appropriate candidate or target item is not selected until much later. This is referred to as “late selection” (84). The BIA model proposes a language node level at the top of the structure. The language node level contains language membership information in the form of language tags. Apart from being able to activate words from the target language, the language level can also exert top-down inhibition on word candidates from the non-target language (De Angelis, 2007:105). The assumption is that language nodes “collect activation of all words from one lexicon and suppress all words in the other lexicon” (Dijkstra & van Heuven, 1998[2010]:64). The BIA model consists of four hierarchical but interconnected levels of representations. The lowest level contains letter features. These are visual features of letters at different positions within a word. They include script type, diacritical markers (e.g. accents, as in French or to a lesser degree in Italian) or onset capitals (as in German). The second level contains letter representations, the third level consists of words. As mentioned previously, the highest processing level is occupied by the so-called language node. There are connections between
54
Language Processing
various levels of representation and between the representations or nodes at each level10. It is assumed that items at the word level can inhibit each other. In addition, the most highly activated words can send activation back to the letter level and (re)activate their constituent letters (Dijkstra & van Heuven, 1998[2010]:64). A target’s neighbours (words which differ from the target in just one letter) are predicted to inhibit other words which share fewer features with the target. This “gang effect” (Dijkstra, 2003:13) facilitates word recognition. The assumption is that neighbours will eventually also begin to suppress each other by way of lateral inhibition (ibid.). As is typical of psycholinguistic comprehension models the BIA model predicts that word recognition is mapped from the bottom. It proceeds from the lowest level to the highest, i.e. letter features in different positions within a word may provide a first indication as to what language is spoken. Letter features then activate letter units and these in turn activate those words that contain the given letter(s) in the right position. Words that do not meet the requirements are inhibited, and their level of activation is reduced. The activated words can send activation to the relevant language node and back to their constituent letters. Target items thus receive activation from several sources. Eventually, the most highly activated item “crosses a recognition threshold” and is identified as the target word (Dijkstra & van Heuven, 1998[2010]:64). Unfortunately, the BIA model does not propose any representation of semantics (Kroll & Dussias, 2004:173). This invariably invites criticism since it is unclear at what stage during speech processing the meaning of a word is accessed. However, this weakness is redressed in the revised version of the model, as will be illustrated further down. Word recognition as proposed by the BIA model (and its successor, the BIA plus model) is considered a dynamic process in the sense that language information is continuously exchanged between the various processing levels (Dijkstra, 2007:217). Importantly though, it has to be noted that Dijkstra does not assume any significant dynamic interaction between bilingual processing and contextual factors. Instead, he suggests that lexical processing is not subject to any extreme influence by contextual factors and claims that lexical processing “remain(s) relatively unaffected in different contexts and task situations” (Dijkstra, 2007:223). This stands in contrast to Grosjean’s proposition that language mode is affected by external factors such as situational setting, interlocutor, etc. It also appears to run counter to more recent psycholinguistic theorizing which postulates a range of oftentimes unpredictable effects on multilingual discourse and development. 10 The use of the term ‘node’ for single representations or items at each level is somehow misleading as the term also denotes the language (node) level at the top of the structure.
Language Comprehension
55
Input is the most important cue for selection in visual word recognition (Dijkstra, 2003:20). Language-specific cues such as diacritical markers or onset capitals are particularly powerful since they immediately eliminate those competitors that do not meet the feature criteria. Cues may include “languagespecific cues, language distance, and script type, and linguistic context aspects having to do with morphological, syntactic, and language membership information” (Dijkstra, 2003:25). High frequency words are recognised faster than low frequency words since they have a “higher resting level activation” (2007:232). This makes them priviledged in terms of their relative level of activation and in terms of processing speed (Dijkstra, 2003:14). Lexical items from typologically distant languages are predicted to compete less than candidates from typologically related language systems (Dijkstra, 2003:20). Script differences are expected to facilitate the recognition process (21). The same holds true for selectivity-inducing higher-level linguistic types of information such as morphological markers, sentence context information, and language membership information (23). Morphologically complex words are predicted to have fewer neighbours to compete with in the selection process. Similarly, a syntactically or semantically constraining sentence frame is expected to allow for faster word recognition as it reduces the number of potential readings. Language membership information becomes available as soon as one language node receives more activation than the other node(s). This information is fed back to the lexical level where all the non-target items are inhibited (23–24). As noted previously, the BIA model assumes an integrated lexicon, i.e., words from two or more languages are stored in one single mental lexicon. Access to the lexical units in the system is believed to be language non-selective and automatic, that is, no initial distinction is made between L1 and L2 language input. For obvious reasons the principle of non-selectivity does not apply to the language node level. In other words, integration of lexical items remains in place until language membership comes into play (De Angelis, 2007:106) which occurs when word candidates activate the appropriate language node and language membership is established (Dijkstra, 2007:217). The non-selective nature of bilingual word recognition implies that orthographic, semantic or phonological cross-linguistic similarities can impact the perception and recognition of words, that is, cross-linguistic similarities can facilitate or hamper comprehension by speeding up or slowing down the recognition of lexical items. In order to account for additional factors which have consistently been found to impact word recognition Dijkstra & Van Heuven (2002) modified and extended the BIA model by a so-called task/decision system. The task/decision system can accommodate task demands in visual word recognition. The new BIA plus model is also endowed with a lexical identification system which contains
56
Language Processing
Figure 6: The Trilingual or Multilingual Interactive Activation Model (based on Dijkstra, 2003:17)
orthographic, phonological and semantic representations that interact during the process of word recognition. The lexical identification system accounts for various types of linguistic context impinging on the recognition process. More recently, the BIA plus model has been extended to accommodate trilingual and multilingual word recognition (Dijkstra, 2003:17; see above). Like its predecessor, the Trilingual or Multilingual Activation Model assumes a common, integrated lexicon and parallel access and activation of languages. As in the bilingual version, selection of the intended word or language occurs by way of activation and inhibition of potential candidates (De Angelis, 2007:106). The Multilingual Activation Model predicts that competition increases as new languages are added to the system (Dijkstra, 2003:18). More particularly, it is assumed that competition increases as speakers become more proficient and as new words are integrated into the multilingual lexicon (19). This said “there are
Language Comprehension
57
often strong selectivity-inducing constraints available in the input stimulus and its context, and it seems likely that the multilingual will exploit these as soon as possible” (25). Dijkstra (2003) notes that the majority of factors that have been shown to impact visual word recognition can be expected to apply for auditory word recognition, too. A modality-dependent modification would then have to consider sublexical cues of a different nature from those implicated in visual processing. Sublexical cues would in this case comprise phonemes, tone, aspiration etc. (Dijkstra, 2003:25). Auditory processing is the focus of the following section.
3.4.2 Auditory language processing in bilingual and multilingual speakers: The BIMOLA model for auditory speech comprehension Auditory language processing differs from the processing of visually presented language in at least one crucial aspect. Auditory language processing, or the perception and comprehension of spoken language, is dependent on the provision of speech signals or language-specific sounds which are not available in printed text, for example. The ensuing section presents the so-called BIMOLA model of lexical access, and outlines the mental operations and mechanisms which are thought to mediate the perception and comprehension of spoken language. The Bilingual Model of Lexical Access was first presented in 1988. In contrast to the BIA model, which accounts for visual word recognition, the BIMOLA model delineates the processes involved in spoken word recognition (Grosjean, 1998 [2010]:24). The BIMOLA model was originally designed to account for bilingual word recognition, but can, as pointed out by its author, be extended to accomodate more than two languages. The BIMOLA model postulates that the speaker can be in a monolingual or bilingual language mode. The monolingual mode entails that one language network is strongly activated while the other is only slightly activated. By contrast, the bilingual mode assumes that both languages are activated, albeit to different degrees (Grosjean, 1997:248). Like the BIA and MIA model, the BIMOLA model thus assumes parallel activation of words from the different languages. As indicated, the BIMOLA model consists of three levels, with features at the lowest level, phonemes at the subsequent level and words at the highest level. The feature level is shared by all languages. Phonemes and words are said to be organised both, independently because they are stored in separate subsystems, and interdependently because they belong to one and the same system (Grosjean, 1997:248; 2008:203).
58
Language Processing
In the BIMOLA model connections between the various levels can be bottomup and, to some extent, top-down. Connections between features and phonemes are unidirectional, viz., from features to phonemes. Bottom-up activation occurs between features and phonemes and phonemes and words. Features activate phonemes and phonemes activate words. By contrast, connections between phonemes and words are bidirectional. Phonemes activate words and words can activate phonemes. Above the word level there is a further level or component containing higher linguistic information which can send top-down activation to words and then on to phonemes. This component for global language information is positioned outside the processing system itself. A language node such as proposed in the BIA model becomes redundant because the global language information component takes over its functions. An obvious weakness of the BIMOLA model consists in the fact that it does not account for word meaning. The same holds true for the BIA model, as the reader will remember. As already mentioned, many of the here outlined processing models and theories have been challenged on the grounds that they do not really differ in any substantial way from structuralist accounts of language processing. In particular, it has been argued that the assumption of a special language processor in the brain is unmistakably redolent of Chomskyan theorizing. The author of the present volume endorses this criticism. As stated previously, cogent alternatives to traditional conceptualisations of linguistic representation and processing have been submitted by scholars working in the field of Dynamic Systems or Complexity Theory. These researchers question the idea that linguistic operations are based on linguistic content or representations in the brain (LarsenFreeman & Cameron, 2008:107–109). Instead, they argue that complex dynamic processes operate on non-representational internal states (108) suggesting that linguistic knowledge is not ‘given’ or located in the brain but ‘achieved’ (109). The implication is that knowledge or meaning is not represented in the mind but that it emerges in the dynamics of interaction (ibid.). From a Dynamic Systems Theory perspective no distinction is required between language representation and language processing operations since all language-related procedures emerge and self-organise from frequently occurring patterns of language use (111). This conceptualistion of language contrasts sharply with most of the above outlined models of representation and processing. The section that follows provides a detailed description of a dynamic systems approach to language representation and processing. The purpose is to help the reader get a more concrete idea of dynamic conceptualisations of language.
Language representation and processing from a dynamic systems perspective
59
3.5 Language representation and processing from a dynamic systems perspective In their recent publication on the dynamics of multilingual processing, Lowie & Verspoor (2011) contrast traditional processing models with a dynamic systems view of language processing. The authors contend that the notion of separate and static modules with a conceptualiser, formulator, lexicon and articulator in the brain is incompatible with a dynamic systems approach. In opposition to early linear models, the authors propose a continuous model within a continuity-of-mind framework. The continuous model assumes no fixed entities in the mind but only processes (278). The assumption is that all mental activity consists of cognitive routines and is “constructed over time” (274) as it emerges from use (276; cf. Larsen-Freeman & Cameron, 2008:107–109). In other words, what is commonly referred to as ‘representations’ are, from a DST perspective, “processes in and of themselves, sparsely distributed patterns of neural activation that change nonlinearly over the course of several hundred milliseconds, and then blend right into the next one” (Spivey, 2007:139 in Lowie & Verspoor, 2011:278). Lowie & Verspoor (2011) suggest that representations can best be understood as “highly variable, dynamically changing and context dependent distributed mental subsets that do not inherently belong to a particular language” (280). Instead of discrete, sequential processing stages, Lowie & Verspoor (2011) propose dynamic and constantly changing subsystems, i.e., sets of related cognitive routines (276). According to the authors, these subsystems “are interconnected and interact constantly, and while interacting they affect and change each other continuously” (277). From this it follows that the processing ‘components’ are not conceived as separate, impermeable modules, but that, instead, there is constant interaction (ibid.). In particular, this means that at the conceptualiser level, there are no languagespecific representations but emergent, usage-based entities or concepts which “are not initiated by language-specific information, but by experience” (276). As regards the multilingual lexicon, Lowie & Verspoor (2011) advocate a dynamic view of lexical knowledge (281). More specifically, lexical knowledge is here envisaged as dynamic, context-based, cognitive routines. The implication is that the meanings of words are not stable, but constantly changing. Word meanings are said to be dependent on individual and contextual factors. Word meanings from different languages are seen as having ‘fuzzy boundaries” (281) because they overlap to various degrees. Vocabulary acquisition is explained in terms of ‘gradual discovery’ of meaning (ibid.).
60
Language Processing
Crucially, grammar is not seen as distinct or separate from the lexicon (cf. Bleckner, et al., 2009:7). Early suggestions that the lexicon and the grammatical encoding component discretely store and process lexical and syntactic information respectively, are not supported by the DST approach. Therefore, in contrast to traditional psycholinguistic models, Lowie & Verspoor (2011) posit that there is no clear border between the lexicon and syntax (281). More particularly, it is assumed that “lexicon and grammar form a continuum” and that therefore “grammar is itself meaningful, just as lexical items are” (Langacker, 2008:67 in Lowie & Verspoor, 2011:281). This, according to Lowie & Verspoor, is corroborated by the frequent occurance of so-called formulaic expressions or chunks in everyday communicative exchanges (282). In the same way as conceptual, lexical and syntactic processing are seen as integrated routines, phonological processing, too, is explained in dynamic terms of combined and coordinated actions between the various subsystems in the mind. Lowie & Verspoor reject the idea of an articulator containing declarative information11 in the form of fixed representations which are processed in a sequential manner (284). Rather, phonological processing and acquisition are thought to be dependent on the speaker’s perceptual abilities and are held to be correlated with the frequency with which particular phonological patterns occur in the immediate linguistic environment (283). It is additionally suggested that L2 learners initially rely on the sound structure of their L1. Dissimilarities between languages are said to hamper the acquisition process and/or phonological processing. In contrast, similarities beween the phonological systems of the learner’s languages appear to facilitate phonological acquisition and processing (284). In Chapter 4 further considerations on multilingual processing are proposed and complemented by a comprehensive review and discussion of recent developmental models of second and third language learning.
11 The commonly drawn distinction between declarative and procedural knowledge is also seen as problematic since it presupposes that mental representations are static (278).
4 Developmental models of second and third language learning This chapter explores developmental models of second and third language learning. In contrast to the models outlined in the previous chapter, the acquisitional models presented in this section give broader scope for multiple language learning and development since the principal focus is not on linguistic processing but on general cognitive and psycholinguistic factors implicated in multilingual performance. Of the ensuing models the Dynamic Model of Multilingualism (2002) is by far the most complete and comprehensive.
4.1 William & Hammarberg’s model of L3 production Focusing on multiple language acquisition, Williams & Hammarberg (1998) propose a developmental model of L3 production. The authors argue that earlier speech processing models, though valuable contributions to the field, failed to address a range of crucial issues including the impact of previously learned languages (326) and the differential roles of L1 and L2 in L3 acquisition and production (295, 303). According to Williams & Hammarberg, the learner’s L2 plays a fundamental role in L3 acquisition and development (295). In particular, Williams & Hammarberg state that learners’ “L2s appear more likely to be activated than the L1 as supplier language during the early stages of L3 acquisition” (323). The authors assume differential acquisition mechanisms for L1 and L2 hypothesising that L3 acquisition may reactivate the acquisition mechanisms that were operational during the acquisition of L2 (ibid). Williams & Hammarberg observed that, in the early stages of L3 production processes, learners tend to draw on their L2 for support. This leads the authors to conclude that speakers’ L2 may be “activated in parallel to the L3 interlanguage, underlying L3 production” (295). William & Hammarberg’s model predicts role assignment to each of the languages implicated. Role assignment is thought to take place at a higher level of processing and is held to render processing procedures smooth and economical (325). This view entails that, whenever three languages are involved, one language (generally the speaker’s L2) functions as default supplier, while the other (generally the L1) plays only a negligible role in L3 lexical construction processes. Williams & Hammarberg found that speakers’ L1 is commonly assigned an instrumental role and is mainly used to comment on the form or function of linguistic items (295) or on the communicative situation (307). Conversely, the L2
62
Developmental models of second and third language learning
is generally assigned the role of default supplier, thus supplying or contributing most of the cross-linguistic items (324). As explicated by the authors, the (default) supplier language is more highly activated than the L1 (325). The authors further note that L2 influence can extend to L1 productions causing the speaker to generate utterances which may appear to deviate from the L1 norm (313-316; 322). Williams & Hammarberg identify four variables which they presume to be implicated in assigning a given language the default supplier role. These include proficiency, typology, recency of use and L2 status. In addition to these, a fifth contributing variable is proposed, namely individual learner situation. It is suggested that the default supplier role is assigned to the language that best matches or conjoins these variables (322). To recapitulate, we may say that the novelty of Williams & Hammarberg’s model consists in “the incorporation of L1 versus L2 status” (323) into a model of speech production. As intimated by the authors, L2 status appears to override any influence of L1 during L3 lexical construction attempts. Importantly, as the speaker’s proficiency in L3 increases, the role of L2 and L1 diminishes (325). It thus follows that the effects of L2 and L1 on L3 speech processing correlate with the speaker’s level of L3 mastery.
4.2 The Dynamic Model of Multilingualism A radically novel perspective on multilingualism and multiple language acquisition has been presented by Herdina & Jessner (2002). The authors purport to provide a scientific means of predicting multilingual development (87). In their investigation and discussion of multilingual systems the authors propose a new conception of multilingual development by endorsing a systems-theoretic perspective and by applying a ‘multilingual norm’ to linguistic behaviour (Jessner, 2008a:270). The Dynamic Model of Multilingualism (henceforth DMM) provides a psycholinguistic framework for the modelling of multiple language-related processes (154). DMM conceives of language development as a complex and dynamic process (Jessner, 2008a:270; cf. Larsen-Freeman & Cameron, 2008, Lowie & Verspoor, 2011). In the attempt to remedy the shortcomings of previous theory formation the authors aim to accomodate a wide range of commonly neglected parameters including negative growth patterns, language maintenance, metalinguistic abilities and autocatalytic changes arising from the system itself. As specified by Herdina & Jessner (2002), an adequate description of multilingualism needs to account for cross-linguistic phenomena, for language attrition and loss, for learner variability and for the positive effects of being multilingual.
The Dynamic Model of Multilingualism
63
The positive consequences of multilingualism are said to comprise, in particular, speakers’ enhanced metalinguistic and metacognitive abilites, divergent thinking and cognitive flexibility (Herdina & Jessner, 2002:75). The DMM is premised on the assumption that multilingual systems are not merely the sum of a number of language systems but that they are quantitatively and qualitatively unique. This is in accord with Kecskes & Papp (2000) who (with regard to bilingualism) note that the bidirectional influence between L1 and L2 results in a language use which differs from that of monolingual speakers of either language (in Kecskes, 2006:222). Kecskes (2006) proposes that as new L2 knowledge merges with existing L1 knowledge, a new and complex language system emerges (235). Herdina & Jessner (2002) therefore argue that multilingual proficiency is “not reducible to monolingual competence” (57). Rather, multilingual proficiency is seen as observing its own unique principles (ibid.). This is echoed by Bernaus et al (2011) who, relating to Herdina & Jessner’s (2002) holistic conceptualisation of plurilingual (i.e. multilingual) competence, state that “the specificity of any single language coexists with the mutual influences and connections between the components of the individual repertoires” (17). The present author proposes to call this unique constellation and the interaction between the speaker’s multiple language systems dynamic connectivity and complex interactivity. Herdina & Jessner (2002) argue that bilingualism constitutes a variant form of multilingualism and suggest that many of the findings yielded by bilingualism research can be applied to multilingualism. At the same time, they assume that certain features of multilingualism are distinct from bilingual phenomena and that multilingualism research therefore requires a differential, viz. holistic approach (4; see also Jessner, 2006 and De Angelis & Jessner, 2012). This coincides with the suggestion made by Ellis (2008:381) that a dynamic perspective (such as provided by Emergentism, Chaos Complexity Theory, and Dynamic Systems Theory) is required if one is to gain a more complete understanding of multilingualism. DMM conceives of human beings as complex biological systems (Herdina & Jessner, 2002:154) and of languages as inherently dynamic, constantly changing (74) and highly idiosyncratic. Multilingual speakers are seen as complex psycholinguistic systems comprising a number of individual language systems (LS1, LS2, LS3, etc.) (Herdina & Jessner, 2002:3). Parameters implicated in the development of multilingual systems include non-linearity, connectivity, variability, irregularity, unpredictability, self-organisation and self-renewal (74–84). The dynamic model is based and draws extensively on concepts and notions from related fields of science such as biology and cognitive psychology (85) where Systems Theory, Complexity Theory and Chaos Theory have come to occupy a
64
Developmental models of second and third language learning
prominent role in the discussion and investigation of dynamic systems (cf. Larsen-Freeman & Cameron, 2008). Dynamic systems have been described as “systems with many components that interact, meaning that they codetermine each other’s time evolution” (van Geert, 2008:181). More specifically, dynamic systems are seen as developing “in a process of interaction between the dispositions of the system and input from the environment” (Herdina & Jessner, 2002:86). Herdina & Jessner (2002) contest traditional theories in linguistics pointing to their inadequacies and weaknesses in dealing with the complexities of multiple language acquisition and/or learning (2). UG or Universal Grammar theories, in particular, are seen as “a hindrance rather than a help in developing a realistic view of multilingualism” (152). In accord with systems theory, the dynamic model conceives of progress in language learning as “non-linear in contrast to other approaches where this progress is interpreted as an ordered sequence of individual steps suggesting a steady upward motion where one step follows on the other” (89). The dynamic systems approach stands in stark contrast to nativist theories which posit that language learning is a linear process based on a set of predetermined developmental stages which cannot be skipped and are unilinear (35). According to UG theory all learners have the same underlying language competence and pass through the same developmental stages. In UG terms, competence is invariable and seen as underlying performance. It is not subject to variations observed in performance (Herdina & Jessner, 2002:40). Hence, the notion of language competence as employed in UG is considered incompatible with DMM. UG comprises a set of inborn or genetically determined principles and parameters. Parameters are thought to have a limited number of settings, and the particular settings of all the parameters that apply to a particular language determine the structure of that language. Generative theory is concerned with the ideal speaker-listener, and accumulation of linguistic knowledge is seen as leading to ideal native-speaker competence12. As regards the acquisition of second or third languages, UG based theories assume parameter resetting. The argument goes that L2 learners have access to the same principles as native speakers but that parameters have to be reset. It is important to keep in mind that the unitary view of language competence propounded by generativist theories contrasts with the notion of multilingual proficiency in DMM. The term ‘competence’, as employed in UG theory, wrongly implies some sort of “clearly definable entity” (Herdina & Jessner: 2002:54). In DMM the notion ‘multilingual proficiency’ is therefore preferred to 12 The DMM, by contrast, is not interested in native-like attainment but in the process of multiple language learning.
The Dynamic Model of Multilingualism
65
that of ‘competence’. According to DMM, multilingual proficiency is “by necessity a derived quantity” (57) in the sense that the multilingual speaker develops linguistic and cognitive skills “that differ substantially from those required by and expected of the monolingual speaker” (144). In other words, multilingual proficiency is seen as a multicomponential construct consisting of a number of interrelated abilities, skills and strategies. According to the DMM, multilingual systems are dynamic and therefore characterised by constant change or growth. In DMM growth is non-linear and is seen as potentially taking two directions, an up- or forward direction as in the increase and accumulation of language knowledge and a down- or backward direction as in the case of language attrition or loss (91). One variable effecting gradual loss of language and thereby compromising the stability of the multilingual system is identified in the limited availability of learner resources (91) and the (perceived) disproportionate amount of General Language Effort or GLE (GLE is defined as the sum of LME and LAE; LME=language maintenance effort; LAE = language acquisition effort) required for the maintenance of a given language system. This line of argument is compatible with Zipf’s law of least effort (Herdina & Jessner, 2002:102). In DMM system stability is seen as a function of individual and psycholinguistic factors such as perceived communicative needs, motivational factors, self-esteem, perceived language competence, and anxiety amongst others (138; see also chapter 5). The relationship between these factors is described as one of complex interdependences and is seen as quite unpredictable due to the autodynamic and autocatalytic (i.e. relying on no external cause) behaviour of the system and its components (138). A central proposition advanced by the authors relates to the development of special skills or “dispositional properties” in the multilingual speaker which are not found in monolingual speakers. These properties are seen as promoting the learning and retention of additional languages or language systems (143). Chapter 1 and Part II of this book elaborate on the importance of these properties for multilingual learning. Herdina & Jessner (2002) describe DMM as “a psycholinguistic model which sees language change on an individual level as a function of time” stressing that “the focus is placed on the variability and dynamics of the individual speaker system”, an aspect of multilingual learning which had previously been largely ignored (2). In contrast to the monolingual speaker, the multilingual user is seen as possessing a unique set of skills comprising language learning skills, language management skills and language maintenance skills. The development of these skills in multilingual speakers is accompanied by qualitative changes in the system and results in a heightened metalinguistic awareness in multilingual users (129). In addition, it is assumed that multilinguals possess an enhanced
66
Developmental models of second and third language learning
multilingual monitor (EMM). The development of the EMM is thought to correlate with the number of language systems available to the speaker (64) and with the frequency of use of each of these systems (129). The EMM is part of the so-called M-factor which, according to the authors, exerts facilitative effects on the L3 acquisition process. An emergent property, the M-factor is defined as “a function of the interaction between more than one language system” (130) and as a “dispositional effect which will have a priming or catalytic effect in TLA” (129). The assumption is that the M-factor is linked to profound changes in language awareness and in language strategy development. Hobbs (2012) notes that the “multilingual literature has widely accepted the dynamic model of multilingualism (DMM) theory of Herdina & Jessner (2002), and relevance of the theory has been demonstrated” (207).
4.3 The Biotic model of Multilinguality Also advocating a dynamic perspective on multilingualism, Aronin & O’Laoire (2004) propose the Biotic Model and introduce the notion of ‘multilinguality’ to describe the “inner constructs of a single speaker” (16) and the “inherent, intrinsic characteristic” of the multilingual user (17). As specified by the authors the term multilinguality does not merely relate to speakers’ languages and their linguistic knowledge and behaviour (18) but also comprises their cognitive behaviour, their abilities, resources, metalinguistic awareness, their identity, personality traits, self-image and the plethora of socio-cultural influences they are exposed to (18–20). In accord with Herdina & Jessner (2002), the authors espouse a holistic and dynamic view of multilingual language development highlighting “the impenetrability, complexity and unity of multilingual phenomena” (Aronin & O`Laoire, 2004:18). The Biotic Model of Multilinguality focuses on “the essential dynamics of growth, change, fluctuation, input, absorption and decay” (19). The languages known to the multilingual speaker are seen as constituting “a kind of ecosystem or biosystem” (20) whose distinguishing features include complexity, interrelatedness, fluctuation, variation, inconsistency, multifunctionality, selfbalance, self-extension and non-replication (20). As in the Dynamic Model of Multilingualism, the single parts of the multilingual system are seen as interacting with one another in diverse and mostly unpredictable ways (20). This interaction is accompanied by fluctuations and alterations on all levels. In other words, multilingual speakers’ knowledge and their use of languages, the speaker’s cognitive resources, their emotional state and personality are all subject to modifications over time (22).
A dynamic model of expert-novice co-adaptation
67
4.4 A dynamic model of expert-novice co-adaptation during language learning and acquisition More recently van Geert, Steenbeek & van Dijk (2011) have proposed a highly relevant dynamic approach to language development and acquisition. Van Geert et al. (2011) conceive of language learning as a complex and dynamic process of self-organisation and emergence. As specified by the authors, the language learner travels “through a state space” (238; cf. Lowie & Verspoor, 2011:277) in which s/he occupies a given point according to her/his mastery of the L1 and L2. The learner’s position in this state space is not fixed but changes over time as a consequence of the complex interplay of many interacting components and variables. In this framework, the L2 learner’s knowledge is construed as a complex dynamic system (237). As expounded by Van Geert et al., complex systems consist of many interacting components whose dynamics automatically result in the emergence of novel properties (ibid.; see also Herdina & Jessner, 2002; Jessner, 2008a). According to the authors, two major characteristics of L2 learner systems are fluctuation and adaptation. Fluctuation is held to be greater in the early stages of L2 learning than in the later stages (244). It obtains in the form of interand intra- individual variability (ibid.). Adaptation occurs because the learner automatically adapts her/his language to that of an expert or teacher in her/his immediate environment. Adaptation is said to be dependent on the learner’s current level of knowledge (251). Importantly, adaptation leads to language growth in the novice. According to van Geert et al. (2011), language acquisition is dependent on socially embedded interaction (235). As regards the formal teaching of languages and the relationship between pupils and teachers, it is suggested that “students and teacher are interacting, autonomous, intentional agents and that learning and teaching processes obtain their form from the interaction between those intentional agents” (257). A complex form of causality is assumed to drive the learning process. In other words, as the participants’ (teachers’ and students’) behaviour, interests and emotions dynamically interact they mutually influence each other and determine the learning outcomes. In particular, it is intimated that “a previous action of the student has an influence on the subsequent action or reaction of the teacher, which influences the subsequent action of the student and so forth” (243).
5 Memory, monitoring and fluency In this chapter we examine concepts and issues related to memory capacity, monitoring and fluency. An overview of extant theories and models is given and differential theoretical perspectives are considered and discussed. Linguistic processing relies heavily on memory (Ellis & Sinclair, 1996:234; Fortkamp & Bergsleithner, 2007:40). The hypothesis is that “[a]ll input and output in language comprehension and production must transit through memory systems” (Towell & Dewaele, 2005:214). Working memory appears to play a particularly important role in linguistic processing. Working memory has been defined as “a specialised cognitive system for the short term storage and processing of information” (Temple, 2000:291; cf. Fortkamp & Bergsleithner, 2007:41). Levelt (1989) describes Working Memory as the “limited-capacity resource in conceptualizing and monitoring” (21) specifying that only a few concepts or bits of internal speech can be attended to at a time (ibid.). While this is clearly congruent with the widely held perception that cognitive resources are limited (Temple, 2000:291; see also Ellis & Larsen-Freeman, 2006:560, Fortkamp & Bergsleithner, 2007:41 and Green, 1998), one also needs to bear in mind that the notion of specialised, modular memory systems contradicts dynamic views of processing such as embraced here. With regards to individual differences in L2 performance or attainment, Segalowitz (1997) suggests that these differences be studied “as a function of practice, training, and general language learning history” (97). A somewhat different proposition comes from Dewaele (2002) who argues that inter- and intra-individual variation in L2 speaker achievement may stem from subjects’ differential working memory capacity and L2 proficiency (231). Dewaele refers to recent studies in the field of neurobiology which strongly suggest that individual differences in L2 performance may be due to “inter-individual variation in the networks mediating the L2” (230), in particular those mediating short term memory (229) and memory span (228). The underlying assumption is that verbal performance in L1 and L2 is, at least in part, a function of short term memory capacity and processing mode. While L1 speech processing is largely automatic, processing in L2 is typically controlled and effortful requiring constant monitoring and intervention. It is hypothesised that L2 processing places greater demands on short term memory (230) because it necessitates constant attention on the part of the speaker “at virtually every stage of processing” (231) to “store and coordinate fragments processed by the formulator, before the next stage of processing” (Temple, 1997
Memory, monitoring and fluency
69
in Dewaele, 2002:231). Dewaele (2002) speaks of ‘interventions’ when referring to monitoring or corrective measures. He explicates that . . . interventions [. . .] may involve particular ‘subtasks’. During the conceptualisation phase of Levelt’s production model, the L2 speaker might need to decide whether a particular register is appropriate. Another intervention might be needed during the formulation phase: is the word or structure retrieved from the mental lexicon socio-pragmatically appropriate? Some crucial diacritic features attached to the activated lemma might be missing, requiring attention. The L2 speaker would need to solve these morphological or syntactical problems as quickly as possible. The pronunciation of some difficult phonemes might again need conscious intervention during articulation (223).
The assumption is that when parallel processing operations prove too effortful, the system reverts to serial processing, and this can disrupt fluency of speech (Temple, 1997 in Dewaele, 2002:231). As noted by Levelt (1989), automatic processes can run in parallel, but controlled processing is typically serial and slow because it requires attentional resources (21; cf. Kecskes & Papp, 2000:75). More particularly, it is assumed that controlled processing “demands attentional resources, and one can attend to only a few things (the items in Working Memory) at a time” (Levelt, 1989:21). It has also been suggested that working memory efficiency may be affected by subjects’ capacity to ignore distractions of various kinds (Rosen & Engle, 1998:419, 432; cf. Bialystok, 2001, 2005). The implicit assumption is that if attention is split, resources will be depleted. L2 learners are commonly believed to rely more on declarative or explicit knowledge or memory than on procedural or implicit knowledge or memory (Dewaele, 2002:230; but see Lowie & Verspoor, 2011 for a DST perspective on declarative and procedural knowledge13). This is deemed to increase the load on the short term memory system (ibid.). In contrast, fluent speech, as found in proficient users of an L2, is said to rely on procedural knowledge (Towell & Dewaele, 2005:21; Levelt, 1989). It is widely assumed that procedural knowledge is called upon automatically and without conscious awareness. Procedural knowledge is held to run on its own resources, without sharing processing capacity with other processes (Levelt, 1989:20). In the ensuing sections the author enlarges upon a range of issues related to speech monitoring, automaticity and fluency.
13 From a DST view, the distinction between declarative and procedural knowledge is highly problematic (Lowie & Verspoor, 2011:278).
70
Memory, Monitoring And Fluency
5.1 Monitoring A central assumption in language processing research is that speakers possess a monitoring system which allows them to monitor or inspect their speech and take corrective measures if necessary. Vigliocco & Hartsuiker (2002) define selfmonitoring as “the process with which a speaker inspects his or her speech plan for well-formedness. These processes include detection of errors and other anomalies, implementing an interruption in speech, and planning a correction.” (448). Similarly, Hartsuiker & Kolk (2001) conceive of speech monitoring as the “process of inspecting one’s own speech and taking appropriate action when errors are made” (113). The assumption is that speakers continuously monitor their own speech. As specified by Levelt (1989), the speaker is “his own listener.” He has access to his internal and his overt speech and can compare “what was said or internally prepared to what was intended” (13). As soon as an error is detected, speech production is halted and corrective measures are taken. A new attempt at generating a given word or sentence is then undertaken. This is generally referred to as a “repair” (Hartsuiker & Kolk, 2001:113). In DMM the idea of monitoring is closely connected with the notion of metalinguistic awareness (Herdina & Jessner, 2002:63). The Dynamic Model of Multilingualism assumes an enhanced multilingual monitor (EEM) in multilingual systems. The DMM predicts that the enhanced monitor “is used by the multilingual speaker to watch and correct her or his language(s) in a multilingual context which increases the use of such a monitor” (Herdina & Jessner, 2002:64). The monitor is seen as expanding its functions whenever new languages are added to the system. In particular, the multilingual monitor is seen as serving the following functions: 1) fulfilling the common monitoring functions (i.e. reducing the number of performance errors, correcting misunderstandings, developing and applying conversational strategies); 2) drawing on common resources in the use of more than one language system and 3) keeping the systems apart by checking for possible transfer phenomena and eliminating them and thereby fulfilling a separator and cross-checker function (64).
5.2 Automaticity in L2/L3 acquisition Cognitive psycholinguistics conceives of the learner as having limited processing capacity and attentional resources (cf. Ellis & Larsen-Freeman, 2006:560). It is
Automaticity in L2/L3 acquisition
71
generally assumed that in order to “understand and react to new information, the learner uses controlled processes that make heavy requirements on attention and consequently demand cognitive effort and time. . .” (Sanz, 2000:35). According to Segalowitz & Hulstijn (2005) automaticity designates the “absence of attentional control in the execution of a cognitive activity” (371). It is ascribed a ballistic (that is unstoppable) and unconscious nature (372), and it is not seen as a “unitary construct” but as consisting of several related components such as rapidity and effortlessness. It is not assumed that all these components are necessarily involved concurrently when automatic processes are at work. Speakers may, for instance, display fluency and automaticity in word recognition but a given task may still require a substantial amount of effort and resources (372). Thus, while the criterion rapidity may be met, that of effortlessness may not. Segalowitz & Segalowitz (1993) emphasise that the notion ‘automatic’ can not simply be equated with ‘fast’ but must instead be understood in terms of some kind of restructuring and reorganisation of the processes underlying and subserving speech. This assumption implies that with increased practice and experience speech processing not only becomes faster but also undergoes cognitive change(s) (Segalowitz & Hulstijn, 2005:375; see also Herdina & Jessner, 2002; Bialystok, 2005). Following Bialystok & Sharwood Smith (1985), Ringbom (1987) suggests that amount of practice co-determines the degree to which processing operations become automatised (25). Ringbom notes that when L2 procedures are automatised, knowledge retrieval occurs fast and efficiently with the result that attention can be directed at other tasks (1987:25). Automatisation of retrieval procedures is thought to play a particularly important role during oral communication (ibid.). According to Segalowitz & Hulstijn (2005), “automaticity cannot really be talked about without also talking about attention” (379). Attention-based mechanisms and attention-focusing are seen as crucial factors and indispensable for language learning and automatic speech processing. While automatic processes relate to fluency, rapidity and accuracy, attention-focusing is said to be called upon when learners focus their attention on a particular linguistic feature or task (378; cf. Schmidt, 2001). In fluent speech the automatic and rapid succession of uttered words can (at least to some extent) be said to be dependent on selective attention. Selective attention enables the speaker to engage in the communicative act even under noisy and distracting conditions (Segalowitz & Hulstijn, 2005:378; see also Bialystok, 2001).
72
Memory, Monitoring And Fluency
Segalowitz & Hulstijin (2005) further intimate that noticing may be a cognitive prerequisite for attaining fluency in the L2 (see also Schmidt, 199014, 1993, 2001 and Ellis & Larsen-Freeman, 2006:569). The authors also discuss the possibility that successful L2 acquisition may be correlated with an increased availability of pre-assembled linguistic units, so-called chunks, as these are deemed to reduce the amount of effortful online computations (Segalowitz & Hulstijn, 2005:385). The availability of such pre-assembled linguistic units is, amongst other things, predicted to depend on frequency and recency of L2/L3 use. More recently, Green (2007) has stated that cognitive systems need “energy to work” (371). If such energy is not available, or is insufficient, “control will be imperfect” (ibid.). In consequence, if resources are not replenished at an appropriate pace, control is predicted to be impaired (375) and this is bound to affect fluency.
5.3 L2/L3 fluency In very broad terms, L2/L3 fluency denotes a speaker’s ability to adequately communicate an idea in a second language. Speakers are generally perceived as fluent when they use a large vocabulary, when they make few grammatical errors and when they speak with little or no foreign accent (Segalowitz, 2010:4). Attainment of fluency is said to involve the acquisition of memorised sequences of language (Ellis & Sinclair, 1996:234, 246). Fillmore (1979) proposed four aspects of fluency including a) the ability to engage in conversation and speak without having to pause all the time, b) the ability to generate meaningful utterances without drawing on excessive filler items such as ‘you know’, c) the ability to adapt to different socio-pragmatic contexts, and d) the ability to use language in creative ways, such as in puns, metaphors, etc. (in Segalowitz, 2010:4; see also Fillmore, 1991: 52–59; cf. Nation, 1990:377). The latter clearly overlaps with notions of metalinguistic awareness and skills (cf. MAT 3: Pinto et al., 1999). To the above outlined characteristics of fluency scholars have added speech features such as speech rate, hesitation and pause phenomena, density of clause usage, and combinations of these, amongst others (5). According to Segalowitz (2010), L2 fluency is an observable characteristic of real-time speech behaviour. This behaviour [. . .] reflects the cognitive and emotional states of the speaker at the time of speaking. Fluency [. . .] reflects the developmental history and the current state of the speaker. (6) 14 Schmidt (1990) holds that the subjective experience of ‘noticing’ is an essential prerequisite to convert input into intake (Schmidt, 1993:209).
L2/L3 fluency
73
Dysfluencies can obtain at various critical moments during speech processing (Segalowitz, 2010:9). Segalowitz (2010) has recently proposed two psycholinguistic models: a model of the L2 speaker which he adapted from Levelt (1999) and De Bot (1992), and a fluency model (see Fig. 7 below). The fluency model comprises four components which impact the L2 speaker’s fluency to a greater or lesser extent. The four components include the speaker’s perceptual and cognitive systems, their motivation, the communicative or social context and the speaker’s perceptual and cognitive experiences (which are seen as the result of the sociolinguistic context). The model of the L2 speaker identifies various points of possible vulnerability with regard to speakers’ fluency in L2 speech performance. So-called { f } or fluency vulnerability points in the L2 processing system mark the points at which fluency can be compromised due to underlying processing difficulties.
Figure 7: Fluency Model (based on Segalowitz, 2010:21)
74
Memory, Monitoring And Fluency
Segalowitz (2010) argues that the efficiency of the speaker’s cognitive and perceptual system affects her/his fluency in L2 and that motivation impacts the individual’s determination to attain fluency. In the same way he assumes that the sociocultural context exerts its influence on the speaker’s linguistic productions and language behaviour (see also Fillmore, 1991:63). Additionally, as suggested by Segalowitz (2010), speakers form a set of perceptual and cognitive experiences and this impacts their perceptual and cognitive systems (see Fig. 7). All of these variables are seen as interacting in dynamic and complex ways. The influence between social context and cognitive/perceptual systems is held to be bidirectional (23). A clear weakness of the L2 speaker model and the fluency model relates to the fact that they assume hierarchical and modular processing. As already pointed out, modular accounts of language processing are incompatible with a DST perspective because they cannot accomodate “non-static scenarios, such as linguistic change, language development and multilingualism” (Schmid & Lowie, 2011:2).
5.4 Psycholinguistic factors contributing to fluency in L2/L3 L2/L3 speech is generally more hesitant, contains more pauses, errors and repairs and is produced at a slower rate than L1 speech (Temple, 2000:288). There appears to be general consensus that fluency in L2/L3 is (to a certain extent) determined by factors such as aptitude, cognitive style, motivation, but also by the learner’s linguistic and social environment, their social status, their educational background, age, exposure, or use of strategies (cf. Gibson & Hufeisen, 2011:75). This section focuses on the psycholinguistic variables impinging on L2/L3 fluency. We know that there is considerable inter-individual variation in speakers/ learners of an L2/L3. There is also a certain amount of intra-individual variation since L2/L3 users’ performance may change from one situation to the next (Dewaele, 2002:221; cf. van Geert et al. 2011:260–61). Moreover, it has been noted that some individuals seem to be forming “more effective mental links than others” (Cook, 2002:219). This raises the question as to what causes these links to be more effective in some learners relative to others. Clearly, the speaker’s psycho-social identity can be seen as central to her/his communicative success and performance in L2/L3. There are indications that the speaker’s personality, their self-confidence, their attitudes and their perceived competence all impact L2 processing and fluency (Dewaele, 2002:222). Extravert individuals have been found to display greater stress resistance thus performing better under very
Psycholinguistic factors contributing to fluency in L2/L3
75
stressful conditions or pressure (233; cf. Hoffmann, 2001). Introvert subjects, by contrast, tend to be negatively affected by stress showing impaired performance due to longer retrieval times (233), delayed processing operations (234), and loss of automaticity (235). Dewaele (2002) relates to recent studies in the field of neurophysics and biophysics which link these behaviours to biochemical processes in the two populations. These studies intimate that extraverts are typically under-aroused, while introverts are commonly over-aroused (232). Dewaele explains that a moderate level of cortical arousal is ideal and that introverts, due to their already high level of cortical arousal, are less stressresistant because they will soon exceed an optimal level of arousal, getting even more aroused. Extroverts, by contrast, are able to compensate for their lower cortical arousal levels and can therefore deal better with stressful situations than introverts (233). Feelings of anxiety, worry, failure, low self-esteem and extreme emotionality appear to inhibit and debilitate performance in L2 (236–240; cf. Bleyhl, 2005:59– 60). As reported in Eysenck (1992), it is “well established that anxious individuals have reduced processing capacity” (21). It has been intimated that some forms of anxiety may be more permanent or stable than others. However, no matter what form anxiety takes, the argument goes that it will interfere with information processing (Dewaele, 2002:238) and reduce working memory capacity (239). There are strong indications that excessive self-concern and selfrelated cognition, i.e., an exaggerated preoccupation with one’s fears and insecurities can divert valuable cognitive resources from verbal tasks (222, 236, 241). Segalowitz (1997) suggests that a speaker’s mastery of the L2 is a function of how effectively her/his perceptual and cognitive processes can meet the demands imposed by the communicative situation. These demands are said to comprise the ability to recollect “linguistic information under difficult conditions and to have a high degree of attentional flexibility for handling the many different stimulus dimensions that carry significant communicative information” (107). Segalowitz intimates that L2 attainment may be affected by how “one’s perceptual and cognitive resources are organized” adding, however, that “everyone has the potential to attain a high level of mastery” (108) and that L2 competence can not be explained by way of a single determining factor or even some innate predisposition (107). Segalowitz warns that “[i]t would be incorrect . . . to expect that a particular change in one component15 would lead 15 As outlined earlier, the fluency model comprises four components including the speaker’s perceptual and cognitive systems, their motivation, the communicative or social context and the speaker’s perceptual and cognitive experiences (which are seen as the result of the sociolinguistic context).
76
Memory, Monitoring And Fluency
invariably to a simple proportional change in some aspect of fluency for all speakers in all circumstances” (2010:24). Endorsing a dynamic perspective of fluency and language development the author posits the need to take into account the whole system and not just single parts (ibid.). In the following chapter selected aspects of child language and child multilingualism are considered. The chapter acts as an introduction to theories and conceptions which are of relevance to the present investigation and which will be developed further in Part II of this book where the empirical findings of the study are presented and analysed, and where the author expands in more detail on morpho-syntactic and phonological awareness, code-mixing, and crosslinguistic interaction and sensitivity in young multilingual learners.
6 Multiple language acquisition in children The present book with its specific focus on early multilingual learning and development would not be complete without a description and analysis of how small children acquire first and second languages. The ensuing sections therefore review some of the recent literature on child monolingual, bilingual and trilingual acquisition. A variegated range of perspectives, informed by different theoretical backgrounds, is sketched and contrasted. The chapter lays the ground for the second part of the book where the focus is on young learners’ experiences in high-immersion multilingual programmes.
6.1 Cognitive development in children As young children learn their native language, they are also learning about the world (Ellis, 2008:378). Bialystok (2002) identifies the moment when “mental representations of knowledge become more explicit and more structured” as a crucial moment in the child’s cognitive development (153). The author argues that it is then that children are increasingly able to “comprehend not only unrelated facts, but also the relationships between various concepts and ideas” (ibid). While “[i]n the early stages of development, knowledge is represented as implicit routines, embedded in concrete contexts, and unrelated to other representations that share some abstract similarity” the child’s mind changes and “[as] mental representations become analysed, knowledge can be organised around abstract categories and details retrieved independently of their contexts” (ibid). These developmental changes in the mind are seen as fundamental for the emergence of thought in the young child (153). With regard to language representations Bialystok (1991b) suggests that in small children these are initially organised along semantic and pragmatic lines (116). As the child gains more practice and becomes more proficient language representations become increasingly stronger and more structured. More specifically, the child’s language representations in the brain develop as they pass through three stages. These stages are conceived as a “continuum, not discrete stages” (118). The assumption is that in the first stage or phase children develop conceptual representations. At this point the child’s attention is focused on her/ his immediate contextual environment. In the process of building up a preliminary knowledge base the child relies on contextual and situational clues since it is not yet in a position to comprehend language. In the second phase, the child develops so-called formal representations of language which include some explicit although still very rudimentary knowledge of language. The child is at this stage not capable of carrying out abstract operations. Only at the third
78
Multiple language acquisition in children
stage, the so-called symbolic stage does the child form abstract representations which can be accessed and retrieved without any concrete context. Bialystok (1993) notes that the child is now able to make abstractions and grasp the arbitrary nature of linguistic signs or symbols (222–224).
