Language Attrition
Studies in Bilingualism (SiBil) The focus of this series is on psycholinguistic and sociolinguisti...
93 downloads
1706 Views
3MB Size
Report
This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
Report copyright / DMCA form
Language Attrition
Studies in Bilingualism (SiBil) The focus of this series is on psycholinguistic and sociolinguistic aspects of bilingualism. This entails topics such as childhood bilingualism, psychological models of bilingual language users, language contact and bilingualism, maintenance and shift of minority languages, and sociopolitical aspects of bilingualism.
Editors Kees de Bot
University of Groningen
Thom Huebner
San José State University
Editorial Board Michael Clyne
University of Melbourne
Kathryn A. Davis
University of Hawaii at Manoa
Joshua A. Fishman Yeshiva University
Francois Grosjean
Université de Neuchâtel
Wolfgang Klein
Max Planck Institut für Psycholinguistik
Georges Luedi
University of Basel
Christina Bratt Paulston University of Pittsburgh
Suzanne Romaine
Merton College, Oxford
Merrill Swain
Ontario Institute for Studies in Education
G. Richard Tucker
Carnegie Mellon University
Volume 33 Language Attrition. Theoretical perspectives Edited by Barbara Köpke, Monika S. Schmid, Merel Keijzer and Susan Dostert
Language Attrition Theoretical perspectives
Edited by
Barbara Köpke Université de Toulouse - Le Mirail
Monika S. Schmid Rijksuniversiteit Groningen
Merel Keijzer University of Technology Delft
Susan Dostert Heinrich Heine Universität Düsseldorf
John Benjamins Publishing Company Amsterdam / Philadelphia
8
TM
The paper used in this publication meets the minimum requirements of American National Standard for Information Sciences – Permanence of Paper for Printed Library Materials, ansi z39.48-1984.
Library of Congress Cataloging-in-Publication Data Language attrition : theoretical perspectives / edited by Barbara Köpke ... [et al.]. p. cm. (Studies in Bilingualism, issn 0928-1533 ; v. 33) Includes bibliographical references and index. 1. Language attrition. I. Köpke, Barbara. P40.5.L28L35 2007 306.44'6--dc22 isbn 978 90 272 4144 3 (Hb; alk. paper)
2007022630
© 2007 – John Benjamins B.V. No part of this book may be reproduced in any form, by print, photoprint, microfilm, or any other means, without written permission from the publisher. John Benjamins Publishing Co. · P.O. Box 36224 · 1020 me Amsterdam · The Netherlands John Benjamins North America · P.O. Box 27519 · Philadelphia pa 19118-0519 · usa
Table of contents
Acknowledgment
ix
Bilingualism and attrition Monika S. Schmid and Barbara Köpke
1
Language attrition at the crossroads of brain, mind, and society Barbara Köpke
9
Understanding attrition within a MOGUL framework Mike Sharwood Smith
39
Dynamic systems theory, lifespan development and language attrition Kees de Bot
53
The grammatical profile of L1 speakers on the stairs of potential language shift Carol Myers-Scotton
69
First language attrition from a minimalist perspective: Interface vulnerability and processing effects Ianthi Maria Tsimpli
83
(Psycho)linguistic determinants of L1 attrition Ayşe Gürel L1 attrition features predicted by a neurolinguistic theory of bilingualism Michel Paradis The role of L1 use for L1 attrition Monika S. Schmid
99
121 135
vi
Table of contents
Critical periods in language acquisition and language attrition Christophe Pallier A hidden language: Recovery of a ‘lost’ language is triggered by hypnosis Rosalie Footnick
155
169
Identity, immigration and first language attrition Petra Prescher
189
Language attrition and ideology: Two groups of immigrants in Israel Miriam Ben-Rafael and Monika S. Schmid
205
Stimulated recall methodology in language attrition research Antonio F. Jiménez Jiménez
227
Name index Subject index
249 255
Acknowledgment
This volume was made possible by the generous financial support of the Dutch National Science Foundation NWO, the Dutch Royal Scientific Academy KNAW, the Faculty of Arts of the Vrije Universiteit Amsterdam, the Laboratoire Jacques Lordat, Toulouse and the Université de Toulouse – Le Mirail. We would also like to thank Patricia Leplae and Kees Vaes of John Benjamins Publishing Co. for their support and advice, and the anonymous reviewers who kindly assisted us with the process of selecting and revising the papers collected here.
Bilingualism and attrition Monika S. Schmid and Barbara Köpke
Rijksuniversiteit Groningen, The Netherlands / Laboratoire de Neuropsycholinguistique “Jacques Lordat” (Octogone), Université de Toulouse – Le Mirail, France
I believe in the fundamental interconnectedness of all things. (Douglas Adams, Dirk Gently’s Holistic Detective Agency)
At the heart of virtually all research on bilingualism, if not on language in general, is the recognition that the acquisition and knowledge of a first language occupy a privileged status in the human mind. There is great speculation on why this may be the case: are the reasons biological and, if so, is the difference between L1 and L2 acquisition due to the fact that the human brain contains something that is specifically equipped to acquire and hold a first language, or do certain biological properties of the brain change in the process of maturation? Is there such a thing as Universal Grammar and, if so, is it (wholly or partly) available to L2 learners? Such controversies notwithstanding, it is unanimously recognized in linguistic science that the L1 is different in many ways from any other abstract system of knowledge human beings possess, and different in specific ways from any language learned later in life. However, for a long time, the differences between L1 and L2 were viewed and researched in a somewhat biased way: L1 was assumed to be the stable and unchanging baseline from which acquisition, knowledge and use of the L2 deviated in some ways. The proficiency of L2 learners was compared to that of native speakers, and was perceived to fall short of the target (Cook 2003). Under such a perspective, what is of interest to linguistic research is the transfer from L1 that L2 learners, and particularly those in the early stages of acquisition, experience on all linguistic levels (Fig. 1). During the ongoing process of L2 acquisition, this traffic from L1 is reduced as far as possible. This may be easier on some levels than on others, but the general observation is that, as the learner becomes more advanced, the overall level of L1 influence decreases. Investigations of bilingualism and SLA have typically sought to describe what this reduction process looks like, and explain it in terms such as
Monika S. Schmid and Barbara Köpke
Figure 1. L1 influence on L2 in SLA
Figure 2. L2 influence on L1 in attrition
access to UG in SLA, the stability of phonetic systems after the so-called Critical Period, psycholinguistic models of speech production, etc. Similarly, in L1 attrition, the traditional idea is that, as the L2 becomes prevalent in everyday usage and dominant in the speaker’s mind, what is of interest to research is traffic which goes the other way (see Fig. 2). Investigations of L1 attrition therefore typically are not interested in what the attriter’s L2 system looks like, and to what extent there still is L1 transfer to be perceived here. Instead, research focusses on the description and explanation of L2 interference on L1 (for an overview of research on L1 attrition see Köpke & Schmid 2004).
Bilingualism and attrition
Figure 3. An integrated view of bilingualism
Only recently has it been recognized that bilinguals may not have one ‘normal’ language (in which they are indistinguishable from monolinguals – the L1 in the case of L2-learners and the L2 in the case of L1 attriters) and one ‘deviant’ one (in which knowledge is less extensive than that of monolinguals, and also tainted by interference from L1 in SLA and from L2 in attrition). Psycholinguistic research has established that bilinguals process language in a way which is slightly different from that of monolinguals, in that corresponding lexical items in all language systems are always active to some degree, no matter which language is being used or accessed (e.g. Van Hell & Dijkstra 2002). Similarly, it has been established that bilinguals have an ‘in-between’ way of processing sentences (Hernandez, Bates & Avila 1994) and of structuring their phonetic space (Cutler, Mehler, Norris & Segui 1989; Flege 1987). If traffic is always bi-directional, as suggested by these findings, then L1 attrition may not be the special and rare ‘condition’ as which it is often perceived. What is usually assumed is that a number of necessary conditions have to be satisfied in order for L1 attrition to set in: emigration, extensive use of the L2 in daily life, extremely reduced use of the L1 in daily life, plus a fairly long time span (decades). However, it is possible that this particular mix of circumstances brings about a more immediately visible version of a process of change in the L1 that all bilinguals undergo to some degree. Such a view of L1 attrition is compatible with recent views on overall multilingual language competence which challenge traditional assumptions that traffic between language systems is normally one-way. As Cook (2003) points out, it is probably the case that with the acquisition of an L2 at any point in an individual’s lifetime, the L1 system is also fundamentally and irrevocably changed. That would argue for an integrated view of bilingual development (see Fig. 3).
Monika S. Schmid and Barbara Köpke
It should be emphasized that such a view is not inconsistent with the assumption that L1 knowledge and development is different from other knowledge systems. The difference between L1 interference on L2 on the one hand and L2 interference on L1 on the other is perceptible in both SLA and L1 attrition. Moreover, this difference is not a merely quantitive one. While interference phenomena on the level of the lexicon, for example, may look very similar across these two processes, the phonological system of a mature L1 is probably so stable that it is impervious to L2 influence – and, on the other hand, many very advanced L2 speakers still experience phonological traffic from L1. However, it does follow from an integrated view of bilingual development (such as the holistic view of bilingualism argued for by Grosjean (1992) or what has been referred to as a ‘multicompetence’ perspective, e.g. Cook (2003)) that it makes little sense to study the development of one language system in isolation. We would argue that here the study of changes of the L1 under the influence of L2 can create added value for linguistic research at large. The fundamental difference of the native L1 system from anything else we know can best be explored from a perspective which investigates not only how this system affects others, but also how the L1 itself is subject to influences from outside. If the objective of the study of bilingual development is to integrate investigations of L1 and L2 development, then the theoretical and methodological challenge is to integrate the research fields of attrition and SLA. In practical terms, this would argue for an approach by which investigations of attrition routinely also investigate the development of the L2, and vice versa. The implications are that investigations of L1 attrition should avail themselves more fully and more fundamentally of the theoretical frameworks that have been formulated with respect to SLA and bilingualism. We see this as one of the major stumbling blocks in current attrition research: significant recent developments (both theoretical and methodological) notwithstanding, the field of attrition is still far less extensive, less theoretically sophisticated, and more descriptively oriented than SLA research. The problem often appears to lie in the application of theoretical approaches in a way that leads to the formulation of testable hypotheses. There are a considerable number of studies on L1 attrition which provide a solid introduction to a particular theoretical framework, and then proceed to analyse and discuss data in a well-founded way – however, the intermediate and necessary step of arriving at
. There is some evidence for phonetic attrition, e.g. Major (1992). However, there are no studies which have found any indication of even the most minor restructuring of the phonological system.
Bilingualism and attrition
a clearly-stated hypothesis which derives logically from the theoretical framework is often either omitted altogether or taken in a somewhat unsatisfactory manner. Arguably, the problems involved in re-assessing theories of bilingualism in a way that will lead to principled predictions for language attrition are linked to the bias of second language research pointed out above: where L1 and L2 are taken to be different in any way, L1 is often assumed as the (undescribed) default or baseline. On such a basis, hypotheses can only easily be derived for the deviant or different system, the L2. Turning them around to make predictions for the development of the L1 in a situation of language contact or language dominance reversal is often a problem which proves to be more complex than anticipated. It is for these reasons that we invited a number of researchers who are experts in particular disciplines of bilingualism to provide an in-depth perspective on L1 attrition which will make the translation of theory to hypothesis easier for future research. The present volume is the result of these efforts. The volume starts with a set of three papers which consider the impact of determining factors for the process of language attrition, arguing for a more integrated perspective than has previously been taken. Köpke gives an overview of the impact of neurobiological and cognitive processes as well as extralinguistic factors such as language use and attitudes on language acquisition and attrition. Sharwood Smith presents MOGUL, a model for understanding these two processes in an overarching framework with the ability to integrate features that traditionally are the domain of formal linguistics with processing factors. De Bot takes a step further and introduces a perspective from chaos theory. Language, he argues, can best be considered and studied as a Dynamic System which is in constant flux, and within which a large number of diverse factors conspire in complex (and often unpredictable) ways to bring about changes. The following three papers introduce more established linguistic frameworks and theories, and demonstrate how these can be applied and interpreted with respect to language attrition. Myers-Scotton shows how the 4-M model can account for grammatical aspects of language maintenance or shift in a bilingual Xhosa-English community. Tsimpli introduces the minimalist approach and exemplifies it on the basis of data from near native Greek-English and Italian-English bilinguals. Gürel then presents how data from Turkish-English and English- Turkish bilinguals can be interpreted both on the basis of Government and Binding Theory and within a psycholinguistic framework in the context of Activation Thresholds. These latter notions are also at the heart of the subsequent papers. Paradis presents an introduction to psycho- and neurolinguistic aspects of bilingualism in general and their impact on L1 attrition in particular. Schmid takes up some
Monika S. Schmid and Barbara Köpke
of these notions in a data-based approach, investigating the impact of L1 use and language modes in a corpus of German-English and German-Dutch data. The following two papers present investigations of L1 attrition in early bilinguals. On the basis of a group of Korean adoptees in France, Pallier demonstrates the case of a first language which has apparently vanished from memory to the degree that neither psycho- nor neurolinguistic techniques are able to detect any trace. He discusses the concepts of maturation and the Critical Period and puts into question Penfield’s idea of L1 acquisition leaving indelible traces in the brain. Footnick then presents a startling case-study of a language which had been hidden from conscious memory, but which the speaker (an early Ewe-French bilingual) was able to partly recover through hypnosis. She argues that the inability to access a language may sometimes be triggered by emotional conflicts and presents a ‘conflictual hypothesis’ to explain this type of attrition. The concepts of emotions, attitudes and identities are also prominent in the last set of papers. Prescher discusses theories of identity and identification and illustrates these with examples from a corpus of German-Dutch bilinguals. Ben-Rafael & Schmid present an example of the different impact of integrational vs. instrumental motivation for language acquisition and its impact on attrition among two different groups of migrants (French-Hebrew and Russian-Hebrew bilinguals). Finally, Jiménez Jiménez, taking a Vygotskian perspective on L1 attrition from Sociocultural Theory, introduces the concept of the Stimulated Recall Protocol for attrition research.
References Cook, V. J. (2003). “The changing L1 in the L2 user’s mind”. In V. Cook (Ed.), Effects of the Second Language on the First (pp. 1–18). Clevedon: Multilingual Matters. Cutler, A., Mehler, J., Norris, D. & Segui, J. (1989). “Limits on bilingualism”. Nature, 340, 229– 230. Flege, J. E. (1987). “The production of ‘new’ and ‘similar’ phones in a foreign language: Evidence for the effect of equivalence classification”. Journal of Phonetics, 15, 47–65. Grosjean, F. (1992). “Another view of bilingualism”. In R. J. Harris (Ed.), Cognitive Processing in Bilinguals (pp. 51–62). Amsterdam: North Holland. Hernandez, A. E., Bates, E. A. & Avila, L. X. (1994). “On-line sentence interpretation in Spanish-English bilinguals: What does it mean to be ‘in-between’?” Applied Psycholinguistics, 15, 417–446. Köpke, B. & Schmid, M. S. (2004). “First language attrition: The next phase”. In M. S. Schmid, B. Köpke, M. Keijzer & L. Weilemar (Eds.), First Language Attrition: Interdisciplinary perspectives on methodological issues (pp. 1–43). Amsterdam: John Benjamins. Major, R. C. (1992). “Losing English as a first language”. The Modern Language Journal, 76 (2), 190–208.