6.2 Theories of Child Language acquisition Theoretical positions on how language is acquired and processed remain deeply divided. This divide between the various diverging conceptions of language and language development is perhaps best illustrated by the stark opposition between structural theories and functional theories of language (Bialystok, 2002:147; Karmiloff & Karmiloff-Smith, 2001:104). While functional linguistics “holds that language is an instrument for human communication that can best be understood by an analysis of the interaction of its various levels of organisation – phonology, (morpho-)syntax, semantics – and its contexts of use” (Perdue, 2002:124), structural linguistics posits an inborn Universal Grammar which underlies all human languages and allows the child to acquire the native language by simply setting a fixed set of parameters according to the specific characteristics of the language exposed to (in Karmiloff & Karmiloff-Smith, 2001:106). This perspective entails that the environment is seen as playing a negligible role, functioning merely as a sort of trigger for the setting of parameters because the child is already born with a set of abstract linguistic structures. Linguistic structures are, according to this viewpoint, not learned by exposure to a given language but are built into the human brain and are “there from the start” (see Karmiloff & Karmiloff-Smith, 2001:104). Chomsky (1960) views language acquisition as the product of rule formation (Larsen-Freeman & Long, 1991:57). According to Chomsky (1986), humans are endowed with a “distinct ‘language faculty’” (4), a “language acquisition device” which is seen as an “innate component of the human mind” (3). The underlying assumption is that humans are equipped with a set of invariant principles common to all languages (Karmiloff & KarmiloffSmith, 2001:106) and a fixed set of parameters which differ from language to language (108–109). As noted by Karmiloff & Karmiloff-Smith (2001), Chomskyan theory limits itself to investigating “what is learned as the child acquires grammar”, while failing to explain “how grammar structures are learned” (111). In contrast, functional linguistics takes an entirely different approach. Functional linguistics assumes that children learn language from the input they receive (Tomasello & Abbot-Smith, 2002:208). This is in line with Karmiloff & Karmiloff-Smith (2001) who conceive of language knowledge as “the complex product of the interaction between some initial, domain-relevant (not domain-
Early bilingual acquisition: One system or two?
79
specific) predispositions and the rich structure of the linguistic input” (7). Instead of assuming a pre-wired child equipped with an innate abstract linguistic structure, Karmiloff & Karmiloff-Smith (2001) hold that very early on humans possess a sufficiently developed hearing system. This hearing system allows the fetus to process “some of the sounds that filter through the amniotic liquid” (1). At 6 months of gestation, the fetus pays attention to linguistic sounds and to the sound qualities of the mother’s voice. The fetus also becomes sensitive to the intonation and rhythm that characterise the mother’s speech productions (2). These early experiences with the sounds of noises and language are said to be critical because they prepare the newborn for the language learning task. According to Karmiloff & Karmiloff-Smith (2001) these early “intrauterine experiences prime the newborn for linguistic input and can therefore be viewed as playing an important role in the overall process of language development” (2). This proposition stands firm as a valid alternative to the structuralist assumption that an innate universal grammar guides the infant in the acquisition of her/his native language. Karmiloff & Karmiloff-Smith (2001) conclude that language acquistion and development have to be seen as dynamic processes which can not satisfactorily be accounted for by rigidly adhering to either domain-specific or domaingeneral theories of language. Rather, these authors intimate that both domainspecific and domain-general mechanisms must be accepted as playing a role in the acquisition of language (126). In a similar line of argument, Bialystok (2002) states that it “is simplistic to assume that either of these approaches [formal vs functional approaches; note added] alone could adequately account for the monumental task of mastering an entire language” (149).
6.3 Early bilingual acquisition: One system or two? A frequently articulated concern relates to the misconception that children growing up with two or more languages from an early age “might be confused linguistically, cognitively, emotionally, and possibly even morally” (Meisel, 2004:91; Genesee & Nicoladis, 2007:1). Similar doubts are often raised with regards to bior multilingual education for it is still widely assumed that “monolingualism represents the natural or normal case of language development” (1). This attitude, so Meisel, contributes to the general perception that “deviating from this norm implies risks which had better be avoided” (1). Meisel dismisses all above mentioned concerns as unwarranted stating that “[c]learly, the human language faculty has an endowment for multilingualism” (112).
80
Multiple language acquisition in children
In the recent past, linguists have shown an increased interest in investigating language differentiation in young bilingual children. Their main purport has been to determine whether very young bilinguals are capable of differentiating their language in a systematic way or not (Poulin-Dubois & Goodz, 2001:104). Inextricably intertwined with the issue of language differentiation is the question of whether the child’s linguistic systems develop independently of each other or whether there is initially a single system which contains items from both languages. This latter position has famously been defended by Volterra & Taeschner (1978). These authors proposed that, in the early stages, children possess a single unified system in which the languages are mixed. Then, the bilingual child gradually develops two differentiated lexicons which initially, however, draw on the same syntactic rules. Only later, so the argument, do bilingual children develop separate systems for both, lexicon and syntax (Volterra & Taeschner, 1978[2007]:318; Barnes, 2006:11). The one- or single-system hypothesis, as proposed by Volterra & Taeschner (1978) is contrasted by the so-called “dual system hypothesis” (Meisel, 2004:98) or “Independent Development Hypothesis” (Bergman, 1976 in Barnes, 2006:14) which entails the supposition that the child develops two separate systems for the two languages. The implication is that the child is perfectly capable of differentiating the two languages and of selecting the appropriate language according to the situational requirements. As noted by Barnes (2006), the prevailing view today is that of two separately developing systems (see also Genesee & Cenoz, 2001:3; Paradis & Genesee, 1996[2010]:487 and Genesee & Nicoladis, 2007). A major preoccupation for many parents and educators relates to the rate and onset of language development in children who are raised with two or more languages. More particularly, parents and educators are worried about whether young bilingual children “will show the same ages of onset for the major language milestones and whether their rate of language development will be comparable to that of monolingual children” (Poulin-Dubois & Goodz, 2001:95; see also Genesee & Nicoladis, 2007). Although research on these issues has been scarce, there are strong indications that language development in children growing up with more than one language is not delayed or compromised by the addition of a second language (Poulin-Dubois & Goodz, 2001:95; cf. Genesee & Nicoladis, 2007). It has, however, been pointed out that the pace at which a child’s languages develop may not be the same for both languages (Paradis & Genesee, 1996 in Barnes, 2006:15; Paradis & Genesee, 1996[2010]:505). This said, these initial disparities in language proficiency seem to be of minor consequence since it is to be expected that the child will catch up in time and will grow up to command two language systems, provided that s/he is suffi-
Early bilingual acquisition: One system or two?
81
ciently motivated and receives sufficient input and support. Both, quality and quantity of input (particularly parental input) have been found to impact the child’s linguistic behaviour especially in terms of code-switching and mixing habits (Barnes, 2006:14, 17, 20, 21). Genesee (2002) notes that “[b]ilingual children clearly demonstrate different language behaviours in comparison to monolingual children; most noticeably, they mix elements from two languages in the same utterance or conversation. This has often been interpreted as a sign of incompetence or deficit. . .” (Genesse, 2002:193; see also Genesee & Nicoladis, 2007). Such interpretations have commonly been linked to the perception that the cognitive resources available to bilingual children are split between two languages. Children’s mixing habits have typically been construed as indicative of the child’s failure to differentiate the two languages and as evidence for the existence of a unitary language system (Genesee, 1989:320; Poulin-Dubois & Goodz, 2001:96). Crucially, such a unitary system was held to consist of undifferentiated phonological, lexical and syntactic subsystems (323). The implicit assumption was that the young child (under the age of 3) is not really bilingual yet since she/he is unable to differentiate between the two linguistic codes. Only after the age of 3, so the proponents of this theory, when mixing diminishes as a direct consequence of language separation, can the child become truly bilingual (see Genesee, 2002:171). Recently, scholars have contested this theory on the grounds that no sufficiently convincing evidence has as yet been provided to justify the unitary system hypothesis (Genesee, 2002:173). Genesee (1989) particularly objects to the assumption that the observed decline in mixing with increasing age is seen as implying that the child has initially got one fused linguistic system which eventually develops into two separate systems ([2007]:324; Paradis & Genesee, 1996[2010]: 486). Genesee argues that mixing decreases not because the supposedly unitary system is evolving into two separte systems but because the child is building a more refined and resourceful language system which does no longer require it to draw on the other language for support (ibid; see also Comeau & Genesee, 2001:232). Genesee (1989) maintains that if the unitary system hypothesis were right this would entail that all “bilingual children use items from both languages indiscriminately in all contexts of communication” and that there is “no differential distribution of items from the two languages as a function of the predominant language being used in different contexts” ([2007]:323). This leads Genesee to dismiss the above conjectures as untenable. Hoffmann & Stavans (2007) comply with Genesee’s proposition stating that “[w]e go along with the separate language acquisition hypothesis that has been proposed (. . .), and widely accepted in the literature on bilingual language acquisition and suggest
82
Multiple language acquisition in children
that it extends to simultaneous trilingual acquisition as well” (57). In opposition to the one-system hypothesis, Genesee (2002) affirms that bilingual child language exhibits a certain systematicity and functional separation of linguistic elements from the two codes (192). This, so the author, renders the unitary system theory untenable and redundant (1978 [2007]:328). Genesee further argues that it is inappropriate to talk of linguistic confusion at this early stage of bilingual language development and that the “language that bilingual children acquire is not deviant” (2002:192). Instead, so Genesee, the “linguistic competence that underlies [. . .] [children’s] performance in both of their languages reveals the same underlying linguistic competence as that of monolinguals in most significant respects” (ibid; see also Meisel, 2004:96). Most linguists today concur that young bilingual children are able to utilise “their developing languages differentially and appropriately from the one-word stage onward, and certainly from the stage when there is evidence of syntax in their spoken language” (Genesee, 2002:173). Meisel (2001:15–16 in Barnes, 2006:16), for instance, refutes the one-system hypothesis positing that bilingual children develop autonomous linguistic systems from an early age and that “early separation of systems is not the exception but the rule in the simultaneous acquisition of various languages” (Meisel, 2004:98). The author (2004) argues that early mixes, as found in young simultaneous bilinguals, do not provide support for the fusion hypothesis “since they do not involve combinations of structural properties of the two languages” (97). Hence, language mixing is not seen as conflicting with the view of language separation but rather as indicative of an enhanced sensitivity to the linguistic behaviour of the environment (98). This complies with Genesee et al. (1995) who found that young bilingual children are sensitive to the linguistic proficiency of their interlocutor and that they can adjust their language accordingly (Comeau & Genesee, 2001:233).
6.4 The acquisition of morphosyntax in bi- and multilingual children This section investigates the acquisition of grammar in bi- and multilingual children. The two theoretical positions (Generative – UG-based – Grammar and Cognitive-Functional – usage-based – Linguistics) introduced in section 6.2 are here examined in more detail. As indicated, Cognitive-Functional Linguistics posits that children learn the words, rules and constructions of a language from the input they receive, i.e., from their linguistic environment (Tomasello & Abbot-Smith, 2002:208). Generative Grammar, by contrast, postulates that
The acquisition of morphosyntax in bi- and multilingual children
83
children possess an innate set of grammar rules. Generativist theory holds that the acquisition of language entails parameter setting and the linking up of all the words and constructions of a given language with this abstract inborn universal grammar (207). According to Chomsky, the setting of parameters permits humans to take a shortcut to grammar because the setting of parameters reduces the amount of exposure deemed necessary for the successful acquisition of language (in Karmiloff & Karmiloff-Smith, 2001:109). The assumption is that, once a parameter is set, it automatically activates other related constraints operating on that particular language (ibid.). Significantly, Chomsky does not see language acquisition as “something in which the child is actively engaged” but merely as “something that happens to the child when he is placed in an appropriate setting” (139). In contrast, Slobin (1973) maintains that language acquisition presupposes an “active child mind” (247). He suggests that the child is guided by general cognitive-perceptual strategies as it actively attempts to understand speech (246). According to Slobin, [i]nner linguistic structures change with age as computation and storage space increase, as increasing understanding of linguistic intentions leads the child into realms of new formal complexity, and as internal structures are interrelated and re-organized in accordance with general principles of cognitive organization (247)
This is consistent with Tomasello (2003) who posits that children slowly construct grammar (4). Tomasello’s construction-based approach implies that humans possess a mixture of language-relevant16 and more general mechanisms which allows them to acquire the grammar of their native tongue (ibid.). The child is seen as learning from concrete examples of linguistic input and not on the basis of any abstract linguistic structures or principles (140). Hence, children’s linguistic productions are seen as attesting to their knowledge of specific words or syntactic structures. They do not, as postulated by structuralists, reflect the supposedly underlying abstract categories and rules (ibid.). According to the functionalist position, children’s productions do not consist of “ready-made, abstract linguistic structures” but instead reflect children’s ability to rapidly process language and to recognise and apply recurring (slot-and-frame) patterns. This ability permits them to gradually build up a grammar of more complex and sophisticated constructions. Slobin (1971, 1973, 1982) proposes a range of linguistic operating principles which aid the child in the process of building up a basic grammar (Slobin,
16 The notion of ‘language-relevant’ must not be taken to signify ‘language-specific’.
84
Multiple language acquisition in children
1985:1159). These principles ensure that input from the child’s socio-linguistic environment is stored and subsequently organised in the child’s mind. These stored representations can then be drawn upon and retrieved at later stages of development (1160). Contrary to what is postulated by Chomskyan theory, Slobin’s principles are not inborn and genetically specified. In other words, Slobin does not presume an innate universal grammar (Slobin, 1985:1158). Though conceding that the infant’s mind may be predetermined to process language, Slobin nevertheless posits that this “is not to say that the grammatical system itself is given as innate knowledge, but that the child has innate means of processing information and forming internal structures” (Slobin, 1971 in Aitchison, 2008:141). Similarly, Tomasello & Abbot-Smith (2002) maintain that children do not simply activate an internal grammar but that they learn the syntax and structures of language from their linguistic environment (209). Tomasello (1995) expounds that “. . . the fact that all cultures have a language does not mean that the basic structure of language is innate.” (137) These authors’ positions stand in stark constrast to the strong generativist credo that “[o]nce a child is able to parse an utterance such as ‘Close the door!’, he will be able to infer from the fact that the verb ‘close’ in English precedes its complement ‘the door’, that all verbs in English precede their complements” (Radford, 1990 in Tomasello & AbbotSmith, 2002:209). The view is taken here that it is highly unlikely that young children are equipped with such abstract linguistic categories. Rather, as noted by Tomasello (2000), young children do not appear to be very productive with their early language and this can be taken to indicate that humans are not born with abstract linguistic categories (Tomasello & Abbot-Smith, 2002:207). MacWhinney (1975) and Tomasello (2000) both argue that the acquisition of syntax involves the learning and use of item-based constructions (MacWhinney, 2005:53). These authors maintain that the child first learns a given item-based construction and that by virtue of increased exposure to language the child will then learn to extract patterns from the input and will eventually be able to produce increasingly more complex constructions (MacWhinney, 2005:53). It is generally acknowledged that simpler items are mastered earlier than complex ones and that the salience of an item determines how fast the item is acquired (Hoffmann, 1991:70). The frequency with which a particular linguistic item is encountered in the input is also predicted to impinge on the pace of acquisition (Ellis & Larsen-Freeman, 2009:95). In bi- and multilingual children (psycho)typology is thought to play an important role when it comes to learning certain linguistic structures. If the child’s two or more languages draw on similar constructions in order to express the same conceptual idea then this may facilitate and speed up the acquisition
The acquisition of morphosyntax in bi- and multilingual children
85
process17. If, by contrast, the languages differ in the way they verbalise a conceptual idea and if their syntactic structures diverge in terms of their perceived or relative complexity or similarity, then the child may initially resist the acquisition of a given structure, particularly if it perceives this latter to be too complex (Hoffmann, 1991:70). With regards to interference and transfer at the level of syntax, Hoffmann (1991) notes that some items such as prepositions are “particularly susceptible to interference” (98). This, according to the author, may be due to the fact that prepositions serve different functions in different languages. Similarly, Hoffmann & Stavans (2007) found that young trilingual children mix prepositions with non-target utterances18, as in “Will you play mit me?”, “Nina pinta mit lapiz”, and “Nina Auto fahren con Mami” (57, italics in the original). Hoffmann & Stavans (2007) intimate that the child knows that a given slot needs to be filled with a given preposition, but that “its language-specific lexical shape has not yet become fully contextualized”, a phenomenon the authors propose to term ‘permeability of languages’ (57). The three-stage model presented earlier (Volterra & Taeschner, 1978) assumes that bilingual children’s grammars develop in three successive steps. Initially, the child is seen as having one unitary language system where lexical items from both languages are stored. Then, gradually, the child is said to develop two separate lexicons but syntactic structures are still represented in one single system. Finally, after having gained sufficient mastery, it is assumed that the child’s languages come to form two separate systems ([2007]:304). As said above, the one-system hypothesis has come under attack from various sides. Meisel (2004:97) dismisses the unitary system hypothesis arguing that bilingual children utilise language-specific syntactic constructions as soon as they start using grammar structures. In the same way, De Houwer (2005:33) rejects the single-system theory on the grounds that such a theory would imply that bilingual children are unable to keep their two languages apart, when, in fact, it has been found that young bilingual children’s language productions are mostly unilingual (35) containing only very few mixed utterances. A similar point is made by Lanza (1997) who found that her two-year-old bilingual subjects were clearly able to differentiate their languages (319 in Barnes, 2006:17). De Houwer (1990) argues that “the morphosyntactic development of the one language does not have any fundamental effect on the morphosyntactic development of the other” (De Houwer, 2005:33). According to De Houwer (2005), 17 This said, it is also thinkable that (perceived) similarity at the lexical or structural level may result in more interference and an increment in the occurance of errors. 18 But see Myers-Scotton, 2006:235 for a different proposition.
86
Multiple language acquisition in children
no “systematic morphosyntactic influence from one language on the other” has ever been evidenced (43). In consequence, no negative cross-language influences on the child’s developing language systems are to be expected (40). The author suggests that the morphosyntactic systems of children who are exposed to two languages from birth develop independently, regardless of the “input condition” (33) (see Lanza, 2001 for a somewhat different view). Errors occuring at this early stage in the child’s linguistic development have been accounted for in different ways, depending on researchers’ theoretical background and affiliation. Cognitive-Functional Linguistics, postulating a usagebased learning process, posits that errors occur because the young child experiments with morphological items and grammatical structures. No inborn universal grammar is assumed to mediate the acquisition process. According to Tomasello & Abbot-Smith (2002), children do not possess abstract linguistic categories and schemas which help them generate infinite numbers of grammatical sentences, as is postulated by researchers in the UG-oriented camp (207). Rather, children are predicted to acquire rule-based grammatical constructions by learning them from the sociolinguistic environment. As the young child is exposed to more sophisticated language it will eventually be able to deduce and apply the appropriate linguistic forms and structures. In contrast, Generativist Theory posits that the child makes errors because it has not yet fully constructed its internal grammar. In other words, the parameters and principles have not been set yet. The child is still learning the system of rules underlying the speech patterns of the respective language. Hence, UG can not be fully exploited. Chomsky’s propositions of an innate grammar and language device are of course irreconcilable with dynamic and holistic approaches to language learning such as proposed by the DMM.
6.5 Code-mixing in bilingual and trilingual children It has been found typical of most bi- and multilinguals to mix elements from various languages (cf. Lüdi, 2003:174). Poulisse (1997) posits that it is characteristic of L2 speech productions to carry traces of L1, particularly when speakers’ level of proficiency in the L2 is low (206). A distinction is commonly made between code-mixing and code-switching. Code-mixing has been described as “the conflation of various linguistic units [. . .] within the sentence” (Hoffmann & Stavans, 2007:57). Code-switching, by contrast, “involves mixing linguistic elements [. . .] across sentences” (ibid; italics in the original). Switching and mixing are often seen as indicative of deviant language behaviour but one needs to be cautious when assessing the value of mixed
Code-mixing in bilingual and trilingual children
87
productions. According to Martin (2003), it is high time to “dispel the myth that such bilingual practices are a substandard form of communication” (84). Herdina & Jessner (2002) similarly state that Whilst there is in some instances of one language system interfering with, and possibly its mere existence restricting the accessability of, the other language system, the interpretation of crosslinguistic effects as primarily negative in terms of reducing the respective language achievements of the multilingual speaker represents not only a very one-sided view of the effects to be expected but also constitutes a misunderstanding of the nature of the multilingual’s language system (61).
In the same line of argument Genesee (2002) notes that code-mixing in bilinguals is frequently misinterpreted as a language deficit when instead it ought to be acknowledged as a communicative strategy with clearly defined functional and formal properties (192–3; cf. Hoffmann & Stavans, 2007:55 and Edwards & Deweale, 2007:221). A similar point is made by Cook (2002) who describes codeswitching as “a natural ability in L2 users and not as a sign of confusion or incompetence”, adding that this also holds for small children (18). Poplack (1980) argues that there is “evidence that code-switching is a verbal skill requiring a large degree of linguistic competence in more than one language, rather than a defect arising from insufficient knowledge of one or the other” ([2007]:240). In unison with a growing number of scholars, Poplack postulates that code-switches are not indicative of deviant behaviour but that they are instead “a suggestive indicator of degree of bilingual competence” (240; cf. Garcia, 2009:50). Before turning to discuss code-mixing and switching in children we first examine how, according to current theorising, code-switching/ mixing is accomplished online in the first place. The models presented here below are based on linear and modular assumptions of linguistic processing. To this author’s knowledge, no comprehensive dynamic account of switching and/or mixing behaviour is available at this stage. One of the first accounts of code-switching was presented by Poulisse & Bongaerts (1994) and Poulisse (1999). These authors argue that bilingual speakers are very skilled at separating and mixing their languages according to their communicative needs. The model of lexical access devised by Poulisse & Bongaerts (1994) assumes that the preverbal message in the conceptualiser contains the specification [+L1] or [+L2]. If the speaker wishes to use a particular L2 word during an otherwise monolingual L1 discourse situation, the language cue or specification in the preverbal message causes the L2 lemma to be activated (Poulisse, 1997:217; 1999:62). If, by contrast, an L2 item is accidentally selected instead of an L1 lemma, the switch is said to be unintentional. In this case, Poulisse & Bongaerts (1994) speak of a slip of the tongue (Poulisse, 1997:217). According to
88
Multiple language acquisition in children
the authors, lexical intrusions into L2 are more frequent than in the other direction. This is explained in terms of the higher frequency of L1 items. Frequency and exposure are thought of as the major determining factors in non-intentional code-switching. An important contribution to this field also comes from Myers-Scotton (see also chapter 3). Myers-Scotton (2005) defines bilingual speech as “utterances that include surface-level morphemes from two or more language varieties” and code-switching as one type of bilingual speech that “comes in several forms” (327). The author specifies that bilingual speech need not necessarily contain surface-level words from two languages, but may comprise surface-level words from only one language and abstract rules (but not actually words) from the other (Myers-Scotton, 2006:234). According to Myers-Scotton, bilingual speakers switch languages whenever they perceive one language to be better suited to convey a particular idea or intention than the response language (MyersScotton, 2005:327), or else, when the use of a given non-target word or phrase allows for greater “pragmatic impact” (Myers-Scotton, 2006:266). As explicated by the author, “[t]here is nothing random about how [. . .] morphemes are organized in a [bilingual] clause” (249) since code-switches build upon “a principled basis” (2006:249). Hence the claim that code switches are not indicative of poor language competence or “broken or bad language” (Myers-Scotton, 2006:249; italics added). The languages implicated in CS do not (always) contribute equally to the switched utterance (235, see also Muysken, 1997[2010]:70). Instead, there is an “asymmetry between what the Matrix Language and the Embedded Language can supply to the bilingual clause” (Myers-Scotton, 2005:331; MyersScotton, 2006:235,243). The assumption is that certain grammatical elements or morphemes can only be supplied by the Matrix Language which is also the language that provides the morphosyntactic frame (Myers-Scotton, 2005:329). The Matrix Language thus plays a more important role in relation to the so-called Embedded Language which supplies various types of embeddings or islands (ibid.). The Matrix Language is typically the speaker’s L1, although this does not always have to be the case. While both languages are switched on at all times the Matrix Language is the more activated language (ibid.). However, when embeddings are produced and integrated into the ML frame, the Matrix Language’s level of activation is momentarily decreased (ibid.). Crucially, embeddings must obey the structural norms or requirements of the Matrix Language (ibid.), that is, they must not violate the well-formedness of the utterance to be generated. They must be compatible with the predicate-argument structure of the sentence and with its morphological realisation patterns (331). Myers-Scotton (2006) explains wellformedness constraints in terms of grammatical rules which indicate how the
Code-mixing in bilingual and trilingual children
89
various elements in a given sentence can be legally combined (238). Nouns generally meet these criteria easily as they do not (in contrast to most verbs) assign but only receive thematic roles. This explains why non-target language nouns tend to agree easily with the morphological and syntactic requirements of the Matrix Language frame. It also explains the high occurance of nouns in code-swiched productions (330; see chapter 20 for pertinent qualitative findings yielded by the present investigation). As specified by Myers-Scotton (2006), almost 100% of all borrowed words are content words (337) with nouns the most frequently borrowed lexical category (226; see also Hoffmann & Stavans, 2007:63). Verbs, by contrast, (especially when inflected) have been found to occur less frequently in code-switched utterances. This is mainly due to the fact that verbs are involved in complex operations, such as assigning roles in the sentence, establishing syntactic relations, and providing subcategorization frames (331; but see Hoffmann & Stavans, 2007:63 for a different proposition). Function words are, according to Myers-Scotton, only rarely borrowed (2006:235). Word order differences between languages do not appear to reduce or impede switching in bilingual speakers in any significant way (331). It is posited that when two languages have differential word order, the order of the Matrix Language dominates (Myers-Scotton, 2006:255). Embedded Language islands constitute an exception to this rule in that they retain their original word order (257). In view of what has been said above, it is safe to assume that language mixing is an integral part also of early bilingual development (De Houwer, 2005:42) and an important aspect of bilingual acquisition (Lanza, 2001:202). Bilingual children are known to resort to code-mixing for a range of communicative and pragmatic purposes (Genesee & Nicoladis, 2007). As has been suggested, children code-switch to compensate for linguistic inadequacies in one of their languages (see Genesee & Nicoladis, 2007), or to “capitalize on the typological efficiency of one linguistic system versus the other so as to achieve successful communication” (Hoffmann & Stavans, 2007:56). If they are hungry or thirsty, children need to get their message across and be understood fast. Young children growing up bilingually may initially not command a sufficiently large stock of vocabulary in each of their languages. To make up for this lexical gap the child may opt to borrow from her/his other language(s) (Genesee, 2002:190) thereby drawing on lexical and grammatical devices in the stronger language(s) because they are not yet available in the weaker language(s). However, as stated by De Houwer (2005), mixed-language utterances can not simply be reduced to a lack of language skill or mastery. Neither are they a manifestation of “language confusion” (42). Instead, mixed utterances have to be construed in terms of the bilingual child’s linguistic environment and the “sociolinguistic norms” that exist in that environment. This claim is backed up by empirical
90
Multiple language acquisition in children
studies which indicate that young bilingual children are “keenly attuned to the specific linguistic environments” in which they are raised (4). The implication is that bilingual children’s switching habits reflect the practices of their environs. More particularly, children “are socialized into community norms for language use and language choice through everyday interactions” (Lanza, 2001:201, italics in the original). Hence, parent-child interaction and parental responses to their child’s mixing affect the child’s linguistic behaviour and development in significant ways (225). When bilingual children mix their languages or codes, they do so by mobilising all the linguistic resources at their disposal (Genesee, 2002:192). In this sense early mixing, far from being random, must be seen as a highly strategic and deliberately deployed tactic testifying to the child’s developing communicative competence (Genesee, 2002:190) and to her/his resourcefulness rather than to incompetence (ibid.) or confusion (187). Hoffmann & Stavans (2007) specifically emphasise the “linguistic resourcefulness of multilinguals” and the “linguistic versatility” of young trilingual users (55), while Lanza (1992:636) intimates that bilingual children possess a “bilingual awareness” which allows them to establish whether mixing is appropriate or not in a particular situational context. In addition, it has been suggested that bilingual children possess an early awareness of their interlocutor’s requirements. As noted by Barnes (2006), “[i]t seems that children are able to distinguish between those addressees they view as monolingual and those that are bilingual” (20–21). In the same vein, Genesee (2009) argues that bilingual children are capable of adjusting their choice of language code(s) to match their interlocutor’s linguistic profile (5–6; see also Genesee & Nicoladis, 2007). To sum up, there appears to be general consensus that mixing in bilingual children can not be adequately investigated if not in relation to the child’s sociolinguistic environment. Lanza (2001) points to the importance of “examining the interrelationship between the child’s linguistic development and his language socialization within which this development is embedded” (225). Similarly, Hoffmann & Stavans (2007) hypothesise that the sociolinguistic conditions shaping the child’s linguistic landscape affect not only the development of each language but also the kind of switches that the child produces (55). Lanza (1997) found that children mix their languages when they perceive mixing to be socially accepted, and that children refrain from mixing whenever mixing appears inappropriate (319 in Barnes, 2006:17). Goodz (1989:42) suggested that mixed input from adults encourages mixed output on the part of the child (Genesee, 1989[2007]:327 and in Barnes, 2006:19). This is supported by Meisel (1989:15) who argues that parental choice of codes and parental code mixing influence linguistic behaviour in the child. A similar point is made by Genesee
Phonological development in bilingual/multilingual children
91
(1989) who claims that parents’ language mixing habits constitute a kind of model to which bilingual children are particularly susceptible (327).
6.6 Phonological development in bilingual/multilingual children Young children are often said to be better language learners than adults. Studies on young children’s phonological development indicate that language-general perception patterns in the very young infant become language-specific later on and that this transformation may occur as early as 6 months of age. In other words, while very young infants are sensitive to a vast range of non-native phonological structures, this sensitivity to non-native sound patterns diminishes within the first year of life (Walley et al. 1999:308). It is widely acknowledged that as age of exposure increases, learners tend to find it ever more difficult to negotiate cross-language vowel and consonant differences, and to perceive and produce L2 sounds accurately (Walley et al. 1999:310). As noted by Singleton (2001), “[f]ew L2 researchers challenge the proposition that those L2 acquirers whose exposure to the L2 begins early in life for the most part attain higher levels of proficiency than those whose exposure begins in adolescence or adulthood” (84). Similarly, Flege (1999) argues that “[q]uite clearly, earlier is better. . .”19 since “as more of what is initially uncommitted vowel space becomes devoted to the L1, the formation of new vowel categories is more difficult” (Walley et al. 1999:311). Scholars in the generativist tradition and proponents of the Critical Period Hypothesis have argued that after a certain critical age the human brain loses its flexibility and capacity for acquiring languages. In particular, it has been maintained that “whatever the nature of the endowment that allows humans to learn languages, it undergoes a very broad deterioration as learners become increasingly more mature” (Johnson & Newport, 1991:256). The critical period hypothesis has never been satisfactorily proved and remains as controversial as when it was first advanced by Lenneberg in 1967. Flege (1999), for instance, posits that the “critical period hypothesis does not appear to provide the best explanation” of age-related differences in L2 attainment (101). In the same way, Singleton (2001) argues that the assumption “that L2 age effects are exclusively neurologically based, that they are associated with absolute, welldefined chronological limits, and that they are particular to language looks less and less plausible” (85). Johnson & Newport (1991) by contrast suggest that some marked decline in the ability to learn a language may obtain as early as 4 19 Flege is mainly referring to L2 pronunciation.
92
Multiple language acquisition in children
to 7 years of age. Language acquisition capacity in adulthood is, according to these authors, severely diminished since “the changes that occur between childhood and adulthood in language learning seem to affect all aspects of grammar acquisition, including access to UG, the ability to set parameters, and the ability to master the language-specific details of the grammar” (256). The authors thus conclude that “[t]he outcome of a late second language learning . . . is an acquired language which [. . .] is imperfectly mastered and sometimes even violates universal constraints on human languages” (256). According to the Critical Period Hypothesis, attainment of a native-like accent in the L2 is compromised if L2 acquisition does not commence at an early enough age (Bongaerts et al., 1997:448). This, so Scovel (1988), is due to the fact that pronunciation is “the only aspect of language performance that has a neuromuscular basis” (101). The implication is that neuromotor development affects the learner’s capacity to acquire native-like pronunciation. Conversely, Flege (1999:101–132) posits perceptual differences in young children and older learners of an L2. This appears corroborated by more recent studies which found that infants are highly sensitive to speech sounds from different languages (see Genesee & Nicoladis, 2007). Crucially, this early sensitivity to foreign-language phonetic units is predicted to diminish sharply at the age of 6 to 12 months (Kuhl et al., 2003:9096). If young children are only exposed to one language their perception is increasingly tuned to the sounds of that language. Phonetic category representations are then established for vowels and consonants in that particular tongue (Flege, 1999:105; but see Lowie & Verspoor, 2011:278 for a critical the notion of mental representations). If children grow up bilingual from a very young age, they will establish categories for L2 vowels and consonants, too (for more on phonological sensitivity, see Part II, section 20.1.). It is suggested that learners who “establish a phonetic category representation for an L2 sound may be more accurate in producing it than those who do not” (Flege, 1999:126). According to Flege (1999), there are strong indications that “correlations exist between the accuracy with which vowels and consonants are produced and perceived” (126; cf. Lowie & Verspoor, 2011:283–4). In older learners category formation for L2 is predicted to be less successful than in young children. In older learners, accuracy in L2 pronunciation may be reduced due to missing or imprecise L2 phonetic representations. It appears that older “bilinguals are unable to fully separate the L1 and the L2 phonetic system” (Flege, 1999:106). Scholars working within the DST framework have suggested that so-called dynamic attractors (which are the result of physical laws of articulation) may affect and determine the L2 user’s pronunciation. It is posited that the L2 user’s subjective experience and perception, the frequency with which certain sound patterns occur in the linguistic environment, and the perceived
Phonological development in bilingual/multilingual children
93
similarity or dissimilarity between the implicated languages all influence the said attractors (see Lowie & Verspoor, 2011:283). In L2/L3 acquisition, competing L1 attractors may, according to Lowie & Verspoor (2011), be overcome by targeted awareness raising exercises. The authors also recommend pronunciation teaching strategies which employ holistic approaches and methodologies such as “paying attention to overall articulatory settings rather than focusing on the correct pronunciation of individual sounds” (284). With regards to phonological representations in children, Genesee & Nicoladis remark that it is as yet not clear whether bilingual children have two languagespecific phonological repertoires for their two languages (2007). Similarly, Bosch & Sebastian-Galles (2001) note that the processes underlying perceptual differentiation in bilingual children are not yet fully understood. As pointed out by these authors, researchers are at present not able to provide conclusive answers as to whether bilingual infants form separate representations for their phonologies (90). This said, there are indications for an early perceptual differentiation. Exploratory experiments conducted by Bosch & Sebastian-Galles in 1997 and 2001 speak to an early discrimination in bilingual infants (see also PoulinDubois & Goodz, 2001:106). Moreover, their findings showed no delay in language discrimination abilities compared to monolingual infants (84). Bosch & Sebastian-Galles (2001) even suggest that “simultaneous exposure to two languages could lead to an enhanced language discrimination capacity” because bilingual infants “may have increasing opportunities to detect inconsistencies in their somewhat complex input, in which certain incongruent cues or heterogeneous sound patterns might often be present” (79). This is in accord with Kuhl’s (2000) suggestion that language experience changes infants’ discriminative abilities and “results in a ‘mapping’ that alters perception” (11853). To some extent it also complies with dynamic conceptions of language development according to which multilingual systems are evolutionary systems that undergo constant alterations and/or adjustments to changes in the environment (Herdina & Jessner, 2002:86). In this chapter the focus has been on how infants and young children acquire and cope with multiple languages. In the next chapter the perspective widens as we examine issues related to multilingual speaker performance and proficiency in various age groups.
7 Multilingual Proficiency In multilingual communities or contexts there tends to be a strong focus on the communicative function of languages with the result that the notion of proficiency comes to occupy a subordinate role (cf. Olsshtain & Nissim-Amitai, 2004:37). Not surprisingly, multilingual proficiency is seen as distinct from monolingual proficiency or standards since, in multilingual contexts, proficiency does not imply attainment of perfect mastery in each language. Instead, multilingual proficiency is primarily related to communicative success and is held to be dependent on individual and societal needs, particularly on perceived individual requirements (36). Jessner (2008b:92) reports two broad phases or traditions in the history of the investigation of language proficiency. The first phase, up to about 1980, conceived of language proficiency as a single, global ability, the second phase saw scholars re-evalutate and revise these early (mis)conceptions. One of the results of their theorizing was the proposition of a radically novel conceptualization of language proficiency which conceives of multilingual competence as a multicomponential construct. The thus achieved paradigm shift marked the beginning of a new phase in the scientific debate on multilingual proficiency. Some of the more recent contributions to this field of research (cf. De Angelis & Jessner, 2012; Hobbs, 2012; Gibson & Hufeisen, 2011; Hufeisen, 2011; Cook, 2002) are particularly valuable because they describe multilingual proficiency as a multifaceted and multidimensional component of multilingual systems and do not reduce proficiency to the speaker’s capacity to generate errorfree utterances. Multilinguals are typically defined as individuals who can communicate in two or more languages (Cenoz et al., 2003:2). Multilingual speakers differ from monolingual and bilingual speakers in that they have not one or two but three or even more different language systems. The presence and interaction of these multiple languages in the mind makes multilingual systems and multilingual processing infinitely more complex than monolingual or even bilingual systems and/or processing (cf. Herdina & Jessner, 2002; Kecskes, 2006:222; Stavans & Swisher, 2006: 193; Bono, 2011:25). There is general consensus that multilingual processing and performance are influenced by a range of factors including individual differences, cognitive resources, language typology, frequency and recency of use, language dominance, etc. (Grosjean, 2001; Cook, 2002; Herdina & Jessner, 2002; Forkamp & Bersgsleithner, 2007). With regards to L2/L3 attainment, Cenoz, Hufeisen & Jessner (2003) note that so-called balanced bi- or multilinguals are not frequently found and that “a balanced level of proficiency in several languages is not to be expected if we take into account the different dimensions
Multilingual Proficiency
95
of communicative competence including linguistic, pragmatic, sociolinguistic, discourse and strategic competence” (2). Early research on bilingual proficiency tended to regard bilingualism as some sort of double monolingualism, and bilingual speakers as underachieving or even cognitively deficient individuals. This perception can in part be attributed to the impact of the Chomskyan conception of language as a uniform and pure instance of an innate Universal Grammar (Chomsky, 1986:17). As stated by Chomsky, it is not the ‘larger picture’ that is the object of investigation in UG based research but merely “certain elements” relating to the language faculty (3) and only the “speech community with no dialect diversity and no variation among speakers” (17), viz. the “ideal speaker-listener, in a completely homogeneous speech-community, who knows its language perfectly and is unaffected by such grammatically irrelevant conditions as memory limitations, distractions, shifts of attention and interest, and errors (random or characteristic) in applying his knowledge of the language in actual performance” (in Jessner, 2008b:91). Such a view wrongly implies that language competence stands in isolation from other more general cognitive abilities. The UG perspective appears all the more inadequate if one considers that a robust body of literature today indicates that “[t]he linguistic system is interconnected with non-linguistic cognitive systems” (Barlow & Kremer, 2000 in Segalowitz, 2010:54; see also Green, 1998, Herdina & Jessner, 2002, Bialystok, 2005; Larsen-Freeman & Cameron, 2008 and Lowie & Verspoor, 2011). Contrastive Analysis Hypothesis, which saw its heydays in the early second part of the 20th century, proved to be similarly inadequate as a comprehensive theory of L2 acquisition. Based on Behaviourist Theory which conceived of language learning as the product of habit formation, Contrastive Analysis Hypothesis held that “where two languages were similar, positive transfer would occur; where they were different, negative transfer, or interference, would result” (LarsenFreeman & Long, 1991:53). In this sense proponents of CA systematically compared languages in the attempt to identify points of similarity or difference so as to anticipate difficulties and reduce the occurance of errors to a minimum (55). Native-like attainment, it appeared, could only be achieved through the systematic elimination of errors. According to CA, errors resulted from the intrusion of L1 forms or structures. The underlying assumption was that learners had little or no control over the habits formed in their L1 (61). L2 learning was seen as the process of adopting and internalising the habits of the new language while at the same time suppressing the habits formed in the native language. It was claimed that contrastive analysis of the L1 and L2 would allow linguists and educators to predict and thus prevent the formation of wrong habits and the occurance of errors. The CAH is today regarded as having serious flaws. It is
96
Multilingual Proficiency
argued that CAH is inadequate for the purpose of explicating the psycholinguistic processes underpinning L2 learning (56). In fact, CA was presently abandoned in favour of Error Analysis which differed from CA in that it did not envisage the L2 learner as “a passive recipient of TL” (61) but ascribed her or him a more active role in the learning process. Like CA, however, Error Analysis soon proved unsatisfactory as a comprehensive theory of secondary language acquisition. Researchers adhering to both, CA and EA, focused primarily on errors as their objects of investigation. This restricted focus invariably renders both approaches untenable as a theory of L2 development. As pointed out by LarsenFreeman & Long (1991), Error Analysis was presently complemented by Performance Analysis (62). PA was committed to investigating learner performance and did not limit itself to analysing learner errors. Scholars in the tradition of PA studied learners’ language productions and linguistic idiosyncracies, developmental sequences, and subjects’ learning strategies. It soon became clear, however, that certain utterances could not be accounted for without extending the scope of linguistic investigations so as to include the wider discourse in which utterances are embedded (69). Crucially, it also became clear that bilingual and multilingual speakers could not simply be measured against monolingual standards. Cook was one of the first scholars to condemn the widespread ‘monolingualist’ norm and the deep-rooted belief in the supremacy of the native speaker (Cook, 2002:10; see also Herdina & Jessner, 2002, Jessner, 1999, 2008b). According to Cook, the L2 user “is a different kind of person, not just a monolingual with added extras” (4; see also De Angelis & Selinker, 2001:45). The implicit assumption is that “L2 users can perfom specific activities that L1 users cannot” (4) since they move between two languages skillfully and draw on the resources of both (5). As stated by Cook (2002), the notion of complete, native-speaker competence is redundant and “meaningless” since “competence is whatever it is” (10). Hence, Cook (1991) proposed the term ‘multi-competence’ to relate to the “knowledge of more than one language in the same mind” (Cook, 2002:10), suggesting that the developing language systems of learners of an L2 are systems in their own right with their own idiosyncratic features. Such a claim clearly contrasts with the traditional fractional or monolingual view of learner systems as incomplete and deficient reflections of native grammars (Cook, 2002:9). The multilingualist viewpoint propounded by Cook is echoed by Olshtain & Nissim-Amitai (2004). The authors propose to adopt a multilingual perspective of proficiency emphasising particularly the role of the speaker’s perceived linguistic needs and requirements. Olshtain & Nissim-Amitai state that as multilingual speakers use their languages in distinct contextual settings, the need to master each of these languages is clearly dependent on everyday patterns of
Multilingual Proficiency
97
language use (32; cf. Cenoz, 2012 paper presented at the 8th International Conference on Third Language Acquisition and Multilingualism in Castellon, Spain). In other words, as multilingual speakers employ different languages in different domains and encounters of their daily life, it is felt that command of these languages need not be the same in all languages (32). Along very similar lines, Jessner (1999), too, relates speakers’ multilingual proficiency to their perceived communicative needs, stating that these needs are subject to constant change (207). According to Cook (2002:8), the L2 system is “a creation of its own” (9). More particularly, it is assumed that “[l]inguistic competence is whatever it is at the particular moment that it is being studied, not a partial imitation of what it might become one day” (ibid.). It follows that L2 systems are “neither a defective version of the monolingual native grammar nor a partial transfer from the second, even if there are elements of both within it” (ibid.). In accord with Cook (2002), Herdina & Jessner (2002) reject the notion of the ideal native speaker. The authors also dismiss the long-standing Chomskyan dissociation between language competence and performance as it has proved inadequate for the discussion of multilingual development and proficiency. Instead, Jessner (2008b) describes the multilingual speaker as “somebody who has a different way of using and knowing her or his languages in contrast to the native speakers of the respective languages” (100). Such a perspective clearly entails a readiness to focus on the benefits of multilinguality and to “accredit a less prominent role to the linguistic deficits” of multilingual users (ibid.). An analogous proposition comes from Grosjean who calls for a (w)holistic perspective on bilingual phenomena (Jessner, 2008b:93). In particular Grosjean (1994) laments that bilinguals are frequently described and evaluated in terms of the fluency and balance they have in their two languages; language skills in bilinguals have almost always been appraised in terms of monolingual standards; research on bilingualism has in large part been conducted in terms of the bilingual’s individual and separate languages. (2)
The author argues that bilingual subjects must be seen as developing differential competencies (in their various languages) according to their communicative needs in every-day contact situations20 (see also Ofelia Garcia, 2012 paper given at the 8th International Conference on Third Language Acquisition and Multilingualism). Grosjean notes that “bilinguals rarely develop equal fluency in their languages” (2), specifying that bilingual speakers are “people who use two (or 20 The reader is referred to DMM according to which the speaker’s (perceived) communicative needs represent “the driving force of language learning and use (Jessner, 2008b:95).