Bilingualism and attrition
Van Hell, J. & Dijkstra, T. (2002). “Foreign language knowledge can influence native language performance in exclusively native contexts”. Psychonomic Bulletin & Review 9 (4), 780– 789.
Language attrition at the crossroads of brain, mind, and society Barbara Köpke
Laboratoire “Jacques Lordat” (Octogone), Université de Toulouse – Le Mirail, France
This chapter is an attempt to survey the predictions of some of the most important factors involved in attrition and to discuss them with respect to empirical data from L1 attrition. The factors discussed include biological aspects (plasticity, activation thresholds, inhibition and emotional implication), cognitive aspects (memory, aptitude, literacy and task dependency) and external factors (the role of language use, the cultural context and attitudes). The discussion shows that (a) many of these factors are closely linked, (b) some of them may be considered cluster factors which are more influent and more strongly interrelated than others, and (c) none of the factors may alone be considered responsible for attrition. This may be accounted for by a multicomponential view of language attrition. Keywords: interdisciplinary approaches; language control mechanisms ; multicomponential view
Introduction The growing interest in language attrition over the past twenty-five years cannot be explained solely on the basis of the fact that most of us are in one way or another affected by the phenomenon. Rather, the focus in attrition has become more and more theoretical since research in this domain has contributed to sheding light on the dynamics of bilingual and multilingual language competence and processing (see de Bot, this volume). This may be due to the fact that attrition research is closely linked to other research within this field. First of all, there is the link with studies in language acquisition and learning – which may be considered the reverse of attrition if we expand Jakobson’s regression hypothesis (1941) – as documented by a growing number of studies attempting to apply theories of L2
10
Barbara Köpke
acquisition/learning to L1 attrition (e.g. Gürel 2002; Keijzer 2004; Montrul 2004; Rieussec & Köpke 2005, 2007; Sharwood Smith & van Buren 1991). Secondly, attrition has been related to research in language pathology as both are concerned with some kind of language dysfunction (de Bot & Weltens 1991). Other than Dressler (1991) there are no empirical studies comparing aphasia and attrition (or shift) data, but as we will see in this chapter, a growing number of attrition studies are based on theoretical notions developed to account for bilingual aphasia. Finally, attrition has also been related to diachronic language change and might benefit from parallels drawn with that field (Schmid 2004). One of the specifics of attrition research might well be that it also – and perhaps in a more striking way than the study of language acquisition or pathology – cannot be grasped without taking external factors into account. Attrition, albeit clearly defined as an individual phenomenon (de Bot 2001; Köpke & Schmid 2004), is intimately linked to social aspects of language use. Consequently, language attrition is a promising research issue for the exploration of links between the brain, mind and external factors that are also of particular interest for research in multilingualism. Starting from this assumption, the aim of the present chapter is to discuss the predictive values of various factors involved in attrition. Among these are brain mechanisms, cognitive processes and external factors, some of which have already been extensively discussed in the attrition literature. Other factors are discussed here for the first time. The list of factors proposed below is neither exhaustive, nor can the discussion of each be more than embryonic. The approach is mainly psycholinguistic, although there will be frequent references to notions arising from the cognitive neuropsychology of bilingualism (Paradis 1997) and to sociolinguistic concepts around bilingualism. I will further follow Nespoulous (1994) in the distinction of brain mechanisms (discussed in the first part) which are – at least in their original substance – physiologically grounded, and cognitive processes (discussed in the second part) which are involved in the functional architecture of language in the human mind (Nespoulous 1994: 397).
1. Brain mechanisms 1.1 Plasticity Plasticity is the most often cited reason for a biologically-constrained critical period for language acquisition and/or learning (see the discussion of this distinction in Ellis 1994; Pallier, this volume; Paradis 1994, 2004; Zobl 1995). It is based on the idea that synaptic connections are not fully mature in the first years of life, facilitating quick adaptation to new situations (see the detailed discussions in
Attrition at the crossroads
Pallier, this volume; Ventureyra 2005). This is easily transposable to the immigrant situation: a radical change in the linguistic environment requires adaptation to the new language situation, including learning the host language, in combination with reduced L1 use. The concept of brain plasticity predicts that this adaptation will be easier and quicker the younger the subject (Lenneberg 1967; Penfield & Roberts 1959). Faster language learning due to greater plasticity might also imply strong L1 attrition in young immigrants, whereas in older immigrants, reduced brain plasticity would both hinder the adaptation to the L2 environment and prevent L1 attrition. For L2 acquisition and/or learning, such maturational effects have been amply documented (e.g. Birdsong 1992; Flege, Yeni-Komshian & Liu 1999; Hyltenstam & Abrahamsson 2003; Johnson & Newport 1989; see the descriptions in Pallier, this volume). L1 attrition data appear to corroborate the hypothesis that age is one of the most predictive factors in attrition (see the literature reviews in Francis 2005; Köpke 2004a; Köpke & Schmid 2004; Ventureyra, Pallier & Yoo 2004) since an L2 learned early in life has been repeatedly reported to quickly replace another language, while in later bilinguals the L1 appears largely impervious to erosion. Whether this is due to greater stabilization of linguistic competence in older subjects or not, however, still needs to be demonstrated.
1.2 Activation thresholds The activation mechanism (Luria 1966, 1974) is one of the best known brain mechanisms and has been used, metaphorically, for the cognitive modelling of language processing, especially monolingual and bilingual lexical access (for monolingual access e.g. Dell 1986; McClelland & Rumelhart 1981; McClelland & Elman 1986, for bilingual access e.g. Grainger & Dijkstra 1992; Grosjean 1988). Activation was invoked as a neuronal basis of memory (comprising concepts and linguistic sound forms) half a century ago: “Each passage of a stream of neuronal impulses leaves behind it a persisting facilitation, so that impulses may pass that way again with greater ease” (Penfield & Roberts 1959: 246). The idea that frequency of use entails facilitation of activation is further developed in Paradis’ Activation Threshold Hypothesis (Paradis 2004, this volume) which assumes that a certain amount of neural impulses is needed in order to reach activation. This framework has been shown to account for differential inhibition in polyglot aphasia, and also for the control over two competing language systems in healthy bilinguals (Paradis 1985, 1993, 2004). Since the activation threshold of any item stored in memory is dependent upon frequency and recency of previous activation, the most important predic-
11
12
Barbara Köpke
tive factor for language attrition within this framework is language use (Paradis, this volume). To be more precise, those L2 items or rules used more frequently will be more easily activated when they are in competition with less frequently used L1 items or rules. This principle has been shown to be valid for the lexicon in Köpke (2002). Concerning grammatical parameters, it has been clearly demonstrated with respect to Turkish and English binding properties by Gürel (2004, this volume) that attrition occurs where both languages have equivalent forms which may compete, but not where there are no equivalent forms in both languages, so any competition between them is precluded. These findings underscore the validity of the notion of frequency of use in attrition, but also the necessity to interpret findings within a rigorous linguistic framework in order to account for grammatical attrition. The lack of the latter might be the reason why there is not more experimental evidence for the role of language use, although this factor is systematically presumed to play a major role.
1.3 Inhibition Inhibitory neural cells inhibit the electric activity of the neural cells with which they establish contact (Fabbro 1999: 70). As an important means of controlling neural activity, the inhibition mechanism is ubiquitous in the nervous system (Boujon 2002; Green 1998b). Inhibition and activation have been linked in cognitive models of language processing, within unilingual (see Berg & Schade 1992 for an overview) and bilingual models alike (e.g. Green 1986, 1998b; Paradis 2004, this volume). It has been argued that “… a combination of excitatory and inhibitory processes is more efficient in achieving selection than excitatory processes alone” (Green 1998b: 102). From a cognitive point of view, inhibition can be defined as “… the negative effect the processing of a given item has upon the activation value of another” (Berg & Schade 1992: 406). It helps to prevent activation from spreading to too many parts of the system and allows for temporarily stable representations. As such, it plays a key role in any instance of mental control (Green 1986) since it is needed in order to cancel out competitors that are likely to interfere with a selected item, structure, or language (Green 1986: 217, 1993: 264). Consequently, inhibition is likely to be a key mechanism in situations where maximum interference is to be expected such as L2 learning, L1 attrition, or polyglot aphasia (Ijalba,
. Note however that in this definition inhibition appears as a kind of side-effect of the activation of another item, unlike Green’s framework which clearly sees inhibition as an active and resource-consuming mechanism.
Attrition at the crossroads
Obler & Chengappa 2004). However, the inhibition mechanism has received less attention than activation in these fields. In the context of polyglot aphasia, Green claims that the patients’ inability to use one language spontaneously is not due to lack of activation but to the inability “… to suppress sufficiently the activation of the other system” (Green 1986: 220), in other words, to lack of inhibitory resources. It is possible that a similar mechanism is at work in the case of L1 attrition. Two means of deselecting a language can be hypothesized in non-pathological contexts: either through a lowering of activation (which is the automatic consequence of non-use and hence dependent on time rather than on resource) or through inhibition (which can be obtained rapidly at the expense of resources). Consequently, for the L1 attriter, the use of the L1 may be doubly impeded: by lack of activation of L1 on the one hand and the need to strongly inhibit the highly active L2 on the other. The precise impact of inhibition mechanisms, however, would depend on several factors, namely: (a) Actual language use including use of each language, language mode, codeswitching, and probably also attitudes to mixed language, etc. (see 3.3. below); and (b) Typological relatedness of the languages: the closer the languages and the more structures they share, the more interference is to be expected. The predictions that can be inferred from this imply that inhibition, as a crucial mechanism allowing the ‘separation’ of languages, may account for different outcomes of attrition: – In immigrants who have no contact with other immigrants and who never use L1, that language will be completely dormant (Green 1986; Norman & Shallice 1986) resulting in both accessing and processing difficulties arising from the fact that a large proportion of the available resources will be consumed for the inhibition of the very active L2. Conversely there will not be much interference in terms of errors since both languages have very different activation states. – In immigrants who have frequent contact with other bilingual immigrants and who are used to operating in a bilingual mode, both L1 and L2 are active. Hence L1 will be less difficult to access but its use will contain much more interference (see Köpke 2001 for an illustration of both cases and Schmid, this volume, for further discussion of this point). – In addition, inhibition might explain why recent immigrants are sometimes reported to experience difficulties in L1: since they are still in the process of learning the L2, they have to inhibit their more dominant L1 very strongly, leading to processing difficulties which will appear similar to attrition. Some evidence supporting such an explanation comes from studies on language switching where a higher cost has been observed when individuals have to
13
14
Barbara Köpke
switch into the dominant language compared to switches into the non-dominant language (Green 1998a; Meuter & Humphreys 1997).
1.4 Involvement of subcortical structures Language processing is not only a matter of the cortical structures of the left (grammar) and right (pragmatics) hemispheres, subcortical brain structures have also been found to participate in language processing (Crosson, Novack & Trenerry 1988), though their role is still difficult to define. The limbic system – as one of the most ancient parts of the brain – is the centre of the communicative functions network in primates and also plays a major role in human communication (Damasio 2003). It is assumed that parts of it, e.g. certain structures of the left basal ganglia and of the left thalamus, participate in the first stages of verbal expression and, most importantly, that the limbic system plays a major role in many control functions, both motor and cognitive (Fabbro 1999: 79). Moreover, subcortical structures have also been shown to be the seat of emotion in the brain (Fabbro 1999; Pavlenko 2005). Clinical evidence suggests that the subcortical structures of the right hemisphere play a role in the regulation of automatic speech of high emotional significance (e.g. prayers, see Speedie, Wertman, Ta’ir & Heilman 1993), while Pavlenko (2005: 154) claims a participation of the limbic system in spontaneous emotional speech. Hence, communication and language on the one hand and emotion on the other share some neuroanatomical structures. The relation between emotion and language and the consequences this may have for a specific language is of primary interest for attrition research. Pavlenko (2005) claims that this link should best be studied in bilingual or multilingual subjects because they provide the unique opportunity to compare verbal behaviour and emotional reactions across languages. She stresses (2005: 153) that “… depending on their linguistic trajectories, bi- and multilinguals may have different neurophysiological responses to their respective languages, or, at least, to emotion-related words” (such as taboo expressions or swear words). Words of a specific language are linked to personal fears, or experience of prohibition, punishment or reward, and these contribute to the individual’s “… perception of language embodiment, whereby words invoke both sensory images and physiological reactions” (Pavlenko 2005: 155). Research on such aspects of bi- or multilingualism, however, is still scarce (see Pavlenko 2005 for an extensive review). Emotional implication has received more attention with respect to language acquisition, or learning. For Schumann (1997), language acquisition is favoured by positive emotional values which allow the establishment of links with deeper brain structures. This is also the point of view of Lamendella (1977) and Paradis
Attrition at the crossroads
(1994, 2004) who suggest that the L1 is generally based more solidly on the limbic system and its emotional values, whereas L2 learning, particularly when it takes place in the classroom, is largely based on the mere cognitive processes of the neocortical areas (cf. also the discussion on procedural vs. declarative memory below). Similarly, Pavlenko (2005: 156) distinguishes between embodied languages, acquired within an emotional and contextualized process, and disembodied languages, learned in a decontextualized process such as FL classroom learning. But she does stress that both constitute the extremes of a continuum and that emotional implication is not restricted to the L1, particularly when the L2 becomes the basis of a person’s affective life. She suggests, therefore, that affective linguistic conditioning is a life-long enterprise (Pavlenko 2005: 157). This point is of particular interest for attrition research. Even though the relationship between emotional impact and attrition still remains unknown, there is data suggesting that language learning and attrition may depend on a similar biological base. Especially cases where particularly traumatic circumstances obtain (see Schmid 2002; or the studies conducted with adopted children: de Bode 1996; Isurin 2000; Nicoladis & Grabois 2002; Pallier, Dehaene, Poline, LeBihan, Argenti, Dupoux & Mehler 2003; Ventureyra 2005) suggest that strong emotional events may have serious consequences. This allows the assumption that the motivation for maintaining a language is similar to the motivation for language learning, and that both these situations entail high emotional values which may be related to subcortical brain structures. Emotion is most likely a key factor in any case of attrition (cf. also the lifespan perspective taken by de Bot, this volume), and it may well play an even bigger role in language attrition than in language acquisition or learning. The possible prediction arising from this is that L1 attrition would occur only in cases in which either the L2 is strongly emotionally ‘invested’ (which does not necessarily entail more success in L2 learning), or the L1 is strongly rejected (after traumatic experiences such as for example persecution, abandonment, betrayal, or rape). In sum, attrition seems to be dependent upon four principal brain mechanisms: – plasticity, which is a function of the age of onset of bilingualism and/or attrition – activation, which is a function of frequency of language use – inhibition, which is a function of type of language use – subcortical involvement, which is a function of emotional implication These mechanisms seem also to be reflected in the cognitive processes implied in language attrition.
15
16
Barbara Köpke
2. Cognitive processes 2.1 Memory Memory can be related to attrition in a number of ways. The first notion which has to be considered is long term memory (LTM) where linguistic knowledge is stored over time just as many other kinds of knowledge (e.g. arithmetic, historical facts, playing chess, driving a bicycle). If the information stored in LTM is not activated regularly, it might be forgotten. But forgetting may be caused by multiple mechanisms which can contribute in different ways to an explanation of the forgetting of a language. This aspect will be discussed in the first Section 2.1.1. Second, as the kind of information stored in LTM is of diverse nature, cognitive psychology has further divided LTM into structurally and functionally distinct subcomponents according to the kind of material which will be stored, the way this is done and how the information will be accessed. Among these, the distinction between procedural and declarative memory (Paradis 1994; Ullman 2001) is of particular interest for attrition research (see below Section 2.1.2). Third, the role of memory in language processing in general and in attrition in particular is not restricted to the long term retention of linguistic information in LTM. Memory processes are also involved in online processing of information and control of two language systems (see 1.2 and 1.3). Such functions rely on working memory (WM) – usually defined as “… a limited capacity system responsible for the temporary storage and processing of information while cognitive tasks are performed” (Collette, van der Linden & Poncelet 2000: 46). In Baddeley’s original conception (1986), WM is composed of two peripherally-based storage systems (the phonological loop and the visuospatial sketchpad) and a central administrator – called central executive in more recent developments of the model (Baddeley 2000, 2003) – whose role is, roughly speaking, to coordinate the whole thing and to allocate attentional resources. The third section (2.1.3) is an attempt to elaborate on a possible role of WM, and more specifically the central executive functions, in attrition.
2.1.1 Forgetting MacLeod (1976) used the savings paradigm focussing on the kind of residual information (phonological form vs. semantic information vs. information on the language the word was in) in the memory traces of L1 and L2 words which could not be recalled five weeks after memorisation (see also de Bot & Stoessel 2000 for L2 attrition). The results seem to suggest that, over time, meaning is retained more easily than form and language information. However, this and other similar studies (see Ammerlaan 1996 for a detailed review) are not really instructive for
Attrition at the crossroads
L1 attrition research because they do not so much investigate the forgetting of a language, but, rather, the retention of linguistic material (see 2.1.2 below). From another perspective, closer to attrition, Bahrick (1984) investigated the retention of school Spanish in Anglophones who had stopped learning Spanish between 1 and 50 years earlier. The results indicate that forgetting slows down as time passes. Moreover, for participants who had higher L2 proficiency at the end of their Spanish curriculum, there was an initial period where L2 knowledge remained stable before slowly starting to vanish. Written comprehension was less affected by forgetting than all other skills. However, this study also focused on the retention of formally-acquired L2 knowledge. The only studies concerning an L1 are based on hypnosis data (As 1963; Fromm 1970, see also Footnick, this volume). The fact that As and Fromm show that their patients are able to recover an otherwise completely forgotten L1 under hypnosis suggests that, in some cases, even total L1 attrition remains a question of retrieval failure, i.e. attrition at the performance level. What all these studies show is that language attrition might be tackled as a memory problem. In his detailed discussion of different forgetting theories within the context of L1 attrition, Ecke (2004) distinguishes different processes: decay, interference, regression and suppression, distortion, retrieval slowdown and failure, cue dependency, and interaction and dynamic systems. Of course, many of these processes are likely to be interrelated (decay can cause retrieval slowdown or failure, etc.). The predictions arising from these theories are discussed in detail, suggesting that the L1 attrition literature provides plausible examples for all of these processes. So, the most important point of this discussion is that, similar to forgetting, there is not just one possible process of attrition, but rather a variety of processes which are likely to be influenced by a multitude of factors allowing for many different predictions.
2.1.2 Declarative and procedural memory Despite the extensive discussion of the distinction between declarative and procedural LTM and its relevance for bilingualism by Paradis (e.g. 1994, 2004) and Ullman (2001), this distinction has not received the attention it deserves in all domains of research on bilingualism. The only study which applies this concept does so in the context of polyglot aphasia (Ijalba, Obler & Chengappa 2004: 80). Following Ullman’s definitions (2001: 106), declarative memory is involved in the learning and use of knowledge about facts and events and seems to be particularly suited for arbitrarily related information via associative or contextual binding. Procedural memory is involved in the learning and control of cognitive and motor skills and habits, which are typically based on sequences. Both memory systems rely on distinct neurofunctional and neuroanatomical structures
17
18
Barbara Köpke
(Paradis 1994; Ullman 2001). With respect to bilingualism, the declarative/procedural model predicts a difference in the memory structures involved in the processing of grammatical structures in earlier vs. later-learned languages. Whereas the lexicon is stored in the associative structures of declarative memory for all languages, the implication of procedural memory in the processing of grammar would depend on age of acquisition (since declarative memory plays an increasing role in later language learning) and proficiency (since a language that has been less automatized by the speaker is more likely to be processed declaratively). Roughly, the lexicon of an L1 and a later-learned L2 are likely to be located in the same memory structures (declarative memory) whereas the grammars of an earlier-learned L1 and a later-learned L2 are likely to depend on different memory systems (procedural memory for L1, declarative memory for L2). There would, however, be no differences between the two languages of early bilinguals, as corroborated by data from brain imaging studies with bilinguals (see Köpke 2004a; Ullman 2001), from polyglot aphasia (Ijalba et al. 2004) and specific language impairment (Ullman 2001: 109), among others. If we take into account the assumption that interference or competition arises only between structures that are similar in nature, this would allow the following predictions with respect to L1 attrition: (a) vocabulary, which is declarative for both languages, is a good candidate for interference in both early and late bilinguals; (b) as far as grammar is concerned, in contrast, interference is expected to be more pronounced in early bilinguals. For late bilinguals, the largely declarative L2 grammar is less likely to interfere with the procedural L1 grammar. These predictions are not incompatible with L1 attrition data so far: studies with late bilinguals suggest that the lexicon is affected more strongly by L1 attrition than the grammar (e.g. Köpke 2002). Furthermore, it is striking that investigations of early bilinguals (mostly case studies) frequently focus on morphology or syntax (e.g. Håkansson 1995; Kaufman & Aronoff 1991; Levine 1996; Schmitt 2004; Seliger 1991; Turian & Altenberg 1991; Vago 1991), compared to studies of late bilinguals, which are mainly concerned with more general features (e.g. Ben Rafael 2004; Gross 2004; Köpke 1999; Yağmur 1997) or focus on the lexicon only (e.g. Ammerlaan 1996; Hulsen 2000; Olshtain & Barzilay 1991; Pavlenko 2003).
2.1.3 Working memory The role of working memory in language processing is still controversial. Apart from a possible role as an emergency resource in case of cognitive overload (McCarthy & Warrington 1994: 345), it is generally assumed that higher working memory capacity is related to higher comprehension skills, both in listening and reading comprehension (Daneman & Carpenter 1980; Just & Carpenter 1992; McCarthy & Warrington 1994), but its role in language production has received
Attrition at the crossroads
much less attention, at least as far as unilingual language production is concerned. The most likely assumption is that WM – and specifically the central executive of WM (Baddeley 1986, 2000, 2003) – plays a role in tasks involving high levels of cognitive control. Among the mechanisms associated with the central executive is inhibition (van der Linden & Collette 2002) which has already been discussed in 1.3. In bilinguals the need for a control mechanism in online speech production is much more obvious than in unilinguals (e.g. for the L2 learner: Bialystok & Sharwood Smith 1985) since the use of the two languages needs to be controlled according to language mode (Grosjean 1992). Recall that inhibition (see above 1.3) has been claimed to be the mechanism exercising control over the two competing language systems in bilingual language processing (Green 1986, 1998a, 1998b). This might point to the central executive of WM having a possible role in the control functions involved in bilingual speech, especially in situations where higher control demands arise, e.g. when one language is much more accessible than the other. In a similar vein, it has recently been proposed that interference would not so much be a side-effect of learning, but arise from mental and behavioural control (Anderson 2003: 416). In this view, forgetting would result from “…inhibitory control mechanisms recruited to override prepotent responses” (Anderson 2003: 415). It is possible that this also has implications for predictions concerning L1 attrition. Strong cognitive demands arising from reduced access to the less-used L1 and from its competition with the more accessible L2 would put strong demands on executive control mechanisms, resulting in processing difficulties similar to those found in L2 learners: an inability to activate automatic procedures (Norman & Shallice 1986), thereby triggering controlled strategic behaviour which is much slower and characterized by dysfluencies. Attrition would thus, to a large extent, be a processing issue, as has been suggested repeatedly in discussions of the competence/performance issue (e.g. Sharwood Smith, this volume). Although this is among the oldest (and probably most complex) theoretical issues in attrition research (e.g. reviews in Köpke & Schmid 2004; Sharwood Smith 1983) experimental studies investigating attrition from that angle are still scarce (among the exceptions see Dussias 2004).
2.2 Language aptitude A further cognitive skill which may be relevant to language attrition but has not so far been investigated in this context is language aptitude. Within research on L2 acquisition, this concept has been defined as the potential or talent a person has
19
20 Barbara Köpke
for learning foreign languages (e.g. Carroll 1973; Krashen 1981). It has been found to vary considerably between individuals, while remaining relatively independent of other cognitive factors such as general intelligence (Abrahamsson & Hyltenstam 2005). Language aptitude is a composite measure established through a variety of independent variables including phonetic encoding, grammatical sensitivity and phonetic memory (e.g. Meara, Milton & Lorenzo-Dus 2003) which have been found to correlate with success in language learning. Recently, Abrahamsson & Hyltenstam (2005) investigated whether language aptitude could account for exceptional L2 achievement in late learners by correlating it to grammaticality judgement scores in Spanish L2 speakers of Swedish who were carefully screened for a native-like command of L2 Swedish. The results confirmed that a high level of language aptitude appears to be a necessary condition for near-native L2 attainment in adulthood. This finding is interpreted as evidence for maturational constraints which can only be overcome by exceptional language learning aptitude. The implications of aptitude for L1 attrition have not yet been discussed, all the more since L2 proficiency is not usually assessed in attrition research. The concept of language aptitude is nevertheless promising for attrition research and might be better suited to capture individual variation in attrition than e.g. education level (see the discussion in Köpke & Schmid 2004: 10–12). The predictions ensuing from this for attrition are that greater language aptitude, which is generally assumed to lead to higher L2 proficiency, could prevent attrition – at least to some degree. Note that such a prediction is in contradiction with proficiency-based accounts (Opitz 2005) e.g. within the multicompetence model (Cook 1992): these accounts predict more L1 attrition in more proficient L2 speakers, whereas an aptitude-based account predicts less L1 attrition in these cases. This issue should remain one of the challenges in L1 attrition research for the next years.
2.3 Literacy From a more psycholinguistic perspective, it has been proposed that literacy might contribute to the cognitive organisation of language and contribute to age effects observed in L1 attrition together with other factors, such as brain plasticity and type of memory involved in language learning (Köpke 2004a). Literacy can be seen as a factor which might prevent attrition in several ways. On the sociolinguistic level, literacy (a) allows the speaker to maintain contact with the L1 by reading, which might be an important source of confirming evidence for the L1 (see Sharwood Smith & van Buren 1991; see below 3.3 for a more detailed discussion of this aspect); and (b) the wish to have access to written input may enhance motivation for maintaining the L1. But more importantly, from the neurofunc-
Attrition at the crossroads
tional and psycholinguistic perspective, (c) it is likely that literacy contributes to the grounding of a language in memory as it adds orthographic representations and new synaptic connections. However, these effects are closely related to age, and distinguishing the factors literacy and age in an experimental setting raises methodological problems that might be impossible to overcome. It can nevertheless be claimed that less attrition is to be expected in subjects who have had the opportunity to become literate in the L1, especially if they frequently use that skill, a suggestion made by Olshtain (1986) in order to explain her findings in L2 attrition. While the methodological difficulties associated with the actual testing of this factor may be insolvable, this aspect should nevertheless be kept in mind.
2.4 Task dependency Last but not least, attrition of oral production skills is largely a question of online processing (see above) and as such is directly dependent upon task demands. Different language tasks with different cognitive demands do not yield the same results, even if exactly the same linguistic structures are tested. It has been shown, for example, that lexical access in a picture/word matching task is easier than lexical retrieval in a naming task (Ammerlaan 1996; Hulsen 2000), two tasks with very different requirements regarding the level of activation needed for their accomplishment. Similarly, Major (1992) found that L1 Voice Onset Time (VOT) was closer to native standards in a more formal read-aloud task than in informal spontaneous speech. Köpke (1999) showed that interference was stronger in a grammaticality judgement task than in an oral sentence generation task testing the same grammatical structures. The issue of task dependency certainly is a methodological question, but it also raises important theoretical issues. Paradis (2004: 11) relates differential behaviour observed between online and offline language tasks to the type of memory structures involved (procedural vs. declarative) in the resolution of a task. So, in some cases, task dependent differences in linguistic behaviour are due to the fact that some tasks allow for the use of metalinguistic knowledge, whereas others do not. But variation in linguistic behaviour may be accounted for in different ways, depending on the theoretical framework of the analysis: the psycholinguistic account relates variation to different task demands (e.g. leading to different activation levels required as in Ammerlaan 1996) or to different task goals as perceived by the subject (e.g. focus on form in a formal read-aloud task as in Major 1992). . The question of what exactly is actually measured by grammaticality judgements has furthermore been extensively debated (Altenberg & Vago 2004) in the context of attrition.
21
22
Barbara Köpke
The linguistic account would be to take variation as indicative of some sort of optionality developing in the linguistic system of the attriter (e.g. Tsimpli, this volume). It is perhaps worth considering the possibility that these different accounts may not merely be the outcomes of different theoretical frames, but that they may reflect the complementarity of several factors within a more global approach of attrition. Such considerations bring us back to the competence and performance distinction (Chomsky 1965; Fodor & Garrett 1966; Sharwood Smith & van Buren 1991) discussed below. Since precise predictions arising from the issue of task dependency need to be discussed with respect to each specific task, further discussion of this issue is beyond the scope of this chapter. In sum, it appears reasonable to assume that attrition is cumulatively predicted by a. internal cognitive factors arising, for example, from characteristics of the memory structures underlying linguistic competence, b. structural organisation of the linguistic system (in relation namely with literacy) and c. linguistic aptitude of the subject together with more task dependent factors. Many of these cognitive aspects are closely linked to brain mechanisms and depend on constraints due to the characteristics of the human brain and mind. But the claim in what follows will be that the human mind does not develop in complete isolation, but rather in permanent interaction with the subject’s social environment.