98
Multilingual Proficiency
more) languages (or dialects) in their everyday lives” for differential functions, in distinct domains and with varying degrees of proficiency (1). This leads the author to posit that bilinguals have to be evalutated in terms of their “total language repertoire” (2). The comparison with the high hurdler has been employed to illustrate this point. The high hurdler combines the disciplines of jumping and sprinting in one person and his achievements are acknowledged for what they are, but it would not make any sense to compare the high hurdler to either a sprinter or a jumper (Jessner, 2008b:93). In the same way, it makes little sense to compare a monolingual speaker to a bilingual or multilingual speaker. The DMM specifies multilingual proficiency as a “cumulative measure of psycholinguistic systems in contact” (Jessner, 2008b:95) and as an indirect function of general language effort (GLE) (Herdina & Jessner, 2002:131) whereby GLE comprises LAE (language acquisition effort) and LME (language maintenance effort). Language proficiency is not seen as a steady state (160) but as dynamic and therefore changing all the time. This is consistent with Grosjean (1994) who notes that [a]s the environment changes and the needs for particular language skills also change, so will the bilingual’s competence in his or her various language skills. New situations, new interlocutors, new language functions will involve new linguistic needs and will therefore change the language configuration of the person involved. (2)
Meisel (2004) concurs with Grosjean and states that “a person who uses two languages regularly, is not necessarily equally at ease in each of them in all communicative contexts and does not even have to be able to use them equally well” (93). In accordance with Herdina & Jessner (2002), Meisel notes that language dominance or proficiency is subject to constant change as it “can shift repeatedly [. . .] over the entire life span, depending on a person’s communicative needs” (2004:94). The idea that the ultimate goal of L2/L3 learning must be to attain native-like mastery of a target language and to pass as a native speaker of that language (cf. Cook, 2002:10) thus appears wholly inappropriate. A significant contribution to this area of reseach comes from Cummins (1979) who proposed to distinguish between two types of language competence in speakers of an L2. According to Cummins, language competence comprises two distinct skills, basic interpersonal communication skills (BICS) and cognitive/ academic language proficiency (CALP). BICS relate to basic language skills such as comprehension and speaking, pronunciation, vocabulary and grammar (Cummins & Swain, 1992:152–156; Baker, 2006:175). BICS are said to be employed in everyday communicative and context-embedded situations. CALP, by contrast, denotes “the level of language required to understand academically demanding
Multilingual Proficiency
99
subject matter in a classroom” (185). CALP makes use of the cognitive functions of language (Herdina & Jessner, 2002:16) and is seen as occurring in contextreduced academic environments (Baker, 2006:174). BICS and CALP are pictorially represented as an iceberg with basic communicative skills above the surface and CALP skills of analysis and synthesis below the surface (Baker, 2006:175). The distinction between BICS and CALP allows Cummins to account for the observed discrepancy between some bilinguals’ apparent communicative competence on the one side and their relative failure in school or academic contexts on the other (Baker, 2006:176). As pointed out by Herdina & Jessner (2002), Cummins’ contribution carries important implications insofar as it indicates that “language competence is a much more complex phenomenon than originally assumed” (17). As part of the MIVAS project (Inferential Methods for System Evaluation) recently carried out in the province of Bolzano/Bozen, De Angelis & Jessner (2012) examined the written performance of trilingual students of L1 Italian, L2 German and L3 English. Unlike traditional approaches to language development which typically measure multilinguals’ proficiency or productions “from a reductionist or deficit perspective”, the authors took great care to adopt a holistic approach to multilingual proficiency and learners’ multilingual knowledge repertoire. Following Cummins’ Interdependence Hypothesis (1991), the authors claim that multilingual users’ language systems build upon a common underlying proficiency and interact in complex, dynamic and unpredictable ways (cf. Bono, 2011:25). Hence, with regards to multilingual testing, the authors posit that the “sole focus” on performance in one language does not suffice to judge a pupil’s overall linguistic knowledge (see also Hobbs, 2012:20621). Instead, they specify, one needs to look at the whole multilingual system, which comprises not only the speaker’s languages, but also takes into consideration so-called emergent properties (such as meta- and cross- linguistic awareness) which obtain as a function of the interaction between the multiple languages implicated. Moreover, as stated by De Angelis & Jessner, the adoption of a “monolingual yardstick” does not do justice to multilingual users’ knowledge of language, since such knowledge includes more than knowledge of the individual languages. According to De Angelis & Jessner, multilingual knowledge or proficiency is characterised by an enhanced multilingual awareness which guides multilingual speakers’ cognitive processes. Similarly, Cenoz (2012) posits that highly successful multilinguals are aware that mastering languages helps them tremendously to acquire additional languages and improve proficiency (8th Conference on Third Language Acquisition and Multilingualism). It thus follows that multilingual knowledge 21 Hobbs (2012) suggests that multilingual children need to be assessed in more than one language (206; cf. Garcia, 2009:378).
100
Multilingual Proficiency
is unique and certainly not comparable to monolingual competence of a given language. Multilingual learning and teaching is the focus of the next chapter where we consider different ideological positions, different types of multilingual education and different classroom arrangements or practices. The chapter also examines some of the gains deriving from multilingual instruction and makes suggestions for curricular considerations.
8
Multilingual Learning In The Classroom
As noted by Jessner (2008b), the influence of Contrastive Analysis Hypothesis22 in the 1960s led to a stigmatisation of L1 usage in the foreign language classroom (96). Cross-language interference, commonly perceived as negative transfer has since been feared to hamper progress in L2/L3 acquisition. For a long time, the wide-spread assumption has been that cross-language interference would have detrimental effects on the learner’s linguistic and cognitive development. In the 1940s and up to the 1960s, linguists and pedagogues working within the framework of CA assumed that by conducting systematic contrastive analyses of learners’ L1 and L2 productions they would be able to identify problematic structures and reduce interference effects of the learners’ L1 on their L2(s). As explicated earlier, Contrastive Analysis Hypothesy posited that “where two languages were similar, positive transfer would occur; where they were different, negative transfer, or interference, would result” (Larsen-Freeman & Long, 1991:53). Only recently has there been a change of perception. This has led to an increase in interest in multilingual education programmes on the one side (see Garcia, 2009 for a comprehensive overview of bi- and multilingual education programmes), and to the recognition on the other, that cross-linguistic interaction and consultation can have facilitative effects on L2/L3 learning processes. In particular, the idea has been winning ground that building on prior linguistic knowledge as “a cognitive basis for further language learning” (96) can immensely benefit the acquisition of second or third languages (cf. Jessner, 1999:207). With regards to multilingual schooling, Cummins23 (2001) states that [w]hen children [. . .] develop their abilities in two or more languages throughout their primary school years, they gain a deeper understanding of language and how to use it effectively. They have more practice in processing language, especially when they develop literacy in both, and they are able to compare and contrast the way in which their two languages organize reality. (2)
It has been intimated that multilingual education is “[o]ne of the most challenging problems of applied linguistics” (Herdina & Jessner, 2002:159). Clearly, multilingual education presupposes an informed understanding of multilingualism and multiple language acquisition (cf. Hobbs, 2012:204–207 and Nicole Marx,
22 Contrastive Analysis was developed by Weinreich and Haugen in order to explicate what happens in the bilingual mind as languages come into contact (James, 1996:143). 23 Cummins is here particularly concerned with the importance of mother-tongue instruction for children who attend educational programmes in an L2.
102
Multilingual Learning In The Classroom
2012 paper given at the 8th International Conference on Third Language Acquisition and Multilingualism). Multilingual education and multilingual instructional programmes are discussed in detail in Garcia (2009) and Cenoz (2009; see also Blackledge & Creese, 2010). Conceiving of multilingual education as the teaching of more than two languages in schools which aim to foster multilingualism and multiliteracy, and drawing on Hornberger’s notion of continua (2007), Cenoz (2009:32) has proposed the ‘Continua of Multilingual Education’ with the intent to provide a means of representing, describing and contrasting differential types of multilingual programmes (34). Without “establishing closed categories and hard boundaries” (56), Cenoz’s model considers a wide range of educational, linguistic and sociolinguistic factors (including language/s of instruction, contact or linguistic distance between the languages involved, subject- and teacher-related variables, school context, number, status and general vitality of the languages implicated, the pupils’ close social networks, etc.) and allows for a ‘dynamic’ classification of educational programmes (as ‘more’ or ‘less’ multilingual), depending on their degree of multilingualism and on their specific position on the given multilingual continua. As stated by Cenoz (2009), the ‘Continua of Multilingual Education’ highlight the inherent complexity and the dynamics of multilingual education and consider the impact of various highly relevant variables on a particular educational programme. Garcia (2009) criticises that even in multilingual education programmes learners are often expected to have separate and full competencies in each of their languages (7). She argues that such idealised expectations take no account of how languages are utilised in multilingual communities or contexts (7; see also Garcia & Sylvan, 2011:389). Garcia (2009) therefore calls for a reconceptualization of bi- or multilingual education (9). Her call is echoed by Garcia & Sylvan (2011) who identify an urgent need to put an end to the restrictive language arrangements which are in place in many classrooms worldwide and which impose on pupils and teachers which language they have to use in a particular context with a particular interlocutor (Garcia & Sylvan, 2011:386). Advocating dynamic multilingual programmes which teach for “singularities in pluralities” rather than simply promoting one particular language or standard variety, Garcia & Sylvan (2011) point to the necessity of acknowledging the linguistic experiences of the individual student in order to build on their differential language practices (386). This, according to the authors, presupposes an acceptance of each individual student’s (trans)languaging (i.e. individual students’ flexible and dynamic use of multilingual practices) and requires flexible pedagogical approaches which center on the uniqueness of individual experiences (391). Garcia & Sylvan (2011) propose seven principles (including heterogeneity,
Multilingual Learning In The Classroom
103
collaboration, learner-centeredness, language and content integration, language use from students up, experiential learning, and local autonomy and responsibility), all aimed at fostering dynamic plurilingual practices in instruction so as to facilitate learning through the promotion of students’ multilingual/multiliterate competence (398). Building on translanguaging for the purposes of sense-making and learning forms one of the core components of Garcia & Sylvan’s dynamic model of bi – and multilingual education. As explicated by Jessner (2008b:91), multilingual education “can only be successful if the cognitive potential of multilingualism is explicitly acknowledged on the societal level”, that is, if there is a willingness, on all sides, to abandon the long-held view that bi/multilingualism represents an obstacle or hindrance to learners’ linguistic and cognitive development, if there is acknowledgement that multilingual speakers’ linguistic and cognitive resources are not divided or reduced on account of their multilingual experience, and if there is acknowledgement that multilinguals are not incompetent speakers of their various languages. A further significant aspect to bear in mind when discussing multilingual education is its socio-political dimension. The present author may be forgiven for speculating that the bias against multilingual schooling reflects the political climate in a particular region or country (cf. Garcia, 2009:12, Orwenjo, 2011: 294–317 and Hobbs, 2012:206). This issue will be elaborated upon in more detail in the second part of this book. Herdina & Jessner (2002) have convincingly argued that the “development of multilingual proficiency leads to an enrichment of the individual language system” (160; cf. Wandruska, 1979:16, 18; Bono, 2011:49). This has been supported by a number of recent studies evidencing cognitive advantages of biand multilingualism (see Bialystok, 2001, 2005, 2007; Jessner, 2006; De Angelis & Jessner, 2012; Bono, 2011). While it is probably true to say that the majority of multilingual subjects have domain-specific (or even modality-specific) competence in their two or more languages, it has been pointed out that also “partial competence in a particular language” is to be seen as “meaningful” in that it “forms part of a multiple competence which it enriches” (Coste, 1997 in Herdina & Jessner, 2002:160; cf. Cook, 2002 and Vetter, 2012). Researchers working in a holistic, dynamic framework have for some time now been calling for traditional monolingual norms of language acquisition and development to be replaced by multilingual norms (see for example Cenoz, 2009, Garcia, 2009, Blackledge & Creese, 2010, Garcia & Sylvan, 2011). It is hoped that such a change of perspective will allow for the transition from fractional to holistic conceptions of multilingual acquisition and proficiency which take into consideration that “the minds, languages and lives of L2 users are different from those of monolinguals” (Cook, 2002:9). With regards to multilingual instruction this may require a
104
Multilingual Learning In The Classroom
thorough reevaluation of traditional teaching approaches (cf. Hobbs, 2012:204) and the adoption of new language teaching concepts with a firm grounding in multilingualism research. In ‘Teaching third languages: Findings, trends and challenges’, Jessner (2008) discusses the benefits of crosslinguistic approaches to language teaching. She proposes to bridge the languages in the curriculum and to include also those languages which, although not part of the school syllabus, still play a role in the pupils’ linguistic biographies (44). Bridging languages, so Jessner, allows learners to “profit from the synergies and to exploit the resources that many [. . .] already have available through their prior language knowledge” (44; see also Allgäuer-Hackl & Jessner, 2013; Garcia, 2009; Krumm & Reich, 2013). Advocating a holistic approach to multiple language teaching, Herdina & Jessner (2002:160) call on educators to focus on language learning skills, language management skills and language maintenance skills in the classroom. Contributing to learners’ metalinguistic awareness and forming part of the Mfactor, these skills have been found to develop in the multilingual speaker as a result of profound qualitative changes in the multilingual system (Jessner, 2006:88-118). As noted earlier, the M-factor or Multilingualism factor is a set of skills or abilities which multilingual users or learners develop thanks to their previous linguistic and metacognitive knowledge and/or experience (Jessner, 2008a:275). This accumulated knowledge and the resulting (meta)linguistic awareness are thought to exert a strong influence on the acquisition of additional languages (ibid.; see also Bono, 2011). Herdina & Jessner (2002) posit that explicit teaching of meta-knowledge will increase learners’ metalinguistic and metacognitive awareness (cf. OtwinowskaKasztelanic, 2011:15; Allgäuer-Hackl & Jessner, 2013). According to Herdina & Jessner, awareness-raising can be achieved by drawing on knowledge of previously learned languages, by pointing out commonalites between language systems and by familiarising learners with targeted language learning strategies. From the perspective of the DMM, language awareness, including awareness of the forms and functions of linguistic structures, and awareness of cross-linguistic similarities and/or differences, thus come to occupy a central role in the language learning classroom. The recommendation therefore is to design teaching materials with the explicit purpose of promoting metalinguistic and crosslinguistic awareness in the multilingual learner (2002:161; cf. Marx, 2012 paper given at the 8th International Conference on Third Language Acquisition and Multilingualism). Jessner (2008b) points to the centrality of metalinguistic awareness in effecting broader cognitive benefits. The author stresses the importance of fostering “synergy effects and cross-fertilization through cooperation between the languages and the language subjects in a classroom” (96; cf. Jessner, 1999:207;
Multilingual Learning In The Classroom
105
Hobbs, 2012:223–4; Allgäuer-Hackl & Jessner, 2013 and Hufeisen, 2011). Exploiting the etymology of English is proposed as one way of drawing pupils’ attention to commonalites in their various languages (40). Portfolios or other forms of linguistic background documentation are seen as equally powerful means to raise learners’ awareness of language(s) in particular and of multilingualism in general (2008:44; cf. Garcia, 2009). Similarly, Hawkins (1999) envisages a language curriculum that links all the languages and provides a kind of language apprenticeship. Ideally, so Hakwins, there is close and constant cooperation and exchange between all the (language) teachers (140). Hawkins (1999) proposed a 5 point catalogue for the implementation of an “ideal language curriculum” (140). Hawkins particularly recommends to 1) strengthen the mother tongue so as to give pupils confidence and motivation to learn (138,128; cf. Orwenjo, 2011:312) 2) include language awareness in the curriculum, encourage co-operation between the language teachers and focus on awareness in teacher training courses (139; cf. Hobbs, 2012:213; Jessner, 2008) 3) train pupils’ ear by teaching them to listen (139) 4) make pupils and teachers see multilingualism as a positive asset (140) 5) talk more about language(s) in the classroom, engage in cross language comparisons and discuss language learning strategies (140; cf. Allgäuer-Hackl & Jessner, 2013; Krumm & Reich, 2013; Schöpp, 2013). This is echoed by Hufeisen (2011; 2005) who also stresses the importance of complementing second or foreign language learning with learning about languages. In the same way as Hawkins (1999), Hufeisen (2011) proposes a language syllabus of instruction that involves teachers across the curriculum. The author points to the importance of sensitising learners to the forms and functions of their various languages and to foster their (foreign) language learning skills. Hufeisen proposes to link all the learners’ languages in a plurilingual whole school curriculum (266). It is probably true to say that the wide-spread misconception that only the strict separation of languages in the classroom can prevent confusion in beginning learners is slowly being supplanted by a growing acceptance that cross-linguistic consultation and comparison can facilitate and even accelerate language learning processes in the L2 and/or L3. As the beneficial effects of cross-language comparative approaches on learners’ linguistic and metalinguistic abilities become more and more apparent, linguists and educators are beginning to turn their back on traditional uni-lingual teaching methodologies. A number of innovative projects and initiatives in several European countries testify to this
106
Multilingual Learning In The Classroom
profound change of paradigms on the language education front (cf. Jessner, 2008:34–38; Hufeisen & Lutjeharms, 2005; Garcia, 2009). As James (1996) points out, traditional CA completely failed to acknowledge the benefits of between-language comparisons. In opposition to CA, James therefore suggests that “within-language and cross-language associations can become the new content” in the language classroom (147; cf. Hobbs, 2012:206). The author urges educators to engage in classroom-based contrastive comparisons by making structural and semantic relations between native and foreign language(s) the objects of study (145). James proposes translation as a particularly effective means of raising cross-linguistic awareness, “since, uniquely, in the act of translation two manifestations of MT 24 and FL are juxtaposed” (James, 1996:147; see also Malakoff & Hakuta, 1991:150). This clearly accords with OomenWelke (2006) who posits that multilingual learning occurs to a large extent through comparisons adding that “awareness of language learning strategies can build on the constructive potential of comparing language” (317; translation in Jessner, 2008b:100). Following this line of thought educators at one of the primary schools participating in the present research study have opted for the integration into the curriculum of a one-hour weekly session termed Riflessione Lingua (see 13.3. in part II of the book). The lesson aims at fostering pupils’ awareness of the structural properties, the idiosyncracies and the commonalities between their L1 and L2. Unfortunately, the L3 is not represented during this lesson. Riflessione Lingua broadly corresponds to what Hawkins (1981) had in mind as he advocated the introduction of “LA25 as a subject bridging the ‘space between’ MT and FL studies” (James, 1996:142). During Riflessione Lingua the Italian and the German teacher are both present in the classroom and take it in turns to point out and explicate various formal aspects of the pupils’ L1 and L2. In this sense, the lesson provides “a unified ‘whole-language’ reflective experience” (cf. James, 1996:142) enabling pupils to “take stock of” (ibid.) what they already know about their languages, and showing them how to exploit this knowledge as they build their language systems. The recommendation here is to include L3 English (and any other language spoken by the pupils) in these weekly crosslinguistic sessions since this will provide the pupils with a truly multilingual experience. As the classroom observations have revealed, the children receiving bilingual education (plus L3 English for 3 hours a week), already engage in a fair amount of trilingual cross-language comparisons (see part II of this book). Integrating the L3 (and other languages) into the weekly awareness class can 24 MT = mother tongue; FL = foreign language 25 LA = language awareness
Is there anything to be gained from multilingual learning?
107
be expected to enhance pupils’ already remarkable cross-language awareness and abilities even further. A last important point relating to language awareness in the classroom is made by Wright & Bolitho (1993). Emphasising the central role of the teacher in the language learning classroom, the authors state that “the more a teacher is aware of language and how it works, the better” (292; cf. Jessner, 2008:44). According to these authors “a linguistically-aware teacher will be in a strong and secure position to accomplish various tasks” (ibid.). Wright & Bolitho even go as far as saying that “successful communicative teaching depends [. . .] on a high level of language awareness in a teacher” (ibid.). Similarly, Hobbs (2012:213) points to the critical importance of teacher training and competence. Hobbs states that teachers need adequate metalinguistic awareness training and high fluency in the target language (211). This is in full accord with Cenoz & Lindsay (1996) who found that their respondents’ performance in L3 English correlated strongly with their teacher’s competence and confidence in that language (88). Cenoz & Lindsay conclude that the “role of the teacher, [. . .], is certainly crucial in Primary English” (93; see also Otwinowska-Kasztelanic, 2011:15). In the South Tyrolean context, the issue of primary teachers’ mastery of L3 English is particularly relevant since it is an open secret that a number of primary teachers do not have an adequate command of the English language. It is this author’s contention that the study of language(s) as part of primary teachers’ university education should be intensified and complemented by language awareness training (see Otwinowska-Kasztelanic, 2011:15 for a similar proposition; cf. Krumm & Reich, 2013 und Allgäuer-Hackl & Jessner, 2013).
8.1 Is there anything to be gained from multilingual learning? Cenoz (2003a) provides a comprehensive review of studies, conducted in various educational contexts, on the effects of bilingualism on TLA and indicates a tendency towards mixed results which she attributes to the fact that researchers have employed differential methodologies to explore different learning contexts and aspects of proficiency (79). Cenoz (2003a) does, however, state that the majority of studies carried out thus far report positive effects of bilingualism on TLA (particularly with regards to learners’ metalinguistic awareness, communication skills and language learning strategies). The aim of this section is to outline some of the linguistic and/or cognitive gains of early bi- and multilingual learning by reporting on relevant studies conducted in various educational contexts throughout Europe.
108
Multilingual Learning In The Classroom
The studies reported here below all highlight the benefits of bilingual schooling and/or of pupils’ bilinguality on TLA. In their study, Valencia & Cenoz (1992) found that secondary school bilingual students acquiring a third language in the Basque country performed significantly better than their monolingual counterparts. Bilingualism (particularly early and balanced bilingualism) turned out to be a reliable indicator of high achievement in L3 English (439– 440). The authors observed a significant effect of bilingual schooling (440) on third language achievement and concluded that bilingualism has a positive effect on foreign language acquisition (444). Cenoz and Valencia (1994) also examined Basque/Spanish bilingual school children and found that, compared to monolingual learners of an L2, the bilingual children were advantaged with regards to the acquisition of L3 English. Lasagabaster (1998) equally reports positive effects of bilingualism on English L3 acquisition. Lasagabaster observed significant correlations between his respondents’ L1, L2 and L3 intimating that multilingual speakers’ proficiencies in their three or more languages are interrelated. Similar results have emerged from a recent study carried out by De Angelis & Jessner (2012; see also chapter 7 (MIVAS project) and section 1.6. of this book). These researchers investigated trilingual learners at the primary level in South Tyrol and found additive cognitive effects of bilingualism (Italian/German) on learning L3 English. More specifically, they observed that pupils who performed better in L2 German also produced longer texts in L3 English. Their data reveal a strong relationship and interdependence between the pupils’ L1 Italian, L2 German and L3 English. De Angelis & Jessner suggest that L3 learning processes may be facilitated by learners’ knowledge of prior languages. The authors conclude that their results support previous studies in the field which identified similarly positive effects of bilingualism on TLA (64). In Austria, Griessler (2001) investigated 3 cohorts of high school students. The first group was enrolled in an English immersion programme, the second attended a so-called F(rench)-branch and received traditional English instruction (according to the Austrian high school curriculum) plus L3 French from as early as grade 7, and the third group was enrolled in a regular high school programme. Griessler found that the English immersion group attained the highest levels of English proficiency, which, as noted by the researcher, attests to the advantages of immersion education. Griessler further found that the F(rench)branch group outdid the students in the regular programme in producing an oral narrative (based on the “Frog where are you?” story) in their L2 English. Griessler interprets this particular finding as evidence for the supportive effect of third language learning on second language proficiency (50). Charkova (2005) examined partially bilingual and partially trilingual (10 year old) school children in Bulgaria and found that they outperformed mono-
Is there anything to be gained from multilingual learning?
109
lingual children on an L1 word-definition test. Charkova intimates that the findings of her study provide some evidence that early foreign language education has beneficial effects on children’s ability to define nouns (516). More particularly, the researcher suggests, that studying one or more foreign languages early on in school can heighten pupils’ perception and awareness of the semantic features of a formal definition, allowing them to successfully distinguish between relevant and irrelevant characteristics of a given concept (517). This said, Charkova also points out that “the positive effect of early foreign language education was limited to the class of concrete nouns. The quality of children’s [. . .] definitions for abstract nouns did not reflect the same benefits.” (518). She cautions that further studies are needed to verify the results obtained and that her findings may not be generalizable to other populations or contexts. More recently, Andreou (2007) studied bi- and trilingual fourth grade pupils and found that her trilingual group (with L1 Albanian, L2 Greek and L3 English) performed significantly better on a phonological awareness test than the bilingual group (with L1 Greek and L2 English). Andreou suggests that trilingual learners develop a heightened phonological awareness because they are constantly required to pay attention to language form when making decisions about their own language choices (11). Andreou intimates that her trilingual school children’s advantage may be linked to their prior experience with and knowledge of two languages (11). Investigating TLA in a different age group (17–18 year old high school students) Keshavarz & Astaneh (2004) also evidenced positive effects of bilingualism on the acquisition of L3. The researchers report significant differences between their monolingual and bilingual respondents on a foreign language vocabulary test. The results of their study demonstrated that bilingual subjects attained higher scores on a controlled productive ability vocabulary test compared to their monolingual counterparts. Keshavarz & Astaneh propose that the observed difference in foreign language vocabulary achievement may be correlated with the experimental group’s bilinguality. The authors intimate that bilinguals are better and more efficient at foreign language vocabulary learning than monolinguals (300). The above studies all indicate positive effects of early bi- or multilingual schooling and learning. This should not, however, delude us into thinking that we can indiscriminately generalise these findings to other educational and/or sociocultural contexts. It is clear that more, in particular long-term, research is needed on how early bi- and multilingual schooling affects young learners’ linguistic and cognitive development and on how school curricula and syllabi need to be designed so as to maximally benefit pupils’ learning. Clearly, an increased focus on learners’ prior linguistic knowledge and on their (meta- and
110
Multilingual Learning In The Classroom
cross)linguistic awareness and abilities would constitute an important step in the right direction. In the following we examine various aspects related to multilingual awareness and metalinguistic abilities. The aim is to provide important insights into the relationship between multilingual awareness and (third) language learning. The chapter closes with suggestions for future research.
9 Multilingual Awareness This chapter constitutes one of the central sections of the book. The focus is on multilingual awareness in young learners. In light of the many encouraging research findings generated over the past years or so (cf. Garcia, 2009:95; see chapters 1 and 8), there is legitimate reason to assume significant positive correlations between multilingual awareness and third language acquisition. In what follows, multilingual awareness and abilities are therefore explored in greater depth.
9.1 Learning without awareness? In the 1960s and 1970s many researchers in cognitive psychology and in the field of SLA concentrated on the role of implicit learning versus explicit or conscious learning. It was claimed that successful acquisition of a second or foreign language was largely dependent on unconscious acquisition processes. Conscious learning, by contrast, was deemed to contribute only very little to the process of SLA (Robinson, 2008:134–5). Krashen (1982), for instance, postulated two separate learning systems, one for unconscious acquisition and one for conscious learning. According to Krashen, unconscious acquisition of a second language was infinitely more effective than conscious learning (in Robinson, 2008:135). His critics however contested that attention to form and consciousness-raising activities in the classroom can be very beneficial to language learners, and that learning is impossible without some attention or consciousness (ibid.). Today there is overwhelming consensus on the fundamental importance of awareness in SLA and TLA. As stated by Jessner (1999), awareness of linguistic forms and functions “plays a central and facilitating role in the acquisition of additional languages” (207). According to Jessner (1999), metalinguistic awareness is a key component in the cognitive aspects involved in language learning (203; see also Allgäuer-Hackl & Jessner, 2013:114). The ensuing section elaborates on linguistic and metalinguistic awareness in more detail.
9.2 Language awareness and Metalinguistic abilities Early case studies by Leopold (1961) and Ronjat (1913) suggested that growing up with two languages is beneficial for children’s cognitive and linguistic development (Diaz, 1983:38, cf. Malakoff & Hakuta, 1991:147–148). In particular, it was
112
Multilingual Awareness
observed that children’s bilingual experience could promote their awareness of the conventional and arbitrary nature of language. Bilingualism appeared to foster the child’s capacity to distinguish between the word and its referent. In 1962 Vygotsky postulated that early contact with a second language enables the child to “see his language as one particular system among many, to view its phenomena under more general categories, and this leads to an awareness of his linguistic operations” (1962:110). In the same year, Peal & Lambert’s seminal study of bilingual English/French children in Montreal evidenced positive effects of bilingualism on children’s linguistic awareness. The authors found that their subjects’ bilingual experience resulted in greater mental flexibility which was not restricted to language but also extended to other, non-linguistic domains of thought. Subsequent research has supported this notion (for more on this, see chapter 1). Peal & Lambert’s findings marked an important transition in terms of people’s perception of the cognitive consequences of bilingualism. In recent years, the body of literature pointing to bilingual and trilingual advantages has continued to grow (see Bialystok, 2001, 2005, 2007; Bono, 2011; Chen et al., 2008; De Angelis & Jessner, 2012; Gibson & Hufeisen, 2011; Jessner, 1999, 2006; Keshavarz & Astaneh, 2004; Otwinowska-Kasztelanic, 2011; Rahimpour, 2002; Sanz, 2000). Historically, the notion of language awareness has been associated with two British scholars, George Perren (1974) and Eric Hawkins (1987). Perren (1974) wrote a collection of articles which he published under the title The Space Between and in which he encourages language teachers to adopt an interdisciplinary approach within a language awareness curriculum (in James, 1996: 138). Hawkins (1987) intimated that language awareness activities could bridge the “space between” learners’ languages (in James, 1996:138). Perren (1974) and Hawkins (1987) suggested that cross-linguistic contrastive approaches to language teaching would fine-tune learners’ attenae for formal and functional aspects of language and would put them in charge of their own learning progress (cf. Bono, 2011:46; Otwinowska-Kasztelanic, 2011:15). The concept of metalinguistic awareness is not easily defined. The inconsistent and non-systematic use of differential termini employed in this field of research has certainly not helped to throw light on the conceptual and terminological thicket. In English, one commonly encounters the concepts language awareness, knowledge about language, linguistic awareness, metalinguistic awareness, metalinguistic capacity and consciousness (-raising). Translation into other European languages poses further difficulties because additional languagespecific concepts such as Sprachbewusstsein and Sprachbewusstheit (in German), consapevolezza and coscienza (in Italian) or conscience and prise de conscience (in French) have to be considered (Jessner, 2006:41). This said, there appears
Language awareness and Metalinguistic abilities
113
to be general consensus that metalinguistic awareness or knowledge is “more” than linguistic knowledge (Bialystok, 2004:123). Karmiloff-Smith et al. (1996), for instance, note that in contrast to linguistic knowledge, “[m]etalinguistic awareness involves conscious reflection on, analysis of, or intentional control over various aspects of language – phonology, semantics, morphosyntax, discourse, pragmatics – outside the normal unconscious processes of production or comprehension” (198). Similarly, Pinto et al. (1999:14) argue that metalinguistic knowledge is distinct from and beyond the common use of language in production and comprehension. Metalinguistic awareness and bi- or multilingual proficiency can be thought of as separate but related skills (Malakoff & Hakuta, 1991:148). Metalinguistic awareness or ability is not generally conceived of as a unique or homogeneous ability but rather as a set of diverse skills (Pontecorvo et al., 1989:38; Malakoff & Hakuta, 1991:148). Importantly, metalinguistic awareness is held to be the strongest predictor of good proficiency in second or third language acquisition (Pinto et al. 2011). According to Malakoff & Hakuta (1991) metalinguistic awareness is “the ability to successfully approach and solve certain types of problems.” In this sense metalinguistic awareness “is both an awareness and a skill” (148). Jessner (2006) defines metalinguistic awareness as speakers’ ability to “focus attention on language as an object in itself or to think abstractly about language, and consequently, to play with or manipulate language” (42). Metalinguistic abilities can also be thought of as special skills which allow the individual to distinguish “between form and meaning in order to be able to manipulate languages” (Jessner, 2008b:97). As intimated by Bialystok (2001b), “metalinguistic ability describes the capacity to use knowledge about language as opposed to the capacity to use language” (124). Similarly, Gombert (1992) conceives of metalinguistic abilites as “a subfield of metacognition concerned with language and its use – in other words comprising: 1) activities of reflection on language and its use and 2) subjects’ ability to intentionally monitor and plan their own methods of linguistic processing” (13). For Tunmer et al. (1984) metalinguistic awareness denotes the realisation “that the stream of speech, beginning with the acoustic signal and ending with the speaker’s intended meaning, can be looked at with the mind’s eye and taken apart” (12). In a similar way, Galambos & Hakuta (1988) define metalinguistic awareness as “the ability to attend and reflect upon the properties of language” (141). This accords with James & Garrett’s proposition (1991) that language awareness comprises “the ability to think about and to reflect upon the nature and functions of language” (3; cf. Tunmer, Bowey & Grieve, 1983:567). Similarly, Malakoff (1992) states that “[m]etalinguistic awareness allows the individual to step back from the comprehension or production of an utterance in order to
114
Multilingual Awareness
consider the linguistic form and structure underlying the meaning of the utterance” (518). This is in line with Carter (2003) who notes that language awareness “relates to the development in learners of an enhanced consciousness of and sensitivity to the forms and functions of language” (64). Sharwood Smith (2008) suggests that “[b]ecoming aware of linguistic patterns, i.e., being ‘metalinguistically’ aware is [. . .] something that happens to us very early on when, as children, we play games with words” (179). The author specifies that this awareness “appears to be enhanced amongst bilingual or multilingual children who have everyday involvement with different language systems” (ibid.). In children, metalinguistic awareness has to be seen in the context of the child’s overall linguistic and cognitive development (cf. Ranta, 2008:206; Kecskes & Papp, 2000:78; Pinto et al. 1999:15). Emergent metalinguistic awareness and abilities in the young child are believed to be indicative of profound maturational and cognitive changes. Kecskes & Papp (2000) note that [t]he development of basic metalinguistic abilities is connected with the first 11 or 12 years of human development [. . .], during which there is a gradual shift of attention from meaning to structure in tasks requiring deliberate control over language forms. These basic abilities, such as judging synonymity, understanding structural ambiguity, judging grammaticalness and meaningfulness, segmenting sentences and words into their constituents, and manipulating forms by moving morpheme boundaries, develop as the result of interactions of linguistic, cognitive, and metacognitive skills. All these activities require students to have both an awareness of language as a system and the ability to access and manipulate knowledge about that system. (79)
As indicated above, it is hypothesised that early exposure to an L2 or L3 will result in enhanced linguistic and metalinguistic development. Mohanty (1994) explicates that “the bilingual child develops certain coping strategies which boost his metalinguistic development in particular and metacognitive development in general. Better development of metalinguistic and metacognitive processes, in turn, helps the child exercise greater control over his cognitive processes and makes them more effective, improving the level of performance of the child in a variety of intellectual and scholastic tasks.” (92; cf. De Angelis & Jessner, 2012). Mohanty’s position accords with Bialystok’s (1992, 2001) proposition that bilinguals are better at controlling attention and at attending to important information in a given task. According to Bialystok (2001), bilingualism enhances children’s ability “to attend to relevant information in the presence of misleading distractions” (179) while at the same time accelerating the development of “a general cognitive function concerned with attention and inhibition” (2005:417). Bilingual children, so the author (1992), “develop more advanced levels of control [. . .] than monolingual children do. This advantage is traced to
Language awareness and Metalinguistic abilities
115
the enriched opportunity for learning language and learning about language that is available to children who are learning to attend to two linguistic systems.” (501; cf. Bono, 2011:30–31). Bialystok (2004) proposes a cognitive framework for metalinguistic development which considers the “relation among different uses of language and their underlying cognitive requirements” (14). The author postulates two cognitive skills or processesing components of metalinguistic awareness: analysis of representational structure and control of attention. Analysis of representational structure relates to the child’s capability to generate mental representations of linguistic information at increasingly detailed levels of structure. Control of attention refers to the child’s ability to focus attention on specific aspects of a problem. According to Bialystok (1992:505), control of attention is particularly important when a problem contains some sort of conflict or ambiguity. These two cognitive components of metalinguistic awareness are pictorially represented as the axes of a Cartesian plane indicating their degree of involvement in differential language tasks or language usage situations (ibid.). As noted by the author, language tasks or language use situations can be classified according to their level of cognitive demands and positioned at various points on the Cartesian plane. This makes it possible to determine the extent to which the two cognitive components are operational during a given language task or communicative situation. The implicit assumption is that different language tasks or uses require differential reliance on the two processing components. Everyday communicative exchanges, for instance, are thought to place low demands on control and analysis. Metalinguistic tasks26 by contrast place much higher demands on analysis and control (2004:15). Bialystok intimates that analysis and control develop in response to the speaker’s experience. Bilingualism, so the author, may be one such experience (2004:134). According to Bialystok, it is plausible that knowledge of two languages “may make structural patterns more noticeable and hasten the child’s attention to the systematic features of language” (ibid.). Bialystok found that bilingual advantages obtain particularly for tasks which pose high demands on control (2001:169; 1992:506). In the same way, Clark (1978) argued that “learning two languages at once [. . .] might heighten one’s awareness of specific linguistic devices in both” (36). Similarly, Lambert (1990) suggested that “bilingualism provides a person with a comparative, three-dimensional insight into language, [. . .] that the monolingual rarely experiences” (212). This is congruent with Herdina & Jessner (2002) who 26 Bialystok points out that the relative complexity of single metalinguistic tasks can differ substantially (2001b:15).
116
Multilingual Awareness
note that one “must assume that metalinguistic awareness is one of the factors in which multilinguals outperform their matched monolingual counterparts” (65; cf. Bono, 2012:30). The same line of argument is found in Jessner (1999). The author predicts that the development of competence in two or more languages will result in higher levels of metalinguistic awareness (201; cf. Kecskes & Papp, 200:79; Otwinowska-Kasztelanic, 2011:26 and Ranta, 2008:206). Jessner (1999) states that prior language learning experience changes the quality of language learning. This very often results in differing language strategies which the experienced language learner develops in contrast to the inexperienced one. This development of advanced cognitive skills in language learning can lead to the speeding up of the language learning process. (207)
The underlying assumption is that metalinguistic awareness in monolingual speakers can not “be compared in both degree and quality to awareness as developed in bi- and multilingual users” (Jessner, 2008a:277). The reseach study presented in the second part of this book corroborates this assertion. In the DMM metalinguistic knowledge and awareness of that knowledge are key components of the multilingual user’s linguistic proficiency. From a dynamic systems perspective, metalinguistic awareness is seen as mediating the development of individual multilingualism (Jessner 2008a:270). In DMM the existence of an M-factor or Multilingualism factor is assumed. A unique property of multilingual systems, the M-factor emerges through the interaction of several language systems (Herdina & Jessner, 2002:130). As outlined earlier, the M-factor encompasses a number of components including language learning skills, language management skills, language maintenance skills, (multi)language aptitude27, metalinguistic abilities (MLA) and an enhanced multilingual monitor or EMM (Herdina & Jessner, 2002:129,138,160). In speakers of multiple languages metalinguistic awareness is said to manifest itself in enhanced a) divergent and creative thinking (e.g., wider variety of associations, original ideas); b) interactional and/or pragmatic competence (cultural theorems of greeting, thanking, etc.); c) communicative sensitivity and flexibility (language mode); d) translation skills that are considered a natural trait in the majority of multilinguals (Jessner 2008a:277).
27 The authors subsume both, metalinguistic abilities and (multi)language aptitude under the acronym MLA as there appears to be no conclusive evidence that the two notions represent different concepts (Herdina & Jessner, 2002:116).
The emergence of metalinguistic awareness in monolingual children
117
From a dynamic systems perspective, metalinguistic abilities are predicted to have unforseeable catalytic effects on further language learning. Crucially, these effects are deemed to be of a facilitative and accelerating nature (Jessner, 2008a:277). The positive impact of multilingualism on speakers’ (meta)linguistic development is substantiated by Cummins’ theory of a Common Underlying Proficiency which posits that the speaker’s languages do not function separately but instead “operate through the same central processing system” (Baker, 2006:169; Cummins & Swain, 1992:83), possibly augmenting and enhancing the person’s knowledge and understanding. Jessner (1999) relates Cummins’ (1991) Common Underlying Proficiency to multilingual speakers’ increased metalinguistic abilities and enhanced monitoring system. The Common Underlying Proficiency posits the connectedness and interaction of the speaker’s language systems. According to Jessner (1999), this interconnectedness, typically found in multilingual systems, enhances speakers’ monitoring functions and results in heightened levels of metalinguistic awareness (205). Earlier, Ben-Zeev (1977) had come to a similar conclusion affirming that bilingualism or the constant negotiation of languages forces speakers to draw on certain strategies which can accelerate their cognitive development (1009).
9.3 The emergence of metalinguistic awareness in monolingual children There is considerable controversy as to what young children, particularly preschool children know about language and as to whether that knowledge qualifies as metalinguistic awareness or not. Scholars particularly disagree on issues pertaining to the cognitive bases of metalinguistic awareness, on what constitutes metalinguistic awareness, and on the factors (both endogenous and exogenous) mediating and promoting the emergence and development of metalinguistic awareness and abilities in young children (cf. Bialystok, 2001:170). However, there appears to be some consensus that children have to reach a cognitive threshold before they can perform certain metalinguistic tasks and before they can develop an abstract understanding of language. In other words, the emergence and development of metalinguistic abilities are seen as concomitant with the child’s cognitive and linguistic development. According to Karmiloff-Smith (1992), for instance, the young child becomes increasingly skilled at manipulating language because language representations in the mind undergo transformations or redescriptions and thus become available for conscious inspection and reflection (17).
118
Multilingual Awareness
A number of researchers have argued that children only gain metalinguistic awareness proper through the acquisition of reading and writing skills, that is, upon entering school, while prior to that, children lack the capacity to abstract form from meaning. It has been intimated that very young children under the age of four, and in some cases not until the age of 7 or 8, are unable to make explicit, out-of-context judgements on language (Kamhi & Koenig, 1985:199). Various reasons have been adduced to account for the young child’s presumed inability to do so. Kamhi & Koenig (1985) identified the following. According to these authors children encounter difficulties because certain metalinguistic tasks require them to draw on processes that demand high levels of attention in order to access the specific linguistic knowledge tapped by a given task. In such tasks children are typically confronted with linguistic forms that are taken out of context and they are expected to know that language consists of discrete building blocks which can be assembled or taken apart and investigated in isolation (207). Young children, so the authors, find this a particularly daunting task because they tend to focus their attention on the meaning of an utterance rather than on the form (199). In opposition to this it is maintained that already pre-school children can be aware of aspects of language since they are capable of carrying out certain analytical operations on language (Bialystok, 1993, Fox & Routh, 1975, KarmiloffSmith et al. 1996 in Wehr, 2001:45; see also De Houwer, 1990:310). A number of studies have shown that “learners can be ‘aware’ of aspects of language without being able to explicitly articulate that awareness” (Jessner, 2006:53). As specified by Jessner (2006), “[m]etalinguistic abilites can be observed in children as young as two years of age when they are capable of self-corrections of word form and order and pronunciation, show concern about the proper word choice, pronunciation and style, and comment on the language of others” (44). Along similar lines Pinto (1999) asserts that the “acquisition of reading-writing skills at the onset of elementary schooling cannot be considered a principal factor in the development of metalinguistic abilities, but rather a consolidation and an expansion of metacognitive processes already activated” (22). Similarly, Slobin (1978) maintained that children as young as 2 to 6 are capable of carrying out a number of metalinguistic tasks including self-corrections and rephrasings, comments on the speech of others, explicit questions about speech and language, comments on their own verbal productions, and providing a response to questions about language (in Jessner, 2006:45). According to Pinto et al. (1999) the “span of metalinguistic development is much broader, progressive and at the same time structured into significant phases, than what has been explored or even hypothesized up to now. Its lower limit can be identified in those first significant behaviors which appear between the ages of 5–6 [. . .] Its higher limit is constituted by abstract formalizations [. . .]
Metalinguistic awareness in bilingual and multilingual children
119
which only a restricted category of adults (specifically, linguists) is capable of constructing. Between these two extremes there are intermediate forms of metalinguistic awareness” (37). From this perspective, metalinguistic awareness and abilities, conceived as dynamic aspects or components of language competence, can be seen as taking various forms on a continuum. In addition to the theoretical and conceptual inconsistencies outlined earlier, researchers have so far failed to converge on how to measure metalinguistic awareness and abilities in children (see below for more on metalinguistic testing) and on what precisely it is that permits the child to carry out metalinguistic operations. As noted by Jessner (2008a), “metalinguistic abilities still lack the necessary operationalization to be immediately verifiable” (277). The controversy is exacerbated by the fact that the constructs of metalinguistic awareness and abilities mean different things to different people. Depending on whether one adopts a narrow or wider perspective of metalinguistic awareness, the features subsumed under this notion may include anything from self-repairs at the very early stages of language acquisition to explicit knowledge of linguistic structures and functions in later years (see Wehr, 2001:36–37 and Pinto et al., 1999). As pointed out by Bialystok (2001), the first thing to do therefore is to decide “what constitutes metalinguistic ability” (170). A further question meriting closer investigation is whether awareness of linguistic form and function can promote the child’s language development or whether the directionality of effect is the reverse, or else, whether the effect is two-directional (cf. Wehr, 2001:27, Jessner, 2006:45, Jessner 2008:280 and Pinto et al., 1999:15). The answer to this question is bound to carry significant implications for the implementation of educational policies and school curricula. It is the author’s contention that the development of enhanced metalinguistic and metacognitive skills in learners of multiple languages will lead to the speeding up of language learning processes (cf. AllgäuerHackl & Jessner, 2013:132; Bono, 2011:46; Jessner, 1999:207), both, in formal and natural learning contexts. Metalinguistic awareness in learners of multiple languages is the focus of the next section.
9.4 Metalinguistic awareness in bilingual and multilingual children Aspects of metalinguistic awareness and abilities have not only been investigated in monolingual children but also in children growing up with more than one language. A substantial number of studies have reported clear cognitive advantages deriving from children’s bilingualism. Similar advantages have been found for tri- or multilingual children (although it is true to say that the paucity of research data makes it difficult to draw any definitive conclusions).
120
Multilingual Awareness
Bialystok (2001) suggests that “the experience of bilingualism might alter children’s developing conceptions of language” (170; cf. Ranta, 2008:206). This is in accord with Jessner (2008a) who postulates profound qualitative changes in third language acquisition and processing due to multilingual users’ enhanced level of metalinguistic awareness (276). It has been intimated that children’s early experience with diverse language codes may further not only their grasp of the abstract dimension of language but also signal to them “that there is a pragmatic dimension to word meanings as well as a semantic one” (Bialystok, 1992:510). Marsh & Hill (2009) note that there is “an increasing body of evidence about multilingual individuals in various cultural environments that points to enhanced functioning of these individuals when compared to monolinguals” (5). The authors state that “[k]nowledge of more than one language points to the expansion of specific types of competence. Multilingualism appears to help people realise and expand their creative potential. In addition, thinking, learning, problem solving and communicating, all of which are transversal knowledge-steeped skills used in our daily lives, show signs of enhancement through multilingualism”(23). There is compelling research evidence that linguistic awareness develops earlier in bilingual and trilingual children, and that children who are raised bilingual or trilingual demonstrate a more abstract understanding of language (Bialystok, 1992:510; see also Ben Zeev, 1977, Wehr, 2001, Barnes, 2006, and Pinto et al., 1999:31). Bilingual children’s enhanced performance on tasks which require them to focus both on formal and semantic aspects of language has been explained by way of their constant management of two or more language systems which commands the child’s continuous engagement in controlled cognitive processing (Bialystok, 1992, 2001). In bilingual and trilingual children the continuous negotiation of languages thus provides an added dimension to the concept of metalinguistic awareness as it extends across language borders. Therefore, the notion of multilingual awareness (which apart from crosslinguistic awareness also comprises awareness of meta-pragmatic and cross-cultural aspects of language) is proposed here to capture the multiple facets and the complexity and sophistication of the construct in multilingual speakers, because metalinguistic awareness in multilingual speakers differs in degree and nature from metalinguistic awareness in monolinguals (cf. Jessner, 2006:42; OtwinowskaKasztelanic, 2011:14; Gibson & Hufeisen, 2011:75; Bono, 2011:49). In particular, research has shown that bilingual children28 possess an early awareness of the arbitrariness of language (see Galambos & Hakuta, 1988:141) 28 The view is taken here that the same holds true for tri- and multilingual children. The reader will recall that the term ‘bilingual’ employed here renders the word choice found in the original literature.
Metalinguistic awareness in bilingual and multilingual children
121
and that they are advantaged in processing information which requires high levels of control of attention (Bialystok, 1992:507; see below). Literacy skills in two languages are held to additionally bolster bilingual children’s ability to manipulate language (see Thomas, 1988; Bialystok, 1991). With regards to the arbitrariness of language Cummins (1978) suggested that bilingual children have a greater awareness of the arbitrary nature of word-referent relationships. Cummins also found that the bilingual groups in his study exhibited a more flexible and analytic orientation to language input (148). The author concluded that bilingualism can enhance children’s metalinguistic awareness and “promote an analytic orientation to linguistic input” (131). Further evidence for bilingual children’s superior performance on tasks of metalinguistic ability was found by Peal & Lambert (1962), Ianco-Worral (1972), Ben Zeev, (1977) and, most famously perhaps, Bialystok (1991). Peal & Lambert (1962) posited that bilingual children possess an enhanced mental flexibility which is not specific to language but rather extends to all cognitive domains. In unison with Vygotsky (1962:110) the authors argued that this cognitive flexibility is derived from the bilingual experience of having two distinct codes to describe the surrounding world (cf. Bialystok, 1992:510). Ianco-Worrall (1972) reported that her South African bilingual subjects were more aware of the arbitrary nature of words than were her unilingual control group. Similarly, Ben Zeev (1977) found cognitive advantages in bilingual children, suggesting that the constant crosslinguistic interference between language systems requires bilingual children to resort to coping strategies which may accelerate their cognitive development. As outlined earlier, Bialystok (1991) proposes two components of metalinguistic awareness, analysis of linguistic representations and control of attention. The assumption is that different levels of analysis and control are involved when children resolve linguistic and metalinguistic problems (Bialystok, 1992:505). Bialystok argues that the acquisition of two or more languages enhances the control mechanisms involved in the processing of language (Bialystok, 2001:179; Wehr, 2001:61). According to the author, early experience with two distinct languages causes the child to “develop more sensitive means for controlling attention to linguistic input” (Bialystok, 1992:510). In other words, “early childhood bilingualism modifies children’s development of control of attention” and this accounts for bilingual children’s superiority in inhibiting distracting information as encountered in metalinguistic tasks (Bialystok, 2004:338). Galambos & Hakuta (1988) similarly suggest that bilingualism not only enhances “knowledge about the properties of language, but rather alters the approach toward language, from a message-oriented approach to a form-oriented one” (142). This accords with Jessner’s proposition that “linguistic objectivation is the multilingual’s most characteristic cognitive ability” (2006:42).
122
Multilingual Awareness
Multilingual children thus routinely pay attention to linguistic form as they have to attend to the speech stream in order to identify the language being spoken and in order to respond in the appropriate language. This requires high levels of control and attention. As stated above, the bilingual child’s constant negotiation of two or more languages is held to result in an enhanced ability “to selectively attend to appropriate cues and inhibit attending to others” (Bialystok & Martin, 2004:326). It is hypothesised that as children develop cognitively and linguistically, their mental representations undergo important modifications: increasingly abstract categories emerge in the child’s mind as formal and symbolic representations are formed and stored alongside conceptual representations (Wehr, 2001:57; but see Lowie & Verspoor, 2011 for a critical analysis of the notion of mental representations). These abstract and explicit representations allow the child to retrieve information or knowledge independently of any concrete contextual information (Bialystok & Martin, 2004:325). Linguistic knowledge can thus be consciously accessed, analysed and manipulated. Here, control of attention is hypothesised to play a crucial role as it is the process “by which attention is selectively directed to specific aspects of a representation, particularly in misleading situations” (Bialystok & Martin, 2004:325). The assumption of an enhanced control function is clearly in line with the DMM which posits an enhanced monitor for speakers of multiple languages. The EMM detects and repairs errors, performs the role of separator and cross-checker and draws on a pool of common resources as found in multilingual systems (Jessner, 2008a:276). The EMM forms part of the M-factor or Multilingualism factor through which the bi- or multilingual speaker is advantaged when learning an additional language (Herdina & Jessner, 2002:129). In summary, it can be said that bilingual children have consistently been found to exhibit higher levels of control (Bialystok & Martin, 2004:326). Their superior performance on certain metalinguistic tasks, such as grammaticality judgement tasks containing distracting semantic information (ibid.), has been ascribed to these children’s prior experience with and knowledge of language(s). Importantly, the languages of multilingual learners are seen as interacting in complex, dynamic and quite unpredictable ways (Jessner, 2008a:275). This interaction is held to have autocatalytic and accelerating effects on additional language learning and is said to result in the emergence of unique properties such as heightened levels of metalinguistic and crosslinguistic awareness (277). Metalinguistic awareness or the “conscious manipulation of and reflection on the rules of a language” (276) is defined as a “set of skills or abilities that the multilingual user develops owing to her/his prior linguistic and metacognitive knowledge” (Jessner, 2008a:275). Multilingual learners are predicted to attain high levels of metalinguistic awareness and ability due to qualitative changes which
Measuring components of metalinguistic awareness
123
obtain as a function of their prior language learning experience. The assumption is that prior language learning experience sensitises learners to the differences and similarities between different linguistic codes thereby changing the quality of language learning (Jessner, 1999:207). Crosslinguistic “awareness (tacit or explicit) of the interaction between the languages in a multilingual’s mind” (Jessner, 2008a:279) is considered a central component of the construct of multilingual awareness. It is hypothesised that crosslinguistic awareness increases with the number of languages spoken. Kemp (2001), for instance, found a positive correlation (in multilingual users) between grammatical awareness and number of languages spoken (in Jessner, 2006:62 and 2008a:278; see Gibson & Hufeisen, 2011 and Otwinowska-Kasztelanic, 2011 for similar results). Prior knowledge of languages has also been investigated by Andreou (2007; see chapter 1). The author conducted a study of trilingual school children in grade four and concluded that “the experience of three different languages is likely to result in enhanced awareness of the analysis and control components of language processing on the part of trilinguals” (12). In particular, Andreou reported that, compared to the bilingual control group, her trilingual sample exhibited a “heightened sensitivity to the phonological units of words” (8). Andreou’s study is particularly significant because it demonstrates that the benefits yielded by the trilingual experience surpass that of the bilingual, and this is precisely what Jessner (1999) implies as she states that “learning [. . .] a third language sets itself apart from learning a second language as the student can reflect consciously on the strategies used in SLA and apply them to TLA” (207). According to Jessner (1999), the “acquisition of a third language clearly differs from the acquisition of a second language because prior language learning experience changes the quality of language learning. This very often results in differing language strategies which the experienced language learner develops in contrast to the inexperienced one. This development of advanced cognitive skills in language learning can lead to the speeding up of the language learning process.” (207; see also Otwinowska-Kasztelanic, 2011:14 and Gibson & Hufeisen, 2011:83).
9.5 Measuring components of metalinguistic awareness Researchers have employed a variety of different tasks and tests to study metalinguistic skills in adults and children, both monolingual and bilingual. This section reviews some of the test instruments and tasks which have been utilised to measure metalinguistic and/or metacognitive awareness in children (Pinto et al.’s MAT III, delineated further down, is an exception insofar as it investigates
124
Multilingual Awareness
metalinguistic ability in adolescents and adults). According to Bialystok (1992), metalinguistic tasks can be classified for their demands on analysis of linguistic knowledge and control of processing (506; see above). As specified by the author, analysis of linguistic knowledge and control of linguistic processing (also referred to as executive control) are cognitive components or skills which are needed (albeit to differing degrees) to perform metalinguistic tasks. It is supposed that metalinguistic tasks in which “the solution depends on paying attention to some aspect of the language input that is not salient, not usual, or not expected” require the highest level of control of processing (Bialystok, 1992:505). Such tasks include, for instance, counting the number of words in a sentence, cloze tests, symbol substitution tasks such as proposed by Ben Zeev (1977), judgement of sentence acceptability but also more domain-general tasks such as the tower task or the embedded figures task (see below). The so-called c-test or cloze test29 is the task which is thought to pose the greatest challenge to control of processing and selective attention because, during task completion, subjects need to focus on both, form and meaning (Bialystok, 1992:509). In the symbol substitution task children have to substitute an item in a given string of words, thereby generating nonsense sentences (e.g., “They are good children” → “Spaghetti are good children”). The symbol substitution task but also the wordcounting task (which has children count the number of words in a given sentence) place high demands on subjects’ control of processing, i.e., on selective attention. The difficulty in these tasks consists in ignoring the most salient feature, namely the meaning of a particular word or sentence. This is also the case in Piaget’s sun-moon task (1929) which requires the child to determine what the sun and moon are called if they switch names, and which is then up in the sky at night. It is of course ‘the sun’, since sun and moon have only switched names. Instead of relying on the conventional meaning of the word, the child is (as in the above outlined symbol substitution task) required to “dissociate the word from its ususual meaning” (Bialystok, 1992:506). In so-called sentence judgement tests children are presented with grammatically correct sentences which contain a semantic error, such as in “Apples grow on noses” (ibid.). The children then have to decide if the sentence is grammatically acceptable or not. The difficulty here resides in the fact that although the sentence is semantically anomalous or “silly”, there is no syntactic violation. However, children frequently reject such sentences as unacceptable because they fail to ignore the most salient feature, i.e. meaning. Importantly, studies have shown that bilingual children perform higher on such judgement tasks 29 Cloze tests contain a number of gaps. Respondents have to complete the text by filling in the missing letters.
Measuring components of metalinguistic awareness
125
than monolingual children (Bialystok, 1992:506) and that bilingual children are better at solving high control problems (similar to the ones outlined above) than monolingual children. Bialystok found that bilingual children “are better able to intentionally control the focus of their attention on linguistic input than are monolingual children. They are less distracted by salient features of the input, such as meaning, and can more easily isolate specific aspects of the input, such as form, to arrive at various kinds of metalinguistic judgements” (1992:507). The underlying assumption is that monolingual and bilingual children “move in different cognitive worlds, experience different linguistic environments, and are challenged to communicate using different resources” (Bialystok, 2004:88). Bilingual children’s superiority in selective attention or control of processing is explained in terms of “their early experience with two languages” (Bialystok, 1992:510). According to Bialystok (1992), this early bilingual experience leads them to “develop more sensitive means for controlling attention to linguistic input” (510). More specifically, Bialystok argues that bilingual children “are used to hearing things referred to in two different ways, and this can alert them earlier to the arbitrariness of referential forms” (510). Bilingual children, so the author, are “used to attending to the language that is being spoken as well as the meanings, and this can help them attend to forms and to distinguish between relevant and irrelevant linguistic variation” (ibid.). Bilingual children typically “need to address different people in different languages, and this can enhance their flexible access to a variety of forms that have the same meaning” (ibid.). In addition, bilingual children “need to learn different conventions for marking such social relations as politeness, and this can signal to them that there is a pragmatic dimension to word meanings as well as a semantic one” (ibid.). As mentioned above, bilingual advantages for control of processing or selective attention also obtain for non-linguistic tests such as the Tower Test and the Embedded Figures Test (Bialystok, 1992:507,509). These tests are thought to engage very similar types of control processes as, for instance, the metalinguistic tasks outlined earlier. The Tower Test and the Embedded Figures Test do not focus on linguistic problems per se but on general problem solving. In the Tower Task children are shown towers made out of lego blocks and duplo blocks. The duplo blocks are of course much bigger. The children are told that each block is an apartment in which a family lives. The experimenters then build two apartment buildings out of the blocks and ask the children to count the apartements (i.e. the blocks) and say which building has more families living in it. The critical task has children compare a small lego tower with a tall duplo tower in which the higher duplo tower consists of fewer blocks than the lego tower. In this task height is the salient but misleading feature. Here, the children
126
Multilingual Awareness
need to ignore the height of the tower and instead focus on the number of blocks. As indicated by Bialystok (2005:422), bilingual children have been shown to solve this task significantly better than monolingual children. Similarly, in the Embedded Figures Task children need to isolate a specific aspect of the problem to arrive at the resolution. In this case, the child has to detect a particular shape inside a larger pattern. The Embedded Figures Task requires field-independent processing. Field-independent processing involves the ability to “break down a perceptual field into its components and to attend to individual parts of the whole” (Bialystok, 1992:507) while at the same time inhibiting those parts of the whole which are irrelevant. Children who succeed in the Embedded Figures Task are considered to be field independent (ibid.). Bialystok (1992) suggests that bilingual children are more field independent than monolingual children (508). In a similar way, Bialystok & Martin (2004) also argue that bilinguals have better inhibitory control for ignoring perceptual information than monolinguals do (325). One of the most comprehensive test batteries measuring metalinguistic awareness and abilities has been designed by Pinto and colleagues (1999). Pinto et al. (1999) have proposed three tests (MAT I, MAT II, MAT III) to identify and measure metalinguistic skills in different age groups (i.e. from 4-6 years, from 9-13 and in adolescents and adults). One central objective of Pinto et al.’s tests consists in defining “with greater precision the temporal limits within which these abilities are reached” (Pinto et al., 1999:109). Importantly, metalinguistic abilites in preschool children (4–6) are thought to comprise epi- and metalinguistic forms of awareness. The child is, at this stage, not expected to fully verbalise her or his motivations for producing a given answer or solution. However, as specified by Pinto, the task items in the MAT I focus on decontextualised signs or symbols which clearly require analysis and intentionality on the part of the child (40). It can therefore be assumed that certain epi-processes have, at this stage, been outgrown (Pinto et al., 1999:40). The MAT I requires the child to carry out word order corrections, evaluate word length, perform symbol substitutions and lexical segmentations, identify letters, words, numbers and rhyme and the function of written signs, such as punctuation, and to indicate the function of spaces between words in a text. In the MAT II, conceived for children aged 9-13, respondents are asked to identify and analyse the linguistic problem posed and to provide a possibly exhaustive disambiguation thereof. The focus is on children’s awareness of semantic and grammatical relations in sentences, children’s awareness of synonymy, ambiguity and anomaly, and their understanding of isolated phonemes, syllables and free lexemes. The MAT II presupposes a more profound analytical competence than is required for the completion of the MAT I and a minimum of familiarity with grammatical rules and
Measuring components of metalinguistic awareness
127
specific terminology which children generally acquire in their first years of primary school. It is therefore recommended to not administer the MAT II prior to age 9 as children may not yet have encountered the structures and technical terms required for the completion of this test. Primary schooling marks an important developmental transition in the child’s abstract thinking and abstractive abilities and in terms of their knowledge of the world. Birdsong (1989), for instance, posits that the onset of literacy and formal instruction significantly enhance children’s metalinguistic awareness (32). Similarly, Pinto et al. (1999) note that “the study of metalinguistic development in subjects of different ages following the first period of schooling shows how the exposure to metalinguistic terms can activate more abstract levels of consciousness” (14). This progression provides the basis for a more advanced understanding of language and more sophisticated metalinguistic performance (110). Later, with the onset of puberty, speakers develop a more profound understanding of metaphors, puns and riddles and of the arbitrariness of linguistic labels and the manipulative nature of language (110). In adolescents and adults metalinguistic awareness is associated with the ability to consciously reflect upon, play with or manipulate language, to comprehend figurative language, to judge the formal correctness and semantic and/or pragmatic appropriateness of an utterance, and, at the highest level, to employ specific terminology to denote and classify linguistic units. Hence, in the MAT III, the focalisation is on “a deeper assessment of the ability to comprehend a specific linguistic modality, figurative language” (118). In the MAT III respondents are evaluated on their capacity to identify and correct errors of a mostly morphosyntactic nature and on the ensuing justification of their solutions. In addition, subjects are tested on their grasp of figurative language including metaphors, advertising slogans and poetic language (114). The implicit assumption is that the “degree of elaboration of the kinds of metaphorical meanings present in an instrument such as the [. . .] MAT III is an indicator of [. . .] diversity in cognitive-linguistic styles” (118). Pinto et al.’s tests have been employed in various multilingual contexts throughout Europe and Canada. It should be noted, however, that the tests have thus far only been used to assess metalinguistic awareness in native speakers of Italian, English, and Basque/Spanish (see Lasagabaster, 2001b:316), etc. In other words, the tests have always been administered in the respondents’ L1. In order to measure metalinguistic awareness and abilities in multilingual children it might, however, be useful to compile a metalinguistic ability test of a different nature, designed to elicit participants’ awareness of linguistic forms and functions across languages. Such a test battery has yet to be developed. It will be an exciting and challenging task for future research to do so.
128
Multilingual Awareness
Part I of this book has given a comprehensive overview of a wide range of topics pertaining to bilingualism and multilingualism research. Incorporating such a wide spectrum of topics, it was hoped, would make it easier for the reader to gain an understanding of the complexities of multilingualism and multilingual development. In Part II, which follows next, the central theoretical issues and concepts introduced in the first part are taken up and related to the multilingual learning context of South Tyrol. The author links the notions and conceptions presented in Part I to her investigation of young children in highimmersion multilingual programmes, and reiterates the overarching claim of the book that early multilingualism and high-immersion multilingual education can confer huge benefits on young learners.
II A South Tyrol case study Multilingualism appears to effect substantial cognitive and linguistic changes in the speaker. Herdina & Jessner (2002:144)
Experimental findings in a multilingual setting 10 The Aim of the Study The broad aims of the present research study are to test previous findings about tri- or multilingual learners’ superiority with regards to metalinguistic abilities (Kemp, 2001; Jessner, 1999, Lasagabaster, 2001) and to verify claims or assumptions that highly experienced learners of an L2 are better and more successful language learners. An important goal of the study is to establish whether highimmersion multilingual education has positive effects on pupils’ linguistic and/ or cognitive development, in other words, whether there is anything to be gained from high-immersion multilingual education and from increased experience with an L2 and/or L3, and whether there are any specific skills or abilities that are enhanced through early high-level exposure to three languages in a formal setting. The study, which was conducted in South Tyrol, examines whether children in high-immersion multilingual educational programmes differ from children in traditional low-immersion education programmes with regards to their metalinguistic awareness and abilities and in terms of their mastery of L2 German and L3 English. In South Tyrol, the here described high-immersion multilingual programmes are commonly referred to as ‘bilingual’ programmes because pupils receive their education through the medium of two languages (approximately 50% of school subjects are taught in L1 Italian and 50% in L2 German30). L3 English is taught as a foreign language and is provided for 3 hours from 1st grade onwards. Following Llinares (2007), these programmes are here referred to as high-immersion multilingual programmes (as opposed to low-immersion programmes). Two groups of pupils in high-immersion multilingual programmes (n = 40) and two groups of pupils in low-immersion programmes (n = 44) participated in this study. In contradistinction to the high-immersion groups, the control groups in the low-immersion programmes had limited exposure to L2 German (6 hours per week). As regards exposure to L3 English, all the participants (those in the control and in the experimental groups) received 3 hours per week from 1st 30 These bilingual instructional programmes only exist for Italian schools. German schools have, so far, not been allowed to offer bi- or multilingual education. However, new legislation passed in summer 2013 is expected to bring about important changes.
132
The Aim of the Study
grade onwards, which is to say that they received the same amount of L3 English input. A competence test in L2 German and L3 English, and a metalinguistic abilities test (in L1 Italian) were administered to establish whether the groups differ in terms of their proficiency in L2 and L3 and in terms of their metalinguistic awareness. The data obtained were statistically analysed. The results of the statistical analyses revealed that there are significant differences in all three tests between the high-immersion experimental and the low-immersion control groups. As anticipated, respondents in the high-immersion groups performed higher on all the tests. The author is led to conclude that there are significant differences in how experienced learners31 acquire an L3 and in how they approach and process language(s) (cf. Stavans & Swisher, 2006:215). On the basis of these findings it is here argued that the respondents in the high-immersion groups possess a more highly developed metalinguistic awareness because they have at their disposal certain affordances which are not available to inexperienced and less (meta)linguistically aware language learners (cf. OtwinowskaKasztelanic, 2011:12–15; see also Aronin & O’Laoire, 2003:11 and van Lier, 1996:52).
31 In the present case the notion ‘experienced’ relates to those subjects who have received high levels of L2 input in the form of high-immersion multilingual instruction.
11 Historical background South Tyrol is the predominantly German-speaking region to the North-East of Italy. It is home to a German-speaking community of about 290,000 people, an Italian-speaking community of about 110,000 and a small Ladin community of about 18,000 speakers (cf. 2001 Census). The German community thus constitutes the majority group while the Italian and the Ladin community represent minority groups in the region. Depending on their own linguistic background, parents send their children to a German, Italian or Ladin school (see further below). As noted by Baker (2006), “[l]anguage cannot be divorced from the context in which it is used” (5). In this sense, South Tyrol’s language situation and policy can not really be understood if viewed in isolation from its complex historical and socio-political context. The ensuing section therefore takes a closer look at the history of South Tyrol. Prior to 1919 South Tyrol was part of Austria. With the end of World War I and the 1919 Treaty of St. Germain, South Tyrol was ceded to Italy. At the time 86 % of the population were German-speaking, 4% were Ladins and a mere 3% were speakers of Italian (Alcock, 2001:2). The rise of the Fascist regime under Benito Mussolini in the 1920s led to the italianisation of people’s and geographical names resulting in the replacement of German appellations and (private and official) signs or inscriptions with Italian translations. The German language was banned from public offices and educational institutions. German newspapers and radio programmes were strictly forbidden (3). The recruitment of large numbers of Italian teachers, officials and workers from the South of Italy was intended to accelerate the pace of assimilation. By 1939 Mussolini’s policy of italianisation had led to an increment of 22% in the Italian population in South Tyrol. In spite of everything, the German language survived in the homes and families and was secretly taught in the so-called “catacomb” schools which were held in private living rooms, cellars or barns. However, as Alcock (2001) expounds, the closure of German schools, the cessation of the use of the language in the public administration, the departure of Austrian officials and the difficulties in getting Austrian newspapers led to a break with Austrian Hochsprache and written tradition; only the oral tradition, Dialekt, survived in the home and in the valleys. The elimination of the South Tyrolese from the public administration and educational institutions had robbed their community of an administrative class and a generation of intellectual leaders [. . .]. Economically and socially the group was unbalanced, overwhelmingly concentrated on its poor alpine farms, while the Italian administrative and industrial classes, concentrated in the big towns, had been better paid and better housed. (6; see also Egger, 1977:20–23)
134
Historical background
The end of the Second World War and the demise of Fascism led the South Tyrolese to demand the reunification with the Austrian motherland. Their demand was not met. However, the 1946/47 Paris Peace Conference was to pave the way for an internationally recognised and constitutionally guaranteed autonomy for South Tyrol. The 1948 Autonomy Statute constituted an important step towards greater protection of the German- and Ladin-speaking communities. However, public perception was that it failed to deliver on the “parification” promised in the 1946 Paris Peace Agreement32. The 1948 Autonomy Statute stated that German could be used in all domains of public life and that all teaching was to be provided in the pupils’ mother tongue. However, German was not granted the status of co-official language (Alcock, 2001:9). In 1959 a provincial law made it mandatory for new recruits to provincial institutions to be bilingual (9). Candidates’ bilinguality was established and certified by the provincial authorities following the successful completion of an examination held four times a year (20). These examination procedures have recently been challenged before the European Court of Justice. As a consequence, in 1999, a number of modifications were undertaken both to the test procedures and to the test format (16). In 1969, following years of unrest and disputes, the Italian and Austrian Governments signed the so-called “Package”, a catalogue of measures aimed at reviewing and improving the 1948 Autonomy Statute. As Alcock (2001) notes, this second “Autonomy Statute as promulgated in 1972 gave the South Tyrolese much, [. . .], of what they had always wanted” (14). It has to be noted that few, if any, South Tyroleans would subscribe to Alcock’s claim that South Tyrol got what it demanded. Importantly though, article 99 of the New Autonomy Statute (1972) states that “Die deutsche Sprache ist in der Region der italienischen Sprache, die die amtliche Staatssprache ist, gleichgestellt“ (see Das neue Autonomiestatut, 14., Ergänzte Auflage 2009:101), that is, in the region the German language is parified with the Italian language, which is the official language of the State. A system of ethnic proportions, introduced in 1981 with the obligation to declare one’s linguistic origin, regulates the distribution of public offices and public housing (Scochi, 2011:20; Alcock, 2001:19). As specified above, everybody applying for or holding a public office is required to pass an examination (esame di bilinguismo or Zweisprachigkeitsprüfung) attesting one’s linguistic competence (Scochi, 2011:21). The certificate which one obtains following the successful completion of the exam is issued by the Provincial authorities (Alcock, 2001:20). It has been repeatedly criticised that the distribution of public posts occurs on 32 The Peace Agreement is widely referred to as the Gruber-De Gasperi Agreement, after the then Austrian Foreign Minister Gruber and the Italian Prime Minister De Gasperi.
Historical background
135
the basis of candidates’ linguistic origin and not on account of their professional competence. The same holds for the distribution of public housing where eligible persons need to meet the criterion of linguistic origin before anything else. It is not uncommon for people to declare their ethnic origin in such a way as to benefit maximally from this system (cf. Scochi, 2011:24). In other words, a native speaker of one language may declare to be a native speaker of the other language in order to obtain a public post, for instance. Even L2 or L3 teachers do, in order to get access to a teaching position, declare that they are native speakers of the required language, when in fact they are native speakers of the other language (ibid.). The obligation to indicate which linguistic group one belongs to has further been criticised on the grounds that there is a growing number of people who have grown up with two languages and find themselves in a position where they have to declare as one or the other, without being given the possibility to declare themselves as bilingual (ibid.). A third Autonomy Statute laid down in the course of a constitutional reform in 2001, has granted South Tyrol further competences and primary legislative powers in several public domains such as school buildings and school welfare, kindergarten and vocational training (Alcock, 2001:14).
12 The sociolinguistic context: Language education in a bi/trilingual community On account of the region’s agitated history it comes as little surprise that many South Tyrolese hold ambiguous views about multilingual education. Multilingual education is seen as a threat to the German-speaking population’s linguistic and cultural identity and is widely perceived as linguistic and cultural contamination or dilution (Egger, 1977:163). Egger (1977) warned that widespread bilingualism in South Tyrol would generate a mixed language variety (‘Mischsprache’) (162) which would be “sehr unangenehm anzuhören” (94) [which would be very unpleasant to listen to] and which would contain German dialectal and Italian elements. Such mishmash, as maintained by Egger, was to be discouraged (94,163). As early as 1945 the Italian law allowed for bilingual instruction in municipalities with mixed-language inhabitants. The law provided for the possibility “di impartire l’insegnamento paritetico nelle due lingue degli alunni” (to impart bilingual education in the pupils’ two languages) from 4th grade onwards “nei comuni in cui la popolazione fosse mistilingue” (in mixed-language communities) (Egger, 1977:144; cf. Decreto Legislativo Luogotenenziale 27 Ottobre 1945, n. 775, art. 3; translations by this author). However, the necessity to perfect pupils’ competence in German was deemed incompatible with the idea of an integrated school. As a result the proposition of a bilingual model afforded by law 775 was disregarded by the German school authorities. In a similar way, in 1948, the Longon primary school in Bozen was denied the introduction of a bilingual experimental class for which no less than 70 pupils had already enrolled. The local newspaper Dolomiten, catering predominantly for the Germanspeaking community, openly supported the officials’ decision arguing that bilingual instruction would mean “die Zerstörung der Schule in der eigenen Muttersprache”, i.e., the end of mother tongue instruction (translation by this author) (Scochi, 2011:17). This echoes Article 19 of the 1972 Autonomy Statute and the corresponding executive measures of 1983 (DPR33 89/1983) which state that In der Provinz Bozen wird der Unterricht in den Kindergärten, Grund- und Sekundarschulen in der Muttersprache der Schüler, das heißt in italienischer oder deutscher Sprache, von Lehrkräften erteilt, für welche die betreffende Sprache ebenfalls Muttersprache ist. [In the Province of Bozen instruction in kindergarten, primary and secondary schools is provided in the pupils’ mother tongue, that is, in Italian or German, by teachers for whom the respective language is also their mother tongue; translation by this author.] 33 Decreto del Presidente della Republica.
The sociolinguistic context: Language education in a bi/trilingual community
137
Executive measures issued in 1988, but still in place today, aim at averting pedagogical disruptions caused by pupils with insufficient linguistic competence in the language of schooling. In other words, if a pupil does not adequately master the language of the school at which he or she is enrolled, a joint commission consisting of an equal number of representatives from the two linguistic groups (i.e. German and Italian34) will decide whether the child can stay on or has to change school. This way, it is argued, unilingual instruction will not be undermined by the presence of non-native pupils (Scochi, 2011:21; Alcock, 2001:20). Article 19 of the Autonomy Statute also states that the L2 is to be taught from second or third class onwards. It is further specified that the teaching of the L2 is to be provided by native speakers of that language. This also holds for the teaching of the L1 (see above). In addition, as of 2004, language teachers are required to sit a language examination in order to attest their linguistic competence in the language of the school at which they teach (Scochi, 2011:23). A provincial law passed in 2004 decreed the anticipated introduction of L2 Italian in German-language primary schools. The law allotted the Italian language a minimum of 1 hour and a maximum of 2 hours per week in the first grade curriculum (ibid). In addition to the above adduced socio-political objections, bilinguality has generally been regarded as detrimental to pupils’ linguistic and cognitive development. Egger (1977) stated that “[w]enn Kinder in den anderssprachigen Kindergarten geschickt werden, wird ihnen die Gelegenheit geboten, die zweite Sprache zu lernen; genommen wird ihnen aber die Möglichkeit, ihre erste Sprache im Kindergarten weiterzuentwickeln” (93) [When children attend kindergarten in the other language they may be given the chance to learn a second language but they are also robbed of the opportunity to develop their first language in kindergarten; translation by this author]. In a similar way, Egger voiced strong concerns regarding the increasing number of Italian children attending German schools to improve their L2: Die deutsche Sprachgruppe ist mit dieser Entwicklung aus mehreren Gründen nicht einverstanden. Dadurch wird nämlich den deutschen Kindern das vom Autonomiestatut gewährte Recht auf den deutschen Unterricht geschmälert, weil so das Italienische zum Teil Gebrauchssprache in der Schule wird. Es wird auch der Unterrichtserfolg vermindert, weil Kinder mit ganz unzureichenden Kenntnissen der Unterrichtssprache auch die anderen Kinder aufhalten (96). [The German-speaking community do not agree with this development for a number of reasons. Firstly, German children are curtailed the right to German instruction as guaranteed by the Autonomy Statute, because this will entail the increased use of Italian in school. Secondly, pupils’ academic success will suffer as a consequence of the Italian children’s inadequate mastery of the language of instruction; translation by this author.] 34 The Ladin school system is not considered here.
138
The sociolinguistic context: Language education in a bi/trilingual community
Egger thus concluded that either the German or the Italian pupils will miss out since schools can not cater for both, German and Italian children, at the same time (98). Arrangements of this type, the author assessed, are therefore best abandoned. To lend support to his claims, Egger referred to a number of scholars who had all identified negative effects of bilinguality on the development of children’s L1 (146) on the one hand, and negative effects on their psychological well-being on the other (151). Egger noted that “Untersuchungen an zweisprachigen Schulen haben aber gezeigt, daß die starke Förderung einer zweiten Sprache fast immer mit Verzögerungen in der Entwicklung der Muttersprache gekoppelt ist“ (146). [Empirical studies in bilingual schools have shown that the promotion of a second language is almost always accompanied by a delayed development of the mother language; translation by this author.]
To substantiate his claim that bilingualism carries negative implications for children’s psychological well-being, Egger related to Francescato who reported neurotic disturbs in children growing up in bilingual environments [“neurotischen Störungen bei Kindern, die in einem zweisprachigen Milieu aufwachsen”]. A similar view was propounded by Diebold (1968) who noted that bilingualism could cause “emotional maladjustment” and “psychodynamic conflict” (in Genesee, 2003:206).
13 Multilingual Educational Programmes in South Tyrol As competence in two or more languages is taking on increased value, many countries around the world are eager to adopt new forms of language instruction (immersion or multilingual programmes in Canada, Hong Kong, the Basque Country, Belgium, Luxembourg, etc.). For various, but especially for sociopolitical and historical reasons, the integration of second and foreign languages into the curriculum presents a special challenge in bilingual or trilingual communities. It seems curious that in South Tyrolean schools there is strict segregation of linguistic groups. German families have traditionally been sending their children to German-speaking kindergarten and school while Italian-speaking families have sent their children to Italian institutions. Mixing has not been encouraged by either the school authorities or the political class. Immersion programmes, bilingual education, or Content and Language Integrated Learning (CLIL) have generally been viewed with scepticism as they are deemed incompatible with Art. 19 of the Autonomy Statute. The author may be forgiven for intimating that rethinking second and foreign language teaching policies at all levels of schooling in South Tyrol is not merely desirable but long overdue (see also Final recommendations in Chapter 21). In spite of the many controversies surrounding bi- and multilingual forms of instruction, there has, in the past few years, been increasing parental pressure to develop bilingual and trilingual education programmes. In March 2008 a sizable number of Italian and migrant families in Bolzano/Bozen demanded permission to register their children at the German kindergarten. To calm the waters and to preempt a large influx of non-German children into German schools, the German ruling party SVP decided to grant Italian kindergartens and primary schools in the region greater autonomy in their language teaching policies (Scochi, 2011:48). This newly-gained freedom in curricular decisionmaking can today take the form of team-teaching in kindergarten (in which case an Italian and a German teacher are present) and of bilingual or trilingual education models at the primary and lower secondary level. First attempts to increase L2 teaching hours in Italian schools and kindergartens had been undermined by the SVP in 1992/1993 (53). Today, Italian elementary schools are free to teach up to 50% of the total hours in the L2 (54). Several Italian primary schools in South Tyrol now offer multilingual education with a minimum of 11 and a maximum of 14 hours of L2 German instruction (plus 1 to 3 hours of L3 English). A number of other (Italian) schools have adopted similar models but they provide fewer hours of L2 tuition (8.5 to 10 hours). Assessment of performance in
140
Multilingual Educational Programmes in South Tyrol
these programmes is undertaken by the Italian teachers. German teachers are not permitted to mark pupils’ performance (in subjects other than L2 German) since this would violate Art. 19 of the Autonomy Statute (55). As for the teaching of L3 English, all Italian primary schools have, since 2007/08, been offering a minimum of 1 hour of English from first grade onwards (58). In most schools English is taught for at least 2 or even 3 hours per week. Until a few years ago, the regional government had been stressing that these multilingual programmes were to be seen as an experiment (with a beginning and an end) rather than a gateway to the institutionalised implementation of multilingual education. To reinforce their position, politicians and officials liked to have recourse to Art. 19 of the New Autonomy Statute which states that in the province of Bolzano/Bozen all teaching is to be delivered in the pupils’ mother tongue. In consequence, multilingual education programmes are still the exception in South Tyrol, not the rule35 (even in Italian schools). The local government, in particular the Freedom Party and the Union for South Tyrol, but also the majority party SVP, are opposed to the implementation of integrated schools (this is particularly true for German schools). In Canada, where French and English co-exist, Genesee (1987) criticised that the politically imposed “Englishonly legislation continues to plague the acceptance of diversity in our own country” (in Robinson, 1998:42). An analogy with the South Tyrolean context clearly suggests itself.
13.1 Subject Matter Instruction in L2 German An important consideration needs to be made here with regards to the problems encountered by native speakers of Italian intent on learning L2 German in South Tyrol. The German spoken by the South Tyrolean community is a dialect which differs considerably from the standard German spoken in Germany. In South Tyrol standard German is only utilised in the written mode, in the school context, by the media or in very formal settings. In everyday communicative exchanges Italian learners of L2 German will almost exclusively be confronted with local dialect forms (which vary from town to town and from valley to valley). This discrepancy between written and spoken German renders acquisition of L2 German rather difficult for Italian speakers. As a consequence, the Italian school authorities have, for some years now, been aiming to raise the number of L2 teaching hours by promoting the introduction of bilingual education 35 This does not apply to Ladin schools where the teaching of Ladin, German, Italian and English can take various forms.
Subject Matter Instruction in L2 German
141
models at the primary and secondary level. In German schools, similar undertakings have thus far been forestalled by the majority party SVP (Südtiroler Volkspartei) who, as representatives of the German-speaking minority in Italy, are determined to hold on to mother-tongue instruction. Recent legislation, passed in July 2013, however, is bound to bring about some significant changes. The new regulation (BLR36 1034) allows for the temporary realisation of bilingual projects in the form of CLIL (Content and Language Integrated Learning) from the elementary to the upper secondary level in German schools. In addition to the above outlined difficulties regarding the characteristic features of German as spoken in South Tyrol, there is some disagreement amongst educational authorities and lay persons as to the efficiency of teaching certain contents or subjects in a language which is not the pupils’ native tongue. The most prominent case is that of mathematics. It has been argued repeatedly that teaching geometry and arithmetic in a non-native language may negatively affect the learning process and delay children’s cognitive development. However, empirical evidence seems to imply that this scepticism is unfounded. Bournot-Trites & Reeder (2001), for instance, examined the effects of teaching mathematics in L2 French on pupils’ achievement in mathematics. The authors administered a mathematics test to two groups of French immersion students. The test was administered in L1 English. The authors found that their test group who received 80% of the core curriculum (including mathematics) in L2 French outperformed the control group who received only 50% of the curriculum in French and was taught mathematics in L1 English. The authors noted that higher intensity in the form of increased time per day of instruction in the second language was associated with an increase in French academic proficiency level. Bournot-Trites & Reeder argue that high levels of exposure and proficiency in L2 French enhance pupils’ understanding of mathematical concepts taught in that language. They conclude that “increasing the intensity of French in a French immersion program by teaching math in French is strongly associated with a positive effect on mathematics achievement” (40). According to Bialystok (2002) the symbolic nature of numbers resembles the symbolic function of words (157). Following Bialystok, one could therefore speculate that the handling of two or more language codes and sets of linguistic symbols will enhance children’s grasp and understanding of number concepts (see Bialystok, 2005:419-422 for a detailed discussion of mathematical abilities in bilingual children). This also seems to be in line with Robinson (1998) who reports a number of studies all evidencing positive effects of bilingualism on pupils’ performance in mathematics (39, 40). 36 BLR = Beschluss der Landesregierung
142
Multilingual Educational Programmes in South Tyrol
At the Longon and Manzoni schools, the teaching of arithmetic or mathematics, has been the object of an ongoing debate and controversy. Over time, two major positions have emerged. On the one side the propounded view is that the teaching of mathematics must be delivered in the pupils’ native tongue (i.e. Italian). In the other camp, the prevailing conviction is that it shouldn’t really matter whether the discipline is taught in children’s L1 or L2. No common ground has as yet been found on this issue. However, both schools have come up with viable solutions that have been more or less accepted by all the parties involved, i.e. school authorities, principles, teachers and parents. The Longon primary school has chosen to teach mathematics in German, which the reader will recall, is most pupils’ L2. Pupils have a German-language textbook and acquire the specific language of arithmetic in their L2. During the maths lesson, the Italian teacher is also present in class and helps whenever children encounter difficulties with the content or language. In addition, on certain days, the class is divided into two groups and while one group practises German or music with the German teacher, the other group does maths with the Italian teacher. After an hour the groups switch. At the Manzoni school, by contrast, mathematics is taught in (what is most) pupils’ L1 Italian. Teachers and pupils work with an Italian textbook. In a special maths workshop the children recapitulate the same and/or cover new contents in their second language, German. Hence, while most of the teaching of mathematics is delivered in Italian, pupils are also familiarised with the specific German terminology. The issue as to whether the teaching of mathematics ought to occur in pupils’ L1 or L2 remains as yet unresolved. It will be a task for future research to examine whether teaching maths in the L2 is conducive to the learning process or whether, instead, it interferes with pupils’ academic progress. A third possibility, a null or neutral effect, must also be considered. As regards the teaching of Physical Education and Drawing in L2 German, problems of a slightly different nature from those afflicting the maths classroom have been identified: critics argue that the instructional discourse during the sports and drawing lessons is largely one-way, that pupils just follow the instructions of the teacher and that they do what they are told without contributing anything linguistically. In other words, it is being intimated that the pupils’ role during these lessons is that of the passive receiver and that therefore teaching PE and Drawing in the L2 will not benefit the children in any significant way. This objection is, of course, unfounded, since it is generally acknowledged that increased input (even when there is little output) can promote learners’ mastery of the L2.
Multilingual education and L1 and L2 development
143
13.2 Multilingual education and L1 and L2 development A concern frequently voiced by parents, teachers and education authorities is that the teaching of two or more languages at an early age may interfere with children’s cognitive growth and development bringing forth speakers who do not master either of their languages adequately. In particular, parents and teachers are often worried that multilingual education may have detrimental effects on the development of the child’s first or dominant language. However, a large body of research has evidenced that this is not the case (see Hobbs, 2012:204; Bono, 2011:30–1). Instead, bi- or multilingualism has been found to enhance not only learners’ first language (cf. Bournot-Trites & Tallowitz, 2002:11,30–31; De Angelis & Jessner, 2012:62) but also their general cognitive capacities and their linguistic awareness (Kecskes & Papp, 2000:29; Bialystok, 2005; Bleyhl, 2005:60; Bono, 2011:49; Gibson & Hufeisen, 2011:75). The author joins Cummins (2003) in proposing that no detrimental effects on the pupils’ L1 need to be feared when school subjects are divided between or delivered in two or more languages. Instead, as indicated by Cummins (2003), [m]ore than 150 research studies conducted during the past 35 years suggest that when children [. . .] develop their abilities in two or more languages throughout their elementary or primary school years, they gain a deeper understanding of language and how to use it effectively. They have more practice in processing language, especially when they develop literacy in both, and they are able to compare and contrast the ways in which their two languages organise reality. The research also suggests that bilingual children may also develop more flexibility in their thinking as a result of processing information through two different languages. (61–2; cf. Kecskes & Papp, 2000:81)
As for young learners’ competence in L2 or L3 it is important to bear in mind that a distinction needs to be made between basic communicative skills on the one side and academic language skills on the other. BICS (basic interpersonal communicative skills) are activated in context-rich communicative situations where there is additional, mostly non-linguistic, support in the form of gestures, facial expressions, head nods, pointing to objects and other contextual clues. CALP (cognitive/academic language proficiency), in contrast, is required in context-reduced academic settings and presupposes higher-order thinking and abstractive skills. Little or no contextual information is available, and language is said to be ‘disembedded’ from any meaningful context (Baker, 2006:174; Cummins, 1991:70–72; Cummins & Swain, 1986: 153–156; Cummins, 2001:67– 68). CALP is cognitively and academically much more demanding than BICS. As a consequence, it takes the child longer to master CALP than to tackle everyday communicative situations. Cummins explicates that it generally takes the
144
Multilingual Educational Programmes in South Tyrol
child about two years to attain context-embedded fluency in the L2. In contextreduced settings, by contrast, the child takes anything between five to eight years to acquire the same level of proficiency as a monolingual (Baker, 2006:178). Educators need to be aware of this because children who have not yet moved beyond the context-embedded stage may not be ready to engage in higher-order cognitive processes such as synthesis, analysis, interpretation, etc. (179) and may appear to lag behind in their cognitive development when in reality their L2 competency has just not yet developed to a stage which allows them to cope with the demands of the curriculum (177). In consequence, as predicted by Cummins (2000), there may be a temporary lag in achievement when the curriculum is taught through the L2 (Baker, 2006:172).
13.3 Riflessione Lingua or Reflection on Language The last few decades have seen scholars increasingly call for cross-linguistic approaches to foreign language teaching (Jessner, 1999:207; Jessner, 2006:127; Jessner, 2008:15-45; Otwinowska-Kasztelanic, 2011:15; Bono, 2011:49; Hobbs, 2012; Hufeisen, 2005,2011; Krumm & Reich, 2013; Allgäuer-Hackl & Jessner, 2013). In 1974 Hawkins first put forward his proposal of ‘Language Awareness’ as a new school subject (see chapter 8). Hawkins conceived of ‘Language Awareness’ as a “bridging subject” which should enable learners to meaningfully link and contrast their native tongue with foreign language(s) (Hawkins, 1999:124; James, 1996:142). The notion of language awareness as a school subject in its own right slowly began to take ground. Two decades later, James (1996), echoing Hawkins’ plea, postulated that within-language and cross-language associations and comparisons should become the new content and focus in the language classroom (147). Recognising and exploiting one’s knowledge of language(s), so the author, will “provide a unified ‘whole-language’ reflective experience” (142) enabling learners to transfer their knowledge and skills to the foreign language context. More recently, a similar proposal has come from Jessner (1999). Jessner emphasises the fundamental importance of comparative learning, the focalisation on commonalities between different language systems and the integration of specific language learning strategies (207). According to Jessner, a cross-language approach entails conscious reflection on form and function, contrastive learning and building on (an) already (more or less) established language system(s) so as to bring linguistic features, which might otherwise go largely unnoticed, to the learner’s attention (cf. Hufeisen, 2005, 2011; Allgäuer-Hackl & Jessner, 2013; Schöpp, 2013).
Riflessione Lingua or Reflection on Language
145
While language teachers at both the Longon and the Manzoni school take great care to draw pupils’ attention to linguistic forms and structures, the Manzoni school has gone the extra step further by introducing a novel school subject which they called ‘Riflessione Lingua’ (Reflection on Language). During this lesson, which is held by the Italian and German teacher together, the focus is on the exploration of L1 and L2 linguistic structures. Pupils are encouraged to reflect on language form and engage in cross-linguistic comparison. This accords with what Hawkins had in mind when he called for “more attention to cross language comparisons and more talk about language [. . .] assess, and encourage, pupils’ growing awareness of how the foreign language compares with the mother tongue and how foreign languages are learned” (1999:140). One of the main challenges of such language awareness classes resides in the careful planning and the strategic use and management of two (or more) languages in one single lesson (cf. Baker, 2006:298). In future, the Manzoni school might want to consider involving the L3 in their ‘riflessione lingua’ class so as to engage in a truly multilingual analysis. This desideratum appears supported by Clyne (2003) who states that the “transfer of metalinguistic awareness and intrinsic motivation across languages may benefit both language maintenance and the subsequent acquisition of other languages” (44).
14 Research Questions and Hypotheses Previous research suggests that the early acquisition of second or third languages may bolster children’s cognitive development by sharpening their awareness of structural properties and their understanding of the arbitrary nature of language (Diaz, 1983:48; Lasagabaster, 2001b:313). With this in mind, the present study aimed to ascertain the role of multilingual education and the effects of multilinguality on subjects’ linguistic and metalinguistic performance. Metalinguistic awareness is here construed as “an influential cognitive component which shows its effect on the way multilinguals use and learn their languages” (Jessner, 2006:117). Importantly, awareness of language is also held to affect, if not determine, “the organisation of the multilingual mental lexicon because levels of awareness show influence on the activation of the individual languages in multilingual production” (116), and, according to Singleton & Aronin (2007), the higher the level of language awareness “the more effectively languagerelated possibilities are likely to be perceived and capitalised upon” (85). In light of the central role that (meta)linguistic awareness is known to play in language learning (Galambos & Hakuta, 1988; van Lier, 1996, Rampillon, 1997; Rahimpour, 2002; Herdina & Jessner, 2002; Jessner, 2006, 2008, 2008a, 2008b; Oberhofer, 2008; De Angelis & Jessner, 2012; Hobbs, 2012) the study of metalinguistic awareness and abilities in trilingual school children seems a worthwhile and valuable contribution to the field of multilingualism research. As indicated, the concern of this research study was two-fold. First, the aim was to establish whether children who are exposed to extensive L2 (and L3) input in an instructional setting attain higher levels of competence in these languages than children with limited L2 input (but same amounts of input in L3). Second, the study aimed to examine whether this early multilingual experience impacts young learners’ metalinguistic awareness and abilities. For the purposes of the present study the following main research questions were formulated: Research Question 1: Do children in high-immersion multilingual educational programmes differ from children in low-immersion programmes in terms of their metalinguistic awareness and metalinguistic abilities? Research Question 2: Do children in high-immersion multilingual educational programmes differ from children in low-immersion programmes in terms of their mastery of the L2 and L3?