3. External factors Linguistic development over the lifespan follows internal constraints related to the characteristics of the human brain and mind, but these constraints merely represent opportunities which may or may not be realised as a response to environmental conditions. However, since neuro- or psycholinguistic factors and social conditions are closely linked, this is the point where the snake starts biting its tail: the discussion of most of these aspects and the predictions they allow has been initiated above. The discussions that follow will thus be restricted to the aspects that have not been developed yet.
Attrition at the crossroads
3.1 Language use The issue of language use has already been considered extensively in relation with activation and inhibition (see 1.2 and 1.3). Language use has often been assumed to be an important predicting factor for language attrition and is often implicitly invoked by empirical work (see Köpke 2004 and Köpke & Schmid 2004 for detailed reviews), even though it is difficult to measure. It has previously been proposed to be the key variable at the interface of the psycholinguistic and sociolinguistic aspects of attrition (de Bot 2001). However, the focus has been on quantitative aspects of language use, e.g. amount of contact or frequency of L1 use (but see Schmid, this volume). Intuitively, it makes sense that individuals who make little use of a language experience more attrition than speakers who use the language more frequently, and frequency of L1 use is also a crucial aspect in theoretical models in the context of the Activation Threshold Hypothesis. The discussion of the inhibition mechanism, however, clearly shows that frequency of use is not the only aspect of use that matters. As stressed also by Schmid (this volume) quality of use, or type of contact, is equally important, and this aspect has received hardly any attention. First of all, a distinction needs to be applied to the use of productive skills and receptive skills. In other words, is active language use (i.e. in production) necessary in order to prevent the L1 from deteriorating, or is regular input sufficient to maintain a language once completely mastered? This question is reminiscent of the discussion around input and output in SLA research (e.g. Swain 2000) and was first invoked in that context by Sharwood Smith and van Buren, who propose that “…the native speaker not only needs evidence for developing an L1 system but also needs evidence to maintain his or her L1” (1991: 23). In this view, input would be crucial for the maintenance of a language. This implies also that input might be sufficient, for example in the case of a speaker who has no opportunity to interact with other L1 speakers but may nevertheless maintain the L1 by written exchanges, or simply by input received from books, films or the Internet. Secondly, this raises the question of the quality of that input, which may to some degree be dependent on the sociolinguistic environment of the immigrant (see below). Predictions arising from qualitative aspects of contact are discussed in detail by Schmid (this volume).
3.2 The cultural context of the immigrant The relationship between cognition and culture is illustrated in Sperber and Hirschfeld’s (2007) theory, where it is claimed that culture arises from Social
23
24
Barbara Köpke
Cognitive Causal Chains (SCCC) establishing semantic relationships between people. These SCCCs allow for stabilizing representations and practice within a given community. This view implies that these representations concern – among many other things – language status (see 3.3 below), but also linguistic and pragmatic norms. In other words, SCCCs would include linguistic and pragmatic input as discussed by Sharwood Smith and van Buren (1991). Adopting Sperber and Hirschfeld’s view, it can be claimed that immigrants are cut off from the SCCCs of the L1 community and no longer have any support in stabilizing their L1 representations and practice. This resembles Sharwood Smith and van Buren’s suggestion that (1991: 23): “…the L1 changes not because of lack of use but because of lack of confirming evidence that L1 is the way it is in the L1 community” (my emphasis). The input, or lack thereof, from the L1 community would hence be the critical factor in L1 attrition, and this is likely to be particularly true with respect to the attrition of pragmatic norms (an area of language attrition which is very much under-researched). This raises the question of the L1 community which, in an immigrant setting, can take various forms: either a) the immigrant has no contact with other immigrants in the host country and any confirming evidence comes from the home country, or b) she is a member of an immigrant community which will be the source of at least part of the confirming evidence and new SCCCs. It has been proposed recently (Köpke 2004b) that these two sorts of cultural contexts may have different consequences for attrition: a. In the case of completely isolated immigrants, L1 use is reduced to a necessarily limited amount of contact with L1 speakers in the home country and, possibly, books, TV, and Internet. The SCCCs related to L1 will become weaker and weaker, but they remain those of the L1 community in the home country. The linguistic norms on which the L1 competence is based will not change, but L1 will be more and more difficult to access since its activation level will decrease as L1 becomes dormant (Green 1986); hence this case of attrition will mainly be a question of control over intact knowledge, i.e. performance attrition. b. If the immigrant is part of an immigrant community in the host country, she will be exposed to different kinds of input: possibly to input from the L1 community in the home country as in case a), but also to input from the immigrant community, including attrited speech and different learner varieties . Following Sperber and Hirschfeld (2007), causal chains of culture consist in an alternation of mental and public episodes, both of which are linked through imitation and communication.
Attrition at the crossroads
from 2nd and 3rd generation speakers. The amount of contact will probably be greater than in case a) due to regular L1 input in the migrant setting via clubs, churches, contact with other migrants, community newspapers, etc., but the input will comprise much more structural variation. Moreover, the contact with other migrants who live in close contact with another language is likely to lead to new SCCCs (better adapted to the L2 culture) and to the development of a contact variety, as has been observed by Grosjean & Py (1991) for Spanish immigrants in Neuchâtel, or by Montrul (2004) or Silva-Corvalan (1991) for Spanish speaking immigrants in the USA. In such a case, attrition will involve a restructuring of L1 competence due to the development of new L1 norms in the contact situation. This could furthermore be enhanced by the possibility to communicate in a bilingual mode (Grosjean 1992) characterised by massive codeswitching which may also have consequences on inhibitory control in a largely bilingual context (see 1.3 above). The issue of language use and contact certainly needs to be investigated further in the context of attrition (see Schmid, this volume). Nevertheless, the predictions arising from these thoughts on the type of language use as a function of the cultural context of the immigrant are that attrition may be either more performance-oriented or more competence-oriented (Köpke 2004b: 1341), depending on whether the L1 contact is mainly with members of the L1 community in the country of origin, or with other immigrants in the host country. But this is obviously also a question of attitudes to the L1 norms and to the community.
3.3 Attitudinal factors Attitudinal factors include a complex variety of values attributed by the community or the social environment to the languages (L1 and L2) as well as issues like bilingualism, immigration, integration, etc. (see Ben-Rafael & Schmid, this volume). These values may or may not be identical with the internal values the subject subscribes to (see 1.4 and Schumann 1997), but they will in all cases have some influence on these internal values and on the motivation to acquire L2 and/ or to maintain L1. Motivation usually arises from external factors such as the socio-economic or ideological context. One example illustrating how integrative or instrumental motivation (Gardner & Lambert 1972) are dependent on this kind of context is given by Ben-Rafael & Schmid (this volume) for different groups of immigrants in Israel: whereas the Francophone immigrants arrived in Israel during a period where the exclusive use of Hebrew was an ideological imperative, language maintenance is much higher among more recent Russian immigrants who have a more
25
26 Barbara Köpke
economic and hence instrumental motivation, allowing for the maintenance of Russian as a family language. Other aspects of motivation may be directly linked to age: for example, adults differ from children in that L1 is an important part of their identity that cannot easily be abandoned. On the other hand, a very young child does not have the same motivation with respect to language as a child of school age who is much more oriented to the peer group than to the family. School children should therefore be most prone to attrition since they are the age group most motivated to integrate into the L2 environment. However, age and previously-acquired literacy skills in L1 might counteract such a tendency. Still another type of attitude and motivation may depend on the migrant’s view of the immigrant situation, and will be directly linked to the cultural context as discussed in 3.2. It includes the immigrants’ attitudes with respect to language competence and bilingualism, their origins, and their integration into the L2 community: is a need felt to be a ‘perfect’ speaker of each language? Does the speaker accept codeswitching as a bilingual communication mode or is she opposed to language mixing? Is L1 attrition seen as a logical consequence of integration into the L2 community or is it felt as the loss of some aspect of an immigrant’s personality? These and other questions (see Pavlenko 2005 for more examples) contribute to a very complex picture of the immigrants’ relation with their languages, which will influence language embodiment. Matters are further complicated by the fact that the predictions arising from this are as numerous as the personal configurations are variable. It appears nonetheless that external factors are of crucial importance in attrition. Even if attrition as a process is based on brain mechanisms and cognitive processes, external factors are most likely to play a major role in determining whether there will be attrition, to what extent, and what type of attrition will occur.
4. General discussion What emerges clearly from the discussion of these various predictions is that attrition is dependent on a large variety of factors. Further complications arise from the fact that many of the factors involved weaken or even cancel each other out. Nevertheless, two strong tendencies emerge:
. Of course it cannot be excluded that other factors, as for example group size, interact with motivation in this case.
Attrition at the crossroads
1. Internal and external factors may have an impact on either amount of attrition, type of attrition, or both: it is proposed here that it is crucial to distinguish factors with quantitative effects, which will play a role in the extent of attrition, and others which will be involved in the qualitative distinction of different attrition phenomena. 2. Moreover, among the most important factors influencing attrition, it is possible to isolate some which seem to be more central than others because they appear to be composite factors related to many others. For example, age and cultural context are complex variables related to many of the other factors. It is further suggested that none of these cluster factors alone may account for L1 attrition.
4.1 Quantitative vs. qualitative influence on attrition The first observation is that some factors influence amount of attrition whereas others may have an impact on type of attrition. Obviously, many cases do not allow for clear-cut distinctions – age, for example, may be found in both categories since it is one of the most important cluster factors. The distinction between quantitative and qualitative impact is nevertheless useful because of its theoretical implications. Among the factors that will have an impact on amount of attrition (i.e. how strongly is the subject’s L1 affected?) are: – age in relation to plasticity and literacy (children generally have more difficulties and undergo attrition more quickly); – frequency of language use in relation to its consequences for activation thresholds (people who make less use of L1 are supposed to be more vulnerable to attrition); – emotional impact in relation to motivation may strengthen attrition effects. Still, the same factors and some others may also account for qualitative distinctions between different attrition types: – age in relation to type of memory involved (declarative vs. procedural) will have an impact on the parts of the linguistic system that may be involved: in older subjects, attrition will mostly affect those structures which are subserved by declarative memory, in younger subjects attrition will also extend to implicit linguistic competence (i.e. grammar and phonology) subserved by procedural memory;
27
28
Barbara Köpke
– type of language use and its consequences on inhibition patterns will favour attrition patterns in this respect: no or little L1 use in purely monolingual settings will lead to processing difficulties, i.e. performance attrition; L1 use in bilingual settings may bring about less performance attrition but there may be competence attrition involving a restructuring of linguistic competence, otherwise observed only in children; – the cultural context may furthermore strengthen this distinction between competence and performance attrition in relation to the characteristics of the immigrant community (if there is one) and the prevailing attitudes with respect to bilingualism, codeswitching, attrition, the respective language communities etc. Attrition research so far has focussed on the quantitative effects of extralinguistic factors. It appears crucial to realize that different attrition types may arise, depending on whether the immigrant is a child or an adult at the onset of bilingualism and/or attrition, and whether she is part of an immigrant community or not. Arguably the term ‘attrition’ in the sense of an individual phenomenon related to psycholinguistic processes such as language acquisition, learning or pathology is justified only with respect to attrition arising in isolated immigrants (children or adults) – everything happening within the context of a community is likely to be the development of a contact variety which is a sociolinguistic phenomenon related to language change and should be distinguished from individual attrition.
4.2 Cluster factors The second observation is that some factors seem to constitute clusters composed of several factors. Age is one of the best examples of a cluster factor encompassing others. It has been shown that age has consequences on (a) brain plasticity, (b) on the type of memory structures involved in language learning and language processing (declarative or procedural), (c) on the cognitive grounding of language through its close correlation with literacy and education (involving higher metalinguistic knowledge which is of course also linked to b), (d) on motivation, since the relations and expectations between the subject and her environment are also partly related to age, and (e) language use, which also varies as a function of age. The same is likely to apply to a number of other factors related to attrition. This is perhaps the reason why maturational constraints have been shown to play such an important role in attrition (e.g. reviews in Francis 2005; Kaufman 2000). The linguistic and cultural environment can also be considered a cluster factor. Attrition is a response to a change in the linguistic environment of the emigrant. It is the linguistic environment that will provide the opportunities to use
Attrition at the crossroads
a language – whatever the person chooses to do with that opportunity. So the linguistic environment provides a certain amount of contact with the L1, but it also delivers a certain type of input in the language. We have seen that both will have consequences for activation and inhibition mechanisms. Furthermore, the linguistic environment will also transmit language attitudes and cultural values that will influence the attitudes, motivation and general emotional value systems of the immigrant. If there is less evidence for these latter factors in the attrition literature than for maturational aspects, the main reason is that the linguistic and cultural environment, and, a fortiori, emotional aspects are more difficult to compare across studies or in literature reviews than the distinction between children and adults. At this level, measures developed for other fields – e.g. Ethnolinguistic Vitality Theory measures developed for ethnic group comparisons (Giles, Bourhis & Taylor 1977), Social Network Theory (Milroy 1987), or motivational and attitudinal measures largely developed for language learning in classroom settings (Gardner & Lambert 1972) – are usually not suited for the attrition situation and the development of specifically adapted instruments is needed (for a more detailed discussion see Köpke & Schmid 2004; Schmid, this volume). Despite this gap, it has been shown that emotional factors may outweigh even such well-established variables as age in attrition (e.g. Schmid 2002). It can therefore be concluded that, even if some factors form clusters that are closely linked to a great number of others, there is no one individual factor, or cluster, that might be considered dominant. Rather, it appears that attrition should be considered within a multi-componential view, outlined in Figure 1. Each factor, cluster, or type of factor (e.g. biological, cognitive, or external as proposed here by way of example), contributes with its specific weight arising from the individual situation. The sum of the weight of each will determine whether or not there will be attrition and if so to what degree. Note that, following such a conception, levels of proficiency that are similar on the surface level may rely on completely different underlying neuropsychosociological configurations. In some cases (as in Language Proficiency B), biological factors (such as plasticity) may be unfavourable to the maintenance of L1 (in the sense of great plasticity in a young child predicting high attrition), but external factors will be highly favourable to L1 maintenance (e.g. very positive attitudes and stimulating L1 use); in other cases (as in Proficiency Level A) an inverse relationship may produce a similar result. Only when several components are unfavourable will there be a perceptible . Note that I have excluded ‘aptitude’ from this discussion, not so much because it is not important, but because there are no studies at present which investigate this factor in the context of language attrition, and it seems not to be related to other factors discussed here.