On the basis of previous research (De Angelis & Jessner, 2012; Lasagabaster, 2001b; Cenoz & Valencia, 1994; Hakuta & Diaz, 1985) the author hypothesised that
Research Questions and Hypotheses
147
Hypothesis 1: Early multilingualism, as found in children attending high-immersion multilingual education programmes, carries positive implications for the child’s linguistic and metalinguistic awareness.
and that Hypothesis 2: Early and extensive contact with an L2, as provided by high-immersion multilingual education programmes, carries positive implications for the acquisition of additional languages.
15 Participants The participants in the present study were 84 primary school children drawn from the Longon and Manzoni schools in Bolzano/Bozen, South Tyrol. At both schools teaching is traditionally delivered in Italian. The schools are therefore best classified as Italian-language schools as opposed to German-language schools where tuition is provided in German. In 2005, after years of lengthy and controversial debates, the green light was finally given for the institution of a multilingual class at the Manzoni elementary school. A year later, the Longon elementary school followed suit. Parental interest in multilingual schooling has been high ever since. However, only a limited number of new pupils can be admitted at each of the two schools every year. Of the 84 subjects participating in the present study 44 children attended traditional Italian tuition classes while 40 received multilingual education. All 84 subjects were enrolled in their fourth year and were on average 9 years old (the author is not able to indicate pupils’ mean age as one of the participating schools did not wish to provide the pupils’ birthday). The gender distribution can be described as balanced with 44 female and 40 male subjects taking part in the research project. It should be noted that most of the subjects in the experimental group, i.e. in the multilingual classes, came from bilingual families and were thus exposed to two languages (i.e. Italian and German) on a daily basis. Nevertheless, most of these children were dominant in Italian and had a marked Italian accent when speaking German. As already mentioned, the German commonly spoken in South Tyrol is a dialect variety (not very dissimilar from some Austrian varieties) which differs considerably from the standard German employed in Germany. A small number of subjects (4 in total) had a discernable standard German accent (typical of Germany) when speaking German. From this the author inferred that one of these pupils’ parents must be of German origin. This was confirmed by the data obtained through the background questionnaires distributed to the respondents’ families. The questionnaire sought to elicit information on the children’s (and their parents’) language biography, in particular on the children’s use of each language in- and outside of the home and school, their language proficiency and dominance, their mixing habits and their attitudes towards their three languages. It emerged that a handful of children in the experimental and in the control groups had come into contact with some Spanish (during a 5-week family holiday), Ladin (through the mother, who sometimes used her native language with her child), French (through the mother, who used to sing to the child in her native French), Chinese (immigrant
Participants
149
family), Urdu (immigrant family) or Russian (the child in question was adopted and said he did not recall any Russian). The 84 participating subjects were grouped into four primary school classes, two of which constituted the control groups and two of which made up the experimental groups. The two control classes were taught all the school subjects in Italian, with the exception of L2 German (6 hours per week) and L3 English (3 hours per week). Both, the L2 and L3 were taught from first grade onwards. The two experimental classes, by contrast, received 50% of the core curriculum in L2 German plus a total of 3 hours L3 English per week. All participants had acquired their L3 in the school context consecutively to the development of their L1. A small number of pupils had come into contact with some English in a kindergarten playgroup. Children in both, the experimental and the control groups used mostly Italian among themselves. However, during German lessons pupils were expected to use German. This was particularly the case in the high-immersion classes where pupils were strongly encouraged to address both, the teacher and their peers, in German. Of a total of initially 85 pupils, one child (in one of the experimental groups) had to be discounted since her parents did not grant her permission to participate in the study. In both, the experimental and the control groups, most children’s dominant language was Italian. In the background questionnaire some parents indicated that their child is raised bilingually or trilingually. However, it turned out that the parents were referring to their child’s multilingual schooling and not to the languages spoken in the home and family. In such cases, the information elicited through that particular question was deemed unreliable. Similarly, some of the questions in the background questionnaire were found to elicit very diverse responses and at times inconclusive or even contradictory information. Interviews with teachers, pupils or parents helped to clarify most points at issue.
16 Theoretical and methodological framework and study design As has been previously stated, the main objective of the present study is to ascertain the role of multilingual instruction and the effects of early and extensive contact with an L2 on third language learning and on children’s metalinguistic awareness and abilities. For this purpose the following three (‘pencil and paper’) tests were administered, each in a different language. An adapted, viz. abbreviated version of Pinto’s Metalinguistic Ability Test 2 (henceforth MAT-2) was utilized to measure subjects’ metalinguistic awareness and ability. The MAT-2 was administered in Italian which is most subjects’ dominant language. A general competence test in German measured pupils’ mastery of their L2 German and an English test examined subjects’ knowledge of and performance in L3 English. For the completion of the MAT-2 test subjects were given a total of 60 minutes. The German and English tests had to be completed in 45 and 30 minutes respectively. For practical reasons the original MAT-2 had to be abbreviated so as to allow pupils to complete the test in an hour’s time. The abbreviated version consisted of 4 (instead of the 6 original) sections including Acceptability, Ambiguity, Grammatical Function and Phonemic Segmentation (this latter was subdivided into three further subsections assessing syllable scansion, identification of repeated phonemes and formation of words). The Comprehension and Synonymy sections included in the original version of the MAT-2 did not form part of the test battery utilised in the present study. Whether or not this had any influence on the test results can at this point not be established. It must be noted that while the original MAT-2 is a standardised test and as such highly reliable (cf. Pinto et al. 1999 and Lasagabaster, 2001b), the abbreviated version employed here may be less so. As mentioned above, participants also performed an L2 German and an L3 English test. The German test included a listening comprehension, a grammatical judgement task, a task on verbal inflections and a reading comprehension. The English test featured a reading/listening comprehension (i.e., a short text was read to the subjects while they followed the text on the test sheet in front of them), a matching exercise on everyday routine activities, a true/false task on prepositions and a sentence completion test. An overview of the single test items is presented in the table below:
The MAT-2
Test
Task
MAT- 2
Acceptability (LQs + MLQs)
MAT- 2
Ambiguity (LQs + MLQs)
MAT- 2
Grammatical Function (LQs + MLQs)
MAT- 2
Phonemic Segmentation: – Syllable Scansion (LQs), – Phoneme Repetion (LQs), and – Word Formation (LQs + MLQs)
German
Listening Comprehension
German
Grammatical Judgement Test
German
Verbal Inflections
German
Reading Comprehension
English
Reading/Listening Comprehension
English
Matching Exercise
English
True/False Test
English
Sentence Completion Test
151
Data collection took place in March and April of 2011 and was preceded by interviews with the principals and teachers at both institutions as well as classroom observations during the autumn of 2010. The author is aware that while advocating a multilingual approach to language learning and development, the instruments utilised for this research follow, to some extent at least, traditional monolingual paradigms. This is, in part, due to the fact that no suitable instruments for testing multilingual proficiency and multilingual awareness were available at the time the study was carried out and may be considered a shortcoming of this study. This said, the inclusion of a metalinguistic abilites test goes a considerable way towards a holistic multilingual approach. In addition, there was an important reason for choosing monolingual-based instruments for the purposes of the study, stemming from the author’s intent to evaluate the three languages separately so as to show whether high exposure to L2 impacts pupils’development of L1 and L3.
16.1 The MAT-2 The MAT-2 is one of three tests devised by Pinto and collegues to assess metalinguistic ability in individuals from different age groups. The Metalinguistic
152
Theoretical and methodological framework and study design
Ability Test 2 measures metalinguistic ability in children aged 9–13 by evaluating their reasoning abilities (Pinto, et al. 1999:109) and by focusing on “the type of conceptualization which the subject reveals through the justification of his answer” (54). The MAT-2 is based on D. Hakes’ test battery (1980) as outlined in The Development of Metalinguistic Abilities. Drawing on Piaget, Pinto et al. (1999) distinguish between epi-processes and meta-processes (cf. Gombert, 1992 for a similar proposition). The former reflect an enhanced intuition with regard to the use of language. The latter, in contrast, relate to the intentional reflection on language or the deliberate attention directed towards language. This distinction between epi- and meta-processes is reflected in the MAT-2 test design where subjects are required to attend to two distinct types of questions and elaborations, one epilinguistic and the other metalinguistic in nature (Pinto, et al. 1999:54). In other words, the test employs two different sets of questions and answers. The first type of questions, called Linguistic or L questions, examines pupils’ general knowledge of rules of language use. The second type of questions, termed Metalinguistic or ML questions, seeks to elicit a more or less elaborate and sophisticated explanation of why a particular L answer was given. This second type of questions indicates the extent to which the subject is able to analyse, classify and explain the use of a given linguistic item or structure. Hence, while the focus in LQs lies on general knowledge of language usage, MLQs clearly go beyond the epilinguistic form of elaboration required in the L question, i.e., MLQs examine the subject’s ability to consciously and/or deliberately focus on linguistic form rather than on meaning only. As specified by Ranta (2008), explaining grammar rules, as in the case of MLQs, is one of “the most demanding of metalinguistic tasks” and requires specific metalinguistic knowledge and “technical terminology like noun and verb” (205). Because of the strong and unequivocal metalinguistic focus of the MAT-2, Pinto et al. (1999) recommend to utilise the test for subjects aged nine and above (55). This has been one of the motivations for the author of this volume to study 4th graders (who are all nine years old). Empirical data seem to indicate that very young children, below the age of 5 or 6, while possessing some sort of epilinguistic awareness lack the fully developed metalinguistic awareness and terminology required for the completion of tasks such as those in the MAT-2 (but see van Lier, 1996:74 for a different view). In addition, given that one of the objectives of this research study was to ascertain the impact of multilingual instruction on primary school children’s linguistic and metalinguistic performance, it seemed to make sense to investigate the language behaviour in subjects who had already had several years of multilingual education. This, we surmised, would allow us to obtain more clear-cut results. Given that no sufficient number of multilingual 5th graders (who are in their final year of primary school) was available, the choice fell on 4th graders.
Metalinguistic Ability Test 2: A modified version
153
16.2 Metalinguistic Ability Test 2: A modified version The modified version employed for the purposes of the present study comprises four of the original six sections. The two sections not included in the present research study are those on Comprehension and Synonymy. In addition, the task items for the section Acceptability, Ambiguity, Grammatical Function and Phonemic Segmentation had to be reduced in number so as to make sure that the entire test could be completed in 60 minutes. The reduced version with each individual section will be outlined in the following. The metalinguistic ability test was, as indicated earlier, administered in Italian since Italian was the dominant language for the overwhelming majority of the participating children. The examples given here are reported in English. They are a translation of the Italian test items used in the study. 16.2.1 Acceptability The modified MAT-2 opens with the Acceptability test which consists of two task items or, more precisely, sentence couples. In both items the second sentence contains a semantic anomaly which violates general rules of lexical compatibility or acceptability, or works on the dichotomy human ↔ non-human or animate ↔ inanimate. The LQ requires subjects to indicate whether a given sentence is acceptable or not. The MLQ, which immediately follows the LQ, invites participants to justify their choice and explain why the previous L answer was given. The following example serves to illustrate this. Sentence 1: The cat was playing with the string. LQ: Can you say this? L Answer: ______________ Sentence 2: The string was playing with the cat. LQ: Can you say this? L Answer: ______________ MLQ: Why did you give these answers? ML Answer: ____________________________________________ 16.2.2 Ambiguity As in the case of the Acceptability test, the Ambiguity test consists of two task items. These contain either a semantic ambiguity stemming from the polysemous meaning of a single word or element in the sentence, or from a structural ambiguity in which case the usage of a particular word can alter the meaning of the entire structure. Again, the LQ examines general knowledge of rules of language
154
Theoretical and methodological framework and study design
usage, while the MLQ seeks to transcend the epilinguistic level by encouraging the subject to elaborate on the preceding L answer. The following examples illustrate this point. Sentence 1: To get a car. To get a haircut. To get old. LQ: Do you think the expression “to get” has the same meaning in all three cases or not? LA: ______________________________________________________________________ ______________________________________________________________________ MLQ: Try to say these sentences differently. For example: “To get a car”. MLA: ______________________________________________________________________ ______________________________________________________________________ MLQ: What’s another way of saying “To get a haircut”? MLA: ______________________________________________________________________ ______________________________________________________________________ MLQ: What’s another way of saying “To get old”? MLA: ______________________________________________________________________ ______________________________________________________________________ Sentence 2: John is easy to please. LQ: Does John please others or is he the one who is pleased? LA: ______________________________________________________________________ ______________________________________________________________________ MLQ: What makes you sure of that? ML Answer: ______________________________________________________________________ ______________________________________________________________________ The L answer in sentence 1 (above) requires the pupils to focus on and identify the polysemous meaning of the three expressions presented in the task item. The subsequent ML questions measure subjects’ ability to paraphrase the different meanings. The L answer in sentence 2 examines subjects’ ability to detect
Metalinguistic Ability Test 2: A modified version
155
ambiguity in the sentence structure, while the ML question requires the pupils to comment on it. 16.2.3 Grammatical Function The Grammatical Function test includes three task items assessing subjects’ understanding of the grammatical functions of subject, object and predicate and their grasp of adverbial time phrases. Again, LQs work on pupils’ general understanding of roles and action while MLQs aim to dig deeper. It may be of interest that the 3rd task item does not comprise an LQ but directly elicits an MLA (Metalinguistic answer). Pinto et al. (1999) explicate that since “the first half of the section presents a well-defined action with the objective of pointing to a method of sentence analysis (and also that this is the [. . .] penultimate section of the test), it was deemed necessary to arrive directly at the MLQs” (58). An example (comprising both an LQ and MLQ) is given here below: Sentence 1: John broke the glass. LQ: What was broken? LA: ______________________________________________________________________ ______________________________________________________________________ MLQ: What makes you sure of that? MLA: ______________________________________________________________________ ______________________________________________________________________ 16.2.4 Phonemic Segmentation The Phonemic Segmentation test is subdivided into three distinct sections. In contrast to the previous tests, the focus here is not on the sentence and on the syntagmatic relations between sentence constituents but, instead, on phonemes, syllables and lexemes. Thus, the nature and structure of this fourth and last section differ considerably from that of the preceding test sections. Importantly, the first two subsections of the Phonemic Segmentation test do not include any MLQs. This is due to the fact that the task items in question do, according to Pinto et al. (1999), not exhibit a metalinguistic dimension thus rendering any metalinguistic elaboration redundant.
156
Theoretical and methodological framework and study design
16.2.5 Syllable Scansion The first subsection of the Phonemic Segmentation test, the Syllable Scansion test, is a phonetic-phonological task. As specified above, the syllable scansion test renounces the use of MLQs. The test is made up of a total of 6 task items. Subjects are required to count and to note down the individual syllables in each word. Example: Item 1: Children. LQ1: What are the syllables in this word? LA1: ______________________________________________________________________ ______________________________________________________________________ LQ2: How many are they? LA2: ______________________________________________________________________ ______________________________________________________________________ 16.2.6 Identification of Repeated Phonemes The focalization in the second subsection of the Phonemic Segmentation test is on the identification of vocalic and consonantal phonemes repeated in a single word. As in the previously outlined syllable scansion task, no MLQs are included. The number of task items to be completed amounts to three. Example: Item 1: Elementary. LQ1: What sound is repeated? LA 1: ______________________________________________________________________ ______________________________________________________________________ LQ2: How many times? LA 2: ______________________________________________________________________ ______________________________________________________________________
Metalinguistic Ability Test 2: A modified version
157
16.2.7 Formation of Words The third and last subsection of the Phonemic Segmentation test is composed of two task items which require respondents to form as many words as possible by connecting one of five given letters with a word fragment. In addition, subjects need to indicate the grammatical category of each word. The task is further complicated by the fact that some of the words can belong to more than one grammatical class. The test assesses subjects’ capacity to recognise and grasp semantic distinctions on account of minimal phonetic-phonological differences. Example: Item 1:
LQ: Compose all the words you know by connecting the individual letters with the rest of the word. LA: ______________________________________________________________________ ______________________________________________________________________ MLQ: Are they verbs, adverbs, adjectives, nouns or pronouns? MLA: ______________________________________________________________________ ______________________________________________________________________
17 Data collection procedure and test administration Intensive classroom observations took place in October of 2010 to gain as complete a picture as possible of the school situation and teaching methodologies at both institutions. A questionnaire, distributed to the participants’ parents during this period, allowed the author to assess the respondents’ sociolinguistic background. As indicated earlier, data collection took place in spring of 2011. All the tests were administered by the Italian, German and English teachers of the respective schools and classes. This, it was hoped, would minimize the amount of pressure and anxiety generated by the test situation. The author of this book was present during all of these sessions but did not intervene unless the respective teacher asked her to. Prior to the test administration, teachers were familiarised with the test procedures and the single task items. A number of meetings served to provide specific guidelines and clarifications as to the administration of each individual test. Teachers were presented with sample answers for the single training items that preceded each task. This was deemed crucial since Pinto et al. (1999) emphasise the fundamental importance of explaining each of the tasks presented (76) by involving the subjects in a brief discussion. This joint pre-test activity can initially result in various fragmentary or incomplete answers. Therefore, as specified by Pinto et al. (2003), “[s]ara ‘. . . necessario sollecitare il piu’ possible quei tipi di risposte che focalizzano tutti gli aspetti pertinenti del problema, ed eventualmente proporre un completamento della soluzione” (22; italics in the original). In other words, the teacher administering the MAT-2 is required to “synthesize different pertinent contributions into a single, complete answer, confirming to the entire class the final formula(s) obtained” (Pinto, et al. 1999:77). Hence, once an optimal formula or answer is arrived at, the teacher may write this latter on the blackboard so that pupils can relate to it during the actual test (ibid). The MAT-2, the German and the English test were administered on different days in the course of a 10-day period. All three tests were administered in written form to the whole class. Those pupils who, for whatever reason, were absent on the day of testing, were (if possible) administered the test on a different day. When the MAT-2 was first employed in the 1980s the test was administered individually and orally (Pinto, et al. 1999:76). However, for obvious practical reasons, the oral format was soon abandoned in favour of the written mode.
18 Scoring Evaluation and analysis of the collected data are based on specific criteria and guidelines provided by Pinto and her colleagues (1999:81). Great care was taken in adhering to the scoring procedures devised by these authors. For L questions the scoring procedures followed a right/wrong dichotomy, that is, subjects were given 1 point for each correct item or answer and 0 points for answers they got wrong. The final score for L answers was arrived at by adding up the single task item scores. ML questions followed a different pattern. Each item or answer was evaluated according to a three-level scale ranging from 0 to 2 depending on how complete and exhaustive the subject’s answer was. Every single ML answer was evaluated according to a catalogue of psycholinguistic criteria formulated for metalinguistic achievement in subjects aged 9–13 (Pinto et al. 1999:81). Pinto and colleagues (1999:81) delineate the three ML levels as follows: At level 0, the pre-analytical level, subjects do not yet have the “ability to analyse the sum of semantic and grammatical indices in the items presented” (ibid). Respondents may simply not know the answer, they may provide tautological answers such as “Because that is how it is” or “That’s what it says in the sentence”, or they may produce a seemingly sophisticated explanation but use the specific vocabulary in an inappropriate or erroneous manner (ibid.). The implication is that at the pre-analytical level subjects are not able to move beyond a minimal elaboration of the task item. Hence, subjects are given 0 points for their response. At the subsequent level, or level 1, subjects exhibit some degree of pertinent but still insufficient analysis. As Pinto et al. (1999) expound, subjects are, at this stage, able to isolate “at least one semantic/grammatical clue” (81). However, their method of analysis is as yet little refined and does therefore not permit them to “disambiguate the problem posed” in any sufficient way. Only at level 2 do subjects display a complete understanding of “all the pertinent semantic and grammatical indices” (82) in a given task item. This understanding allows them to adequately analyse and disambiguate the problem at hand. At this stage, subjects are therefore credited with the metalinguistic ability of conducting exhaustive and pertinent analyses. The qualitative tripartition into 1) pre-analytical, 2) pertinent but insufficient and 3) exhaustive and pertinent analysis corresponds to a quantitative scale on the basis of which 0, 1 or 2 points are granted. As in the case of LAs (Linguistic Answers), the final MLA (Metalinguistic Answer) score is reached by adding up the single MLA item scores.
160
Scoring
Spelling mistakes were not considered in the test evaluation. It may be of interest to other researchers that in the course of her analyses and evaluations of respondents’ written productions in the MAT-2 the author found herself confronted with the not always easy task of establishing whether a given ML answer qualified as level 1 or level 2 on the 3-level scale. The difficulty consisted primarily in determining whether the ML answer produced by a given respondent was to be considered level 1 (pertinent but still somehow insufficient, therefore earning the respondent 1 point), or whether the ML answer in question evidenced a complete disambiguation of the problem posed in which case the respondent merited to be assigned the highest pointage (viz. 2 points). This difficulty emerged only for ML questions and did so in spite of the exhaustive list of examples provided by Pinto et al. (1999:77–108). In contrast, analysis of L questions (which were evaluated according to a right-wrong dichotomy) proved entirely unproblematic. The categorisation problem stems of course from the fact that ML questions are evaluated on the basis of specific psycholinguistic criteria formulated for ML achievement for this age group (Pinto et al. 1999:77), which necessitates a thorough reading and analysis of each individual item. The author decided that, in cases of doubt, she would resolve the categorisation problem by assigning the pointage we deemed most appropriate in a particular case. She thereby considered not only the pupil’s ML answer but also took into consideration how complete and explicit the teacher administering the test had been in presenting and explaining the training items prior to the actual test. If in the test phase the teacher had failed to provide the pupils with an optimal sample answer and instead accepted pupils’ somewhat fragmentary contributions without really proffering additional clues, it was reasoned that an ML answer on the borderline between level 2 and level 3 should be classified as level 3. In addition, great care was taken to apply the same categorisation criteria if two respondents had incidentally produced identical answers (which happened on a handful of occasions). In the following chapter the test results for the MAT-2, the German and the English test are reported and analysed.
19 Analysis and Results The statistical evaluation of the collected data was carried out by means of the SPSS (Statistical Package for Social Sciences). A one-way ANOVA (Analysis of Variance) procedure was employed to contrast subjects’ performance in the three tests and to determine whether the difference between the experimental and the control groups was statistically significant. The one-way ANOVA allows scientists to establish whether between-group variance is larger than withingroup variance. Dissimilarities between our two groups were identified by applying an F-test. The F-test permitted us to determine the significance of the test results. Significance was taken to obtain at the 0.01 level. Our data display normal sample distribution. This has allowed us to employ parametric procedures which are acknowledged to provide stronger assumptions than non-parametric tests. It is important to note, though, that the test population (case n = 84) was comparatively small and that the results may therefore be generalizable only to a limited extent. This notwithstanding the author is hopeful that the results obtained will contribute to the field of third language acquisition and multilingualism research. The findings show that the respondents receiving multilingual instruction performed significantly better on all 3 tests compared to those in the control groups. The results support both, hypothesis 1 and 2, which state that children in high-immersion multilingual educational programmes differ from those attending regular programmes in terms of their linguistic sensitivity and metalinguistic awareness and in terms of their mastery of the L2 and L3. In addition, the subjects in the high-immersion multilingual programmes showed an enhanced crosslinguistic awareness, which although not measured by any of the tests administered, was quite remarkable (see chapter 20). Investigating cross-linguistic abilities and strategies in early multilingual subjects might constitute an exciting project for future research.
19.1 Results MAT-2: L Questions and ML Questions As indicated earlier, the metalinguistic abilities test was administered to establish the extent to which pupils are able to analyse, classify and explain the use of linguistic structures or items and to measure subjects’ abstract understanding of language. Two different sets of questions and answers were employed for the
162
Analysis and Results
purpose. The first type of questions, the so-called L or Linguistic Questions aimed at testing subjects’ general knowledge of rules of language use. The second type of questions, the so-called ML or Metalinguistic Questions examined pupils’ awareness of linguistic patterns.
19.1.1 L Questions All of the 6 sections of the MAT-2 contained LQs (note that for the evaluation procedure the author divided the phoneme segmentation part into 3 distinct sections). As illustrated in TABLE37 1 below, results obtained for LQs show a significant difference between the experimental and the control groups and a reliable advantage for the high-immersion test population. The subjects in the high-immersion multilingual programmes clearly outperformed their peers in the control groups38 in terms of their general knowledge of rules of language use ( p = .01). A general effect of the variable multilingual education (F (83,2) = 6.850, p < .010) was found for LQs.
ANOVA TABLE 1 Sum of Squares
df
Mean Square
F
Sig.
Between Groups
.150
1
.150
6.860
.010
Within Groups
1.791
82
.022
Total
1.940
83
Note: df = degrees of freedom; F = F value of the F- distribution; Sig.= significance. *p < .010.
37 The TABLES reporting the results for the LQs and MLQs of the MAT-2 and for the German and English test are here indicated in uppercase letters. The tables reporting the results for the individual subtests of the MAT-2 are indicated in lower case characters. 38 Note that the terms monolingual and bilingual employed here in the graphs and tables refer to traditional (monolingual) instruction on the one side and bilingual programmes (which are in fact multilingual programmes) on the other. The terms have been chosen for the sake of simplicity. Note further that the number 41 (indicating 41 bilinguals) reported under the boxplot refers to the original number of participants in that group. Importantly, only 40 multilingual respondents are considered in the statistical evaluation since one pupil had to be discounted (see above). The small circles below and/or above the whisker ends in the boxplots indicate the outliers.
Results MAT-2: L Questions and ML Questions
163
Overall results LQs
19.1.2 ML Questions Four of the 6 sections of the MAT-2 contained MLQs. As anticipated, the findings evidence a significant advantage for the respondents in the experimental groups. As shown in TABLE 2 below, results for the ML questions were highly significant with the high-immersion multilingual subjects clearly outscoring their peers in the control groups (p = .000). A general effect of the variable multilingual education (F (83,2) = 13.308, p < .00) was found for MLQs. This again attests to the superiority of the experimental groups over the control groups. In light of the fundamental centrality of the metalinguistic dimension in the present research study this latter finding appears particularly promising. ANOVA TABLE 2
Between Groups
Sum of Squares
df
Mean Square
F
Sig.
775
1
.775
13.308
.000
.058
Within Groups
4.775
82
Total
5.550
83
Note. df = degrees of freedom ; F = F value of the F- distribution; Sig.= significance. *p < .010.
164
Analysis and Results
Overall results MLQs
The overall score for both, LQs and MLQs demonstrates a clear advantage for the experimental groups. However, upon closer inspection, it emerged that the respondents in the multilingual programmes did not outdo their peers in the control groups in all 6 sections of the MAT-2. The test results (LQs and MLQs) for each of the 6 sections of the MAT-2 are reported below. The reader will recall that of these 6 sections, 4 sections (Acceptability, Ambiguity, Grammatical Function and Word Formation) contained both, LQs and MLQs. By contrast, 2 sections (Syllable Scansion and Phoneme Repetition) consisted only of LQs (see study design above). As it turned out, the experimental and the control groups performed equally well on 5 of the 10 subtests (4 + 4 + 2) of the MAT-2. No difference between the groups was found for LQs in the Acceptability test, for LQs in the Grammatical Function test, for LQs in the Word Formation test, for MLQs in the Word Formation test, and for LQs in the Phoneme Repetition test (see tables 1, 5, 9, 10, 8 below). Same or similar results for low- and high-immersion groups Table 1 below reports the results for L questions in the Acceptability test. There is no difference in task performance between the high-immersion experimental
Results MAT-2: L Questions and ML Questions
165
and the low-immersion control groups. The scatter plot provides a visual representation of these results. Table 1 (Acceptability test: L questions): Report Mean acceptability L-questions Class
Mean
N
Std. Deviation
Monolingual
1.9767
43
.15250
Bilingual
1.9756
41
.15617
Total
1.9762
84
.15337
Note: Mean = Mean value (arithmetic average); N = Number of respondents; Std. Deviation = Standard Deviation (of the Mean).
Tables 5 and 9 below report the results for L questions in the Grammatical Function test and for L questions in the Word Formation test. Again, the high-immersion and the low-immersion groups perform at equal levels in both tests. No difference in task performance between the groups is found for L tasks in the Grammatical Function and Word Formation test. The respective scatter plots provide visual representations of these findings.
166
Analysis and Results
Table 5 (Grammatical function: L questions): Report Grammatical function L-questions Class
Mean
N
Std. Deviation
Monolingual
.9884
43
.07625
Bilingual
.9756
41
.10904
Total
.9821
84
.09335
Note: Mean = Mean value (arithmetic average); N = Number of respondents; Std. Deviation = Standard Deviation (of the Mean).
Table 9 (Word formation: L questions): Report Word formation L-questions Class
Mean
N
Std. Deviation
Monolingual
.9884
43
.07625
Bilingual
.9634
41
.13183
Total
.9762
84
.10712
Note: Mean = Mean value (arithmetic average); N = Number; Std. Deviation = Standard Deviation (of the Mean).
Results MAT-2: L Questions and ML Questions
167
Table 10 below displays the results for ML questions in the Word Formation test. No difference between the high-immersion experimental and the low-immersion control groups emerges for ML questions in the Word Formation tasks (as is illustrated by the visual representation in the subsequent scatter plot). Table 10 (Word formation: ML questions): Report Word formation ML-questions Class
Mean
N
Std. Deviation
Monolingual
1.9186
43
.24227
Bilingual
1.9268
41
.23877
Total
1.9226
84
.23915
Note: Mean = Mean value (arithmetic average); N = Number; Std. Deviation = Standard Deviation (of the Mean).
168
Analysis and Results
As for the results of the Phoneme Repetition test (which consists only of L questions), table 8 below shows that again, there is no statistically significant difference between the performance of the high-immersion experimental and the low-immersion control groups in this particular test section. Table 8 (Phoneme Repetition: L questions): Report Mean Phoneme Repetition Class
Mean
Median
Minimum
Maximum
Variance
Monolingual
1.7597
2.0000
.00
2.00
.162
Bilingual
1.8943
2.0000
1.00
2.00
.041
Total
1.8254
2.0000
.00
2.00
.106
Note: Mean = Mean value (arithmetic average); Median = Middle value; Variance = Variation of the data
169
Results MAT-2: L Questions and ML Questions
ANOVA Table Sum of Squares Mean Phoneme Between Groups (Combined)
df
Mean Square
F
Sig.
3.692
.058
.380
1
.380
Repetition * Class Within Groups
8.448
82
.103
Total
8.828
83
Note: df = degrees of freedom; F = F value of the F- distribution; Sig.= significance. *p < .010.
Low-immersion advantage To the author’s surprise, the low-immersion control groups performed higher on the LQs in the Ambiguity test and on the MLQs in both, the Ambiguity and the Acceptability test. These results are shown in tables 3, 4, and 2 below. Table 3 indicates that the low-immersion groups performed better on the L questions in the Ambiguity test.
170
Analysis and Results
Table 3 (Ambiguity test: L questions): Report Mean ambiguity L-questions Class
Mean
N
Std. Deviation
Monolingual
.5814
43
.28723
Bilingual
.4634
41
.28263
Total
.5238
84
.28942
Note: Mean = Mean value (arithmetic average); N = Number; Std. Deviation = Standard Deviation (of the Mean).
As revealed by tables 4 and 2 (below), the low-immersion control groups also performed higher in the ML tasks of the Ambiguity and Acceptability test sections than their peers in the high-immersion test groups. Again, a visual representation of these findings is provided by the respective scatter plots.
Results MAT-2: L Questions and ML Questions
171
Table 4 (Ambiguity test: ML questions): Report Mean ambiguity ML-questions Class
Mean
N
Std. Deviation
Monolingual
.5465
43
.32393
Bilingual
.4512
41
.29165
Total
.5000
84
.31046
Note: Mean = Mean value (arithmetic average); N = Number; Std. Deviation = Standard Deviation (of the Mean).
Table 2 (Acceptability test: ML questions): Report Mean acceptability ML-questions Class
Mean
N
Std. Deviation
Monolingual
.5814
43
.28723
Bilingual
.4634
41
.28263
Total
.5238
84
.28942
Note: Mean = Mean value (arithmetic average); N = Number; Std. Deviation = Standard Deviation (of the Mean).
172
Analysis and Results
High-immersion advantage The high-immersion experimental groups performed higher on the Syllable Scansion test (see table 7 below) and on the MLQs in the Grammatical Function test (see table 6 below). The reader will recall that the Syllable Scansion test consisted only of LQs. The Grammatical Function test, in contrast, was subdivided into LQs and MLQs. Table 7 (Syllable Scansion: L questions): Report Mean Syllables Class
Mean
Median
Minimum
Maximum
Variance
Monolingual
1.6938
1.8333
.50
2.00
.143
Bilingual
1.9106
2.0000
1.17
2.00
.038
Total
1.7996
2.0000
.50
2.00
.102
Note: Mean = Mean value (arithmetic average); Median = Middle value; Variance = Variation of the data
Results MAT-2: L Questions and ML Questions
173
ANOVA Table Sum of Squares Mean Syllables * Between Groups (Combined)
df
Mean Square
F
Sig.
10.780
.002
.986
1
.986
Within Groups
7.502
82
.091
Class Total
8.488
83
Note: df = degrees of freedom; F = F value of the F- distribution; Sig.= significance. *p < .010.
Table 6 (Grammatical function: ML questions): Report Grammatical function ML-questions Class Monolingual Bilingual Total
Mean
N
Std. Deviation
.3798
43
.36790
1.0813
41
.48178
.7222
84
.55206
Note: Mean = Mean value (arithmetic average); N = Number; Std. Deviation = Standard Deviation (of the Mean).
174
Analysis and Results
ANOVA Grammatical function ML-questions Sum of Squares
df
Mean Square
F
Sig.
56.570
.000
Between Groups
10.327
1
10.327
Within Groups
14.969
82
.183
Total
25.296
83
Note: df = degrees of freedom; F = F value of the F- distribution; Sig.= significance. *p < .010.
19.2 Summary and discussion of MAT-2 results To summarise the results of the MAT-2 we can state that, with regards to MLQs, the experimental groups surpassed the control groups in only one subtest, namely in the Grammatical Function test (see table 6 above; see also section 19.5. for a final discussion of the overall results). The between-group difference for MLQs in this particular subtest was so marked that it resulted in an overall (metalinguistic) advantage for the test groups. Altogether, the experimental groups fared significantly higher on both, LQs and MLQs. However, it is important
Summary and discussion of MAT-2 results
175
to keep in mind that the ‘multilingual’ advantage did not extend to all subtests of the MAT-2. We can, at this point, only speculate as to the reasons for this particular outcome, which may be seen as a limitation of this study. This said, there can be no doubt that the respondents in the high-immersion multilingual programmes have attained a very highly developed level of grammatical awareness. It is at this stage not at all clear, however, why a comparable level of awareness was not also reached with regards to judgement of sentence acceptability and ambiguity (MLQs Acceptability test and MLQs Ambiguity test) and with regards to lexical function and categorisation (MLQs Word formation test). Having said this, it is, of course, also conceivable that we are asking the wrong question. Perhaps the question to ask should not be why the subjects in the high-immersion groups did not excel in all components of the MAT-2, but rather why the low-immersion controls performed so well. The author’s tentative answer to this question is that the respondents in the low-immersion programmes performed so high because they too received substantial amounts of input in L2 (6 hours per week) and L3 (3 hours per week). Like their peers in the high-immersion programmes they had gathered a certain experience in second and third language learning. Certainly, they were no novice learners of an L2 or L3. In addition, several of these pupils also came from bilingual families and several of their parents had applied for their child’s admission into one of the high-immersion multilingual programmes (for which they were however not accepted because of limited availability). We are led to believe that, thanks to family support and encouragement, these children had come to form very positive attitudes towards their second and third languages and towards language learning in general. Hence, attitudinal and/or individual motivational factors may have contributed to the control groups’ high performance in several of the subtests of the MAT-2. The overall findings generated by this study are here interpreted as an encouraging indicator of the general benefits of various differential forms of early second and foreign language learning. On the whole the study attests to a number of advantages that multilingual education programmes confer on young learners. However, as has been shown by the results of our study, different programme types or models of multilingual education may not necessarily benefit learners to the same extent. This latter point appears to hold in particular for L2 and L3 learning given that we identified a highly significant difference between the high-immersion and the low-immersion groups’ performance in L2 German and L3 English. The respondents in the experimental groups scored significantly higher on both, the L2 and L3 test (see further down). This seems to indicate that pupils in high-immersion multilingual programmes are better equipped to learn a second and third language than pupils in low-immersion programmes. It is conceivable that the differences in task
176
Analysis and Results
performance between the high-immersion and low-immersion groups can be traced to “significant changes in the way in which children carry out general cognitive processing as a function of having learned two languages” (Bialystok, 2002:159). Reporting on studies which compared monolingual and balanced bilingual children in terms of their capacity to detect structural errors in semantically meaningful sentences, Galambos & Hakuta (1988) noted that the monolingual children’s focus tended to be on sentence meaning, while the bilinguals39 who were proficient in both languages, displayed a much greater readiness to focus on sentence form or structure (147). This is congruent with our findings which revealed that the more balanced and experienced learners in the high-immersion programmes, i.e. those children who received a substantial part of their tuition in L2, scored higher on the metalinguistic ability test (albeit not in all subtests) than the children attending low-immersion educational programmes with less L2 input. The findings generated by the metalinguistic abilities test clearly require further investigation. A follow-up study employing a larger sample population and a more comprehensive MAT-2 (such as proposed by Pinto et al., 1999) might yield more robust results and a more clear-cut (metalinguistic) advantage for young learners in high-immersion multilingual programmes. In spite of the somewhat mixed results, the ensuing conclusions appear to be clear from the data as it stands: (1) Bi- or multilingualism presents no disadvantage to young learners. On the contrary, (2) knowledge of two (or more) languages facilitates the acquisition of an additional language, and (3) early multilingualism fosters learners’ linguistic and metalinguistic awareness. The findings are supportive of previous research by Cenoz & Valencia (1994), De Angelis & Jessner (2012), Galambos & Hakuta (1988), Jessner (2008), Lasagabaster (1998, 2001).
19.3 Results German and English test The results of the German and English test are reported below in TABLES 3 and 4 respectively. As in the case of L and ML questions in the MAT-2, the highimmersion multilingual respondents scored significantly higher in both, the German and the English test, than their counterparts in the control groups. As in the case of MLQs, reported above, the results obtained for the German and English tests were highly significant (p = .000). A general effect of the variable multilingual education (F (81,2) = 188.174, p < .00) was found for the German Test, and (F (77,2) = 24.524, p < .00) for the English Test. 39 Galambos & Hakuta investigated monolingual and bilingual children‘s performance. The focus here was, of course, on low- and high-immersion trilingual learners.
Results German and English test
177
German Test Results ANOVA TABLE 3 Sum of Squares
df
Mean Square
F
Sig.
188.174
.000
Between Groups
72.681
1
72.681
Within Groups
30.900
80
.386
Total
103.580
81
Note. df = degrees of freedom; F = F value of the F- distribution; Sig.= significance. *p < .010.
English Test Results ANOVA TABLE 4 Sum of Squares
df
Mean Square
F
Sig.
1
7.717
24.524
.000
.315
Between Groups
17
Within Groups
23.915
76
Total
31.632
77
Note. df = degrees of freedom; F = F value of the F- distribution; Sig.= significance. *p < .010.
178
Analysis and Results
19.4 Summary and discussion of the German and English test results The high-immersion groups, although receiving the same amount of L3 input (3 hours per week), significantly outperformed their peers in the low-immersion control groups in the German and English test. This finding appears to indicate that the test groups profit more from their special learning experience and from their prior knowledge of linguistic forms and functions. In addition, we need to consider the possibility that subjects’ proficiencies in the three languages are correlated (cf. Cummins, 1991, 2001; De Angelis & Jessner, 2012) and that high competence in one language allows for or promotes high competence in another (see next section for a detailed discussion of the overall results).
19.5 Discussion of overall results On the basis of the results obtained from the MAT-2, the German and the English test it can be stated that the initially formulated hypotheses have been fully verified. The author predicted that early multilingualism has a positive impact on
Discussion of overall results
179
subjects’ linguistic and metalinguistic awareness and on their capacity to learn a further language. The hypotheses, outlined again here below, have been fully borne out by the statistical evaluation of our data: Hypothesis 1: Early multilingualism carries positive implications for the child’s linguistic and metalinguistic awareness. Hypothesis 2: Early contact with an L2 or L3 has positive effects on the acquisition of additional languages.
The high-immersion groups demonstrated superior skill and ability in the Metalinguistic Abilities Test, in the German, and in the English test. In the following the author relates the findings generated by this investigation to previous studies in the field of SLA and TLA and discusses important correspondences. It has been suggested that “bilingualism provides a form of contrastive linguistics instruction which leads bilingual children to compare and analyse the structural aspects of language in more advanced ways than monolingual children” (Bruck & Genesee, 1995:308; see also Cummins & Swain, 1992:21–32). The findings of the present study go some way to supporting this claim. First, as shown by the results of the MAT-2, the high-immersion children display a profound knowledge of language structures in their L1 Italian. This is in line with Cummins (1991) who asserts that high proficiency in a second language has positive effects on the development of the first language (see also De Angelis & Jessner, 2012). Second, in comparison to their peers in the low-immersion programmes, the high-immersion groups appear to possess a more developed rule system in their L2 and L3 and a more developed awareness of those rules. As has emerged from the German and English test, the high-immersion groups are astutely aware of language forms and functions in both, L2 and L3, and they are very skilled at putting this knowledge to use. Third, as shown by the qualitative findings reported in chapter 20, they demonstrate skill at perceiving structural similarities and differences between their languages and they show a certain readiness to engage in comparisons between their three languages. In light of the foregoing the author is led to hypothesise that language processing and use in very experienced multilingual learners (as in the case of the respondents in the high-immersion groups who are exposed to high levels of L2 and L3 input on a regular basis, and who have accumulated a large repertoire of linguistic and metacognitive knowledge) differs from language processing and use in less experienced learners who do not receive such extensive amounts of input and who do not have at their disposal the same (multilingual) resources. Following Herdina & Jessner (2002), it is here assumed that language processing and usage in highly experienced learner-users is qualitatively different from and
180
Analysis and Results
more efficient than processing and use in monolinguals or inexperienced bilinguals learning an L3. The combined advantages deriving from a more developed rule system (in all languages), from heightened levels of meta- and crosslinguistic awareness, and from more effective processing procedures, seem to permit highly experienced learners to solve certain language-related tasks with greater accuracy. According to Bialystok (2002) bilingual children (to which the present author also adds multilingual children) are better at directing their attention to specific aspects of a problem, particularly in cases when the problem contains some sort of conflict or ambiguity (153). When the resolution of a problem or conflict requires the child to ignore or inhibit misleading information very experienced language learners seem to perform higher than monolingual children (and, it is here believed, higher than learner-users with limited exposure to a second or third language). The results yielded by the Metalinguistic Abilities Test, which required respondents to focalise and solve various tasks on ambiguity, acceptability, grammatical function, etc., and which evidenced a significant advantage for the high-immersion groups, are supportive of Bialystok’s proposition. Cummins (1976, 1991) maintains that a threshold level of proficiency must be attained in order for bilinguals to gain cognitively from their bilingual experience (cf. Lasagabaster, 1998; cf. Galambos & Hakuta, 1988:141). While the highimmersion multilingual respondents in our study have clearly reached high levels of competence in L1, L2 and L3, the subjects in the low-immersion control groups (who received only 6 hours of L2 German) appear to have attained this threshold level in only one of their languages, namely their native tongue. It is to be assumed that their proficiency in L2 (and most definitely also in L3) is insufficient to allow for any significant cognitive or linguistic advantages to obtain. In fact, no (statistically) significant positive effects have been found for the informants in the control groups. However, it is important to note, that no negative effects are apparent either. On the contrary, in some of the subtests of the MAT-2 the subjects in the low-immersion programmes performed better than the respondents in the experimental groups (see Summary of MAT-2 test results above). As postulated by Cummins (1979, 1991) multilingual speakers’ languages are interconnected and based on a common underlying proficiency (CUP). From this it follows that knowledge of one language invariably affects knowledge of all other languages. De Angelis & Jessner’s (2012) investigation of trilingual learners in South Tyrol substantiate Cummins’ claim. The authors found important associations between the respondents’ proficiency in L1, L2 and L3, stating that their research “shows clear support for the interdependence between the languages” (2012:64). The results of the present study provide some support for their findings. The study seems to indicate that high levels of proficiency in L2 German
Discussion of overall results
181
are linked to high levels of proficiency in L3 English. In addition, positive correlations appear to exist between the respondents’ metalinguistic abilities (measured in L1 Italian) and their proficiency in L2 German and L3 English. This is illustrated by the fact that the high-immersion multilingual cohorts, who, as outlined earlier, performed higher in the Metalinguistic Abilities Test than the low-immersion groups, also performed higher in the L2 German and L3 English tests. Bialystok (2002) famously posits that children’s bi- or multilingual experience and the continuous handling of two or more language systems affect processing in more general cognitive domains (160). According to Bialystok, early acquisition of two languages modifies cognitive processing not only with regard to language but also with regards to solving non-verbal tasks (ibid.). Here, the multilingual context of this research study necessitates a specification to be made. There can be no doubt that the trilingual experience of the subjects under investigation in the present study impacts the dynamics of both, linguistic and more general cognitive processing beyond that of the bilingual experience (cf. Herdina & Jessner, 2002; Jessner, 2006; Jessner, 2008a,b). The benefits of being tri- or multilingual can be anticipated to be far greater than the benefits of being bilingual (Hoffmann & Stavans, 2007:55). The multilingual advantage has to be seen in relation to tri- and multilingual users’ prior linguistic knowledge and in relation to their extensive language use and language learning experience. The complexity added by the multilingual dimension is best explained within the framework of the Dynamic Model of Multilingualism (2002). The theoretical framework of the DMM provides a sound explanatory basis for the results yielded by the present study. The DMM predicts that catalytic effects deriving from additional language learning and emergent properties, which are unique to dynamic systems (multilingual systems are construed as such), result in “the development of new qualities in the multilingual system” (33) and in a radical “change of quality in the language learning process” (34). More specifically, experienced learners who have already been in contact with a second and third language develop special enhanced skills and abilities which monolingual learners of an L2 do not possess (Jessner, 2006:34). Amongst these we find language management and maintenance skills (which help the learner to allocate resources such as time and energy in order to maintain adequate levels of proficiency) and metaand crosslinguistic abilities (which allow for contrastive comparisons between the languages, for activation of elements across language codes and for transfer of skills and knowledge from one language to another). From the perspective of the DMM, these improved properties and capabilities contribute to a metasystem (consisting of an enhanced multilingual monitor, and heightened levels of metacognitive, metalinguistic and crosslinguistic awareness), which is unique
182
Analysis and Results
to multilingual speakers. Experienced learner-users can rely on this metasystem when attending to language input and when preparing and monitoring their linguistic output. The availability of this metasystem and multilingual users’ observed “tendency [. . .] to extend their knowledge, skills and self-application” by transfering aspects and elements from one language to another (Aronin & O’Laoire, 2004:23–24) can be taken to act as a powerful learning facilitator. It is finally noted that the results of the present investigation are in keeping with Galambos & Hakuta (1988) who intimate that bilingualism “alters the approach toward language, from a message-oriented approach to a form-oriented one” (142; italics in the original). In their study of Puerto-Rican Spanish- and English-speaking children Galambos & Hakuta found that balanced bilingualism fosters metalinguistic awareness in so far as it “enhances the ability to direct attention to the form of language and focus on it effectively” (141/160). In light of the results of the present study it is here suggested that the same holds true for highly experienced tri- and multilingual learners. Clearly, the overall findings generated by this study have to be seen as the result of complex and dynamic interactions of a number of diverse factors including not only multilingual instruction but also bi- or multilingual upbringing, exposure, frequency of use, parents’ educational background and the sociocultural and socioeconomic environment in which subjects have been raised. This claim is supported by the data obtained from the background questionnaires distributed to the children’s parents, from which the following picture emerged: 19.5.1 Educational background mother With regard to parents’ educational background it is important to note that while in the control groups 8 of the subjects’ mothers have a university degree, this number is doubled in the test groups (see below). Educational background mother Educational background mother
Class Monolingual Bilingual Total
Middle school
High School
Vocational School
University
Total
1
27
4
8
40
2
17
3
16
38
3
44
7
24
78
Discussion of overall results
183
In a similar study conducted in the Basque context Valencia & Cenoz (1992) highlighted the central or “symbolic” role of the mother in the child’s development (436) thus lending support to the present author’s suggestion that parents’ and in particular mothers’ educational background affect their children’s foreign language achievement in significant ways (see also Cenoz & Lindsay, 1996). 19.5.2 Educational background father As regards the educational background of the subjects’ fathers, the ratio is 9 to 12 in favour of the test groups. In other words, while in the control groups 9 of the pupils’ fathers have a university degree that number amounts to 12 in the experimental groups. This information is represented in the table and graph below. From these and the above findings the author infers that the child’s educational and sociocultural background constitute variables which can not be neglected if one is to gain a complete picture of multilingual development. This line of arguement is in keeping with Bialystok (2002) who pinpoints a number of factors which are thought to impinge on the bilingual (and, as is argued here, the multilingual) child’s linguistic and intellectual development. According to Bialystok, these factors comprise “the education level of the parents, the literacy environment that the child is exposed to, the nature and extent of the child’s
184
Analysis and Results
proficiency in the first (or home) language, the purposes for which the second language is used, the degree and nature of community support for that language, and the extent to which the child identifies with the group who speaks that language” (156). Educational background father Educational background father
Class Monolingual Bilingual Total
Middle school
High School
Vocational School
University
Total
7
19
5
9
40
2
20
4
12
38
9
39
9
21
78
Discussion of overall results
185
19.5.3 Languages spoken in the home Languages spoken in the home Languages spoken in the home Total Class Monolingual Bilingual Total
One language
More languages
35
5
40
15
24
39
50
29
79
An equally insightful picture emerges with regards to the languages spoken in the home and family. As indicated in the above table, 35 children in the control groups use only one language in the home while only 5 children have been reported to use more than one language on a regular basis. This stands in stark contrast to the data retrieved from the experimental sample according to which no less than 24 subjects use more than one language in the family and merely 15 subjects use only one language in the home. Barrena et al. (2008) point out that the “degree of presence of each language in the family environment seems to have some influence on the linguistic development of the bilingual child” (106). This is in line with the observations of Valencia & Cenoz (1992) who identified regular use and valorisation of the bilingual’s two languages within the family context as crucial variables which impact both, learners’ cognitive and linguistic development (435). A similar point is made by Oberhofer (2008) who notes that “it is likely that parents who would decide to send their child to an English-language kindergarten would have a higher appreciation for language (s), and would impart this to their child at home” (108). Oberhofer’s suggestion coincides with Cenoz & Lindsay (1996:92) who state that parents’ attitudes and expectations are acknowledged to play a fundamental role in foreign language acquisition. 19.5.4 Language learning history In the following we look at our partcipants’ exposure to L2 German in early childhood and kindergarten. In the high-immersion multilingual programmes at the Longon and Manzoni schools, L2 German is used alongside L1 Italian as a medium of instruction. Our respondents’ language learning history reveals that only one subject in the control groups was exposed to German from birth, while that number amounts to an astonishing 20 in the experimental groups. In
186
Analysis and Results
addition, as emerges from the background questionnaires, 19 children in the control groups had had contact with L2 German in kindergarten, while in the test groups that number was 29. This information is displayed in the table and graphs below. Class Monolingual Bilingual Total
Language Learning History German from birth
Language Learning History German in kindergarten
1 20 21
19 29 48
Number of pupils exposed to German from birth
Discussion of overall results
187
Number of pupils exposed to German from kindergarten
It is to be expected that early contact with and amounts of exposure to the L2 correlate with attainment in L2. There are indications that early (and/or) highlevel exposure to an L2 can enhance subjects’ metalinguistic awareness (Oberhofer, 2008:104) and their success in additional language learning (see Valencia & Cenoz, 1992:445). This is supported by the findings of the present study which suggest that amount of exposure to L2 is correlated not only with second but also with third language attainment, and with metalinguistic ability. The high-immersion groups, who received subject matter instruction in L1 Italian and L2 German, outperformed the low-immersion control groups (who received only 6 hours of L2 German per week) in all three tests (MAT-2, German and English test). There is reason to assume that the high-immersion groups have accrued greater benefits from their knowledge of prior languages than the low-immersion groups. Indications are that their extensive experience with languages and language learning provides them with a robust (meta- and cross)linguistic knowledge base and with high levels of language learning awareness which permits them to optimize the use and allocation of their resources and which renders them more autonomous and in control of their learning. The fact that the test groups performed higher in the L3 English test
188
Analysis and Results
(even though they received the same amounts of L3 input as the control groups) provides support for this assertion. The present author joins the authors of the DMM in proposing that minimal changes or slightly altered conditions, such as different amounts of exposure to L2 but not to L3, as in the case of our test populations, can lead to important qualitative modifications and can generate radical new properties (Herdina & Jessner, 2002:82). To sum up, the research study has evidenced an advantage with respect to second and third language acquisition and metalinguistic awareness for the subjects in the high-immersion multilingual programmes at the Manzoni and Longon primary schools in Bolzano/Bozen. The results obtained from our statistical analyses are compatible with previous studies conducted in the field of SLA and TLA (Galambos & Hakuta, 1988; Valencia & Cenoz, 1992; Lasagabaster, 2001b). As anticipated, the findings indicate that early and extensive exposure to an L2 and L3 in the form of high-immersion multilingual education promotes and enhances learners’ metalinguistic awareness and their proficiency in second and third languages (cf. Dickinson et al., 2004:339). Next, the author outlines and discusses important qualitative data and considers some of the idiosyncracies and characteristics of the multilingual respondents’ linguistic productions and behaviour.
20 Some of the things multilingual children do 20.1 Qualitative data on cross-linguistic awareness An important aspect of our high-immersion multilingual subjects’ linguistic behaviour emerged during classroom observations in October 2010. A marked awareness of cross-linguistic structures was observed in the respondents receiving high-immersion multilingual education. The ensuing example serves to illustrate this point. One morning, during English class, pupils were asked to note down, in their workbook, how many brothers and sisters they have. After a brief while, one of the boys raised his hand and inquired whether it was necessary in English, as it was in German, to add grammatical markers to the lexeme ‘one’, in other words whether the lexeme ‘one’ required case or number agreement with the noun as it did in the German language. The pupil in question had written down: (1)
„I‘ve got one brother and one sister“.
He then wanted to know: „Muss man nicht wie in Deutsch . . . da sagt man ‚Ich habe einen Bruder und eine Schwester. . .?’“ [“Do you not have to say like in German. . . there you say ‘Ich habe einen Bruder und eine Schwester. . .?’”].
Even though the pupil lacked the specific terminology, his query clearly testifies to his enhanced crosslinguistic awareness of syntactic relations in sentence constructions. The example illustrates that young multilingual users are able to perceive cross-language commonalities or differences and exploit prior linguistic knowledge. The above observation also confirms the results generated by related studies on the correlation between bi- or multilingualism and syntactic awareness which revealed that “bilingual children show higher levels of syntactic awareness” (Bruck & Genesee, 1995:308; cf. also Galambos & Hakuta, 1988:160 and Gibson & Hufeisen, 2011:83). These latter as well as the author’s own findings are in line with Jessner (2006) who posits that the search for cross-linguistic analogies is a central aspect of metalinguistic reasoning in multilingual speakers (114; cf. Bono, 2011:49 and Otwinowska-Kasztelanic, 2011:1). The above example is just one of a number of instances of this kind. It can be construed as evidence for the interrelatedness of language systems in the mind and for cross-linguistic interaction between these systems. The example
190
Some of the things multilingual Children do
further testifies to young multilingual speakers’ readiness to engage in linguistic analysis and to their heightened perception of cross-linguistic analogies and oppositions. It is also indicative of an Enhanced Multilingual Monitor (EMM) which in multilingual speakers develops as a result of the complex and dynamic interaction of all the linguistic systems involved (Jessner, 2006:117). The constant management of these interacting and, at times, interfering systems leads to the emergence and development of metalinguistic skills and abilities which in turn result in the creation of “an increased monitoring system for all the languages known by the multilingual speaker” (ibid.). It is suggested that as the multilingual speaker’s Enhanced Monitor becomes more developed, the speaker’s metalinguistic awareness and abilities increase too. Cross-linguistic awareness, as observed in the high-immersion multilingual subjects, reflects learners’ “awareness of the links between their language systems” (Jessner, 2006:116). Together, cross-linguistic awareness and metalinguistic awareness are seen as constituting the multilingual speaker’s linguistic awareness (ibid.). Further evidence for cross-linguistic consultation is reported below. One day during the music lesson a rhyming song was sung and the text analysed and discussed in class. Upon encountering the lines (2)
“dein Herzenswunsch, ein echtes Tier, ein Pferd mit feinem Goldgeschirr!“
one pupil confidently pronounced „Aber es reimt sich nicht!“ (“But, it doesn’t rhyme!”), thereby testifying to the general claim that multilingual subjects have a heightened sensitivity to phonological aspects of language “because they must attend carefully to the speech stream in order to distinguish the two languages and to organize their developing lexicons” (Walley, 1993 in Bruck & Genesee, 1995:308). The pupil’s observation regarding the absence of rhyme between the two words ‘Tier’ and ‘Goldgeschirr’ are indicative of a profound awareness of the phonological properties of these lexemes which although sounding similar do indeed not rhyme. It has been affirmed that bi- and multilinguals show an enhanced verbal flexibility and originality (cf. Cummins & Swain, 1992:10-17). The following examples have been selected because they attest to a remarkable verbal agility in some of the multilingual subjects participating in the present study. One day, during the German lesson (the reader will recollect that German is most pupils’ L2), the class was reading a text about a treasure hunt on a remote island. In the ensuing discussion the teacher posed the following question: (3)
„Was ist ein Moor?“ [What is a moor?]
Qualitative data on multilingual productions
191
One of the pupils raised his hand and began to elaborate: (4) „Moor ist ein Gebiet wo‘s sumpfig ist und nass und schwül.“ [A moor is an area where it is swampy and wet and humid.] Some time later, a request for clarification ensued. It came from one of the pupils: (5)
„Was heisst das: ‚Das ist die Höhle der klagenden Seelen’?“ [What does it mean: This is the cave of the wailing souls?]
The teacher, not expecting any of the children to know the answer, was about to give his response when one of the pupils blurted out: (6) „Ich kann das erklären! So wie tote Gespenster, die heulen (halt).“ [I can explain that! Like dead ghosts, wailing (sort of).] In both cases, the boys who were able to produce these sophisticated replies displayed a distinct German accent (as spoken in Germany) which can be taken as an indicator that one of their parents is of German origin, rather than of South Tyrolean descent. As pointed out earlier, the German spoken in Germany differs substantially from the dialect spoken in the province of South Tyrol. The differences relate to features of pronunciation, lexis and syntax. Standard German as spoken in Germany is generally considered superior to and more elevated than the dialectal South Tyrolean variety. Contextual factors such as subjects’ sociolinguistic and sociocultural background have, in past studies, been identified as variables contributing to the complex and multidimensional phenomenon of multilingualism (cf. Galambos & Hakuta, 1988:160). In the case of our respondents, the regular use of standard German in the home might have given them a head-start, since standard German is generally acknowledged to employ more elaborate and sophisticated lexical and syntactical structures than the South Tyrolean dialectal variety.
20.2 Qualitative data on multilingual productions The following section reports a selection of linguistic productions generated by the high-immersion groups during classroom observations in the month of October 2010. All of the ensuing examples are quoted verbatim. The primary aim of this section is to analyse and interpret language behaviour in children who have relatively balanced knowledge of three languages (see also Stavans & Swisher, 2006).
192
Some of the things multilingual Children do
20.3 Mixing/Borrowing In the high-immersion experimental group, mixing, i.e. the insertion of nontarget lexical items has been found to be particularly prevalent. While some instances of mixing may be regarded as occurring on a subconscious level, others will undoubtedly stem from a pragmatic necessity arising from subjects’ imperfect and incomplete mastery of a given language. In particular, respondents have been observed to borrow single words, especially nouns and verbs, or simple noun phrases or idiomatic phrases (as in example 3) from their L1 (see below). This coincides with the observations of Potowski (2009:99) who found that single lexemes are the most commonly codeswitched items (see also Hoffmann & Stavans, 2007:63). (1)
Meine Mami hat mir spiegato, dass. . . [My mum has spiegato (explained) that . . .] spiegato = ITAL. PAST PART. agrees with GERM.AUX. ‘hat’
(2)
Es sind zwei gessi auf den Boden gefallen! [Two gessi (chalks) have fallen on the floor!] gessi = ITAL. PLUR. NOUN agrees in number with GERM. NUMERAL ‘zwei’
(3)
Es ist nicht colpa mia! [It’s not colpa mia (my fault)!] colpa mia = ITAL. COMPLEMENT CONSTR. obeys ITAL. WORD ORDER (embedded island in Myers-Scotton’s sense; intrasentential switch after Edwards & Dewaele, 2007: 227)
As indicated earlier, the activation of non-response lexemes in an otherwise monolingual context as that of the German lesson40 appears to provide strong support for the claim that the language systems of multilingual speakers are activated concurrently and that they interact (Jessner, 2006:100), allowing subjects to draw on a vast range of resources. Here, the insertion of a large number of non-target items from the L1 [(1) (2) (3)] allows us to identify L1 Italian as the preferred supporter language. Sentence (2) sounds slightly awkward but there are no morphosyntactic violations of the matrix language.
40 This said, the German classroom may well fall short of constituting a truly monolingual context (cf. Grosjean, 2001).
Dialectal insertions and lexical inventions
193
20.4 Dialectal insertions and lexical inventions The following sentences include further instances of mixing/borrowing from L1 Italian (1), dialectal influences from L2 German (even though the use of German dialect is generally avoided and discouraged in the German classroom) (1)(2), and examples of word creations on the basis of lexical items in subjects’ L1 (2) (3). The ensuing sentences show relatively complex and sophisticated combinations of the respondents’ L1 and L2 within the same utterance. Sentences (1) and (2) also provide evidence that there is confluence of three languages or language varieties (German, German dialect and Italian) in the same utterance. In both, Italian and German, verbs, adjectives and articles have to agree with the noun in number and gender. (1)
Um halb fünf so machen wir festa erster* und dann gehe ich in die Schule. [At about half past four, we have festa (party) erster* and then I go to school.] festa = ITAL. NOUN (ACC. FEM.) ; erster* = GERM. dialect ADV. (here: wrong word order)
(2)
Meine Oma hatte immer Katzen . . . drei oder vier selvatischen*Katzen und mein Opa tut die Trappeln aufstellen. . . [. . . three or four selvatischen* (wild) cats and my grandpa puts up the Trappeln (mousetraps). . .] ‘selvati’ = ITAL. stem (ADJ.) + ‘schen’ = GERM. ending: ACC. FEM.PLUR. (here: wrong case marker) (from Italian ‘selvatico’) [wild] ‘Trappeln’ = GERM. DIAL. PLUR. NOUN instead of standard German lexeme ‘Mausefalle(n)’
(3)
‚emotioniert‘ (from Italian ‘ emozionato’) [agitated, excited] ‘emozion’ = ITAL. stem (ADJ.) + ‘iert’ = GERM. PAST PART. ending
The above examples may be seen as an indicator of language dominance (Italian in the present case). The dialectal form ‘erster’ used in the first example is commonly employed in the South Tyrolean variety to denote ‘first’ or ‘at first’. It contrasts with the standard German item ‘zuerst’. Given that South Tyrolean dialectal forms are not commonly utilised by the teachers at either of the two schools participating in the study, it can be inferred that the pupil in question had come into contact with this particular form outside the classroom. The same holds true for the lexeme ‘Trappeln’ (in the second example) which is widely used in local dialectal varieties. Since it is unlikely that the pupil had
194
Some of the things multilingual Children do
encountered the word in the school setting the assumption is that she had come across it outside of the school context. This said, it is important to note that the activation of the lexeme ‘Trappeln’ might also be explained in terms of (psycho) typology, i.e. the perception of typological closeness between the L1 and the target item. The fact that the L1 Italian equivalent ‘trappole’ is cognate with the selected L2 item may have facilitated the retrieval of the dialectal form in question and may have caused the child to favour its use over the standard German ‘Mausefalle’. Of course, we also have to consider the possibility that the child may not have been familiar with the standard German item ‘Mausefalle’ at all and that the above alternative (‘Trappeln’) was the only lexical item she was able to access. The third example represents a lexical creation based on L1 Italian (cf. Dewaele, 1998:471). The lexeme ‘emotioniert’ (based on the Italian ‘emozionato’) does, of course, not exist in the German language. The same is true of the lexeme ‘selvatisch(en)’ in the previous sentence, where the pupil invented a novel word on the basis of the Italian lexeme ‘selvatico’. In both cases (1) (2) the pupils in question generated a new word by combining an Italian word stem with a bound morpheme from German.
20.5 L2 influences during L3 discourse Other-language use was also discerned in the form of intrusions from L2 items in L3 discourse situations. The following utterance generated by a pupil in the high-immersion multilingual class at the Manzoni school serves to exemplify this. About five minutes into the English lesson the teacher (native speaker) asked “What‘s the date today?” The pupil whom the teacher had addressed replied “einundzwanzig*”, which is ‘twenty-one’ in German. Instead of using the appropriate target language the pupil had inadvertedly produced the response in his L2. Apart from selecting the wrong language, the pupil had also failed to produce the required ordinal number. The correct answer in German would have been „der einundzwanzigste” (i.e. the twenty-first).
20.6 L3 influences during L2 discourse In the same way, L3 intrusions in L2 communicative contexts were also observed. On one occasion during the German lesson, a pupil sitting next to the experimenter (the present author) addressed the latter with the words: (1)
“Teacher, schau!” [Teacher, look!] ENGL. NOUN + GERM. IMPERATIVE SING.
Trilingual mixing
195
It is important to note that the pupil had no means of knowing that the experimenter spoke any English at all. This led the author to conclude that the pupil’s non-target response was unintentional. The pupil’s reaction when she realised the mistake substantiated this hypothesis. Such instances are, as noted above, a reflection of the interconnectedness and interplay of multilingual systems (cf. Jessner, 2006:116).
20.7 Trilingual mixing Altogether only four instances of trilingual mixing were found. Two such instances (including Italian, standard German and German dialect) are reported above (see section on Dialectal insertions and lexical inventions: sentences (1) and (2)). Perhaps the other two examples, which are reported here below, do not qualify as true instances of trilingual mixing but they are nevertheless included here because they combine three languages in one single sentence. Both utterances (1) (2) provide an insight into the respective speakers’ meta- and crosslinguistic awareness. During an English lesson one pupil pointed out that the word ‘grandpa’ (1)
“have due parole: wie in grandson” [‘have* two words: like in grandson’] ENGL. VERB + ITAL. NUMERAL ‘due’ [two] + ITAL. PLUR. NOUN [words] + GERM. comparison ‘wie in’ [like in] + ENGL. NOUN
The pupil’s metalinguistic comment on the structural properties of the word ‘grandpa’ unmistakably testifies to this young learner’s enhanced meta- and crosslanguage awareness. In the second example the pupil in question had just realised and was now remarking upon the fact that he had combined two languages in an English writing task: (2)
“Ho scritto ‚ein sister‘, ‚ein brother‘“ [I have written ‘ein sister’, ‘ein brother’] ITAL. PASSATO PROSSIMO + GERM. INDEF. ART. + ENGL. NOUN + GERM. INDEF. ART. + ENGL. NOUN
Subjects’ observed preference for mixing two rather than three languages is commented upon in Hoffmann & Stavans (2007:15). These authors note that the largest number of switches in their corpus is bidirectional. This, according to Hoffmann & Stavans, may be due to the increased complexity and difficulty involved in combining elements from three linguistic codes.
196
Some of the things multilingual Children do
20.8 Learner language Our subjects’ idiosyncratic use of language betrays their linguistic insecurity in the face of certain L2 grammatical features. This is illustrated in the examples listed here below. When, on one occasion, the German teacher invited the pupils to select a mammal from a range of pictures lying on the desk, one of the pupils inquired (1)
“Können wir das Zebra, die* Elephant und die Giraffe? Na, das* Löwe. . .“ [„Can we (take) the zebra, the elephant and the giraffe? No, the lion. . .”] die Elephant*, das Löwe* = wrong gender/case marker
only to repeat the same noun phrase two minutes later, this time employing the (again erroneous) definitive article denoting neuter: (2)
“und das* Elephant.“ = wrong gender/case marker [and the elephant].
The correct rendition of the above utterance should read as follows: “Können wir das Zebra, den Elephanten und die Giraffe nehmen? Nein, den Löwen. . .” [“Can we take the zebra, the elephant and the giraffe? No, the lion. . .”] It is clear from this example that the child’s mastery of German articles and morphological markers is as yet incomplete and that gender and case marking still pose difficulties to fourth form pupils even when they receive a substantial amount of their instruction in the L2 (cf. Cummins, 2000 on BICS and CALP). From this example it can also be inferred that languages (construed as complex dynamic systems) and language development are characterized by variability and by unpredictable non-linear growth patterns and spurts, from which it follows that growth and progression in second or third language learning are “far from a straightforward linear transition from a complete lack of knowledge to full command” (Caspi & Lowie, 2010:45; cf. Herdina & Jessner, 2002:89). Instead, as the above examples illustrate, the complexity and dynamics of additional language learning can give rise to entirely unanticipated developments, which can not be reliably forecast.
20.9 Multilingual interaction On one occasion a pupil asked the examiner the following question:
Multilingual interaction
(1)
197
Bist du es nicht italienisch*? Bist du es deutsch*? [Are you not it Italian? Are you it German?]
Importantly, the linguistic structure employed by the pupil can not be traced to his L1, L2 or L3. It is clearly not reducible to any of these languages. The above usage of the personal pronoun ‘es’ (NEUTER SING.) is not typical of any of the three languages known to the child. As indicated, this goes to show that multilingual output possesses a complexity and dynamism of its own. The productions of our respondents can be said to emerge in the process of dynamic multilingual interaction. They can not always be related to linguistic structures or constraints in their other languages. Another such example is reported below. One morning a pupil addressed the German teacher with the following words: (2)
Christine, kann ich eine Frage dir machen*? [Christine, can I make a question you?]
As in the previous example, the above structure can not be ascribed to any particular language known to the child in question. The pupil had probably come across the target structure “Kann ich dir eine Frage stellen?” but was unable to produce it correctly during this particular exchange. It can be hypothesized that the language-specific lexico-grammatical shape of this given construction had not yet been fully internalized. This said, it is important to bear in mind that multilingual phenomena and growth patterns (of the sort we have seen in the above examples) are (even though they may appear abrupt and uneven) not to be interpreted as negative effects of multilingualism but as an indication that the learner system is restructuring and evolving. As explicated, in DMM terms, language growth and development do not follow a neat and linear upward progression but instead also comprise phases of zero growth and even transitional deterioration and setbacks (Herdina & Jessner, 2002:142). With this said, the author concludes her discussion of multiple language use in young learners and directs the reader’s attention to chapter 21 which proposes a final reconceptualization of multilingualism and multilingual education, and provides last recommendations for curriculum considerations and for future research.
21 Reconceptualising language learning and teaching in South Tyrol. Final Recommendations. South Tyrol is caught between two opposing positions. On the one hand, there is an obsession to preserve the status quo, i.e., the monolingual habitus formalised by Art. 19 of the Autonomy Statute (the author borrows the term from Gogolin, 1994) so as to safeguard the German minority language and cultural heritage and ensure education in L1 German. On the other, there is, in the light of a rapidly changing society and global economy, also the desire to promote multiple language learning so as not to fall behind and remain competitive at the European and global level. Unfortunately, it is the former position which prevails and has the strongest supporters. It is important to bear in mind that the negative attitudes towards all things multilingual are linked to a number of factors, including unknowingness and the deep-seated fear that investing in multilingual learning and education will go to the detriment of L1 German. More specifically there is the apprehension that fostering second and foreign language learning (beyond the necessary minimum) may result in a significant loss of German competence not only in the individual speaker, but at the societal level. In combination with a concomitant loss of cultural and historical identity this has, so the critics’ prediction, the potential to compromise social cohesion in the province, to tip the balance of power and to weaken the position of the German community. We do of course need to take these fears seriously, anything else would be complacent41. However, as researchers we need to go further asking what can be done to dispel such profound and debilitating concerns, and we need to provide answers and concrete solutions, always in the hope that the results generated by our research will come to inform educational policies and practices (cf. De Angelis, 2007:138). In the following the author attempts to formulate some propositions on how to counter the existing prejudices against multilingual learning. It is the author’s contention, and it has been argued throughout this book, that a paradigm shift and radical change of perspective is required among scholars investigating multiple language learning and multilingual education. This becomes particularly apparent if one considers that traditional theories and approaches have largely failed to provide satisfactory answers relating to
41 This said, one cannot overlook the fact that such fears are frequently the result of intricate political power games (cf. Garcia, 2009; Shohamy, 2006).
Reconceptualising language learning and teaching in South Tyrol
199
important issues in TLA and multilingualism research. Recent developments within the scientific community, which have seen the number of scholars converging on the need to approach multilingual phenomena from a holistic and multiperspective angle increase, are therefore to be welcomed. As observed by Aronin & Hufeisen (2009), indications suggest that the holistic view on multilingualism “has finally gained currency” (155). The adoption of a holistic and multiperspective view implies rethinking and reconceptualising many of the traditional ideas and core tenets in the field. In particular, it entails rethinking the native speaker norm (Cenoz & Jessner, 2009:133; Garcia, 2009:48) and the replacement of the latter with a multilingual standard (Herdina & Jessner, 2002; Jessner, 2006, 2008), which acknowledges the multilingual learner/speaker as a different but competent (and predominantly domain-specific) user of several languages or linguistic varieties. In addition to what has been stated above, a change in public perception is needed. In order to bring about this change it is crucial to raise public awareness as to the linguistic, cognitive and social benefits of multilingual learning and education42. The general public, and stakeholders and decision makers, in particular, need to gain a better understanding of the cognitive and linguistic implications of multiple language learning. It is paramount that they comprehend the basic dynamics and complexities of multiple language acquisition and development. It is further contended here that teachers need to be given specific and extensive training grounded in research (ideally as part of their university education), as well as in-service consultancy and easier access to relevant research findings. This desideratum is in line with Jessner (2008:45) and Krumm & Reich (2013:36) who both argue that there is an urgent need for teachers to become “experts on multilingualism” (see also Krumm, 2005:35; Meißner, 2005:130–133; Otwinowska-Kasztelanic, 2011:15; Allgäuer-Hackl & Jessner, 2013: 136). It is equally in accordance with Garcia (2009) who points to the central role of the school in the development of a positive attitude towards multilingualism and who calls upon schools and teachers to meet their responsibilities (54; see also Cummins, 2007 and Blackledge & Creese, 2010). These demands are consonant with those of numerous other scholars (Grosjean, 1985,2001; Herdina & Jessner, 2002; Cook, 2002; Olshtain & Nissim-Amitai, 2004; Krumm, 2005; Larsen-Freeman & Cameron, 2008; Cenoz, 2009; Cenoz & Jessner, 2009; Hufeisen, 2005,11; Hobbs, 2012; Vetter, 2012; Allgäuer-Hackl & Jessner, 2013) who, over the last few years, have repeatedly emphasized the necessity to abandon traditional reductionist 42 This could be achieved through public lectures in schools or libraries, parents’ evenings, or through relevant publications.
200
Reconceptualising language learning and teaching in South Tyrol
conceptions and theories of L2/L3 learning, calling instead for the adoption of more holistic and multiperspective approaches to multilingual learning and development. We note that these efforts are in full accord with the Council of Europe’s Common European Framework of Reference for Languages. The Common European Framework makes significant and unequivocal claims concerning plurilingual approaches to language learning and education and about how such approaches have the potential to change people’s ideas and notions about multilingual proficiency and education. In compliance with the above stated and in compliance with the central tenets of this volume, the document postulates that the plurilingual user (to utilise the Council’s terminology) does not keep her/his “languages and cultures in strictly separated mental compartments, but rather builds up a communicative competence to which all knowledge and experience of language contributes and in which languages interrelate and interact” (Council of Europe, 2001:4). The document goes on to state that “[f] rom this perspective, the aim of language education is profoundly modified. It is no longer seen as simply to achieve ‘mastery’ of one or two, or even three languages, each taken in isolation, with the ‘ideal native speaker’ as the ultimate model. Instead, the aim is to develop a linguistic repertory, in which all linguistic abilities have a place” (ibid.:5). Crucially, this entails acknowledging the multilingual user as someone who is not the sum of discrete (balanced or complete) monolingual competencies, but as someone who has a multiple linguistic repertoire with domain-specific competence in L1, L2, L3, Ln. Educational authorities and policy makers in South Tyrol are advised to take serious account of the Council’s recommendations and its sociolinguistic and educational implications. Although still negotiating its position, research in the last few years has developed considerable sophistication in investigating TLA and multilingual development. A brief outline is given below of some of the most relevant findings produced in this field in the recent past. It is intended to reiterate the benefits of L3 learning and education, and to underscore the principal scope of the present volume which consisted in investigating and identifying some of the effects of multilingual learning. Importantly, the recent literature indicates that learners/speakers of multiple languages – process and use language(s) differently and more effectively (Cook, 2002; Herdina & Jessner, 2002) – have already acquired an L2 and are therefore more experienced, more efficient and more successful language learners (Jessner, 2006, 2008, 2008b; Rahimpour, 2002; Valencia & Cenoz, 1992; Keshavarz & Astaneh (2004) – are better at exploiting their linguistic resources (Jessner, 2006; Rutgers, 2012)
Reconceptualising language learning and teaching in South Tyrol
– – – –
– – –
201
adopt a range of targeted strategies (McLaughlin, 1990; Mohanty, 1994; Gibson & Hufeisen, 2003) which allows them to progress faster (Safont Jorda’, 2005) reach higher levels of metalinguistic awareness and cross-linguistic ability and skill (Jessner, 1999, 2006; Gibson & Hufeisen, 2011) develop marked analytical competencies with regard to linguistic forms and structures (Kemp, 2001; Griessler, 2001) display a remarkable cognitive flexibility and metacognitive skill (Bialystok, 2011; but see also Marsh & Hill, 2009; Garcia, 2009; Peal & Lambert, 1962; Ben-Zeev, 1977; Lambert, 1991) display an enhanced ability to overcome lexical and syntactic difficulties across languages (Gibson & Hufeisen, 2003) are better communicators and tend to have more self-confidence, i.e. lower levels of communicative anxiety (Dewaele, 2002 in De Angelis, 2007) can function across languages and cultures, which translates into great advantages at the interpersonal, socio-cultural and pragmatic level (Herdina & Jessner, 2002; Safont Jorda`, 2005)).
The above compilation is far from exhaustive. It is also clear that many more, ideally longitudinal, studies and robust, reliable results are needed before any definitive conclusions can be drawn. The author holds the view that South Tyrol constitutes an exciting and fruitful arena for the further investigation of bi- and multilingual phenomena. Research in South Tyrol should particularly be directed at augmenting our knowledge and understanding of a. the bi/multilingual mind b. bi/multilingual learning and teaching and bi/multilingual educational programmes c. the effects of learners’ attitudes and their motivation(s) for learning d. the impact of sociocultural and sociopolitical factors on individual development More specifically, future research in South Tyrol should address the following questions: Which are the cognitive and linguistic effects of learning several languages at the institutional level from an early age? What role do metalinguistic awareness and abilities play in (multiple) language acquisition? How can prior linguistic knowledge be exploited so as to promote multilingual learning in school?
202
Reconceptualising language learning and teaching in South Tyrol
How can (multiple) language learning in South Tyrol be improved? How do second and third languages affect L1 development and vice versa? How can individual multicompetence further South Tyrol’s competitiveness on the European and global market? To conclude the present discussion and to reiterate her standpoint the author cites De Angelis (2007) who writes “Clearly, adding a language to the mind is bound to enrich us linguistically and culturally, and what is gained can generate effects of various kinds.” (138). In this sense it is anticipated that future research into multilingual learning and education in South Tyrol will generate robust and meaningful data evidencing that L2/L3 learning is linked to significant advantages at the linguistic, cognitive/academic and societal level, also in language minority contexts such as South Tyrol.
22 Conclusion The chapters in this book have expanded on various issues pertaining to (early) multilingual acquisition and development, including the representation and processing of (multiple) languages, child acquisition, multilingual schooling and multilingual awareness. Conceptually, this volume was informed and guided by a dynamic and holistic view of multilingualism. In particular, the present author has drawn on the theoretical framework of DMM which conceives of multilingual speakers as complex psycholinguistic systems (Herdina & Jessner, 2002:3). A dynamic and holistic perspective on tri- and multilingualism, as adopted here, entails the acceptance that multilingual behaviours are not reducible to simple linear explanations or solutions but that they emerge quite unpredictably through the complex and dynamic interaction of a multitude of interconnected variables. Looking at these variables in isolation will (if it is at all possible) prove a futile undertaking since it is the complex interplay of multiple factors, and not single variables in isolation, which impacts and determines the development of the overall system. As anticipated, the outcomes of this research project illustrate that early acquisition of an L2 and L3 does not harm or confuse young learners. Instead, the results obtained demonstrate that early multilingual learning enhances subjects’ cognitive and linguistic abilities. The study compared four groups of fourth form pupils (in high-immersion and low-immersion programmes) with regards to their performance in L2 and L3 and with regards to their metalinguistic awareness and abilities. The results indicate that the respondents in the highimmersion programmes possess a more enhanced linguistic sensitivity and metalinguistic awareness than their counterparts in the low-immersion groups. The present study also implies that the testees in the high-immersion groups are better equipped for learning an L3 than their peers in the low-immersion programmes. The findings of this study are important for at least two reasons. First, the results evidence the positive implications of high-immersion multilingual learning on children’s knowledge of L2 and L3 and on their metalinguistic awareness and cross-linguistic sensitivity. Second, it is hoped that the outcomes of the present study will contribute to a better understanding of multilingual acquisition and development and result in greater acceptance of multilingual programmes in South Tyrol. Crucially, as children learn a second or third language, they also learn about the culture, the values or norms, the beliefs and the traditions associated with a particular language. The acquisition of additional languages therefore not only
204
Conclusion
benefits the child’s linguistic development but also expands their socio-cultural knowledge and widens their perception of the world. It follows that the gains involved in multilingualism are by no means restricted to language alone but that they are of a more general nature because they allow the speaker to move and function in very diverse linguistic and social contexts. Bearing in mind the socio-historical background of the Italian and German community in South Tyrol, it is important to note that multilingual speakers’ ability or willingness to move in two or more cultural worlds must not, as it frequently happens, be misinterpreted as relinquishing one’s own identity and culture but may rather be viewed as indicative of the speaker’s ability to adopt and integrate those aspects of the other language or culture that are conducive to successful communication and interaction beyond the boundaries of one’s own language community. A final point to be made here concerns the political objections and hurdles which stand in the way of a more widespread and systematic implementation of (high-immersion) multilingual education schemes, particularly in German schools. The present author contends that German as a minority language on Italian territory will not suffer from the introduction of multilingual schooling. The German language will not be lost on account of multilingual education but will instead reinforce its status. If it is true that multiple language acquisition fosters speakers’ metalinguistic and metacognitive awareness, then German children and young adults in multilingual programmes will come to cherish their L1 more, and this appreciation will lead them to cultivate and promote their native tongue. They will be especially motivated to maintain their language and culture (see also Garcia, 2009:100). From this it follows that the fears and misgivings surrounding multilingual instruction are wholly unfounded. They certainly have no scientific basis. Multilingual education therefore needs to be acknowledged for what it is, namely a powerful means of promoting learners’ mastery of their first and second languages and linguistic awareness (cf. Lasagabaster, 2001:313; Cummins, 2007:238; Jessner, 2008; Garcia, 2009).
Bibliography Aitchison, J. 2008. The Articulate Mammal. An Introduction to Psycholinguistics. 5th ed. London/New York: Routledge. Alcock, A. 2001. The South Tyrol Autonomy. A short Introduction. County Londonderry, Bozen/ Bolzano. Allgäuer-Hackl, E. (in preparation) The development of metalinguistic awareness in multilingual learners. How effective is multilingual training in the classroom? Allgäuer-Hackl, E. & Jessner, U. 2013. Mehrsprachigkeitsunterricht aus mehrsprachiger Sicht. Zur Förderung des metalinguistischen Bewusstseins. In Vetter, E. (ed.) Professionalisierung für sprachliche Vielfalt. Perspektiven für eine neue Lehrerinnenbildung. Baltmannsweiler: Schneider Verlag Hohengehren. 111–147. Andreou, G. 2007. Phonological Awareness in Bilingual and Trilingual Schoolchildren. Linguistics Journal, Vol. 3, Issue 3, 8–15. Anstatt, T. (ed.). 2007. Mehrsprachigkeit bei Kindern und Erwachsenen. Erwerb. Formen. Förderung. Tübingen: Attempto. Aronin, L. 2012. Current multilingualism and new developments in multilingual research. Paper presented at the 8th International Conference on Third Language Acquisition and Multilingualism, 13.–15. September 2012, Castellò, Spain. Aronin, L. & O’Laoire. 2004. Exploring Multilingualism in Cultural Contexts: Towards a Notion of Multilinguality. In Hoffmann, C. & Ytsma, J. (eds.) Trilingualism in Family, School and Communitiy. Clevedon: Multilingual Matters. 11–29. Aronin, L. & Hufeisen, B. 2009. On the genesis and development of L3 research, multilingualism and multiple language acquisition. In Aronin, L. & Hufeisen, B. (eds.) The Exploration of Multilingualism. Amsterdam: John Benjamins Publishing Company. 1–9. Athanasopoulos, P. 2006. Effects of the grammatical representation of number on cognition in bilinguals. Bilingualism: Language and Cognition, 9, 1, 89–96. Baddeley, A. D., Thomson, N. & Buchanan, M. 1975. Word Length and the Structure of ShortTerm Memory. Journal of Verbal Learning and Verbal Behavior, 14, 575–589. Baker, C. 2006. Foundations of Bilingual Education and Bilingualism. 4th ed. Clevedon (et al.): Multilingual Matters. Balzaretti, L. & Montagna, F. 2009. Easy English with games and activities (4). Recanati. ELI. Bardel, C. & Falk, Y. 2007. The role of the second language in third language acquisition: the case of Germanic syntax. Second Language Research, 23, 4, 459–484. Barnes, J. D. 2006. Early Trilingualism. A Focus on Questions. Clevedon: Multilingual Matters. Barrena, A., Ezeizabarrena, M. J. & Garcia, I. 2008. Influence of the Linguistic Environment on the Development of the Lexicon and Grammar of Basque Bilingual Children. In Perez-Vidal, C., Juan-Garau, M. & Bel, A. (eds.) A Portrait of the Young in the New Multilingual Spain. Clevedon: Multilingual Matters. 86–110. Beardsmore, H.B. 2003. Who is afraid of bilingualism? In Dewaele, J.M., Housen, A. & Wei, L. (eds.) Bilingualism: Beyond basic principles. Clevedon, England: Multilingual Matters. 10–27. Beardsmore, H.B. 2008. Multilingualism, Cognition, and Creativity. International CLIL Research Journal, Vol. 1, 1, 4–19. Ben-Zeev, S. 1977. Mechanisms by Which Childhood Bilingualism Affects Understanding of Language and Cognitive Structures. In Hornby, P.A. (ed.) Bilingualism. Psychological, Social, and Educational Implications. New York: Academic Press. 29–55.
206
Bibliography
Ben-Zeev, S. 1977a. The Influence of Bilingualism on Cognitive Strategy and Cognitive Development. Child Development, 48, 1009–1018. Bernaus, M., Furlong, À., Jonckheere, S. & Kervran, M. 2011. Plurilingualism and pluriculturalism in content-based teaching: A training kit. Council of Europe. Bhatia, T. K. & Ritchie, W. C. 2004. The Handbook of Bilingualism. Malden, MA: Blackwell. Bialystok, E. 1991a. Introduction. In Bialystok, E. (ed.) Language Processing in Bilingual Children. Cambridge: Cambridge University Press. 1–7. Bialystok, E. 1991b. Metalinguistic dimensions of bilingual language proficiency. In Bialystok, E. (ed.) Language Processing in Bilingual Children. Cambridge: Cambridge University Press. 113–139. Bialystok, E. 1992. Selective Attention in Cognitive Processing: The Bilingual Edge. In Harris, R. J. (ed.) Cognitive Processing in Bilinguals. Elsevier Science Publishers B.V. 501–513. Bialystok, E. 1993. Metalinguistic awareness: The development of children’s representations of language. In Pratt, C. & Garton, A. F. (eds) Systems of Representation in Children: Development and Use. Chichester: John Wiley & Sons. 211–223. Bialystok, E. 2001. Metalinguistic Aspects of Bilingual Processing. Annual Review of Applied Linguistics, 21, 169–181. Bialystok, E. 2002. Cognitive Processes of L2 Users. In Cook, V. (ed.) Portraits of the L2 User. Clevedon: Multilingual Matters. 147–169. Bialystok, E. 2004. Bilingualism in Development. Language, Literacy, & Cognition. Cambridge: Cambridge University Press. Bialystok, E. 2005. Consequences of Bilingualism for Cognitive Development. In Kroll, J. F. & De Groot, A. M. B. (eds.) Handbook of Bilingualism: Psycholinguistic Approaches. Oxford, New York: Oxford University Press. 417–432. Bialystok, E. 2007. Cognitive Effects of Bilingualism: How Linguistic Experience Leads to Cognitive Change. The International Journal of Bilingual Education and Bilingualism, 10, 3, 210– 223. Bialystok, E. 2009. Bilingualism: The good, the bad, and the indifferent. Bilingualism: Language and Cognition, 12, 1, 3–11. Bialystok, E. & Hakuta, K. 1999. Confounded Age: Linguistic and Cognitive Factors in Age Differences for Second Language Acquisition. In Birdsong, D. (ed.) Second Language Acquisition and The Critical Period Hypothesis. Mahwah, N.J.: Lawrence Erlbaum. 161–181. Bialystok, E. & Martin, M. M. 2004. Attention and Inhibition in Bilingual Children: Evidence from the Dimensional Change Card Sort Task. Developmental Science 7:3, 325–339. Bialystok, E., Craik, F., Klein, R. & Viswanathan, M. 2004. Bilingualism, aging, and cognitive control: evidence from the Simon task. Psychology and Aging, 19, 2, 290–303. Birdsong, D. 1989. Metalinguistic Performance and Interlinguistic Competence. Berlin: Springer. Birdsong, D. 1999. Second Language Acquisition and the Critical Period Hypothesis. Mahwah, N.J.: Lawrence Erlbaum. Birdsong, D. 1999a. Introduction: Whys and Why Nots of the Critical Period Hypothesis for Second Language Acquisition. In Birdsong, D. (ed.) Second Language Acquisition and The Critical Period Hypothesis. Mahwah, N.J.: Lawrence Erlbaum. 1–22. Blackledge, A. & Creese, A. 2010. Multilingualism. A critical perspective. London: Continuum. Bleckner, C., Blythe, R., Bybee, J., Christiansen, M. H., Croft, W., Ellis, N. C., Holland, J., Ke, J., Larsen-Freeman, D., & Schoenemann, T. 2009. Language Is a Complex Adaptive System: Positition Paper. In Ellis, N. C. & Larsen-Freeman, D. (eds.) Language as a Complex Adaptive System. Malden: Wiley-Blackwell. 1–26.
Bibliography
207
Bleyhl, W. 2005. Die Defizite des traditionellen Fremdsprachenunterrichts oder: Weshalb ein Paradigmenwechsel, eine Umkehr, im Fremdsprachenunterricht erfolgen muss. In Gnutzmann, C., Königs, F. G. & Zöfgen, E. (eds.) Fremdsprachen Lehren und Lernen. Tübingen: Gunter Narr. 45–64. Blumenfeld, H.K. & Marian, V. 2005. Covert bilingual language activation through cognate word processing: An eye-tracking study. Proceedings of the twenty-seventh annual meeting of the cognitive science society; Mahwah: Lawrence Erlbaum; 2005. 286–291. Blyth, C. S. 2009. The Impact of Pedagogical Materials on Critical Language Awareness: Assessing Student Attention to Patterns of Language Use. In Turnball, M. & Dailey-O’Cain, J. (eds.) First Language Use in Second and Foreign Language Learning. Bristol, Buffalo, Toronto: Multilingual Matters. 163–181. Bock, K., Dell, G. S., Chang, F. & Onishi, K. H. 2007. Persistent structural priming from language comprehension to language production. Cognition, 104, 437–458. Bongaerts, T. 1999. Ultimate Attainment in L2 Pronunciation: The Case of Very Advanced Late L2 Learners. In Birdsong, D. (ed.) Second Language Acquisition and The Critical Period Hypothesis. Mahwah, N.J.: Lawrence Erlbaum. 133–159. Bongaerts, T., van Summeren, C., Planken, B. & Schils, E. 1997. Age and ultimate attainment in the pronunciation of a foreign language. SSLA, 19, 447–465. Bono, M. 2011. Crosslinguistic Interaction and Metalinguistic Awareness in Third Language Acquisition. In De Angelis, J. & Dewaele, J.-M. (eds.) New Trends in Crosslinguistic Influence and Multilingualism Research. Bristol, Buffalo, Toronto: Multilingualism Matters. 25–52. Bosch, L. & Sebastian-Galles, N. 2001. Early language differentiation in bilingual infants. In Cenoz, J. & Genesee, F. (eds.) Trends in Bilingual Acquisition. Amsterdam/Philadelphia: John Benjamins. 71–93. Bournot-Trites, M. & Reeder, K. 2001. Interdependence Revisited: Mathematics Achievement in an Intensified French Immersion Program. The Canadian Modern Language Review, 58, 1, 27–43. Bowey, J. A. 1986. Syntactic Awareness and Verbal Performance from Preschool to Fifth Grade. Journal of Psycholinguistic Research, Vol. 15, No. 4, 285–308. Broersma, M. & De Bot, K. 2006. Triggered codeswitching: A corpus-based evaluation of the original triggering hypothesis and a new alternative. Bilingualism: Language and cognition, 9, 1, 1–13. Bruck, M. & Genesee, F. 1995. Phonological awareness in young second language learners. Journal of Child Language, 22, 307–324. Brysbaert, M. & Dijkstra, T. 2006. Changing views on word recognition in bilinguals. In Morais, J. & d’Ydewalle, G. (eds.) Bilingualism and second language acquisition. Brussels: KVAB. 25–37. Cameron, L. & Larsen-Freeman, D. 2008. Complex Systems and Applied Linguistics. Oxford: Oxford University Press. Carter, R. 2003. Key Concepts in ELT. Language awareness. ELT Journal Volume 57/1 January, 64–65. Carvalho, A. M., Freire, J. L. & da Silva, A. J. B. 2010. Teaching Portuguese to Spanish Speakers: A Case for Trilingualism. Hispania, 93.1, 70–75. Caspi, T. & Lowie, W. 2010. A Dynamic Perspective on L2 Lexical Development in Academic English. In Chacon-Beltran, R. et al. (eds.) Insights into Non-native Vocabulary Teaching and Learning. Bristol: Multilingual Matters. 41–58.