29
30
Barbara Köpke
Figure 1. A multicomponential view of language proficiency in the context of attrition
change in overall language proficiency. This might explain why attrition in adults is generally so hard to measure: in many cases, the overall proficiency may remain within a native-like range (see Hyltenstam & Abrahamsson 2003 for a discussion of native-like proficiency; and the suggestions made by de Bot, this volume).
5. Conclusion The preceding discussion on different kinds of internal and external factors related to L1 attrition allows some tentative conclusions and indications for the future direction of research: – Attrition in individual settings is a phenomenon different from the development of a contact variety likely to arise in any setting where an immigrant is exposed to a large variety of L1 input influenced by the contact language (L2). The restructuring of linguistic competence observable in such cases entails a modification of the linguistic norms (cf. Py 1986), but bears little resemblance to a psycholinguistic process. It is furthermore suggested that L1 attrition may be of two types: either affecting online processing and performance (as observed in most studies with adults), or entailing a restructuring of the content of the subject’s grammar (as demonstrated in some studies, mostly involving children). – Some factors involved in L1 attrition are stronger than others due to their multiple links with others, forming ‘cluster factors’; but whatever their strength, there is no one individual factor or cluster able to account for the attrition
Attrition at the crossroads
phenomenon on its own. Attrition needs to be considered within a multicomponent view, relying unavoidably on the complementary contribution of several disciplines of the humanities. In other words, what follows clearly from these considerations is that attrition cannot be conceived as an ‘abnormal condition’, characterised by very specific symptoms and ascribable to one principal organic cause. Rather, attrition could be described as a type of variety within the language proficiency of multilinguals, arising from a multiplicity of causes and, more often than not, remaining within the range of perceived native-like proficiency.
Acknowledgement I am grateful to Monika S. Schmid, Jean-Luc Nespoulous, Angelika Rieussec and two anonymous reviewers for their helpful comments on earlier versions of this paper. I would further like to thank Bert Brouwenstijn for his help with the figure.
References Abrahamsson, N. & Hyltenstam, K. (2005). “Grammaticality judgment and language learning aptitude”. International Symposium on Age and Near-nativeness in L2 Acquisition and L1 Attrition. Stockholm: 2nd December 2005. Altenberg, E. P. & Vago, R. M. (2004). “The role of grammaticality judgments in investigating first language attrition: A crosslinguistic perspective”. In M. S. Schmid, B. Köpke, M. Keijzer & L. Weilemar (Eds.), First Language Attrition: Interdisciplinary perspectives on methodological issues (pp. 105–129). Amsterdam: John Benjamins. Ammerlaan, T. (1996). “You Get a Bit Wobbly... Exploring bilingual lexical retrieval processes in the context of first language attrition”. Unpublished Doctoral Dissertation, Nijmegen: Katholieke Universiteit Nijmegen. Anderson, M. C. (2003). “Rethinking interference theory: Executive control and the mechanisms of forgetting”. Journal of Memory and Language, 49, 415–445. As, A. (1963). “The recovery of a forgotten language knowledge through hypnotic age regression: A case report”. American Journal of Clinical Hypnosis, 5, 24–29. Baddeley, A. (1986). Working Memory. Oxford: OUP. Baddeley, A. (2000). “The episodic buffer: A new component of working memory?” Trends in Cognitive Science, 4 (11), 417–423.
. This conception is largely influenced by Hyltenstam & Abrahamsson’s hypothesis of ‘nonperceivable non-native proficiency levels’ with respect to ultimate attainment in L2 learners (Hyltenstam & Abrahamsson 2003: 572).
31
32
Barbara Köpke
Baddeley, A. (2003). “Working memory: Looking back and looking forward”. Nature Reviews Neuroscience, 4 (10), 829–839. Bahrick, H. P. (1984). “Fifty years of second language attrition: Implications for programmatic research”. The Modern Language Journal, 68 (2), 105–118. Ben-Rafael, M. (2004). “Language contact and attrition: The spoken French of Israeli Francophones”. In M. S. Schmid, B. Köpke, M. Keijzer & L. Weilemar (Eds.), First Language Attrition: Interdisciplinary perspectives on methodological issues (pp. 165–188). Amsterdam: John Benjamins. Berg, T. & Schade, U. (1992). “The role of inhibition in a spreading-activation model of language production. I. The psycholinguistic perspective”. Journal of Psycholinguistic Research, 21 (6), 405–434. Bialystok, E. & Sharwood Smith, M. (1985). “Interlanguage is not a state of mind: An evaluation of the construct for second-language acquisition”. Applied Linguistics, 16, 101–117. Birdsong, D. (1992). “Ultimate attainment in second language acquisition”. Language, 68, 706– 755. Boujon, C. (2002). L’inhibition au carrefour des neurosciences et des sciences de la cognition. Marseille: Solal. Carroll, J. B. (1973). “Implications of aptitude test research and psycholinguistic theory for foreign language teaching”. International Journal of Psycholinguistics, 2, 5–14. Chomsky, N. (1965). Aspects of the Theory of Syntax. Cambridge, MA: MIT Press. Collette, F., van der Linden, M. & Poncelet, M. (2000). “Working memory, long-term memory and language processing: Issues and future directions”. Brain and Language, 71, 46–51. Cook, V. (1992). “Evidence for multicompetence”. Language Learning, 42, 557–591. Crosson, B., Novack, T. A. & Trenerry, M. R. (1988). “Subcortical language mechanisms: Window on a new frontier”. In H. A. Whitaker (Ed.), Phonological Processes and Brain Mechanisms (pp. 24–58). New York, NY: Springer Verlag. Damasio, A. R. (2003). Spinoza avait raison. Joie et tristesse, le cerveau des émotions. Paris: Odile Jacob. Daneman, M. & Carpenter, P. A. (1980). “Individual differences in working memory and reading”. Journal of Verbal Learning and Verbal Behavior, 19, 450–466. De Bode, S. (1996). First Language Attrition: Productive morphology disintegration and neurobiological support. A case study. Unpublished MA Thesis, Pomona, CA: California State University. de Bot, K. (2001). “Language use as an interface between sociolinguistic and psycholinguistic processes in language attrition and language shift”. In J. Klatter-Folmer & P. van Avermaet (Eds.), Theories on Maintenance and Loss of Minority Languages: Towards a more integrated explanatory framework (pp. 65–82). Münster: Waxmann. de Bot, K. & Stoessel, S. (2000). “In search of yesterday’s words: Reactivating a forgotten language”. Applied Linguistics, 21 (3), 364–384. de Bot, K. & Weltens, B. (1991). “Regression, recapitulation, and language loss”. In H. W. Seliger, & R. M. Vago (Eds.), First Language Attrition (pp. 31–52). Cambridge: CUP. Dell, G. S. (1986). “A spreading-activation theory of retrieval in sentence production”. Psychological Review, 93, 283–321. Dressler, U. (1991). “The sociolinguistic and patholinguistic attrition of Breton phonology, morphology, and morphophonology”. In H. W. Seliger & R. M. Vago (Eds.), First Language Attrition (pp. 99–112). Cambridge: CUP.
Attrition at the crossroads
Dussias, P. (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. Ecke, P. (2004). “Language attrition and theories of forgetting: A cross-disciplinary review”. International Journal of Bilingualism, 8 (3), 321–354. Ellis, N. (1994). “Implicit and explicit language learning: An overview”. In N. Ellis (Ed.), Implicit and Explicit Learning of Languages (pp. 1–31). New York, NY: Academic Press. Fabbro, F. (1999). The Neurolinguistics of Bilingualism: An introduction. Hove: Psychology Press. Flege, J. E., Yeni-Komshian, G. & Liu, S. (1999). “Age constraints on second-language acquisition”. Journal of Memory and Language, 41, 78–104. Fodor, J. & Garrett, M. (1966). “Some reflections on competence and performance”. In J. Lyons & R. J. Wales (Eds.), The Edinburgh Conference on Psycholinguistics (pp. 135–182). Edinburgh: Edinburgh University Press. Francis, N. (2005). “Research findings on early first language attrition: Implications for the discussion on critical periods in language acquisition”. Language Learning, 55 (3), 491–531. Fromm, E. (1970). “Age regression with unexpected reappearance of a repressed childhood language”. International Journal of Clinical and Experimental Hypnosis, 18, 79–88. Gardner, R. C. & Lambert, W. E. (1972). Attitudes and Motivations in Second Language Learning. Rowley, MA: Newbury House. Giles, H., Bourhis, R. Y. & Taylor, D. M. (1977). “Towards a theory of language in ethnic group relations”. In H. Giles (Ed.), Language, Ethnicity and Intergroup Relations (pp. 307–348). New York, NY: Academic Press. Grainger, J. & Dijkstra, T. (1992). “On the representation and use of language information in bilinguals”. In R. J. Harris (Ed.), Cognitive Processing in Bilinguals (pp. 207–220). Amsterdam: North Holland. Green, D. W. (1986). “Control, activation, and resource: A framework and a model for the control of speech in bilinguals”. Brain and Language, 27, 210–223. Green, D. W. (1993). “Towards a model of L2 comprehension and production”. In R. Schreuder & B. Weltens (Eds.), The Bilingual Lexicon (pp. 249–278). Amsterdam: John Benjamins. Green, D. W. (1998a). “Mental control of the bilingual lexico-semantic system”. Bilingualism: Language and Cognition, 1, 77–82. Green, D. W. (1998b). “Schemas, tags and inhibition”. Bilingualism: Language and Cognition, 1, 100–104. Grosjean, F. (1988). “Exploring the recognition of guest words in bilingual speech”. Language and Cognitive Processes, 3, 233–274. Grosjean, F. (1992). “Another view of bilingualism”. In R. J. Harris (Ed.), Cognitive Processing in Bilinguals (pp. 51–62). Amsterdam: North Holland. Grosjean, F. & Py, B. (1991). “La restructuration d’une première langue: l’intégration de variantes de contact dans la compétence de migrants bilingues”. La Linguistique, 27, 35–60. Gross, S. (2004). “A modest proposal: Explaining language attrition in the context of contact linguistics”. In M. S. Schmid, B. Köpke, M. Keijzer & L. Weilemar (Eds.), First Language Attrition: Interdisciplinary perspectives on methodological issues (pp. 281–298). Amsterdam: John Benjamins. Gürel, A. (2002). Linguistic Characteristics of Second Language Acquisition and First Language Attrition: Turkish overt versus null pronouns. Unpublished Doctoral Dissertation, Montreal: McGill University.
33
34
Barbara Köpke
Gürel, A. (2004). “Selectivity in L2-induced L1 attrition: A psycholinguistic account”. Journal of Neurolinguistics, 17 (1), 53–78. Håkansson, G. (1995). “Syntax and morphology in language attrition: A study of five bilingual expatriate Swedes”. International Journal of Applied Linguistics, 5 (2), 153–171. Hulsen, M. (2000). Language Loss and Language Processing: Three generations of Dutch migrants in New Zealand. Unpublished Doctoral Dissertation, Nijmegen: Katholieke Universiteit Nijmegen. Hyltenstam, K. & Abrahamsson, N. (2003). “Maturational constraints in SLA”. In C. Doughty & M. H. Long (Eds.), Handbook of Second Language Acquisition (pp. 539–588). Oxford: Blackwell. Ijalba, E., Obler, L. K. & Chengappa, S. (2004). “Bilingual aphasia”. In T. K. Bhatia & W. C. Ritchie (Eds.), The Handbook of Bilingualism (pp. 71–89). Malden, MA: Blackwell. Isurin, L. (2000). “Deserted islands or a child’s first language forgetting”. Bilingualism: Language and Cognition, 3, 151–166. Jakobson, R. (1941). Kindersprache, Aphasie und allgemeine Lautgesetze. Uppsala: Almqvist and Wiksell. Johnson, J. S. & Newport, E. L. (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. Just, M. A. & Carpenter, P. N. (1992). “A capacity theory of comprehension: Individual differences in working memory”. Psychological Review, 99, 122–149. Kaufman, D. (2000). “Attrition of Hebrew in the United States: Sociolinguistic perspectives”. In E. Olshtain (Ed.), Immigration, Identity, and Language (pp. 409–420). Jerusalem: Magnes Publications Hebrew University. Kaufman, D. & Aronoff, M. (1991). “Morphological disintegration and reconstruction in first language attrition”. In H.W. Seliger & R. M. Vago (Eds.), First Language Attrition (pp. 175– 188). Cambridge: CUP. Keijzer, M. C. J. (2004). “First language attrition: A cross-linguistic investigation of Jakobson’s regression hypothesis”. International Journal of Bilingualism, 8 (3), 389–393. Köpke, B. (1999). L’ attrition de la première langue chez le bilingue tardif: Implications pour l’étude psycholinguistique du bilinguisme. Unpublished Doctoral Dissertation, Toulouse: Université du Mirail. Köpke, B. (2001). “Quels changements linguistiques dans l’attrition de la L1 chez le bilingue tardif?” TRANEL, 34/35, 355–368. Köpke, B. (2002). “Activation thresholds and non-pathological first language attrition”. In F. Fabbro (Ed.), Advances in the Neurolinguistics of Bilingualism (pp. 119–142). Udine: Forum. Köpke, B. (2004a). “Neurolinguistic aspects of attrition”. Journal of Neurolinguistics, 17 (1), 3–30. Köpke, B. (2004b). “Attrition is not a unitary phenomenon: On different possible outcomes of language contact situations”. In A. M. L. Suarez, F. Ramallo & X.-P. Rodiguez-Yanez (Eds.), Bilingual Socialization and Bilingual Language Acquisition: Proceedings from the Second International Symposium on Bilingualism (pp. 1331–1347). Vigo: Servizo de Publicacions da Universidade de Vigo. (http://webs.uvigo.es/ssl/actas2002/). Köpke, B. & Schmid, M. S. (2004) “First language attrition: The next phase.” In M. S. Schmid, B. Köpke, M. Keijzer & L. Weilemar (Eds.), First Language Attrition: Interdisciplinary perspectives on methodological issues (pp. 1–43). Amsterdam: John Benjamins.