208
Bibliography
Cenoz, J. 2001. The effect of linguistic distance, L2 status and age on cross-linguistic influence in third language acquisition. In Cenoz, J., Hufeisen, B. & Jessner, U. (eds.) Crosslinguistic Influence in Third Language Acquisition: Psycholinguistic Perspectives. Clevedon: Multilingual Matters. 8–20. Cenoz, J. 2003. The role of typology in the organization of the multilingual lexicon. In Cenoz, J., Hufeisen, B. & Jessner, U. (eds.) The Multilingual Lexicon. Dordrecht: Kluwer. 103–116. Cenoz, J. 2003a. The additive effect of bilingualism on third language acquisition: A review. International Journal of Bilingualism, 7, 71–88. Cenoz, J. 2004. Teaching English as a Third Language: The Effect of Attitudes and Motivation. In Hoffmann, C. & Ytsma, J. (eds.) Trilingualism in Family, School and Communitiy. Clevedon: Multilingual Matters. 202–218. Cenoy, J. 2009. Towards Multilingual Education. Basque Educational Research from an International Perspective. Bristol, Buffalo, Toronto: Multilingual Matters. Cenoz, J. 2012. Form monolingual to multilingual perspectives in Third Language Acquisition research. Paper presented at the 8th International Conference on Third Language Acquisition and Multilingualism, 13.–15. September 2012, Castellò, Spain. Cenoz, J. & Genesee, F. (eds.). 2001. Trends in Bilingual Acquisition. Amsterdam/Philadelphia: Benjamins Publishing Company. Cenoz, J. & Jessner, U. (eds.). 2000. English in Europe: The Acquisition of a Third Language. Clevedon: Multilingual Matters. Cenoz, J. & Jessner, U. 2009. The study of multilingualism in educational contexts. In Aronin, L. & Hufeisen, B. (eds.) The Exploration of Multilingualism. Amsterdam: John Benjamins Publishing Company. 121–138. Cenoz, J. & Lindsay, D. 1994. Teaching English in primary school: a project to introduce a third language to eight-year-olds. Language and Education, 8, 201–210. Cenoz, J. & Valencia, J. 1994. Additive Trilingualism: Evidence from the Basque Country. Applied Psycholinguistics, 15, 197–209. Cenoz, J., Hufeisen, B. & Jessner, U. (eds.). 2001. Cross-linguistic Influence in Third Language Acquistion: Psycholinguistic Perspectives. Clevedon. Cenoz, J., Hufeisen, B. & Jessner, U. (eds.). 2003a. The Multilingual Lexicon. Dordrecht: Kluwer Academic Publishers. Cenoz, J., Hufeisen, B. & Jessner, U. 2003b. Why investigate the multilingual lexicon? In Cenoz, J., Hufeisen, B. & Jessner, U. (eds.) The Multilingual Lexicon. Dordrecht: Kluwer. 1–9. Chang, F., Dell, G. S., Bock, K. & Griffin, Z. M. 2000. Structural Priming as Implicit Learning: A Comparison of Models of Sentence Production. Journal of Psycholinguistic Research, 29, 2, 217–229. Charkova, K. D. 2005. Early Foreign Language Education and Metalinguistic Development: A Study of Monolingual, Bilingual and Trilingual Children on Noun Definition Tasks. Proceedings of the 4th International Symposium on Bilingualism, ed. James Cohen, Kara T. McAlister, Kellie Rolstad, and Jeff MacSwan, 506–521. Somerville, MA: Cascadilla Press. Chen, X., Ku, Y. M., Koyama, E., Anderson, R., & Li W. 2008. Development of Phonological Awareness in Bilingual Chinese Children. J Psycholinguist Res, 37: 405–418. Chomsky, N. 1986. Knowledge of Language. Its Nature, Origin, and Use. Westport: Praeger Publishers. Clahsen, H. 1999. Lexical entries and rules of language: A multidisciplinary study of German inflection. Behavioral and Brain Sciences 22, 991–1060.
Bibliography
209
Clark, E. V. 1978. Awareness of Language: Some evidence from what children say and do. In Sinclair, A., Jarvella, R. J. & Levelt, W. J. M. (eds.) The Child’s Conception of Language. Berlin: Springer. 17–34. Clyne, M. 1997. Some of the Things Trilinguals Do. In Li Wei (ed.). 2010. Bilingualism and Multilingualism. Critical Concepts in Linguistics, Vol. I, London and New York: Routledge. 242– 265. Clyne, M. 2003a. Towards a More Language-centred Approach to Plurilingualism. In Dewaele, J.-M., Housen, A. & Wei, L. (eds) Bilingualism: Beyond Basic Principles. Clevedon: Multilingual Matters. 43–55. Clyne, M. 2003b. Dynamics of Language Contact. Cambridge: Cambridge University Press. Cook, V. 1995. Multi-competence and the learning of many languages. Language, Culture and Curriculum 8 (2), 93–98. Cook, V. 1997. The Consequences of Bilingualism for Cognitive Processing. In De Groot, A. M. B. & Kroll, J. F. (eds) Tutorials in Bilingualism. Psycholinguistic Perspectives. Mahwah New Jersey: Lawrence Erlbaum Associates. 279–299. Cook, V. (ed.). 2002. Introduction. Portraits of the L2 User. Clevedon: Multilingual Matters. Cook, V. (ed.). 2003. Effects of the L2 on the L1. Clevedon: Multilingual Matters. Costa, A. 2005. Lexical Access in Bilingual Production. In Kroll, J.F. & De Groot, A.M.B. (eds.) Handbook of bilingualism: Psycholinguistic Approaches. Oxford, New York: Oxford University Press. 308–325. Costa, A. & Caramazza, A. 1999. Is Lexical Selection in Bilingual Speech Production LanguageSpecific? Further Evidence from Spanish-English and English-Spanish bilinguals. In Li Wei (ed.). 2010. Bilingualism and Multilingualism. Critical Concepts in Linguistics. London and New York: Routledge. 120–141. Council of Europe. 2001. Common European Framework of Reference for Languages: Learning, Teaching, Assessment. Cambridge: Cambridge University Press. Cummins, J. 1976. The influence of bilingualism on cognitive growth: A synthesis of research findings and explanatory hypotheses. Working Papers on Bilingualism, 9, 1–43. Cummins, J. 1979. Linguistic interdependence and the educational development of bilingual children. Review of Educational Research, 49 (2), 222–251. Cummins, J. 1991. Interdependence of first- and second-language proficiency in bilingual children. In Bialystok, E. (ed.) Language Processing in Bilingual Children. Cambridge: Cambridge University Press. 70–89. Cummins, J. 2000. Language, power and pedagogy: Bilingual children in the crossfire. Clevedon: England: Multilingual Matters. Cummins, J. 2001. Children’s Mother Tongue: Why Is It Important for Education? Sprogforum, 7, 19, 15–20. Cummins, J. 2003. Bilingual Education: Basic Principles. In Dewaele, J.-M., Housen, A. & Wei, L. (eds) Bilingualism: Beyond Basic Principles. Clevedon: Multilingual Matters. 56–66. Cummins, J. 2007. Rethinking monolingual instructional strategies in multilingual classrooms, CJAL/RCLA, 10, 2, 221–240. Cummins, J. & Swain, M. 1992. Bilingualism in Education. 1986. Aspects of theory, research and practice. London and New York: Longman. De Angelis, G. 2007. Third or Additional Language Learning. Clevedon: Multilingual Matters. De Angelis, G. 2012. The effect of German L2 proficiency on English L3 writing development. Paper presented at the 8th International Conference on Third Language Acquisition and Multilingualism, 13.–15. September 2012, Castellò, Spain.
210
Bibliography
De Angelis, G. & Selinker, L. 2001. Interlanguage Transfer and Competing Linguistic Systems in the Multilingual Mind. In Cenoz, J., Hufeisen, B. & Jessner, U. (eds.) Cross-linguistic Influence in Third Language Acquistion: Psycholinguistic Perspectives. Clevedon: Multilingual Matters. 42–58. De Angelis, G. & Jessner, U. 2012. Writing across languages in a bilingual context: A dynamic systems theory perspective. In: Manchon, R. (ed.) L2 writing development: Multiple perspectives. Berlin/New York: Mouton de Gruyter. De Bot, K. 1992. A bilingual production model: Levelt’s “Speaking” model adapted. In Li Wei (ed.) 2007. The Bilingualism Reader. London and New York: Routledge. 384–404. De Bot, K. 1996. Review Article: The Psycholinguisticsof Output Hypothesis. Language Learning, 46: 3, 529–555. De Bot, K. 2000. Psycholinguistics in Applied Lingusitics: Trendd and Perspectives. Annual Review of Applied Linguistics, 20, 224–237. De Bot, K. 2004. The multilingual lexicon modelling selection. International Journal of Multilingualism 1, 1, 17–32. De Bot, K. 2008a. Review article: The imaging of what in the multilingual mind? Second Language Research, 24, 1, 111–133. De Bot, K. 2008b. Introduction: Second Language Development as a Dynamic Process. The Modern Language Journal, 92, ii, 166–178. De Bot, K. & Mailfert, K. 1982. The Teaching of Intonation: Fundamental Research and Classroom Applications. TESOL QUARTERLY, Vol. 16, No. 1, 71–77. De Bot, K. & Schreuder, R. 1993. Word production and the bilingual lexicon. In Schreuder, R. & Weltens, B. (eds.) The Bilingual Lexicon. Amsterdam: John Benjamins. 191–214. De Bot, K., Lowie, W. & Verspoor, M. 2005. Second Language Acquisition: An Advanced Research Book. London: Routledge. De Bot, K., Lowie, W. & Verspoor, M. 2007. Keynote Article: A Dynamic Systems Theory approach to second language acquisition. Bilingualism: Language and Cognition, 10, 1, 7–21. De Bot, K., Lowie, W. & Verspoor, M. 2007. Authors’ Response: A dynamic view as a complementary perspective. Bilingualism: Language and Cognition, 10, 1, 51–55. De Florio-Hansen, I., & Hu, A. (eds.). 2007. Plurilingualität und Identität. Zur Selbst- und Fremdwahrnehmung mehrsprachiger Menschen. Tübingen: Stauffenburg Verlag. De Groot, A. 1993. Word type effects in bilingual processing tasks: support for a mixed representational system. In Schreuder, R. & Weltens, B. (eds.) The Bilingual Lexicon. Amsterdam: John Benjamins. 27–51. De Groot, A. 2002. Lexicl Representation and Lexical Processing in the L2 User. In Cook, V. (ed.) Portraits of the L2 User. Clevedon: Multilingual Matters. 32–63. De Groot, A.M.B. & Kroll, J.F. 1997. Introduction and Overview. In De Groot, A. M. B. & Kroll, J. F. (eds) Tutorials in Bilingualism: Psycholinguistic Perspectives. Mahwah New Jersey: Lawrence Erlbaum Associates. 1–16. De Groot, A.M.B. & Kroll, J.F. (eds.). 1997. Tutorials in Bilingualism: Psycholinguistic Perspectives. Mahwah New Jersey: Lawrence Erlbaum Associates. De Houwer, A. 2004. Trilingual Input and Children’s Language Use in Trilingual Families in Flanders. In Hoffmann, C. & Ytsma, J. (eds.) Trilingualism in Family, School and Communitiy. Clevedon: Multilingual Matters. 118–135. De Houwer, A. 2005. Early Bilingual Acquisition: Focus on Morphosyntax and the Separate Development Hypothesis. In Kroll, J.F. & De Groot, A.M.B. (eds.) Handbook of Bilingualism: Psycholinguistic Approaches. Oxford, New York: Oxford University Press. 30–48.
Bibliography
211
Dell, G. S., Chang, F. & Griffin, Z. M. 1999. Connectionist Models of Language Production: Lexical Access and Grammatical Encoding. Cognitive Science, Vol. 23, 4, 517–542. Demographisches Handbuch für Südtirol 2010. 2011. Autonome Provinz Bozen – Südtirol Provincia Autonoma di Bolzano - Alto Adige. Landesinstitut für Statistik – ASTAT Dewaele, J.-M. 1998. Lexical Inventions: French Interlanguage as L2 versus L3. Applied Linguistics, 19/4, 471–490. Dewaele, J.-M. 2001. Activation or Inhibition? The Interaction of L1, L2, and L3 on the Language Mode Continuum. In Cenoz, J., Hufeisen, B. & Jessner, U. (eds.) Cross-linguistic Influence in Third Language Acquistion: Psycholinguistic Perspectives. Clevedon: Multilingual Matters. 69–89. Dewaele, J.-M., Housen, A. & Wei, L. 2003. Bilingualism: Beyond Basic Principles. Clevedon: Multilingual Matters. Diaz, R.M. 1983. Thought and Two Languages: The Impact of Bilingualism on Cognitive Development. Review of Research in Education, Vol. 10, 23–54. Dickinson, D. K., McCabe, A., Clark-Chiarelli, N. & Wolf, A. 2004. Cross-language transfer of phonological awareness in low-income Spanish and English bilingual preschool children. Applied Psycholinguistics, 25:3, 323–347. Diebold, A. & Richard, J. R. 1966. The consequences of early bilingualism in cognitive development and personality formation. Paper prepared for the symposium The Study of Personality: An Interdisciplinary Appraisal held at Rice University, Houston, Texas, November 4– 5, 1966. 1–32. http://eric.ed.gov/ERICWebPortal/search/detailmini.jsp?_nfpb=true & _& ERICExtSearch_SearchValue_0=ED020491 & ERICExtSearch_SearchType_0=no & accno=ED020491 Dijkstra, T. & Van Heuven, W. J. B. 1998. The BIA Model and Bilingual Word Recognition. In Li Wei (ed.). 2010. Bilingualism and Multilingualism. Critical Concepts in Linguistics. London and New York: Routledge. 55–89. Dijkstra, T. 2003. Lexical Processing in Bilinguals and Multilinguals. In Cenoz et al. (eds.), The Multilingual Lexicon. Dordrecht: Kluwer Academic Publishers. 11–26. Dijkstra, T. 2007. Task and Context Effects in Bilingual Lexical Processing. In Kecskes, I. & Albertazzi, L. (eds.) Cognitive Aspects of Bilingualism. Dordrecht: Springer. 213–235. Doman, E. 2005. Current Debates in SLA. Asian EFL Journal Quarterly, 7, 4, 131–143. Durgunoglu, A. Y. 1997. Bilingual Reading: Its Components, Development, and Other Issues. In De Groot, A. M. B. & Kroll, J. F. (eds.) Tutorials in Bilingualism. Psycholinguistic Perspectives. Mahwah New Jersey: Lawrence Erlbaum Associates. 255–276. Dussias, P. E. 2004. Parsing a first language like a second: The erosion of L1 parsing strategies in Spanish-English Bilinguals. International Journal of Bilingualism, 8, 3, 355–371. Edwards, J. 2007. Societal multilingualism: Reality, recognition and response. In Auer, P. & Wei, L. (eds.) Handbook of Multilingualism and Multilingual Communication. Berlin: Mouton de Gruyter. 447–467. Edwards, M. & Dewaele, J.-M. 2007. Trilingual conversations: a window onto multicompetence. International Journal of Bilingualism, 11, 2, 221–242. Edwards, H. T. & Kirkpatrick, A. G. 1999. Metalinguistic Awareness in Children: A Developmental Progression. Journal of Psycholinguistic Research, Vol. 28, No. 4, 313–329. Egger, K. 1977. Zweisprachigkeit in Südtirol. Bozen: Athesia. Ellis, N. C. 2005. Introduction to Part I. In Kroll, J.F. & De Groot, A.M.B. (eds.) Handbook of Bilingualism: Psycholinguistic Approaches. Oxford, New York: Oxford University Press. 3–8.
212
Bibliography
Ellis, N. C. 2008. Words and their usage. Commentary on the special issue on the bilingual mental lexicon. The Mental Lexicon, 3, 3, 375–385. Ellis, N. C. & Laporte, N. 1997. Contexts of Acquisition: Effects of Formal Instruction and Naturalistic Exposure on Second Language Acquisition. In De Groot, A. M. B. & Kroll, J. F. (eds) Tutorials in Bilingualism. Psycholinguistic Perspectives. Mahwah New Jersey: Lawrence Erlbaum Associates. 53–83. Ellis, N. C. & Larsen-Freeman, D. 2006. Language Emergence: Implications for Applied LinguisticsIntroduction to the Special Issue. Applied Linguistics, 27/4:558–589. Ellis, N. C. & Larsen-Freeman, D. 2009. Constructing a Second Language: Analyses and Computational Simulations of the Emergence of Linguistic Constructions From Usage. In Ellis, N. & Larsen-Freeman, D. (eds.) Language as a Complex Adaptive System. 90–123. Ellis, N. C. & Sinclair, S. G. 1996. Working Memory in the Acquisition of Vocabulary and Syntax: Putting Language in Good Order. The Quarterly Journal of Experimental Psychology, 49A, 1, 234–250. Ellis, R. 2009. Implicit and Explicit Learning, Knowledge and Instruction. In Ellis, R., Loewen, S., Elder, C., Erlam, R., Philp, J. & Reinders, H. (eds.) Implicit and Explicit Knowledge in Second Language Learning, Testing and Teaching. Bristol, Buffalo, Toronto: Multilingual Matters. 3–25. Eubank, L. & Gregg, K. R. 1999. Critical Periods and (Second) Language Acquisition: Divide et Impera. In Birdsong, D. (ed.) Second Language Acquisition and The Critical Period Hypothesis. Mahwah, N.J.: Lawrence Erlbaum. 65–99. Etxeberria, F. 2004. Trilinguals at Four? Early Trilingual Education in the Basque Country. In Hoffmann, C. & Ytsma, J. (eds.) Trilingualism in Family, School and Communitiy. Clevedon: Multilingual Matters.185–201. Eysenck, M. W. 1992. Anxiety: The Cognitive Perspective. East Sussex: Lawrence Erlbaum Associates. Fabbro, F. 1999. The Neurolinguistics of Bilingualism. An Introduction. Hove: Psychology Press. Fabbro, F. 2002. The Neurolinguistics of L2 Users. In Cook, V. (ed.) Portraits of the L2 User. Clevedon: Multilingual Matters. 197–218. Ferreira, F. & Henderson, J. M. 1991. How is Verb Information Used During Syntactic Parsing? In Simpson, G. B. (ed.) Understanding Word and Sentence. Elsevier Science Publishers B. V. (North-Holland). 305–330. Fillmore, L. W. 1991. Second-Language Learning in Children: A Model of Language Learning in Social Context. In Bialystok, E. (ed.) Language Processing in Bilingual Children. Cambridge: Cambridge Universtiy Press. 49–69. Fit in Deutsch 1. Übungssatz 02. Kandidatenblätter/Prüferblätter.Goethe-Institut. Flege, J. E. 1999. Age of Learning and Second Language Speech. In Birdsong, D. (ed.) Second Language Acquisition and The Critical Period Hypothesis. Mahwah, N.J.: Lawrence Erlbaum. 101–132. Flynn, S. & Lust, B. 2002. A Minimalist Approach to L2 Solves a Dilemma of UG. In Cook, V. (ed.) Portraits of the L2 User. Clevedon: Multilingual Matters. 95–120. Flynn, S., Foley, C. & Vinnitskaya, I. 2004. The cumulative-enhancement model for language acquisition: comparing adults’ and children’s patterns of development in first, second and third language acquisition of relative clauses. International Journal of Multilingualism, I, I, 3–16. Flynn, S., Foley, C. & Vinnitskaya, I. 2005. New Paradigm for the Study of Simultaneous v. Sequential Bilingualism. ISB4: Proceedings of the 4th International Symposium on Bilin-
Bibliography
213
gualism, ed. James Cohen, Kara T. McAlister, Kellie Rolstad, and Jeff MacSwan, 768–774. Somerville, MA: Cascadilla Press. Fortkamp, M. B. M. & Bergsleithner, J. M. 2007. Relationship among individual differences in working memory capacity, noticing, and L2 speech production. Signo. Santa Cruz do Sul, 32, 52, 40–53. Fortkamp, M. B. M., Prebianca, G. V. V. & Pires, A. P. 2006. Lexical Access in Speech Production: A Brief Literature Review. Matraga, 13, 9, 15–32. Rio de Janeiro: Caetes. Francis, W. S. 2005. Bilingual Semantic and Conceptual Representation. In Kroll, J. F. & De Groot, A. M. B. (eds.) Handbook of Bilingualism: Psycholinguistic Approaches. Oxford, New York: Oxford University Press. 251–267. French, R. M. & Jacquet, M. 2004. Understanding bilingual memory: models and data. TRENDS in Cognitive Sciences, 8, 2, 87–93. Gabrys-Barker, D. 2012. Perceptions of space in the multilingual mind. Paper given at the 8th International Conference on Third Language Acquisition and Multilingualism, 13.–15. September 2012, Castellò, Spain. Galambos, S.J. & Hakuta, K. 1988. Subject-specific and task-specific characteristics of metalinguistic awareness in bilingual children. Applied Psycholinguistics, 9, 141–162. Garcia, A. L. 2008. Classroom Bilingualism at an Early Age: Towards a More Natural EFL Context. In Perez-Vidal, C., Juan-Garau, M. & Bel, A. (eds.) A Portrait of the Young in the New Multilingual Spain. Clevedon: Multilingual Matters. 185–199. Garcia, O. 2009. Bilingual Education in the 21st Century. A Global Perspective. Chichester: Wiley-Blackwell. Garcia, O. 2012. Multilingualism in Education: Translanguaging for Social Justice. Paper presented at the 8th International Conference on Third Language Acquisition and Multilingualism, 13.–15. September 2012, Castellò, Spain. Garcia, O. & Sylvan, C. E. 2011. Pedagogies and Practices in Multilingual Classrooms: Singularities in Pluralities. The Modern Language Journal, 95, iii, 385–400. Genesee, F. 1989. Early bilingual language development: one language or two? In Wei, L. (ed.) 2007. The Bilingualism Reader. London and New York: Routledge. 320–335. Genesee, F. 2002. Portrait of the Bilingual Child. In Cook, V. (ed.) Portraits of the L2 User. Clevedon: Multilingual Matters. 170–196. Genesee, F. 2003. Rethinking Bilingual Acquisition. In In Dewaele, J.-M., Housen, A. & Wei, L. (eds) Bilingualism: Beyond Basic Principles. Clevedon: Multilingual Matters. 204–228. Genesee, F. 2009. Early childhood bilingualism: Perils and possibilities. Journal of Applied Research on Learning, Vol. 2, Special Issue, Art. 2. Genesee, F. & Cenoz, J. 2001. First words. In Cenoz, J. & Genesee, F. (eds.) Trends in Bilingual Acquisition. Amsterdam/Philadelphia: John Benjamins. 1–9. Genesee, F. & Nicoladis, E. 2007. Bilingual Acquisition. In Hoff, E. & Shatz, M. (eds.) Handbook of Language Development. Oxford, England: Blackwell. 234–342. Gibson, M. & Hufeisen, B. 2003. Investigating the Role of Prior Foreign Language Knowledge. In Cenoz, J., Hufeisen, B. & Jessner, U. (eds.) The Multilingual Lexicon. Dordrecht: Kluwer. 87–102. Gibson, M. & Hufeisen, B. 2011. Perception of Preposition Errors in Semantically Correct versus Erroneous Contexts by Multilingual Advanced English as a Foreign Language Learners: Measuring Metalinguistic Awareness. In De Angelis, J. & Dewaele, J.-M. (eds.) New Trends in Crosslinguistic Influence and Multilingualism Research. Bristol, Buffalo, Toronto: Multilingualism Matters. 74–85.
214
Bibliography
Gibson, M., Hufeisen, B. & Libben, G. 2001. Learners of German as an L3 and their Production of German Prepositional Verbs. In Cenoz, J., Hufeisen, B. & Jessner, U. (eds.) Cross-linguistic Influence in Third Language Acquisition: Psycholinguistic Perspectives. Clevedon: Multilingual Matters. 138–148. Gogolin, I. 1994. Der monolinguale Habitus der multilingualen Schule. Münster, New York, Waxmann-Verlag. Gombert, J. E. 1992. Metalinguistic Development. New York, NY (et al.): Harvester Wheatsheaf. Goodz, N. 1989. Parental language mixing in bilingual families. Infant Mental Health Journal, 1 (10), 25–43. Green, D. W. 1986. Control, activation, and resource: a framework and a model for the control of speech in bilinguals. In Wei, Li (ed.) 2007. The Bilingualism Reader. New York: Routledge. 371–383. Green, D. W. 1993. Towards a Model of L2 Comprehension and Production. In Schreuder, R. & Weltens, B. (eds.) The Bilingual Lexicon. Amsterdam: John Benjamins. 249–277. Green, D. W. 1998. Mental Control of the Bilingual Lexico-Semantic System. Bilingualism: Language and Cognition, 1, 67–81. In Li Wei (ed.) (2010. Bilingualism and Multilingualism. Critical Concepts in Linguistics. London and New York: Routledge. 31–54. Green, D. W. 2003. Neural basis of lexicon and grammar in L2 acquisition: The convergence hypothesis. In van Hout, R., Hulk, A., Kuiken, F. & Towell, R. (eds.) The Lexicon-Syntax Interface in Second Language Acquisition. Amsterdam: John Benjamins. 197–218. Green, D. W. 2008. Bilingual Aphasia: Adapted Language Networks and Their Control. Annual Review of Applied Linguistics, 28, 25–48. Griessler, M. 2001. The effects of third language learning on second language proficiency: an Austrian example. International Journal of Bilingual Education and Bilingualism, 4, 1, 50–60. Grosjean, F. 1985. The bilingual as a competent but specific speaker-learner. Journal of Multilingual and Multicultural Development 6: 467–477. Grosjean, F. 1997. Processing Mixed Language: Issues, Findings, and Models. In De Groot, A. M. B. & Kroll, J. F. (eds) Tutorials in Bilingualism. Psycholinguistic Perspectives. Mahwah New Jersey: Lawrence Erlbaum Associates. 225–254. Grosjean, F. 1998. Studying bilinguals: Methodological and conceptual issues. Bilingualism: Language and Cognition, I, 131–149. In Li Wei (ed.). 2010. Bilingualism and Multilingualism. Critical Concepts in Linguistics. London and New York: Routledge. 1–30. Grosjean, F. 2001. The bilingual’s language modes. In Nicol, J. L. (ed.) One Mind, Two Languages: Bilingual Language Processing. Oxford: Blackwell. 1–25. Haas, M. 1998. Early vs. late: the practitioner’s perspective. In Met, M. (ed.) Critical Issues in Early Second Language Learning. Glenview: Scott Foresman Addison-Wesley. 43–48. Hakes, D.T. 1980. The Development of Metalinguistic Abilities in Children. Berlin Springer. Hakuta, K. & Diaz, R. M. 1985. The Relationship Between Degree of Bilingualism and Cognitive Ability: A Critical Discussion and Some New Longitudinal Data. In Nelson, K. E. (ed.) Children’s Language Vol. 5. Hillsdale, N. J.: Erlbaum. 319–344. Hall, C. & Ecke, P. 2003. Parasitism as a default mechanism in L3 vocabulary acquisition. In Cenoz, J. Hufeisen, B. & Jessner, U. (eds.) The Multilingual Lexicon. Dordrecht: Kluwer. 71–86. Hammarberg, B. 2001. Roles of L1 and L2 in L3 production and acquisition. In Cenoz, J. Hufeisen, B. & Jessner, U. (eds.) Cross-linguistic Influence in Third Language Acquisition: Psycholinguistic Perspectives. Clevedon: Multilingual Matters. 21–41.
Bibliography
215
Hammarberg, B. 2010. The languages of the multilingual: Some conceptual and terminological issues. IRAL, 48, 2/3, 91–104. Harley, B. & Wang, W. 1997. The Critical Period Hypothesis: Where Are We Now? In De Groot, A. M. B. & Kroll, J. F. (eds) Tutorials in Bilingualism. Psycholinguistic Perspectives. Mahwah New Jersey: Lawrence Erlbaum Associates. 19–51. Harrington, M. 2005. Cognitive Pespectives on Second Language Acquisition. In Kaplan, R. B. (ed.) Oxford Handbook of Applied Linguistics. Oxford: Oxford University Press. 124–140. Hartsuiker, R. J. & Kroll, H. J. 2001. Error Monitoring in Speech Production: A Computational Test of the Perceptual Loop Theory. Cognitive Psychology, 42, 113–157. Hartsuiker, R. J. & Pickering, M. J. 2008. Language integration in bilingual sentence production. Acta Psychologica, 128, 279–489. Hartsuiker, R. J., Pickering, M. J. & Veltkamp, E. 2004. Is Syntax Separate or Shared Between Languages? Cross-Linguistic Syntactic Priming in Spanish-English Bilinguals. Psychological Science, 15, 6, 409–414. Haugen, E. 1977. Norm and Deviation in Bilingual Communities. In Hornby, P.A. (ed.) Bilingualism. Psychological, Social, and Educational Implications. New York: Academic Press. 91–102. Hawkins, E.W. 1999. Foreign Language Study and Language Awareness. Language Awareness, Vol.8, No.3 & 4, 124–142. Helot, C. 2008. Awareness Raising and Multilingualism in Primary Education. In Cenoz, J. & Hornberger, N. H. (eds.) Encyclopedia of Language and Education, 2nd Edition, Vol. 6: Knowledge about Language, New York: Springer. 371–384. Herdina, P. & Jessner, U. 2002. A Dynamic Model of Multilingualism. Changing the Psycholinguistic Perspective. Clevedon: Multilingual Matters. Heredia, R. R., Altarriba, J. (eds). 2002. Bilingual Sentence Processing. Advances in Psychology 134. Amsterdam: Elsevier. Herwig, A. 2001. Plurinlingual Lexical Organisation: Evidence from Lexical Processing in L1–L2– L3–L4 Translation. In Cenoz, J., Hufeisen, B. & Jessner, U. (eds.) Cross-linguistic Influence in Third Language Acquistion: Psycholinguistic Perspectives. Clevedon: Multilingual Matters. 115–137. Hills, A. E. & Caramazza, A. 1995. Representation of Grammatical Categories of Words in the Brain. Journal of Cognitive Neuroscience, 7, 3, 396–407. Hinger, B., Kofler, A., Skinner, A. & Stadler, W. 2005. The Innsbruck Model of Fremdsprachendidaktik: towards an integrated multilingual approach in pre-service teacher eduation. Teacher Trainer, 19, 1, 17–20. Hobbs, R. D. 2012. Diverse multilingual researchers contribute language acquisition components to an integrated model of education. International Journal of Multilingualism, 9, 3, 204–234. Hofer, B. 2013. On The Dynamics of Early Multilingualism. A Psycholinguistic Study of Bilingual Education in South Tyrol. Unpublished Doctoral Thesis. Innsbruck University. Hoffmann, Ch. 1991. An Introduction to Bilingualism. London: Longman. Hoffmann, Ch. 2001. Towards a Description of Trilingual Competence. International Journal of Bilingualism, Vol. 5, No. 1, 1–17. Hoffmann, Ch. & Stavans, A. 2007. The evolution of trilingual code-switching from infancy to school age: the shaping of trilingual competence through dynamic language dominance. International Journal of Bilingualism, 11, 1, 55–72. Hoffmann Ch. & Ytsma J. (eds.). 2004. Trilingualism in Family, School and Community. Clevedon: Multilingual Matters.
216
Bibliography
Hornby, P. A. 1977. Bilingualism: Psychological, Social, and Educational Implications. N.Y.: Academic Press. Hufeisen, B. 2005. Gesamtsprachencurriculum: Einflussfaktoren und Bedingungsgefüge. In Hufeisen, B. & Lutjeharms, M. (eds.) Gesamtsprachencurriculum Integrierte Sprachendidaktik Common Curriculum. Tübingen: Gunter Narr Verlag. 9–18. Hufeisen, B. 2011. Gesamtsprachencurriculum: Überlegungen zu einem prototypischen Model. In Baur, R. & Hufeisen, B. (eds.) „Vieles ist sehr ähnlich.“ – Individuelle und gesellschaftliche Mehrsprachigkeit als bildungspolitische Aufgabe. Baltmannsweiler, Schneider Hohengehren (Bd. 6 Mehrsprachigkeit und multiples Sprachenlernen), 265–282. Hufeisen, B. & Lutjeharms, M. (eds.). 2005. Gesamtsprachencurriculum - Integrierte Sprachendidaktik – Common Curriculum. Tübingen: Narr. Hufeisen, B. & Neuner, G. (eds.). 2003. Mehrsprachigkeitskonzept – Tertiärsprachen – Deutsch nach Englisch. Strasbourg: Council of Europe Publishing. Ianco-Worrall, A.D. 1972. Bilingualism and Cognitive Development. Child Development, 43 Issue 4, 1390–1400. Jackson, C. N. & Dussias, P. E. 2009. Cross-linguistic differences and their impact on L2 sentence processing. Bilingualism: Language and Cognition, 12, 1, 65–82. James, C. 1996. A Cross-Linguistic Approach to Language Awareness. Language Awareness, Vol. 5, No. 3 & 4, 138–148. James, C. & Garret, P. (eds.). 1991. Language Awareness in the Classroom. London: Longman. Jessner, U. 1997. Towards a dynamic view of multilingualism. In Pütz (ed.) Language Choices: Conditions, Constraints and Consequences. Amsterdam: John Benjamins, 17–30. Jessner, U. 1999. Metalinguistic awareness in multilinguals: cognitive aspects of third language learning. Language Awareness, 3 & 4: 201–209. Jessner, U. 2003. A Dynamic Approach to Language Attrition in Multillingual Systems. In Cook, V. (ed.) Effects of the Second Language on the First. Clevedon: Multilingual Matters. 234– 246. Jessner, U. 2003a. The nature of crosslinguistic interaction in multilinguals. In Cenoz, J., Hufeisen, B. & Jessner, U. (eds.) The Multilingual Lexicon. Dordrecht: Kluwer. 45–55. Jessner, U. 2003b. Das multilinguale Selbst: Perspektiven der Veränderung. In De FlorioHansen, I. & Hu, A. (eds.) Plurilingualität und Identität. Zur Selbst-und Fremdwahrnehmung mehrsprachiger Menschen. Tübingen: Stauffenburg. 25–37. Jessner, U. 2006. Linguistic Awareness in Multilinguals: English as a Third Language. Edinburgh: University Press. Jessner, U. 2008. State-of-the-Art Article. Teaching third languages: Findings, trends and challenges. Lang.Teach, 41:1, 15–56. Jessner, U. 2008a. A DST Model of Multilingualism and the Role of Metalinguistic Awareness. The Modern Language Journal, 92, ii, 270–283. Jessner, U. 2008b. Multicompetence Approaches to Language Proficiency Development in Multilingual Education. In Cummins, J. & Hornberger, N.H. (eds) Encyclopedia of Language and Education, 2nd Edition, Vol. 5: Bilingual Education, 91–103. Jiang, N. 2000. Lexical Representation and Development in a Second Language. Applied Lingustics, 21, 1, 47–77. Johnson, J. S. & Newport, E. 1991. Critical period effects on universal properties of language: The status of subjecency in the acquisition of a second language. Cognition, 39, 215–258.
Bibliography
217
Johnson, J. S. & Newport, E. 1989. Critical Period Effects in Second Language Learning: The Influence of Maturational State on the Acquisition of English as a Second Language. Cognitive Psychology, 21, 60–99. Kamhi, A. G. & Koenig, L. A. 1985. Metalinguistic Awareness in Normal and Language-Disordered Children. Language, Speech, and Hearing Services in Schools, 16, 199–210. Karmiloff, K. & Karmiloff-Smith, A. 2001. Pathways to Language. From Fetus to Adolescent. Cambridge, Massachusetts: Harvard University Press. Karmiloff-Smith, A. Grant, J., Sims, K., Jones, M.-C., & Cuckle, P. 1996. Rethinking metalinguistic awareness: representing and accessing knowledge about what counts as a word. Cognition, 58, 197–219. Kecskes, I. 2006. On my mind: thoughts about salience, context, and figurative language from a second language perspective. Second Language Research, 22, 2, 219–237. Kecskes, I. & Albertazzi, L. (eds.). 2007. Cognitive Aspects of Bilingualism. Dordrecht: Springer. Kellerman, E. 2001. New Uses for Old Language: Cross-linguistic and Cross-gestural Influence in the Narratives of Non-Native Speakers. In Cenoz, J., Hufeisen, B. & Jessner, U. (eds.) Cross-linguistic Influence in Third Language Acquistion: Psycholinguistic Perspectives. Clevedon: Multilingual Matters. 170–191. Kecskes, I. & Papp, T. 2000. Foreign Language and Mother Tongue. Mahwah, N. J.: Lawrence Erlbaum. Kemp, C. 2001. Metalinguistic awareness in multilinguals: implicit and explicit grammatical awareness and its relationship with language experience and language attainment. Doctoral thesis submitted to the University of Edinburgh. Kempen, G. & Hoenkamp, E. 1982. Incremental Sentence Generation: Implications for the Structure of a Syntactic Processor. Proceedings of the 9th International Conference on Computational Linguistics. Prague. 1–6. Kempen, G. & Hoenkamp, E. 1987. An Incremental Procedural Grammar for Sentence Formulation. Cognitive Science, 11, 201–258. Keshavarz, M. H. & Astaneh, H. 2004. The Impact of Bilinguality on the Learning of English Vocabulary as a Foreign Language (L3). Bilingual Education and Bilingualism. Vol. 7, No. 4, 295–302. Kharkhurin, A. V. 2007. The Role of Cross-Linguistic and Cross-Cultural Experiences in Bilinguals’ Divergent Thinking. In Kecskes, I. & Albertazzi, L. (eds.) Cognitive Aspects of Bilingualism. Dordrecht: Springer. 175–210. Kim, K., Relkin, N.L. Kyoung-Min & Hirsch, J. 1997. Distinct cortical areas associated with native and second languages. Nature, 388, 171–174. Kirkici, B. 2004. Foreign Language-Medium Instruction and Bilingualism: The Analysis of a Myth. Sosyal Bilimler Dergisi, 2, 109–121. Kocianovà, M. 2005. Metasprachliche Fähigkeiten zweisprachiger Kinder: Zum Zusammenhang von sprachlicher und metasprachlicher Leistungsfähigkeit und die damit einhergehenden Implikationen für eine adequate Förderung der Russisch-Deutsch sprechenden Kinder im Grundschulalter. Doctoral dissertation, submitted at LMU München: Fakultät für Psychologie und Pädagogik. Kormos, J. 2006. Speech Production and Second Language Acquisition. Mahwah, New Jersey: Lawrence Erlbaum Associates.
218
Bibliography
Kovacs, A. M. 2007. Beyond Language: Childhood Bilingualism Enhances High-Level Cognitive Functions. In Kecskes, I. & Albertazzi, L. (eds.) Cognitive Aspects of Bilingualism. Dordrecht: Springer. 301–323. Kroll, J. F. & De Groot, A. M. B. 1997. Lexical and Conceptual Memory in the Bilingual: Mapping Form to Meaning in Two Languages. In De Groot, A. M. B. & Kroll, J. F. (eds) Tutorials in Bilingualism. Psycholinguistic Perspectives. Mahwah New Jersey: Lawrence Erlbaum Associates. 169–199. Kroll, J. K. & Dussias, P. E. 2004. The Comprehension of Words and Sentences in Two Languages. In Bhatia, T. K & Ritchie, W. C. (eds.) The Handbook of Bilingualism. Malden, MA: Blackwell Publishing. 179–200. Kroll, J. K. & Tokowicz; N. 2001. The Development of Conceptual Representation for Words in a Second Language. In Nicol, J. L. (ed.) One Mind, Two Languages: Bilingual Language Processing. Oxford: Blackwell 49–71. Kroll, J. K. & Tokowicz; N. 2005. Models of bilingual representation and processing. In Kroll, J.K. & De Groot, A. M. B. (eds.) Handbook of Bilingualism: Psycholinguistic Approaches. Oxford: Oxford University Press. 531–553. Kroll, J. K. & Linck, J. A. 2007. Representation and Skill in Second Language Learners and Proficient Bilinguals. In Kecskes, I. & Albertazzi, L. (eds.) Cognitive Aspects of Bilingualism. Dordrecht: Springer. 237–269. Krumm, H.-J. 2005. Von der additiven zur curricularen Mehrsprachigkeit: Über die Notwendigkeit der Einbeziehung von Minderheiten-, Migranten- und Nachbarsprachen. In Hufeisen, B. & Lutjeharms, M. (eds.) Gesamtsprachencurriculum Integrierte Sprachendidaktik Common Curriculum. Tübingen: Gunter Narr Verlag. 27–36. Krumm, H.-J. & Reich, H. H. 2013. Das Curriculum Mehrsprachigkeit in der Schulentwicklung und der Aus- und Fortbildung von Lehrerinnen und Lehrern. In Vetter, E. (ed.) Professionalisierung für sprachliche Vielfalt. Perspektiven für eine neue Lehrerinnenbildung. Baltmannsweiler: Schneider Verlag Hohengehren. 21–41. Kuhl, P. K. 2000. A new view of language acquisition. PNAS, Vol. 97, No. 22, 11850–11857. Kuhl, P. K. 2004. Early Language Acquisition: Cracking The Speech Code. Nature Reviews/ Neuroscience, Vol. 5, 831–843. Kuhl, P. K., Tsao, F.-M. & Liu, H.-M. 2003. Foreign-language experience in infancy: Effects of short-term exposure and social interaction on phonetic learning. PNAS, Vol. 100, No. 15, 9096–9101. La Heij, W. 2005. Selection Processes in Monolingual and Bilingual Lexical Access. In Kroll, J. F. & De Groot, A. M. B. (eds.) Handbook of Bilingualism: Psycholinguistic Approaches. Oxford, New York: Oxford University Press. 298–307. Lambert, W. 1977. The Effects of Bilingualism on the Individual: Cognitive and Sociocultural Consequences. In Hornby, P.A. (ed.) Bilingualism. Psychological, Social, and Educational Implications. New York: Academic Press. 15–28. Lambert, W. 1990. Persistent Issues in Bilingualism. In Harley, B., Allen, P., Cummins, J. & Swain, M. (eds.) The Development of Second Language Proficiency. Cambridge: Cambridge University Press. 201–218. Lambert, W. 1991. And Then Add Your Two Cents’ Worth. In Reynolds, A. G. (ed.) Bilingualism, Multiculturalism, and Second Language Learning. Hillsdale, New Jersey: Lawrence Erlbaum Associates. 217–249. Lanza, E. 1992. Can bilingual two-year-olds code-switch? Journal of Child Language 19, 633– 58.
Bibliography
219
Lanza, E. 1997. Language Mixing in Infant Bilingualism: A Sociolinguistic Perspective. Oxford: Clarendon Press. Lanza, E. 2001. Bilingual first language acquisition: A discourse perspective on language contact in parent-child interaction. In Cenoz, J. & Genesee, F. (eds.) Trends in Bilingual Acquisition. Amsterdam/Philadelphia: Benjamins Publishing Company. 201–229. Larsen-Freeman, D. 1997. Chaos/complexity science and second language acquisition. Applied Lingustics, 18, 141–165. Larsen-Freeman, D. & Long, M. H. 1991. An introduction to second language acquisition research. New York: Longman. Larsen-Freeman, D., Schmid, M. & Lowie, W. 2011. Introduction. From structure to chaos. Twenty years of modeling bilingualism. In Schmid, M. & Lowie, W. (eds.) Modeling Bilingualism: From Structure to Chaos. Amsterdam/Philadelphia: John Benjamins. 1–11. Larson-Hall, J. 2008. Weighing the benefits of studying a foreign language at a younger starting age in a minimal input situation. Second Language Research, 24, 1, 35–63. Lasagabaster, D. 1998. The Threshold Hypothesis Applied to Three Languages in Contact at School. International Journal of Bilingual Education and Bilingualism, Vol.1, No.2, 119– 133. Lasagabaster, D. 2001a. Bilingualism, Immersion Programmes and Language Learning in the Basque Country. Journal of Multilingual and Multicultural Development, 22, 5, 401–425. Lasagabaster, D. 2001b. The Effect of Knowledge About the L1 on Foreign Language Skills and Grammar. International Journal of Bilingual Education and Bilingualism, Vol. 4, No. 5, 310–330. Laufer, B. & Hulstijn, J. 2001. Incidental Vocabulary Acquisition in a Second Language: The Construct of Task-Induced Involvement. Applied Linguistics, 22, 1, 1–26. Lee, P. 1996. Cognitive Development in Bilingual Children: A Case for Bilingual Instruction in Early Childhood Education. The Bilingual Research Journal, Vol. 20, Nos. 3 & 4, 499–522. Leung, Y. I. 2007. Third language acquisition: why it is interesting to generative linguists. Second Language Research, 23, 1, 95–114. Levelt, W. J. M. 1989. Speaking: From Intention to Articulation. Cambridge, MA: MIT Press. Levelt, W. J. M. 1993. Lexical Access in Speech Production. In Reuland, E. & Abraham, W. (eds.) Knowledge and Language, Volume 1, From Orwell’s Problem to Plato’s Problem. 241–251. Levelt, W. J. M. 2001. Spoken word production: A theory of lexical access. PNAS, 98, 23, 13464– 13471. Little, D. 1999. Metalinguistic awareness: the cornerstone of learner autonomy. In Mißler, B. & Multhaup, U. (eds.) The Construction of Knowledge, Learner Autonomy and Related Issues in Foreign Language Learning. Essays in Honour of Dieter Wolff. Tübingen: Stauffenburg. 3–12. Llinares, A. 2007. Classroom bilingualism at an early age: Towards a more natural EFL context. In Perez Vidal, C., Juan-Garau, M. & Bel, A. (eds.) A Portrait of the Young in Multilingual Spain. Bristol: Multilingual Matters. 185–199. Lowie, W. & Verspoor, M. 2011. The dynamics of multilingualism: Levelt’s speaking model revisited. In Schmid, M. & Lowie, W. (eds.) Modeling Bilingualism: From Structure to Chaos. Amsterdam/Philadelphia: John Benjamins. 267–287. Lüdi, G. 2003. Code-Switching and Unbalanced Bilingualism. In Dewaele, J.-M., Housen, A. & Wei, L. (eds.) Bilingualism: Beyond Basic Principles. Clevedon: Multilingual Matters. 174– 188. Nation, P. 1990. Improving Speaking Fluency. System, Vol. 17, No. 3, 377–384.
220
Bibliography
MacWhinney, B. 1997. Second Language Acquisition and the Competition Model. In De Groot, A. M. B. & Kroll, J. F. (eds.) Tutorials in Bilingualism. Psycholinguistic Perspectives. Mahwah New Jersey: Lawrence Erlbaum Associates. 113–142. MacWhinney, B. 2001. The competition model: the input, the context, and the brain. In Robinson, P. (ed.) Cognition and Second Language Instruction. Cambridge: Cambridge University Press. 69–90. MacWhinney, B. 2005. A Unified Model of Language Acquisition. In Kroll, J. F. & De Groot, A. M. B. (eds.) Handbook of Bilingualism. Psycholinguistic Approaches. Oxford: Oxford University Press. 49–67. Major, R. C. 2002. The Phonology of the L2 User. In Cook, V. (ed.) Portraits of the L2 User. Clevedon: Multilingual Matters. 67–92. Malakoff, M. 1992. Translation ability: A natural bilingual and metalinguistic skill. In Harris, R. (ed.) Cognitive Processing in Bilinguals. Hillsdale, NJ: Lawrence Erlbaum. 515–530. Malakoff, M. & Hakuta, K. 1991. Translation skill and metalinguistic awareness in bilinguals. In Bialystok, E. (ed.) Language Processing in Bilingual Children. Cambridge: Cambridge University Press. 141–166. Marini, A. & Fabbro, F. 2007. Psycholinguistic Models of Speech Production in Bilingualism and Multilingualism. In Ardila, A. & Ramos, E. (eds.) Speech and language disorders in Bilinguals. Nova Science Publishers Inc., New York, NY. 47–67. Marinova-Todd, S. H., Marshall, D. Bradford & Snow, C. 2000. Three Misconceptions about Age and L2 Learning. TESOL Quarterly, Vol. 34, 1, 9–34. Martin, P. 2003. Bilingual Encounters in the Classroom. In Dewaele, J.-M., Housen, A. & Wei, L. (eds) Bilingualism: Beyond Basic Principles. Clevedon: Multilingual Matters. 67–87. Marx N. 2012. Still Paying Lip Service to Multilingual Pedagogy? Paper presented at the 8th International Conference on Third Language Acquisition and Multilingualism, 13.–15. September 2012, Castellò, Spain. McLaughlin, B. 1990. The relationship between first and second languages: language proficiency and language aptitude. In Harley, B. et al. (eds.) The Development of Second Language Proficiency. Cambridge: Cambridge University Press. 158–174. Meisel, J. M. 1994. Code-switching in young bilingual children: the acquisition of grammatical constraints. In Wei, L. (ed.) 2007. The Bilingualism Reader. London and New York: Routledge. 336–359. Meisel, J. M. 2004. The Bilingual Child. In Bhatia, T. K. & Ritchie, W. C. (eds.) The Handbook of Bilingualism. Malden, MA: Blackwell. 91–113. Meuter, R. F. I. 2005. Language Selection in Bilinguals: Mechanisms and Processes. In Kroll, J. F. & De Groot, A. M. B. (eds.) Handbook of Bilingualism: Psycholinguistic Approaches. Oxford: Oxford University Press. 349–370. Meuter, R. F. I. & Allport, A. 1999. Bilingual Language Switching in Naming: Asymmetrical Costs of Language Selection. Journal of Memory and Language, 40, 25–40. Meyer, A. S. 1992. Investigation of phonological encoding through speech error analyses: Achievements, limitations and alternatives. Cognition, 42, 181–211. Mohanty, A. K. 1994. Bilingualism in a Multilingual Society: Psychosocial and Pedagogical Implications. Mysore: Central Institute of Indian Languages. Mohanty, A. K. & Perregaux, C. 1996. Language Acquisition and Bilingualism. In Berry, J. W., Dasen, P. R. & Saraswathi, T. S. (eds.) Handbook of Cross-Cultural Psychology. Vol. 2: Basic Processes and Human Development. 2nd ed. Needham Heights, MA: Allyn & Bacon. 217–254.
Bibliography
221
Müller-Lancè, J. 2003. A strategy model of multilingual learning. In Cenoz, J., Hufeisen, B. & Jessner, U. (eds.) The Multilingual Lexicon. Dordrecht: Kluwer. 117–132. Munoz, C. 2005. Trilingualism in the Catalan educational system. Int’l. J. Soc. Lang. 171, 75–93. Munoz, C. (ed.). 2006. Age and the Rate of Foreign Language Learning. Clevedon: Multilingual Matters. Munoz, C. 2011. Is input more significant than starting age in foreign language acquisition? International Review of Applied Linguistics (IRAL), 49(2), 113–133. Muysken, P. 1997. Code-Switching Processes. In Li Wei (ed.) (2010. Bilingualism and Multilingualism. Critical Concepts in Linguistics, Vol. I, London and New York: Routledge. 60–75. Myers-Scotton, C. 1988. Code-switching as indexical of social negotiations. In Wei, Li (ed.) 2007. The Bilingualism Reader. London and New York: Routledge. 97–122. Myers-Scotton, C. 2003. Code-Switching: Evidence of Both Flexiblity and Rigidity in Language. In Dewaele, J.-M., Housen, A. & Wei, L. (eds.) Bilingualism: Beyond Basic Principles. Clevedon: Multilingual Matters. 189–203. Myers-Scotton, C. 2005. Supporting a Differential Access Hypothesis: Code Switching and Other Contact Data. In Kroll, J. F. & De Groot, A. M. B. (eds.) Handbook of Bilingualism: Psycholinguistic Approaches. Oxford, New York: Oxford University Press. 326–348. Myers-Scotton, C. 2006. Multiple Voices: An Introduction to Bilingualism. Oxford (et al.). Blackwell. Myers-Scotton, C. & Jake, J. 1995. Matching lemmas in a bilingual language competence and production model: evidence from intrasentential code-switching. In Wei, L. (ed.) 2007. The Bilingualism Reader. London and New York: Routledge. 244–279. Nitsch, C. 2007. Mehrsprachigkeit: Eine neurowissenschaftliche Perspektive. In Anstatt, T. (ed.) Mehrsprachigkeit bei Kindern und Erwachsenen. Erwerb, Formen, Förderung. Tübingen: Attempto. 47–68. Oberhofer, K. 2008. Metalinguistic Ability and Early Foreign Language Learning. Diploma thesis submitted to the University of Innsbruck. O`Laoire, M. & Aronin, L. 2004. Exploring multilingualism in cultural contexts: towards a notion of multilinguality. In Hoffmann, C. & Ytsma, J. (eds.) Trilingualism in Family, School and Communitiy. Clevedon: Multilingual Matters. 11–29. Olshtain, E. & Nissim-Amitai, F. 2004. Being Trilingual or Multilingual: Is There a Price to Pay? In Hoffmann, C. & Ytsma, J. (eds.) Trilingualism in Family, School and Communitiy. Clevedon: Multilingual Matters. 30–50. Oomen-Welke, I. 2006. Sprachunterricht sprachübergreifend-(wie) geht das praktisch, und hat es Effekte? In Timm, J.-P. (ed) Fremdsprachenlernen und Fremdsprachenforschung: Kompetenzen, Standards, Lernformen, Evaluation. Tübingen: Gunter Narr. 303–320. Orwenjo, D. O. 2012. Multilingual education in Kenya: debunking the myths. International Journal of Multilingualism, Vol. 9, No 3, 294–317. Otwinowska-Kasztelanic, A. 2011. Awareness and Affordances: Multilinguals versus Bilinguals and their Perceptions of Cognates. In De Angelis, J. & Dewaele, J.-M. (eds.) New Trends in Crosslinguistic Influence and Multilingualism Research. Bristol, Buffalo, Toronto: Multilingualism Matters. 1–18. Paivio, A. 2006. Dual Coding Theory and Education. Draft chapter for the conference on “Pathways to Literacy Achievement for High Poverty Children,” The University of Michigan School of Education, September 29–October, 2006. Pallier, C., Colomè, A. & Sebastian-Galles, N. (in press) The influence of native-language phonology on lexical access: exemplar-based vs. abstract lexical entries. 1–5.
222
Bibliography
Papadopoulu, D. & Clahsen, H. 2003. Parsing Strategies in L1 and L2 Sentence Processing: A study of relative clause attachment in Greek. Studies in Second Language Acquisition, 25, 501–528. Paradis, J. & Genesee, F. 1996. Syntactic Acquisition in Bilingual Children. Autonomous or interdependent? In Li Wei. 2010. Bilingualism and Multilingualism. Critical Concepts in Linguistics, Vol. I, London and New York: Routledge. 486–512. Paradis, M. 1997. The Cognitive Neuropsychology of Bilingualism. In De Groot, A. M. B. & Kroll, J. F. (eds) Tutorials in Bilingualism. Psycholinguistic Perspectives. Mahwah New Jersey: Lawrence Erlbaum Associates. 331–354. Pavlenko, A. 2005. Bilingualism and Thought. In Kroll, J. F. & De Groot, A. M. B. (eds.) Handbook of Bilingualism. Psycholinguistic Approaches. Oxford, New York: Oxford University Press. 433–453. Pavlenko, A. 2005. Bilingualism and Thought. In Li Wei (ed.). 2010. Bilingualism and Multilingualism. Critical Concepts in Linguistics, Vol. I. London and New York: Routledge. 362– 391. Peal, E. & Lambert, W. 1962. The relation of bilingualism to intelligence. Psychological Monographs, 76, 1–23. Perdue, C. 2002. Development of L2 Functional Use. In Cook, V. (ed.) Portraits of the L2 User. Clevedon: Multilingual Matters. 123–144. Pienemann, M. 1999. Language Processing and Second Language Development. Processability Theory. Amsterdam/Philadelphia: Benjamins Publishing Company. Pienemann, M., Di Biase, B., Kawaguchi, S. & Hakansson, G. 2005. Processing Constraints on L1 Transfer. In Kroll, J. F. & De Groot, A. M. B. (eds.) Handbook of Bilingualism. Psycholinguistic Approaches. Oxford: Oxford University Press. 128–153. Pinto, M.A. 1995. La consapevolezza metalinguistica. Teoria, sviluppo, strumenti di misurazione. Rassegna Italiana di Linguistica Applicata XXVII, 3, 65–134. Pinto, M.A., Candilera, G., & Ilecito, P. 2003. TAM-2 Test di abilita`metalinguistiche n.2 (9–14 anni). La valutazione dello sviluppo metalinguistico tra scuola elementare e scuola media. Roma: Scione Editore. Pinto, M.A., Titone, R. & Trusso, F. 1999. Metalinguistic Awareness. Theory, Development and Measurement Instruments. Pisa and Rome: Istituti Editoriali e Poligrafici Internazionali. Pinto, M. A., Iliceto, P., & Melogno, S. 2011. Argumentative abilies in metacognition and in metalinguistics: a study on university students. Eur J Psychol Educ, published online: 06 July 2011. Piske, T. & Young-Scholten, M. (eds.). 2009. Input Matters in SLA. Bristol, Buffalo, Toronto: Multilingual Matters. Pontecorvo, C., Orsolini, M. & Zucchermaglio, C. 1989. Metalinguistic skills in children: what develops? Infancia y Arprendizaje, 47, 37–54. Poplack, S. 1980. Sometimes I’ll start a sentence in Spanish y termino en espanol: toward a typology of code-switching. In Wei, L. (ed.) 2007. The Bilingualism Reader. London and New York: Routledge. 213–243. Potowksi, K. 2009. Forms and Functions of Codeswitching by Dual Immersion Students: A Comparison of Heritage Speaker and L2 Children. In Turnbull, M. & Dailey-O`Cain, J. (eds.) First Language Use in Second and Foreign Language Learning. Bristol, Buffalo, Toronto: Multilingual Matters. 87–114.
Bibliography
223
Poulin-Dubois, D. & Goodz, N. 2001. Language Differentiation in Bilingual Infants: Evidence from Babbling. In Cenoz, J. & Genesee, F. (eds.) Trends in Bilingual Acquisition. Amsterdam/ Philadelphia: John Benjamins Publishing Company. 96–106. Poulisse, N. 1997. Language Production in Bilinguals. In De Groot, A. M. B. & Kroll, J. F. (eds) Tutorials in Bilingualism. Psycholinguistic Perspectives. Mahwah New Jersey: Lawrence Erlbaum Associates. 201–224. Poulisse, N. 1999. Slips of the Tongue. Speech Errors in First and Second Language Production. Amsterdam/Philadelphia: Benjamins Publishing Company. Price, C.J, Green, D.W. & von Studnitz, R. 1999. A functional imaging study of translation and language switching. Brain, 122, 2221–2235. Quay, S. 2001. Managing linguistic boundaries in early trilingual development. In Cenoz, J. & Genesee, F. (eds.) Trends in Bilingual Acquisition. Amsterdam/Philadelphia: John Benjamins. 149–199. Rahimpour, M. 2002. Bilingualism and L3 Acquisition. Journal of Faculty of Letters and Humanities (Tabriz). 45 (182): 31–44. Rampillon, U. 1997. Be aware of awareness-oder: Be aware of awareness? Gedanken zur Metakognition im Fremdsprachenunterricht der Sekundarstufe I. In Rampillon, U. & Zimmermann, G. (eds.) Strategien und Techniken beim Erwerb fremder Sprachen. Munich: Hueber. 173–184. Ranta, L. 2008. Metalinguistic Knowledge and Oral Production. Introduction. In Cenoz, J. & Hornberger, N. H. (eds.) Encyclopedia of Language and Education, 2nd Edition, Volume 6 : Knowledge about Language. Springer Science+Business Media LLC. 205–216. Rayner, K. & Morris, R. K. 1991. Comprehension Processes in Reading Ambiguous Sentences: Reflections from Eye Movement. In Simpson, G. B. (ed.) Advances in Psychology, 77. Understanding Word and Sentence. Elsevier Science Publishers B. V. (North-Holland). 175–198. Reynolds, A. 1991. The cognitive consequences of bilingualism. In Reynolds, A. (ed.) Bilingualism, Multiculturalism, and Second Language Learning. Hillsdale, NJ: Lawrence Erlbaum. 145– 182. Ricciardelli, L. A. 1992. Bilingualism and Cognitive Development in Relation to Threshold Theory. Journal of Psycholinguistic Research, Vol. 21, No. 4, 301–316. Ringbom, H. 1983. On the Distinctions of Item Learning vs. System learning and Receptive Competence vs. Productive Competence in Relation to the Role of L1 in Foreign Language Learning. In Psycholinguistics and Foreign Language Learning. Papers from a Conference (Stockholm, Sweden and Abo, Finland, October 25–26. 163–173. Ringbom, H. 1987. The Role of the First Language in Foreign Language Learning. Clevedon: Multilingual Matters. Ringbom, H. 2011. Perceived Redundancy or Crosslinguistic Influence? What L3 Learners’ Material Can Tell us About the Causes of Errors. In De Angelis, J. & Dewaele, J.-M. (eds.) New Trends in Crosslinguistic Influence and Multilingualism Research. Bristol, Buffalo, Toronto: Multilingualism Matters. 19–24. Riordain, M. N. & Donoghue, J. O. 2006. Mathematics Teaching in a Second Language in Ireland. Paper presented at the British Educational Research Association Annual Conference, Unitversity of Warwick. 1–11. Rivers, W. P. 1996. Self-Directed Language Learning and Third-Language Learner. Paper presented at the Annual Meeting of the American Council on the Teaching of Foreign Languages (30th Philadelphia, PA, November 22–24, 1996). 1–13.
224
Bibliography
Robinson, D. W. 1998. The Cognitive, Academic, and Attitudinal Benefits of Early Language Learning. In Met, M. (ed.) Critical Issues in Early Second Language Learning. Glenview: Scott Foresman Addison Wesley. 37–43. Robinson, P. 2008. Attention and Awareness. In Cenoz, J. & Hornberger, N. H. (eds.) Encyclopedia of Language and Education, 2nd Edition, Volume 6: Knowledge about Language. Springer Science+Business Media LLC. 133–142. Roelofs, A. 1996. Serial Order in Planning the Production of Successive Morphemes of a Word. Journal of Memory and Language, 35, 854–876. Roelofs, A. 1997. The WEAVER model of word-form encoding in speech production. Cognition, 64, 249–284. Roelofs, A., Meyer, A. S. & Levelt, W. J. M. 1998. A case for the lemma/lexeme distinction in models of speaking: comment on Caramazza and Miozzo (1997). Cognition, 69, 219–230. Rosen, V. M. & Engle, R. W. 1998. Working Memory Capacity and Suppression. Journal of Memory and Language, 39, 418–436. Rothman, J. 2011. L3 Syntactic Transfer Selectivity and Typological Determinacy: The Typological Primacy Model. Second Language Research, 27, 1, 107–127. Rothman, J. & Cabrelli, J. 2009. What variables condition syntactic transfer? A look at the L3 initial state. Second Language Research, 25, 4, 1–30. Safont Jorda’, M.P. 2005. Third Language Learners. Pragmatic Production and Awareness. Clevedon: Multilingual Matters. Sanz, C. 2000. Bilingual education enhances third language acquisition: Evidence from Catalonia. Applied Psycholinguistics, 21, 23–44. Sanz, C. 2008. Predicting Enhanced L3 Learning in Bilingual Contexts: The Role of Biliteracy. In Perez-Vidal, C., Juan-Garau, M. & Bel, A. (eds.) A Portrait of the Young in the New Multilingual Spain. Clevedon: Multilingual Matters. 220–240. Scochi, C. 2011. Wege zur Mehrsprachigkeit in Südtirol: von Sprach- und Schulpolitik über Schulgeschichte zum Sach- und Fachunterricht in der Zweitsprache. Saarbrücken: VDM Müller. Scovel, T. 1988. A time to speak. A psycholinguistic inquiry into the critical period for human speech. Rowley, MA: Newbury House. Schlak, T. 2003. Die Auswahl grammatischer Lernziele: Linguistische, psycholinguistische und didaktische Perspektiven. ZIF, 8, 1, 1–11.[http://zif.spz.tu-darmstadt.de/jg-08–1/beitrag/ schlak4.htm; eingesehen am 16.06.2010] Schmid, M. S. & Lowie, W. (eds.). 2011. Modeling Bilingualism. From Structure to Chaos. Amsterdam: John Benjamin. Schmidt, R. 1993. Awareness and Second Language Acquisition. Annual Review of Applied Linguistics, 13, 206–226. Schöpp, F. 2013. Mehrsprachigkeit als Zeitfresser? – Was Lehramtstudierende über interkomprehensives Arbeiten im Unterricht der romanischen Sprachen denken. In Vetter, E. (ed.) Professionalisierung für sprachliche Vielfalt. Perspektiven für eine neue Lehrerinnenbildung. Baltmannsweiler: Schneider Verlag Hohengehren. 148–176. Schönpflug, U. 2003. The Transfer–Appropriate-Processing Approach and the Trilingual’s Organisation of the Lexcion. In Cenoz, J., Hufeisen, B. & Jessner, U. (eds.) The Multilingual Lexicon. Dordrecht: Kluwer. 27–43. Schwanenflugel, P. J. 1991. Contextual Constraint and Lexical Processing. In Simpson, G. B. (ed.) Understanding Word and Sentence. Elsevier Science Publishers B. V. (North-Holland). 23–45.
Bibliography
225
Schwartz, A. I. & Kroll, J. F. 2006. Language Processing in Bilingual Speakers. In Traxler, M. J. & Gernsbacher, M. A. (eds.) Handbook of Psycholinguistics. London: Elsevier Science. 963– 995. Schwartz, B. D. & Sprouse, R. A. 1998. Back to Basics in Generative Second Langugae Acquisition Research. Paper presented at the Second Language Research Forum. 1998. Segalowitz, N. 1997. Individual Differences in Second Language Acquisition. In De Groot, A. M. B. & Kroll, J. F. (eds) Tutorials in Bilingualism. Psycholinguistic Perspectives. Mahwah New Jersey: Lawrence Erlbaum Associates. 85–112. Segalowitz, N. 2010. Cognitive Bases of Second Language Fluency. London/New York: Routledge. Segalowitz, N. & Hulstijn, J. 2005. Automaticity in Bilingualism and Second Language Learning. In Kroll, J. F. & De Groot, A. M. B. (eds.) Handbook of Bilingualism. Psycholinguistic Approaches. Oxford: Oxford University Press. 371–388. Sharwood Smith, M. 2008. Morphological and Syntactic Awareness In Forgeign/Second Langugae Learning. In Cenoz, J. & Hornberger, N. H. (eds.) Encyclopedia of Language and Education, 2nd Edition, Vol. 6: Knowledge about Language. New York: Springer. 179–191. Shohamy, E. 2006. Language Policy: Hidden Agendas and New Approaches. London: Routledge. Singleton, D. 2001. Age and Second Language Acquisition. Annual Review of Applied Linguistics, 21, 77–89. Singleton, D. 2003. Perspectives on the multilingual lexicon: a critical synthesis. In Cenoz, J., Hufeisen, B. & Jessner, U. (eds.) The Multilingual Lexicon. Dordrecht: Kluwer. 167–176. Singleton, D. 2007. The Critical Period Hypothesis: Some Problems. Interlinguistica, ISSN 1134– 8941, 17, 48–56. Singleton, D. & Aronin, L. 2007. Multiple language learning in the light of the theory of affordances. Innovation in Language Learning and Teaching 1 (1), 83–96. Slobin, D.I. 1973. Cognitive Prerequisites for the Development of Grammar. In Ferguson, Ch.A. & Slobin, D.I (eds.) Studies of Child Language Development. New York: Holt, Rinehart, and Winston. 240–249. Slobin, D.I. 1985. Crosslinguisitc Evidence for the Language-Making Capacity. In Slobin, D.I. (ed.) The Crosslinguistic study of Language Acquisition, Vol. 2: Theoretical Issues. Hillsdale, N.J.: Lawrence Erlbaum. 1157–1256. Slobin, D.I 2003. Language and Thought Online: Cognitive Consequences of Linguistic Relativity. In Gentner, D. & Goldin-Meadow, S. (eds.) Language in mind: Advances in the study of language and thought. Cambridge, MA: MIT Press. 157–192. Smith, M. C. 1997. How Do Bilinguals Access Lexical Information? In De Groot, A. M. B. & Kroll, J. F. (eds) Tutorials in Bilingualism. Psycholinguistic Perspectives. Mahwah New Jersey: Lawrence Erlbaum Associates. 145–168. Soriente, A. 2007. Cross-Linguistic and Cognitive Structures in the Acquisition of Wh-Questions in an Indonesian-Italian Bilingual Child. In Kecskes, I. & Albertazzi, L. (eds.) Cognitive Aspects of Bilingualism. Dordrecht: Springer. 325–362. Sparks, R. & Ganschow, L. 2001. Aptitude for Learning a Foreign Language. Annual Review of Applied Linguistics, 21, 90–111. Teepe, H. 2005. Legitimation der Frühförderung des Zweitsprachenerwerbs. Beitrag zur Sprachenwerkstatt Nordrhein-Westfalen des Ministeriums für Schule, Jugend und Kinder des Landes Nordrhein–Westfalen.
226
Bibliography
Temple, L. 2000. Second Language Learner Speech Production. Studia Linguistica, 54, 2, 288– 297. Thompson, A. S. 2008. Prominent Factors in the Acquisition of Portuguese: Language Aptitude versus Previous Language Experience. Selected Proceedings of the 10th Hispanic Linguistics Symposium, ed. Joyce Bruhn de Garavito and Elena Valenzuela, 134–145. Somerville, MA: Cascadilla Proceedings Project. Titone, R. 1978. Some Psychological Aspects of Multilingual Education. International Review of Education, Vol. 24, No. 3, 283–293. Titone, R. 1996. From communicative competence through bilingualism to metalinguistic development: Some theoretical pointers and research perspectives. In Alatis, J. et al. (eds.) Georgetown University Round Table On Language and Linguistics 1996. Linguistics, language acquisition, and language variation: current trends and future prospect. Washington D.C.: Georgetown University Press. 161–177. Tomasello, M. 1995. Language is not an instinct. Cognitive Development, 10: 131–156. Tomasello, M. & Abbot-Smith, K. 2002. A tale of two theories: response to Fisher. Cognition, 83, 207–214. Tomasello, M. 2003. Constructing a Language. A Usage-Based Theory of Language Acquisition. Cambridge, Massachusetts, and London, England: Harvard University Press. Tremblay, M.-C. 2006. Cross-Linguistic Influence in Third Language Acquisition: The Role of L2 Proficiency and L2 Exposure. CLO/OPL, Vol. 34, 109–119. Tunmer, W. E., Bowey, J. A. & Grieve, R. 1983. The Development of Young Children’s Awareness of the Word as a Unit of Spoken Language. Journal of Psycholinguistic Research, Vol. 12, No. 6, 567–594. Tunmer, W. E., Pratt, C. & Herriman, M. L. (eds.). 1984. Metalinguistic Awareness in Children: Theory, Research and Implications. Berlin: Springer Verlag. Ullman, M. T. 2001. A Neurocognitive Perspective on Language: The Declarative/Procedural Model. Nature Reviews. Neuroscience, 2, 717–727. Valencia, J. F. & Cenoz, J. 1992. The Role of Bilingualism in Foreign Language Acquisition. Journal of Multilingual and Multicultural Development, Vol. 13: 5, 433–449. Van Essen, A. 2008. Language Awareness And Knowledge About Language: A Historical Overview. In Cenoz, J. & Hornberger, N.H. (eds.) Encyclopedia of Language and Education, 2nd edition, Vol. 6: Knowledge about Language. New York: Springer, 3–14. Van Geert, P. 2008. The Dynamic Systems Approach in the Study of L1 and L2 Acquisition: An Introduction. The Modern Language Journal, 92, ii, 179–199. Van Geert, P., Steenbeek, H. & van Dijk, M. 2011. A dynamic model of expert-novice co-adaptation during language learning and acquisition. In Schmid, M. & Lowie W. (eds.) Modeling Bilingualism: From Structure to Chaos. Amsterdam/Philadelphia: John Benjamins. 236– 266. Van Hell, J. G. 2005. The Influence of Sentence Context Constraint on Cognate Effects in Lexical Decision and Translation. ISB4: Proceedings of the 4th International Symposium on Bilingualism, ed. James Cohen, Kara T. McAlister, Kellie Rolstad, and Jeff MacSwan, 2297– 2309. Somerville, MA: Cascadilla Press. Van Hell, J. G. & Dijkstra, T. 2002. Foreign language knowledge can influence native language performance in exclusively native contexts. Psychonomic Bulletin & Review, 9, 4, 780– 789.
Bibliography
227
Van Heuven, W. J. B., Schriefers, H., Dijkstra, T. & Hagoort, P. 2008. Language Conflict in the Bilingual Brain. Cerebral Cortex, 18, 11, 2706–2716. http://creativecommons.org/licenses/ by-nc/2.0/uk Van Lier, L. 1996. Interaction in the Language Curriculum. Awareness, Autonomy & Authenticity. Harlow, England: Pearson Education, Longman Group. Vetter, E. 2012. Exploiting receptive multilingualism in institutional language learning: The case of Italian in the Austrian secondary school system. International Journal of Bilingualism, 16:3, 348–365. Vigliocco, G. & Hartsuiker, R. J. 2002. The Interplay of Meaning, Sound, and Syntax in Sentence Production. Psychological Bulletin, 128, 3, 442–272. Vigliocco, G., Vinson, D. P., Paganelli, F. & Dworzynski, K. 2005. Grammatical Gender Effects on Cognition: Implications for Language Learning and Language Use. Journal of Experimental Psychology, 134, 4, 501–520. Vinnitskaya, I., Flynn, S. & Foley, C. 2003. The Acquisition of Relative Clauses in a Third Language: Comparing Adults and Children. Proceedings of the 6th Generative Approaches to Second Language Acquisition Conference (GASLA 2002), ed. Juana, M. Liceras et al., 340– 34. 5 Somerville, MA: Cascadilla Proceedings Project. Volterra, V. & Taeschner, T. 1978. The acquisition and development of language by bilingual children. In Wei, L. (ed.). 2007. The Bilingualism Reader. London and New York: Routledge. 303–319. Vygotsky, L. S. 1962. Thought and Language. Boston: MIT Press. Wandruska, M. 1979. Die Mehrsprachigkeit des Menschen. München: R. Piper & Co. Wang, X.-l. 2008. Growing up with Three Languages. Bristol: Multilingual Matters. Wartenburger, I., Heekeren, H. R., Abutalebi, J., Cappa, S. F., Villringer, A. & Perani, D. 2003. Early Setting of Grammatical Processing in the Bilingual Brain. Neuron, Vol. 37, 159–170. Weber-Fox, C. M, & Neville, H. J. 1999. Functional Neural Subsystems Are Differentially Affected by Delays in Second Language Immersion: ERP and Behavioral Evidence in Bilinguals. In Birdsong, D. (ed.) Second Language Acquisition and The Critical Period Hypothesis. Mahwah, N.J.: Lawrence Erlbaum. 23–38. Wehr, S. 2001. Was wissen Kinder über Sprache?: Die Bedeutung der Meta-Sprache für den Erwerb von Schrift- und Lautsprache. Bern (et al.): Haupt. Wei, Li (ed.). 2001. The Bilingualism Reader. London and New York: Routledge. Wei, Li (ed.). 2007. The Bilingualism Reader. London and New York: Routledge. Wei, Longxing. 2002. The Bilingual Mental Lexicon and Speech Production Process. Brain and Language, 81, 691–707. Wei, L. 2003. Activation of lemmas in the multilingual mental lexicon and transfer in third language learning. In Cenoz, J., Hufeisen, B. & Jessner, U. (eds.) The Multilingual Lexicon. Dordrecht: Kluwer. 57–70. Weltens, B. 2011. Epilogue. Twenty years of modeling bilingualism. From chaos to structure – and back again. In Schmid, M. S. & Lowie, W. (eds.) Modeling Bilingualism. From Structure to Chaos. Amsterdam/Philadelphia: John Benjamins Publishing. 289–302. Werker, J. F. 1986. The effect of multilingualism on phonetic perceptual flexibility. Applied Psycholinguistics, 7, 141–156. Williams, S. & Hammarberg, B. 1998. Language switches in L3 production: implications for a polyglot speaking model. Applied Linguistics, 19, 3, 295–333.
228
Bibliography
Whitney, P. & Waring, D. A. 1991. The Role of Knowledge in Comprehension: A Cognitive Control Perspective. In Simpson, G. B. (ed.) Advances in Psychology, 77. Understanding Word and Sentence. Elsevier Science Publishers B. V. (North-Holland). 199–216. Winkler, I., Kujala, T., Tiitinen, H., Sivonen, P., Alku, P., Lehtokoski, A., Czigler, I., Csepe, V., Ilmoniemi, R. J. & Näätänen, R. 1999. Brain responses reveal the learning of foreign language phonemes. Psychophysiology, 36, 638–642. Wolff, D. 1997. Zur Förderung von Sprachbewußtheit und Sprachlernbewußtheit im bilinugalen Sachfachunterricht. Fremdspachenlehren und Lernen (FLuL) 26, 167–183. Ytsma, J. 2000. Trilingual primary eductation in Friesland. In Cenoz, J. & Jessner, U. (eds.) English in Europe: The Acquisition of a Third Language. Clevedon: Multilingual Matters. 222–235.
Index acceptability 124, 150–151, 153, 164–165, 169–171, 175, 180 activation 22–88, 146, 181, 192, 194 advantages 3, 7, 9, 13–14, 108, 112, 115, 175, 180, 201–202 – cognitive 103, 119, 121, 125 Alcock, Antony 133–135, 137 Allgäuer-Hackl, Elisabeth 15, 104–105, 107, 111, 119, 144, 199 ambiguity 114–115, 126, 150, 153, 155, 164, 169–171, 175, 180 analysis and control 115, 121, 123 anxiety 19, 65, 75, 158, 201 aptitude 10, 13, 24, 74, 116 Aronin, Larissa 37, 52, 66, 32, 147, 182, 199 articulator 28, 32, 34–35, 47, 51, 59–60 articulatory 35, 38, 93 Autonomy Statute 134–137, 139–140, 198 attention(al) 8, 44–46, 52, 68–71, 75, 77, 79, 93, 95, 105, 109, 111, 113–115, 118, 121–122, 125, 144–145, 152, 180, 182 attitude 41, 74, 149, 175, 185, 198–199, 201 automaticity 70–75 awareness see language awareness, linguistic awareness, cross-linguistic awareness, metalinguistic awareness, multilingual awareness Baker, Colin 5–6, 20, 98–99, 117, 133, 143– 145 base language 25, 41–42 BICS 98–99, 143, 196 Ben Zeev, Sandra 117, 120–121, 124, 201 Bialystok, Ellen 43, 52, 69, 71, 77–79, 95, 103, 112–126, 141, 143, 176, 180, 183, 201 bidirectional 21, 33, 58, 63, 74, 195 bilingualism 3, 5–9, 14, 22, 40–41, 63, 95, 97, 107–109, 112, 114–115, 117, 119–121, 128, 136, 138, 141, 179, 182 – balanced 108, 182 bilingual – development 89 – education 106, 136, 139–140 – programs 162 – norm 15, 31
bilinguality 3, 6, 9, 108–109, 134, 137– 138 borrowing 31, 41, 192–193 CALP 98–99, 143, 196 capacity constrained 46 (auto)catalytic effects 13–14, 62, 65–66, 117, 122, 181 Cenoz, Jasone 4, 15, 50, 80, 94, 97, 99, 102– 103, 107–108, 146, 176, 183, 185, 187– 188, 199–200 Chaos complexity theory 63 Chomsky, Noam 78, 83, 86, 95 CLIL 139, 141 Clyne, Michael 3, 10, 28, 37, 47–52 cloze test 124 code-mixing 76, 86–90 code-switching 23–24, 38, 40, 45, 47, 49, 51, 81, 86–88 cognitive benefits xvii common underlying conceptual base 17 conceptualizer 22, 28, 32, 35, 45, 51, 59, 87, communicative sensitivity 116 competence 4, 7, 9–11, 13, 40–41, 63–65, 74–75, 82, 87–90, 94–100, 103, 107, 116, 119–120, 126, 134–135, 136–137, 139, 143, 146, 150, 178, 180, 198, 200 complexity 10–12, 31, 37, 58, 63, 66, 83, 85, 102, 115, 120, 181, 195, 196–197 comprehension xxi, 18, 29, 30, 32, 34, 37, 46, 52–57, 68, 98, 113, 150–151, 153 connections 16, 20–21, 47, 53, 58, 63 consciousness 112, 114, 127 convergence 47–48 Cook, Vivian 4, 10, 24, 40, 74, 87, 94, 96– 98, 103, 199–200 Critical Period Hypothesis 91–92 creativity xxi, 7 cross-linguistic awareness 106, 179, 189– 190 cross-linguistic interaction 17, 76 Cummins, Jim 3, 9, 16–17, 20, 98–99, 101, 117, 121, 143–144, 178–180, 190, 196, 204 CUP 17
230
Index
De Angelis, Gessica 3, 15–17, 19–20, 24, 31, 34, 36, 38, 41–44, 45, 52–53, 55–56, 63, 94, 96, 99, 103, 108, 112, 114, 143, 146, 176, 178–180, 198, 201–202 De Bot, Kees 10–11, 19, 22, 23–24, 26, 28, 35, 37–38, 45, 47, 49, 50–52, 73 De Groot, Annette 20–23, 25 development, language 5, 10, 30, 36, 62, 66, 67, 74, 76, 78–80, 82, 93, 99, 119, 196 diacritical markers 53, 55 dialect 40, 95, 98, 136, 140, 148, 191, 193– 195 Dijkstra, Ton 20, 23–25, 28, 43, 45–46, 52– 57 directionality of effect 119 distance, language 55, 102 – social 42 divergent thinking 9, 63 dominant language 143, 149–150, 153 Dynamic model of multilingualism (DMM) 10–14, 31, 62–66, 70, 86, 97– 98, 104, 116, 122, 181, 188, 197, 204 dynamic(s) 10–12, 16, 18, 23, 27–28, 31–32, 37, 41, 47, 52–54, 58–68, 74, 76, 79, 86–87, 92–93, 98–99, 102, 116–117, 119, 122, 181–182, 190, 196–197, 199, 203 Dynamic systems theory (DST) 12, 19, 25, 28–29, 52, 59–60, 69, 74, 92 early multilingualism xix, 128, 147, 176, 178, 179 Ecke, Peter 16 Egger, Kurt 133, 136–138 embedded language 40, 88–89 Emergent(ism) 13–14, 59, 63, 66, 99, 114, 181 epilinguistic 152, 154 esame di bilinguismo 134 experienced learners 131–132, 176, 179, 180 explicit, knowledge 69, 77, 104, 111, 118, 122, 123, 160 Fascism 134 Flege, James 91–92 fluency 72–75
foreign language 3–5, 7, 15, 49–50, 101, 105–106, 108–109, 111, 131, 139, 144– 145, 175, 183, 185, 198 formulator 28, 32, 34–35, 39, 47, 50–51, 59, 68 frequency (of use) 13, 16, 44, 50–51, 66, 72, 95, 182 frequency (high/low) 19, 20, 55, 88, 92 gang effect 54 Garcia, Ophelia 3, 11, 87, 97, 99, 102–106, 111, 198–199, 201, 204 Generative Grammar 82 Genesee 4, 5, 8, 79–82, 87, 89–90, 92–93, 138, 140, 179, 190 German 131–134, 136–145, 148–149, 150– 151, 158, 160, 162, 175–187, 189–197 globalization 3 Gombert, Jean Emile 113, 152 grammatical function 150–155, 164–166, 172–174, 180 Green, David 24, 26–27, 38, 43–47, 51, 68, 72, 95 Grosjean, Francois 10, 28, 31, 40–43, 54, 57–58, 94, 97–98, 192, 199 Hakuta, Kenji 3–5, 106, 111, 113, 120–121, 146, 176, 180, 182, 188, 189, 191 Hall, Christopher 16 Hammarberg, Björn 16, 25, 28, 49, 50, 52, 61–62 Hawkins, Eric 105–106, 112, 144–145 Herdina, Philip 7, 10–15, 37, 40, 47, 52, 62– 67, 70–71, 87, 93, 94–99, 101, 103, 104–116, 122, 146, 179, 181, 188, 196– 197, 199–201, 203 Herwig, Anna 49 high-immersion (multilingual education) 77, 128, 131–132, 146–147, 149, 161–188, 189–194, 203–204 high-immersion programmes/low-immersion programmes 131–132, 146, 165, 167– 170, 175–176, 178–181, 187, 203 Hoffmann, Charlotte 11, 14, 81, 85–90, 181, 192, 195 holistic 37, 39, 47, 63, 103, 151, 199
Index
– approach 10, 12, 41, 47, 63, 86, 93, 99, 104, 200 – view 10–11, 52, 66, 199, 203 – perspective 37, 97, 203 – multicompetence 10 – see also wholistic home language 185 Hufeisen, Britta 16, 31, 74, 94, 105–106, 112, 120, 123, 143–144, 189, 199, 201 Ianco-Worrall, Anita 121 identity 5, 47, 66, 74, 136, 198, 204 inhibition 8, 26–27, 43–46, 53–56, 114 interference 24, 33, 38, 41, 43, 51, 85, 95, 101–102, 121 immersion see high-immersion/lowimmersion programmes implicit, knowledge 69, 77, 111 interconnected(ness) 12, 18, 28, 47, 53, 59, 95, 117, 180, 195, 203 Italy 133, 141 Italian 131–132, 133–138, 139–145, 148– 149, 150, 153, 158–159, 179, 181, 185, 187, 192–195, 197, 204 James, Carl 102, 106, 112–113, 144 Jessner, Ulrike 3, 7, 10–17, 24, 31, 37, 40, 47–49, 52, 62–67, 70–71, 87, 93–99, 101, 103–108, 112–123, 131, 143, 144, 146, 176, 178–181, 188–190, 192, 195– 197, 199–204 judgement test 124, 151 Karmiloff-Smith, Annette 77–79, 83 Kormos, Judit 22, 25, 32–33, 35, 40 Krumm, Hans-Jürgen 15, 104–105, 107, 144, 199 Ladin 133–134, 137, 140, 148 Lambert, Wallace 6, 112, 115, 121, 201 language awareness 14, 66, 104–107, 111– 114, 144–146 language behavior 41, 74, 81, 86, 152, 191 language forms and functions 179 language mode 40–43, 47, 54, 57, 116 language separation 81–82 language set inertia 27
231
Lasagabaster, David 3, 108, 127, 131, 146, 150, 176, 180, 188, 204 learner language 196 language loss 12 lexical inventions 193, 195 language matrix frame 39 lexicon 19–29, 35–36, 39, 45, 49, 50–56, 59–60, 69, 80, 85, 146, 190 linguistic awareness 13–14, 112, 120, 143, 204 L2/L3 learning xxi, 5, 98, 102, 108, 175, 200, 202 Lowie, Wander xx, xxi, 9, 11, 18, 22–23, 25, 27–28, 32–33, 37, 52, 59–60, 62, 67, 69, 92–93, 95 maintenance 10, 13, 15, 62, 65, 98, 104, 116, 145, 181 management 8, 13, 65, 104, 116, 120, 145, 181, 190 matrix language 39–40, 88–89, 192 memory 19–20, 30, 44, 52, 68–69, 75, 95 metalinguistic abilities 9, 62, 105, 110–119, 131–132, 146, 152, 161, 176, 179–181 metalinguistic awareness xviii, xix, xx, xxi, 7, 13, 31, 65–66, 70, 72, 104, 107, 111– 127, 131–132, 145–147, 150, 152, 161, 176, 179, 182, 187–188, 190, 201, 203 modular 10, 28, 32–33, 36–37, 49, 52, 68, 74, 87 monitor(ing) 8, 13, 34, 44, 66, 68–70, 113, 116–117, 122, 181–182, 190 monolingual norm xvii, 31, 103 motivation(al) 4, 10, 47, 65, 73–75, 105, 145, 175, 201 multicompetence 10, 202 multilingual acquisition xvi, xviii, xix, xx, 12, 41, 103, 203 multilingual awareness 13, 17, 99, 111, 120, 123, 151, 203 multilingual educational programmes xvii, xviii, 40, 131, 139, 146, 161, 201 multilingual instruction 3, 4, 100, 102–103, 150, 152, 161, 182, 204 multilingual network 38 multilingualism factor 14, 104, 116, 122 multilinguality xvii, 66, 97, 146
232
Index
multilingual system(s) xix, 10–11, 13, 31, 41, 62, 63, 65–66, 70, 93–94, 99, 104, 116–117, 122, 181, 195 native speaker 64, 96–98, 127, 135, 137, 140, 194, 199, 200 node – combinatorial 23 – lexical 21, 27, 45 non-linearity 63 parameters 31, 64, 78, 83, 86, 92 parents 3, 80, 91, 133, 142–143, 148–149, 158, 175, 182–185, 191, 199 Peal, Elizabeth 6, 112, 115, 121, 201 perceived communicative needs 10, 13, 65, 97 perception 40–41, 43, 52–57, 68, 91–93, 109, 190, 194 – public 134, 199 phoneme repetition 164, 168 phonemic segmentation 150–151, 153, 155– 157 phonology 4, 24, 35, 38, 78, 113 Pinto, Maria Antonietta 72, 113–114, 118– 127, 150–157, 160, 176 plurilingual 43, 47–49, 63, 103, 105, 200 preverbal message 26, 35, 38, 87 primary 101, 106–108, 127, 135–137, 139, 140–143, 148–149, 152, 188 production(s) 13, 16, 18, 22, 29, 30–52, 61– 62, 68–70, 74, 79, 83, 85–89, 96, 99, 101, 113, 118, 146, 160, 188, 191, 197 proficiency see competence pronunciation 38, 69, 91–93, 98, 118, 191 psychotypology 16 see also typology qualitative data 189 recency (of use) 16, 44, 50–51, 62, 72, 95 Reich, Hans 15, 104–105, 107, 144, 199 representation xviii, xxi, 9, 11–12, 16–28, 33, 35–36, 42, 45–46, 50, 52–60, 77– 78, 84, 92–93, 115, 117, 121–122, 203 resources 10, 13, 15–16, 43, 46, 65–66, 68– 75, 81, 90, 94, 96, 103–104, 122, 125, 179, 181, 192, 200
retrieval 20, 23, 26–28, 35, 71, 75, 194 Riflessione Lingua 106, 144–145 Ringbom, Hakan 7, 15–16, 71 routines, cognitive 28, 59–60 Scochi, Claudia Bettina 134–137, 139 selection 23, 25–27, 31, 34, 36, 40, 43–45, 50–56 skills 4–5, 7, 9–10, 13–14, 16–17, 31, 65, 72, 97, 98–99, 104–105, 107, 113–126, 131, 143–144, 181–182, 190 Singleton, David 24–25, 49, 91, 147 SLA xix, 3, 13–14, 23, 111, 123, 179, 188 South Tyrol(ean) 40, 107–108, 128, 131, 133– 135, 138–141, 145, 180, 191, 193, 198, 200–204 sound patterns 47, 91–93 speech plan 38–49, 51, 70 spreading activation 32–33, 35–36 storage xxi, 18–20, 28, 47, 49, 68, 83 strategies 11, 15–16, 65, 70, 74, 83, 93, 96, 104–107, 114–117, 121, 123, 144, 161, 201 status 19, 41, 50, 62, 74, 102, 134, 204 Stavans, Anat 11, 14, 81, 85–90, 181, 192, 195 supplier language 61–62 syllable scansion 150–151, 156, 164, 172 syntax 4, 23–24, 35, 60, 78, 80, 82, 84–85, 191 target language 13, 25, 44, 49, 52–53, 98, 107, 194 third language acquisition xviii, 14, 87, 111, 113, 120, 161, 188 third language learning 5, 14–15, 31, 61, 108, 150, 175, 196 threshold level 9, 44, 180 transfer 16, 17, 49–50, 70, 85, 95, 97, 101, 144–145, 181–182 transference 47–48 translanguaging 11, 103 translation 20, 22, 38, 45, 106, 112, 116, 133, 136 transversion 47–49 trilingual education programmes 162 typology 16, 24, 62, 84, 95, 194
Index
UG 64, 82, 86, 92, 95 unpredictability 63 Verspoor, Marjolijn xx, xxi, 9, 11, 18, 22–23, 25, 27–28, 32–33, 37, 52, 59–60, 62, 67, 69, 92–93, 95 vocabulary 4, 24, 42, 59, 72, 89, 98, 109, 159 – learning 4, 7, 16
Vygotsky, Lev 6, 112, 121 Weinreich, Uriel 19–20, 100 word formation 35, 151, 164–167, 175 word recognition 53–57, 71
233