Attrition at the crossroads
Krashen, S. (1981). Second Language Acquisition and Second Language Learning. Oxford: Pergamon Press. Lamendella, J. (1977). “General principles of neurofunctional organization and their manifestation in primary and secondary language acquisition”. Language Learning, 27, 155–196. Lenneberg, E. H. (1967). Biological Foundations of Language. New York, NY: Wiley. Levine, G. (1996). “Elderly second-generation speakers of Yiddish: Toward a model of L1 loss, incomplete L1 acquisition, competence, and control”. Southwestern Journal of Linguistics, 15 (1), 109–120. Luria, A. R. (1966). The Higher Cortical Functions in Man. New York, NY: Basic Books. Luria, A. R. (1974). “Basic problems in neurolinguistics”. In T. A. Sebeok (Ed.), Current Trends in Linguistics (pp. 2561–2593). The Hague: Mouton. MacLeod, C. M. (1976). “Bilingual episodic memory: Acquisition and forgetting”. Journal of Verbal Learning and Verbal Behavior, 15, 347–364. Major, R. C. (1992). “Losing English as a first language”. The Modern Language Journal, 76 (2), 190–208. McCarthy, R. A. & Warrington, E. K. (1994). Neuropsychologie cognitive. Une introduction clinique. Paris: Presses Universitaires de France. McClelland, J. L. & Rumelhart, D. E. (1981). “An interactive activation model of context effects in letter perception: Part 1. An account of the basic findings”. Psychological Review, 88, 375–407. McClelland, J. L. & Elman, J. L. (1986). “The TRACE model of speech perception”. Cognitive Psychology, 18, 1–86. Meara, P., Milton, J. & Lorenzo-Dus, N. Swansea Language Aptitude Test (LAT) v2.0 (Lognostics, Swansea, 2003). Meuter, R. & Humphreys, G. (1997). “The frontal lobe and bilingual switching: Asymmetrical costs of language selection”. International Workshop on the Psycholinguistics of Bilingualism, Université de Provence: 20–21 May 1997. Milroy, L. (1987). Language and Social Networks. Oxford: Blackwell. Montrul, S. (2004). “Convergent outcomes in L2 acquisition and L1 loss”. In M. S. Schmid, B. Köpke, M. Keizer & L. Weilemar (Eds.), First Language Attrition: Interdisciplinary perspectives on methodological issues (pp. 259–280). Amsterdam: John Benjamins. Nespoulous, J-L. (1994). “Le langage: Introduction. Linguistique, neurolinguistique et neuropsycholinguistique”. In M. Jeannerod & X. Seron (Eds.), Neuropsychologie Humaine (pp. 317–319). Bruxelles: Mardaga. Nicoladis, E. & Grabois, H. (2002). “Learning English and losing Chinese: A study of a child adopted from China”. International Journal of Bilingualism, 6 (4), 441–454. Norman, D. & Shallice, T. (1986). “Attention to action: Willed and automatic control of behavior”. In R. J. Davidson, G. E. Schartz & D. Shapiro (Eds.), Consciousness and Self-regulation vol. 4 (pp. 1–18). New York, NY: Plenum. Olshtain, E. (1986). “The attrition of English as a second language with speakers of Hebrew”. In B. Weltens, K. de Bot & T. van Els (Eds.), Language Attrition in Progress (pp. 185–204). Dordrecht: Foris. Olshtain, E. & Barzilay, M. (1991). “Lexical retrieval difficulties in adult language attrition”. In H. W. Seliger & R. M. Vago (Eds.), First Language Attrition (pp. 139–150). Cambridge: CUP. Opitz, C. (2005). “Trade-offs between language proficiencies as a cause of L2 attrition”. Second International Conference on First Language Attrition. Amsterdam: 17–20 August 2005.
35
36
Barbara Köpke
Pallier, C., Dehaene, S., Poline, J.-B., LeBihan, D., Argenti, A.-M., Dupoux, E. & Mehler, J. (2003). “Brain imaging of language plasticity in adopted adults: Can a second language replace the first?” Cerebral Cortex, 13, 155–161. Paradis, M. (1985) “On the representation of two languages in the brain”. Language Sciences, 7 (1), 1–39. Paradis, M. (1993) “Linguistic, psycholinguistic, and neurolinguistic aspects of “interference” in bilingual speakers: The Activation Threshold Hypothesis”. International Journal of Psycholinguistics, 9 (2), 133–145. Paradis, M. (1994). “Neurolinguistic aspects of implicit and explicit memory: Implications for bilingualism and SLA”. In N. Ellis (Ed.), Implicit and Explicit Learning of Languages (pp. 393–419). New York, NY: Academic Press. Paradis, M. (1997). “The cognitive neuropsychology of bilingualism”. In A. M. B de Groot & J. Kroll (Eds.), Tutorials in Bilingualism: Psycholinguistic perspectives (pp. 331–354). Hillsdale, NJ: Lawrence Erlbaum Associates. Paradis, M. (2004). A Neurolinguistic Theory of Bilingualism. Amsterdam: John Benjamins. Pavlenko, A. (2003). “‘I feel clumsy speaking Russian’: L2 influence on L1 narratives of Russian L2 users of English”. In V. Cook (Ed.), Effects of the Second Language on the First (pp. 32–61). Clevedon: Multilingual Matters. Pavlenko, A. (2005). Emotions and Multilingualism. Cambridge: CUP. Penfield, W. & Roberts, L. (1959). Speech and Brain Mechanisms. New York, NY: Atheneum. Py, B. (1986). “Native language attrition amongst migrant workers: Towards an extension of the concept of interlanguage”. In E. Kellermann & M. Sharwood Smith (Eds.), Crosslinguistic Influence in Second Language Acquisition (pp. 163–172). Oxford: Pergamon Press. Rieussec, A. & Köpke, B. (2005). “Conceptual bridges between attrition and L2 learning”. Second International Conference on First Language Attrition. Amsterdam: 17–20 August 2005. Rieussec, A. & Köpke, B. (2007). “Cognitive flexibility in lexical retrieval: Comparing L2 acquisition and L1 attrition in German”. Fourth International Symposium on Bilingualism. Hamburg: 30 May – 2 June 2007. Schmid, M. S. (2002). First Language Attrition, Use, and Maintenance: The case of German Jews in Anglophone countries. Amsterdam: John Benjamins. Schmid, M. S. (2004). “Le rôle du système linguistique: Attrition vs. contact de langues et changement linguistique”. Journée ‘Quand la sociolinguistique rencontre la neurolinguistique: l’exemple de l’attrition’. Toulouse: Université Toulouse-Le Mirail, 28 Mai 2004. Schmitt, E. (2004). “No more reductions! To the problem of evaluation of language attrition data”. In M. S. Schmid, B. Köpke, M. Keijzer & L. Weilemar (Eds.), First Language Attrition: Interdisciplinary perspectives on methodological issues (pp. 299–316). Amsterdam: John Benjamins. Schumann, J. H. (1997). The Neurobiology of Affect in Language. Malden, MA: Blackwell. Seliger, H. W. (1991). “Language attrition, reduced redundancy, and creativity”. In H. W. Seliger & R. M. Vago (Eds.), First Language Attrition (pp. 227–240). Cambridge: CUP. Sharwood Smith, M. (1983). “On explaining language loss”. In S. Felix & H. Wode (Eds.), Language Development at the Crossroads (pp. 49–69). Tübingen: Gunter Narr. Sharwood Smith, M. & van Buren, P. (1991). “First language attrition and the parameter setting model”. In H. W. Seliger & R. M. Vago (Eds.), First Language Attrition (pp. 17–30). Cambridge: CUP. Silva-Corvalan, C. (1991). “Spanish language attrition in a contact situation with English”. In H. W. Seliger & R. M. Vago (Eds.), First Language Attrition (pp. 151–174). Cambridge: CUP.
Attrition at the crossroads
Speedie, L. J., Wertman, E., Ta’ir, J. & Heilman, K. M. (1993). “Disruption of automatic speech following a right basal ganglia lesion”. Neurology, 43, 1768–1774. Sperber, D. & Hirschfeld, L. (2007). “Culture and modularity”. In P. Carruthers, S. Laurence & S. Stich (Eds.), The Innate Mind: Culture and cognition. Oxford: OUP. Swain, M. (2000). “The output hypothesis and beyond: Mediating acquisition through collaborative dialogue”. In J. P. Lantolf (Ed.), Sociocultural Theory and Second Language Learning (pp. 97–114). Oxford: OUP. Turian, D. & Altenberg, E. P. (1991). “Compensatory strategies of child first language attrition”. In H. W. Seliger & R. M. Vago (Eds.), First Language Attrition (pp. 207–226). Cambridge: CUP. Ullman, M. T. (2001). “The neural basis of lexicon and grammar in first and second languages: The declarative/procedural model”. Bilingualism: Language and Cognition, 4 (1), 105–122. van der Linden, M. & Collette, F. (2002). “Attention et mémoire de travail”. In J. Couillet, M. Leclercq, Ch. Moroni, & Ph. Azouvi (Eds.), La neuropsychologie de l’attention (pp. 41–54). Marseille: Solal. Vago, R. M. (1991). “Paradigmatic regularity in first language attrition”. In H. W. Seliger & R. M. Vago (Eds.), First Language Attrition (pp. 241–252). Cambridge: CUP. Ventureyra, V. (2005). A la recherche de la langue perdue: Étude psycholinguistique de l’attrition de la première langue chez des Coréens adoptés en France. Unpublished Doctoral Dissertation, Paris: EHESS. Ventureyra, V., Pallier, C. & Yoo, H.-Y. (2004). “The loss of first language phonetic perception in adopted Koreans”. Journal of Neurolinguistics, 17 (1), 79–91. Yağmur, K. (1997). First Language Attrition among Turkish Speakers in Sydney. Tilburg: Tilburg University Press. Zobl, H. (1995). “Converging evidence for the ‘acquisition-learning’ distinction”. Applied Linguistics, 16 (1), 35–56.
37
Understanding attrition within a MOGUL framework Mike Sharwood Smith
Heriot-Watt University, Edinburgh, Scotland
Linguistically-based accounts of attrition may give us an analysis of the properties of language observed at particular points in time. Then, by comparing states, we may try to explain the transition between them but, still, discussion concerning the actual mechanisms of change is typically left aside. The same may be said of studies of language acquisition. To integrate accounts of linguistic states at different points in time with psycholinguistic explanations about the transition from one state to the next, we require a broader, interdisciplinary approach. Sharwood Smith and Truscott’s MOGUL (Modular Growth and Use of Language) is one such framework, adapting Ray Jackendoff ’s model of the language faculty to show how real-time processes might drive representational change over time. Keywords: modularity; interdisciplinary approach; Jackendoff ’s model of the language faculty
1. Introduction There are a few key issues that persist in the research literature that looks at how languages develop in the individual over time. By ‘develop’ I mean here change in any direction, i.e. in a manner to be interpreted as either acquisition or as attrition. One issue is how to characterise the precise nature of the relationship between these two processes. Another is the problem of reconciling two fundamental dimensions of change, ones that are normally kept strictly apart but which really ought to be more closely and clearly related. I refer here to the dimension which involves abstract knowledge states on the one hand and, on the other, the processing dimension where the focus is on the workings of cognitive, psychological mechanisms operating in real time. We are able to talk in fairly vague terms about how we think the socio-, psycho- and purely linguistic aspects of attrition
40 Mike Sharwood Smith
are all related but to do this in very precise and interesting ways requires an overarching theoretical framework and an associated metalanguage that can make this achievable. An attempt to provide such a framework will be described below.
1.1 Working with different domains of inquiry There are of course, many ways to approach the study of attrition and in every case one or more research fields seem to be involved. Researchers into language development may generally find themselves working from one of two platforms. The first would involve a primarily linguistic focus, studying the linguistic properties of learner/attriter performance. The second would be within a primarily psychological (or ‘psycholinguistic’) framework, where the real-time processing dimension is investigated, and where researchers typically work with lexical decision tasks, reaction times, and the like. The usefulness of working in a field with a rich and long established tradition and with a variety of analytic tools at the researcher’s disposal would seem to make either option attractive. A researcher can also explore attrition (and acquisition) using the precision instruments of one or more research domains. Whether each domain provides sufficiently precise and appropriate instruments for analysing and interpreting the data, and whether the different analyses can be somehow meshed together to provide a more comprehensive picture of development remain open questions: researchers naturally make do with whatever is available.
1.2 Real time: Synchronic versus transitional accounts Within psycholinguistics itself, there is also a problem relating, in particular, to the typical time periods that psycholinguistic research is supposed to involve. To be more specific, working in a language or cognitive processing framework does not typically mean trying to account for the longer term effects of processing. In fact, psycholinguistics, for most people, has exclusively or mostly had to do with the millisecond-by-millisecond processing of current resources, i.e., at a particular stage (final or intermediate or ‘post-final’) of a language user’s career, that is, without reference to what happens later or what happened before. For example, it is interesting to study how, say, an attriting individual’s more recent L2 knowledge currently interacts with their L1 knowledge and to do this according to mode and situational context. Data obtained can cast light on the current state of attrition and on what part L2 linguistic knowledge plays in differentiating the individual’s
Understanding attrition within a MOGUL framework
L1 behaviour from that of standard native speakers of the language. Nevertheless, how acquisition or attrition actually take place over longer periods of time should also be an eminently ‘psycho-’ linguistic, and ‘processing’ concern because a comprehensive theory of language attrition needs to explain exactly how change comes about, how the state in which individuals find themselves at Time Ti actually develops, over periods of time to a different state at Time Ti+1. This issue relates to the discussion of property theories versus transition theories: in the first case we are dealing with the properties of a learner’s (and here, by extension an attriter’s) current version of the language in question. In the second case, we are dealing with the processes that transform one state into another (Cummins 1983; Gregg 1996). The real-time issue discussed above has therefore added a further important distinction, namely that which exists between synchronic processes and transitional processes. The type of explanation to be offered below will make an explicit connection between these two.
1.3 Overarching frameworks One obvious way to deal with theoretical areas that remain stubbornly opaque is to simply continue doing research within a tried and trusted environment and leave some knotty problems (e.g. the above-mentioned transitional issue and how to define ‘native speaker’) for ‘later’. This would seem a safe strategy but ultimately a counterproductive one since there may well be many local problems that can be resolved in a wider context. The potential of a broader perspective seems to be acknowledged in the sense that there is nowadays a stronger interdisciplinary flavour to acquisition and attrition debates. Another option is to adopt a view of development that simply removes awkward distinctions by dismissing them on theoretical grounds. A prime example would be a theory that would make no distinction between ‘knowledge’ and ‘processing’ in the tradition of behaviourism or its latter day descendant, classical connectionism: no separate symbolic, representational system, no mental grammar or mental lexicon would be part of such a perspective. Language change in the individual would be, in some relatively straightforward way, the creation, strengthening, weakening and disappearance of networks of weighted connections in response to stimuli from the environment. Language systems would emerge and develop into new variants in this way and speaking in terms of discrete subsystems or ‘modules’ like phonology and syntax and refer. See Escudero & Sharwood Smith (2001) for a relevant discussion of how the term ‘native speaker’ may be interpreted.
41
42
Mike Sharwood Smith
ring to mental rules and mental representations would simply be a convenient way for us to talk about the outcome, that is, in ways that describe the result but have little or no role in explaining how the result came about. Perhaps we could then rely on neuroscientists to use their own research into neural structure to inform us about how the creation of new networks takes place over time and import their accounts into our own theoretical explanations. The problem is that, although this reductionist approach to knowledge change constitutes a perfectly valid target for researchers to pursue, we have insufficient evidence so far that this is a reliable way of looking at development and, in addition, many reasons for retaining some form of the knowledge/processing distinction as a way of explaining both the nature of language and the way it changes in the individual (Pinker & Prince 1988; see also Marcus 2001 for ways of reconciling the two approaches). I shall now avoid any further discussion of the merits and demerits of connectionist and/or emergentist approaches, and will proceed to apply the perspective called ‘MOGUL’ (Modular On-line Growth and Use of Language) to the issue of language attrition: this will entail a solution that does, at least in some sense, merge the two dimensions of representation and processing but, as will be made clear below, ‘merging’ does not imply eliminating the distinction but rather reconceptualising the distinction between knowledge and processing in terms that allow a processing perspective to dominate the explanation of development (Truscott & Sharwood Smith 2004a; 2004b). In this way, it is hoped that we can interpret existing attrition data in more meaningful ways as well as develop new lines of empirical investigation. In particular, we can produce a ‘transitional’ theory of attrition that accords with and builds on research claims in different domains (see related discussion in Köpke & Schmid 2004).
2. MOGUL 2.1 The theoretical setting A brief exposé of what the MOGUL framework involves might well introduce a variety of familiar terms like ‘modularity’, ‘connections’, ‘networks’, ‘activation levels’ and ‘universal grammar’, concepts which, taken together, might not all seem to be appropriate bedfellows. However, the whole point about casting research questions within a MOGUL perspective is that it makes these various concepts meaningful within single theoretical accounts of attrition and acquisition. In actual fact, although some researchers working within a cognitivist framework, broadly defined, in which notions such as ‘knowledge’ states are standard, might
Understanding attrition within a MOGUL framework
feel uncomfortable talking about activation and the growth of ‘connections’, it is actually the classical connectionists and people adhering to more or less behaviourist positions who will swiftly disassociate themselves from a MOGUL point of view. MOGUL frames questions of knowledge change in processing terms. However, it does so in ways which are partly or fully compatible with various non-connectionist approaches, that is, with approaches that range from ‘general cognitivism’, involving concepts like, for instance, ‘declarative’ and ‘procedural’ knowledge to domain-specific perspectives on language such as generative linguistics within Chomsky’s Minimalist Programme (Chomsky 1995). In actual fact, the ‘modular’ part of MOGUL places it in the latter, domain-specific part of this spectrum although concepts from the approaches to language cognition that deny the existence of a language faculty, and research associated with those concepts, may also be usefully re-interpreted within MOGUL. One of the aims underlying MOGUL is precisely to seek links that knit together strands of research across theoretical and disciplinary boundaries in a theoretically cohesive manner. It is important therefore to understand that many existing research claims and research findings can be re-examined within a MOGUL perspective. There is already some indication that this process can shed new light on old data (see, for example, the reanalysis of the role of metalinguistic ability in Sharwood Smith 2004, the reinterpretation of the continuum-versusstage option when analysing development in Sharwood Smith & Truscott 2005, and the reinterpretation of the Schwartz & Sprouse ‘Full-Transfer-Full-Access’ position, in second language acquisition, in Sharwood Smith & Truscott 2006; see Schwartz & Sprouse 1996).
2.2 The main tenets 2.2.1 Acquisition and attrition as an ongoing process The core of MOGUL as it now stands is ‘A’-by-Processing-Theory (APT; Truscott & Sharwood 2004b) where ‘A’ may stand for both attrition and acquisition. The main idea is that development (in either direction) is a by-product of the languageuser’s actual use of the language system in question and no ‘LAD’, i.e., no separate, dedicated acquisition or attrition mechanisms are needed to explain it. From this one may infer that there is a point at which a language system becomes stable but there is no literal ‘final’ state of language acquisition, where it is fixed in the learner’s mind: you use a language or you lose it. The point of so-called ultimate attainment in the acquisition of a first language (L1) is the point where all the evidence couched in the ambient L1, the language as spoken and written in the acquirer’s
43
44 Mike Sharwood Smith
environment has been so fully and frequently processed as to suggest the language user currently possesses a near identical system to those around him or her. The ‘system’ involves both the relevant linguistic properties and their high degree of accessibility in performance. The language user appears to more or less match those around him or her and therefore can be regarded as a full member of that language community. The reason that the word ‘ultimate’ is chosen is that, for a great number of speakers, this system will remain in this state more or less until they die. The reason for this, which is given a specific interpretation in the MOGUL framework, is simply that they will go on using it. If there are any changes in the structural properties that their performance manifests then we may infer either that the same speakers around them have themselves started to change or that the language user has moved on to another community where that same language is spoken much less frequently or not at all. The presence of a different ambient language normally accompanies and helps to explain the resulting changes. MOGUL puts content into the notion that many have expressed about the close relationship between attrition and acquisition despite the fact that traditionally acquisition researchers work with notions such as ‘evidence’ and ‘input’ and ‘setting UG parameters’ which would seem to pose a challenge in language ‘loss’. To what extent does it make sense to talk of an attriter ‘resetting’ L1 parameters (however one chooses to understand these notions) where ultimate attainment has by definition taken place? And is it possible to talk of attrition taking place in response to ‘evidence’ (see discussion in Sharwood Smith & van Buren 1991)? The architecture currently proposed for MOGUL as far as the core language system is concerned is based on Ray Jackendoff ’s proposals (Jackendoff 1987, 2002). The great strength of the Jackendoff approach is that it aligns explanations of linguistic form and linguistic processing such that one can easily be translated into the other. In MOGUL we take it a step further and elaborate its implications for a transitional account (see Carroll 2001 for a different elaboration of Jackenoff ’s model). It is important to understand that, as Carroll has pointed out in convincing terms (Carroll 1999), environmental stimuli or the lack of them cannot influence the course of development in a straightforward manner. Processing is understood to go through different stages and what may be termed input is different at each stage. For example, we have, as many assume, a special system for processing sound (or rather acoustic/auditory input), perceived sound from the environment that is specifically relevant to language (Jusczyk 1997). This is a system whose operation is evident from an early age and which is progressively tuned to cope with specific types of linguistic sound actually encountered by the individual. We are driven to interpret linguistically relevant auditory/acoustic information because this is all
Understanding attrition within a MOGUL framework
Figure 1. Phonological input processing
part of the more general drive to make sense of sound, i.e. recreate messages that sound may contain. The first step in building an interpretation is, then, phonological. Phonological processing will serve here as an illustration of how MOGUL interprets development in general. The other, syntactic and conceptual-semantic, stages may be understood, for present purposes, to operate in parallel and in essentially identical ways, each system possessing its own special processor and its own memory, and in addition separate interface systems for chaining structures in one system together with structures created in another (Jackendoff 2002: 125, 199). During performance, a number of candidate chains across the various working memories compete until a ‘best-fit’ chain is selected. To return to our example of the phonological system, over time, this process of continually trying to make sense of language sound results in the language user/acquirer building up a bank of phonological structures in long term memory. This process is a collaboration between (a) the phonological processor that runs according to principles that phonologists attempt to explicate, (b) input to the phonological system, and (c) the phonological structures already created and available in phonological memory. The resulting structures serve to support current and future attempts to interpret and produce messages in speech (see Fig. 1). Our actual performance (in comprehension and production) will be affected by the degree to which we can recruit these phonological structures in (long term) memory on line. Here the notion of ‘activation level’ is crucial since some structure will be more accessible than others. Reformulating this in MOGUL terms, . ‘Long term’ in this perspective is a relative term and should be taken here as simply meaning ‘currently resident’ memory. Working memory involves elements in long term memory that are, for a brief moment, undergoing various types of activation. In this way, one can therefore
45
46 Mike Sharwood Smith
some weakly established structures will have a low ‘resting level’ (also known as a high threshold: Paradis 2000, 2001, this volume) making them weakly accessible. The more they are recruited the higher the resting level becomes (or, in Paradis’ terms, the lower the threshold). From this we may infer that highly accessible, strongly established structures will lose their position of strength if they are not used over a period of time, hence the ‘use it or lose it’ dictum. Not only do the resting levels of phonological (or syntactic or conceptual) structures in memory need boosting to make them more readily available but they also need boosting to prevent them sinking downwards again. Note that phonological structures (or, for that matter morphosyntactic ones) are not viewed here ‘simply’ as networks of connections which have been created in response to language experience. There is no direct relationship between exposure to sound and the strength of internal connections. Both representational properties and frequency of use have a part to play (see Fig. 2). The networks do respond to repeated activation but, to the degree that they are modular, each module operating according to its own principles, the response to environmental stimuli is very indirect. The actual properties of phonological structures are determined and constrained by the architecture of (the phonological part of) the language faculty. The constellations of properties which may form a particular linguistic structure can only be of a certain type and these constraints, we assume, are part of what is generally known as universal grammar. Linguistic, and in this case phonological research provides us with insights into what these constraints are. Put crudely, input that is the wrong size, does not run on four wheels of a special type and goes too fast, cannot enter the railway station. It is only what actually gets into the station, i.e., what has a permissible shape, that counts when we talk about the effects of frequency. This is what Jackendoff means when he describes modules as being ‘structurally constrained’. In MOGUL then, phonological structures are assembled from phonologically permissible elements; the more they are formed and reassembled again and again in real time the stronger they become. In this way and within these constraints, the strength, and hence availability of linguistic structures for future production and also interpretation of messages is determined by the frequency and intensity of use. How this differs from classic connectionism is crucial. It is not the frequency of the external environmental stimuli that dictates the strength of connections in a network or, put another way, the strength or weakness of activation levels. There could be highly frequent (linguistic) events in the language user’s enregard working memory as not being separate from long term memory (for discussion of this issue see Miyake & Shah 1999).
Understanding attrition within a MOGUL framework
Figure 2. Two basic factors affecting language development
vironment that make no impact at all because the language user’s current system simply makes no sense of them, i.e., for one reason or another, cannot (or cannot yet) create permissible linguistic representations out of them. This principle is reflected in the concept of learner readiness, i.e. that only once certain pieces of the language system have been formed, can a given structure be processed and acquired even though the evidence for that structure has been loudly knocking at the door, i.e. has, for some time, been highly frequent in the language that the learner has been exposed to. To recapitulate, in acquisition, a structure is created by matching input from an adjacent system to (some combination of) available structures in the given module (e.g. auditory-acoustic structure linked up to structures within the phonological system) and this newly created structural association is thereafter established, via repeated use, more and more strongly over a period of time in long term memory. The same process in reverse counts as attrition. Whether or not something may be subject to total extinction, a long period of disuse will effectively render it less well established and therefore less accessible. Less accessible structures may lose out in the competition to select the best candidate structure in an effort to produce or interpret utterances. Very inaccessible structures are either ones that have only recently been created in memory or are ones that have been in constant use at some point but have now been left unused for a long time. In both cases, they may never appear in actual visible and audible performance and their continuing presence in memory may only be established via indirect means, e.g. priming experiments. Note that in constructing a representation in working memory for unfamiliar input, the learners may be in possession of the newly created structure for a brief moment. If one allows for the possibility of almost immediate extinction, then acquisition and attrition at a micro level may be taking place all the time. The only reason
47
48 Mike Sharwood Smith
why this might not be so, with temporary structures that are never recreated (because the learner analyses the input in a different way on the next encounter), is if we were to view memory as being, at some minimal level, ‘photographic’ (eidetic). This would mean, perhaps implausibly, that nothing experienced is ever entirely forgotten but still may remain at such a low resting level of activation that it is never used, and simply appears to be forgotten. From the above, it can be seen that MOGUL accords well with and can draw support from a range of research findings in the bilingual processing literature. Compatibility with Paradis’ account has already been mentioned; the same goes, for example, for claims relating to the parallel activation of all available language systems in the individual as explained in the Bilingual Interactive Activation Model (BIA) of Dijkstra and van Heuven as also Grosjean’s language mode, etc. (Dijkstra & van Heuven 1998; Grosjean 1998, 2001). Note that ‘drawing support from’ does not necessarily mean espousing the complete theory underlying the basic findings, however. This would make MOGUL an Incoherent Theory of Everything. Take the case of the BIA for example. This is a model designed to explain the operations of the bilingual mental lexicon and research has supported the basic idea that access is non-selective – both or all languages are always activated during language use whether the language user is operating in L1, L2, or L3 (Dijkstra & van Heuven 1998, 2002). The claim is that there is an integrated lexicon for words of different languages. All of this fits very well into MOGUL, except some reinterpretation has to take place. For a start, there is no single lexicon in the traditional sense but rather multiple lexicons: the elements that make up the traditional lexicon in this model are separately housed in respectively the phonological, syntactic and conceptual memories (see Figs. 1 and 2). This shows both how MOGUL modularises the language system (following Jackendoff) and also how it is a representational and not purely connectionist account because the nodes that are linked together to form a lexical chain have symbolic content: they are phonological, syntactic and conceptual structures. However, none of this is an embarrassment to the basic insights and findings that BIA (and BIA+) has yielded and it generally fits in well with the MOGUL view of attrition (cf. Jacquet & French 2002 for an example of a connectionist interpretation of acquisition based on the BIA+). What MOGUL does is to set such mental lexicon research within a wider context enabling us to draw conclusions for the whole language system. In MOGUL, the ‘lexicons’, taken together, contain the crucial syntactic, phonological and conceptual properties that drive the language system as a whole in specific ways. It also elaborates the implications for language growth, attrition and acquisition. The idea of an integrated lexicon and Grosjean’s view of independent but interconnected language networks are not really different points of view, and
Understanding attrition within a MOGUL framework
MOGUL also reflects Grosjean’s idea of a language mode. In line with what was said above, the total deactivation of one language probably never takes place: it is more a relative concept of more or less activation (or inhibition; Green 1986) according to given conditions. What a Jackendoff-based architecture provides is a way of conceptualising how the language operates. Note that the ‘core’ language faculty in this system consists of the syntactic and phonological systems along with the processors that link them together and that link them to outside systems. One external system (or coalition of systems) is what processes auditoryacoustic and visual information via the sense organs. The other one is conceptual structure which puts much of semantics and pragmatics outside the two core language modules. The conceptual system is multiply connected: not only is it interfaced with the core language system but also with everything else that contributes to our mental life. This connectivity may be illustrated by visualising a chain of structures, each stored in their own representational code but all obeying the basic processing principles of activation and competition. At the heart of this is something we recognise as the equivalent of a lexical item. This might be represented as follows (ignoring mediating interface systems and with the core language modules enclosed in square brackets): Auditory structure ↔ [ Phonological structure ↔ Syntactic Structure] ↔ Conceptual Structure ↔ Episodic structure.
The chain above is in fact a gross simplification because it is only linear. In fact, episodic memories are linked together directly with conceptual and auditory structures: the sound of a bell can invoke the memory of an unpleasant or exciting event at school, for example. The chains are in fact networks of connections in which many other elements may play a role such as visual and affective structures (supporting mental images and emotions). Take the sound of a school bell, which is an auditory structure. On hearing it, the episodic memory is activated along with the associated conceptual structure (meanings of bell) and, also, the linguistic structures associated with bell. If we were at school in France, for example, the activated network is going to be stronger for any element associated with France including French phonological and syntactic structures, hence, following the BIA principle of non-selective access, phonological and syntactic structures associated with cloche and bell will be activated in the English-French bilingual but cloche will probably be more strongly activated than its English neighbour. This elaboration, sketchy as it is, should at least be sufficient to indicate how MOGUL, using Jackendoff architecture, can use Grosjean’s insights and integrate them into a much wider psycholinguistic framework. It should also suggest how
49
50
Mike Sharwood Smith
more than one language can exist side by side in an integrated system as is also the case for varieties within a single language (Cook 1991). A spoken French word, a French-rated episodic memory or a conceptual structure that is uniquely associated with, say, French culture will trigger a chain or chains of associated structures that will raise the activation levels of specific structural elements that, notwithstanding, co-exist side by side in the same set of modules. The precise nature of such activation scenarios is, of course, the stuff of future research but using frameworks such as Jackendoff ’s gets us further down the line in the right direction.
3. Conclusion A MOGUL perspective permits us to see knowledge in processing terms but, without invalidating the notion of abstract linguistic knowledge, make more explicit the close relationship that (most of us intuitively accept) exists between acquisition and attrition. This is helped by the fact that Acquisition by Processing theory removes the need to postulate a separate Language Acquisition Device. Attrition research can benefit by reconciling processing explanations with knowledge-representational explanations. Common frameworks are needed plus an associated metalanguage with which to do it. MOGUL is one attempt to provide such a framework. In its current form, it is certainly not theory-neutral but is sufficiently flexible to allow different processing and abstract property explanations.
References Carroll, S. E. (1999). “Putting ‘input’ in its proper place”. Second Language Research, 15, 337– 388. Carroll, S. E. (2001). Input and Evidence: The raw material of second language acquisition. Amsterdam: John Benjamins. Chomsky, N. (1995). The Minimalist Program. Cambridge, MA: MIT Press. Cook, V. J. (1991). “The poverty-of-the-stimulus argument and multi-competence”. Second Language Research, 7 (2), 103–117. Cummins, R. (1983). The Nature of Psychological Explanation. Cambridge, MA: MIT Press. Dijkstra, A. F. J. & van Heuven, W. J. B. (1998). “The BIA-model and bilingual word recognition”. In J. Grainger & A. M. Jacobs (Eds.), Localist Connectionist Approaches to Human Cognition (pp. 189–225). Mahwah, NJ: Lawrence Erlbaum Associates. Dijkstra, A. & van Heuven, W. J. B. (2002). “The architecture of bilingual word recognition: From identification to decision”. Bilingualism: Language and Cognition, 5, 175–198. Escudero, P. & Sharwood Smith, M. (2001). “Reinventing the native speaker or ‘What you never wanted to know about the native speaker so never dared to ask’”. In A. Nizegorodcew & S. Foster-Cohen (Eds.), EuroSLA Yearbook 1 (pp. 275–286). Amsterdam: John Benjamins.
Understanding attrition within a MOGUL framework
Green, D. W. (1986). “Control, activation and resource: A framework and a model for the control of speech in bilinguals”. Brain and Language, 27, 210–223. Gregg, K. (1996). “The logical and developmental problems of second language acquisition”. In W. C. Ritchie & T. K. Bhatia (Eds.), Handbook of Second Language Acquisition (pp. 49–81). San Diego, CA: Academic Press. Grosjean, F. (1998). “Transfer and language mode”. Bilingualism: Language and Cognition, 1 (3), 175–176. Grosjean, F. (2001). “The bilingual’s language modes”. In J. Nicol (Ed.), One Mind, Two Languages: Bilingual language processing (pp. 1–22). Oxford: Blackwell. Jackendoff, R. S. (1987). Consciousness and the Computational Mind. Cambridge, MA: MIT Press. Jackendoff, R. S. (2002). Foundations of Language. Oxford: OUP. Jacquet, M. & French, R. M. (2002). “The BIA++: Extending the BIA+ to a dynamical distributed connectionist framework”. Bilingualism: Language and Cognition, 5 (3), 202–205. Jusczyk, P. W. (1997). The Discovery of Spoken Language. Cambridge, MA: MIT Press. Köpke, B. & Schmid, M. S. (2004). “First language attrition: The next phase”. In M. S. Schmid, B. Köpke, M. Keijzer & L. Weilemar (Eds.), First Language Attrition: Interdisciplinary perspectives on methodological issues (pp. 1–43). Amsterdam: John Benjamins. Marcus, G. F. (2001). The Algebraic Mind: Integrating connectionism and cognitive science. Cambridge, MA: MIT Press. Miyake, A. & Shah, P. (Eds.) (1999). Models of Working Memory: Mechanisms of active maintenance and executive control. Cambridge, MA: CUP. Paradis, M. (2000). “The neurolinguistics of bilingualism in the next decades”. Brain and Language, 71, 178–180. Paradis, M. (2001). “An integrated neurolinguistic theory of bilingualism (1976–2000)”. LACUS Forum, 27, 5–15. Pinker, S. & Prince, A. (1988). “On language and connectionism: Analysis of a parallel distributed processing model of language acquisition”. Cognition, 28, 73–193. Schwartz, B. D. & Sprouse, R. A. (1996). “L2 cognitive states and the Full Transfer/Full Access model”. Second Language Research, 12, 40–72. Sharwood Smith, M. (2004). “In two minds about grammar: On the interaction of linguistic and metalinguistic knowledge in performance”. Transactions of the Philological Society, 102 (3), 255–280. Sharwood Smith, M. & Van Buren, P. (1991). “First language attrition and the parametersetting model”. In H. W. Seliger & R. M. Vago (Eds.), First Language Attrition (pp. 17–30). Cambridge, MA: CUP. Sharwood Smith, M. & Truscott, J. (2005). “Stages or continua in second language acquisition: A Mogul solution”. Applied Linguistics, 22 (2), 219–240. Sharwood Smith, M. & Truscott, J. (2006). “Full Transfer Full Access: A processing-oriented interpretation”. In S. Unsworth, T. Parodi, A. Sorace & M. Young-Scholten (Eds.), Paths of Development in L1 and L2 Acquisition: In honor of Bonnie D. Schwartz (pp. 201–216). Amsterdam: John Benjamins. Truscott, J. & Sharwood Smith, M. (2004a). “Acquisition by processing: A modular perspective on language development”. Bilingualism: Language and Cognition, 7 (1), 1–20. Truscott, J. & Sharwood Smith, M. (2004b). “How APT is your theory: Present status and future prospects”. Bilingualism: Language and Cognition, 7 (1), 43–47.
51
Dynamic systems theory, lifespan development and language attrition Kees de Bot
Rijksuniversiteit Groningen, The Netherlands
The two basic questions in the field of language attrition are: why do people lose their language? and how do they lose their language? The former question concerns the factors which cause attrition, the latter the mechanics of the attrition process. Though the two questions are intimately related and mutually affect each other, I will draw upon two fields of research which are as yet largely unconnected to shed light on them. Where the why-question is concerned, I will argue for a perspective of developmental psychology in which developments over the whole lifespan are studied. The how-question, on the other hand, is best addressed within the framework of a dynamic systems perspective which focuses on the development over time of complex systems such as human cognition and language. Keywords: Dynamic Systems Theory; chaos and complexity theory; lifespan perspective
1. The why of language attrition: A lifespan developmental psychology perspective The main aim of lifespan developmental psychology is to study development as a life-long process. ‘Development’ implies that there may be gains and losses at different ages, so that development is not necessarily unidirectional. What counts as growth or decline is largely dependent on external and basically inter-subjective criteria. In their introduction to research methods in lifespan developmental psychology, Baltes, Reese & Nesselroade (1977) define the goal of research from this perspective as follows: “Life-span developmental psychology is concerned with finding models that are appropriate for the construction of a theory of ontogenetic change over various age changes” (88). Two approaches are distinguished: a continuity-oriented one in which models that have been developed for a specific
54
Kees de Bot
period of life are applied to other phases of the lifespan, for example in the application of the Piagetian model of early childhood to later ages. In the discontinuous approach, on the other hand, qualitatively different models are applied to different age ranges, for example in that a reduction of working memory capacity and peripheral perception are more likely to play a role in language development among elderly people than among adolescents. The study of development over the lifespan covers a wide range of topics, such as cognitive development, perceptual and motor development, social development, personality development and developmental psychopathology (see Magnusson 1996 and Demetriou, Doise & van Lieshout 1998 for overviews). In the past, investigations of these different areas have been treated as specialized subfields with few connections, but in the last two decades the awareness of the interconnectedness of developments on different levels and in different subsystems has grown. Research on brain functions, currently one of the most quickly expanding fields of investigation, has made it clear that cognitive functions interact with neurological functions, but also with physical changes and changes in the environment. This implies that not all functions and subsystems develop in parallel, some support each others’ growth or decline while others are compensatory in nature. In addition, there is evidence that individual developmental differences are task-dependent. The most important distinction in this respect is between context-dependent and context-free tasks: “The stability of differences in cognitive achievements on complex tasks between individuals is surprisingly low during childhood as well as during adulthood [...]. In contrast, individual differences for highly aggregated measures of context free abilities (IQ) tend to be very stable” (Weinert & Perner 1996: 214). A lifespan perspective on development will allow us to link different age ranges and compare the differential impact of different factors in them. It will also help us to see what changes over the lifespan are gradual and which ones are sudden.
2. Language development from a lifespan perspective In this contribution the focus is on language development over the lifespan. Although there is some treatment of language in handbooks on lifespan developmental psychology, the focus is generally narrow, as exemplified by Bernicot (1998) who discusses well-known stages of first language development, such as the prelinguistic period, the one-word period, the two-word period, and the sentence period. There are short sections on second language development in childhood and on communication and language in the elderly, but the overall claim is that “Communication and language during adulthood and old age have not yet
Dynamic systems and attrition
been studied in a systematic way” (162). This may be true for the French-speaking part of the world the author seems to focus on (though Luce Irigaray published her ‘Le Langage des Déments’ in 1973), but it is certainly not true for other parts of Western Europe and North-America. Klein’s (1996) contribution is less narrow in scope, focusing on the comparison between first language development in children and second language development in adults, in particular among migrants in Europe, but he pays no attention to the large body of research on language in the elderly. His treatment of ‘the age-factor’ in second language acquisition research is limited to the so-called ‘critical period’ (see Köpke, this volume; Pallier, this volume; see also Birdsong 2005 for a recent in-depth overview). This text is representative for most of the language development literature in that the ‘age’ factor is limited to the time span between birth and puberty. The same picture emerges in handbooks on language development where the focus is usually on the early stages. There are good reasons for this: the foundations of language develop in those early years. There seems to be an inverse relation between the increase in age of children and the number of studies devoted to language development at that age. After the development of oral skills has stabilized, scientific interest seems to shift to the development of reading and writing skills and developmental disorders such as dyslexia. There are hardly any investigations on language development in the age range from 18 till 55. As far as there is research on language in different age groups, such as age grading in sociolinguistics and the development of ethnolects and youth languages in adolescents, this does not usually extend to a systematic investigation of how language develops in that large age range and what factors play a role. On the contrary, approaches such as the apparent time method in sociolinguistics proceed from the basic assumption that the linguistic system, once acquired, is stable in the individual. Where lifespan developmental psychology is concerned, the field has been influenced by academic heavyweights such as Paul Baltes and K. Warner Schaie. Over the years they have advocated the inclusion of the third age in the study of development, and in some institutions lifespan development has become synonymous with old age. As the elderly are becoming a more and more important sociodemographic group in the Western world, research of cognitive, emotional and physical aging has increased significantly over the last decade and many specialized subfields with their own journals and congresses have emerged. Recently, Baltes (2003) has called attention to a further subdivision of the field that looks at ‘the fourth age’, the old-old group of 80 to 100 years old, a group that is growing rapidly in many parts of the Western world.
55
56
Kees de Bot
3. Language in the elderly There now is a large body of research on language and aging, and some of that research has merged with research on language attrition. Language attrition research has typically focused on the decline of language skills in healthy individuals over time, and in various typologies of attrition research, language in the elderly has been presented as first language attrition in a first language environment. This assumes that language in the elderly and the changes that take place are not caused by any disease. Obviously this all has to do with how we define aging: Age as such is merely an index variable without explenatory power. In their discussion of language and aging, de Bot & Makoni (2005) argue that aging involves changes at three different levels that interact: physical changes, cognitive changes and social changes. Linguistic decline may have its basis in biological changes, but the interaction between psychological and social changes will affect the rate of decline. Biological or neuropsychological changes lead to a reduction of resource capacity, in particular with respect to working memory, which then leads to changes in behavior. This may or may not be decline, objectively. What is important is that these changes are generally interpreted as decline, and this interpretation leads to changes in behavior. The elderly person is confronted with a style of interaction that is less demanding, less complex, easier to understand, and more easy to engage in. There is also social pressure on the elderly person to accommodate in this respect, even when the perception of cognitive decline is totally wrong. Here social stereotypes with respect to elderly people come in which tend to be very strong and persistent (Kemper & Harden 1999). Elderly people are forced into a type of linguistic behavior which robs them of opportunities to use the language in all its complexity. Given the skills character of language, non-use of a part of the sets or registers, words and constructions may lead to a decline in the use of these skills (see Paradis, this volume). If that happens, a catastrophic downward spiral sets in: changes in behavior are interpreted as decline and these lead to simplified language use. The lack of exercise of more advanced linguistic skills that results from this simplified language use leads to a decline of these skills, and accordingly the language used becomes less elaborate and complex, suggesting a decline in cognitive abilities, which will lead to further adaptations and simplifications in linguistic input and a reduction of the quality of the interaction (Ryan, Kennaley, Pratt & Shumovich 2000). There are strong indications that the core of the problem in language decline is in the functioning of the working memory component. Different types of buffers and short term storage systems are needed for language production and perception, and the decline of the capacity of these systems may be very detrimental
Dynamic systems and attrition
for language skills (Radvansky 1999; Brebion 2003). But working memory is more than just a storage system, it is also the workbench where information is prepared for the next processing step, and the tool used with the workbench is language. So memory and language interact in an interesting way: better language skills enhance memory performance, and a decline in working memory capacity can be compensated by high levels of linguistic skills (Miyake & Friedman 1998). Given global demographic changes, the interest in development in aging is understandable, but there is a risk of a similarly narrow focus as in the study of child language: studying one age group without looking at relevant mechanisms that are at work over the lifespan but may have different impacts in different age ranges. The three types of changes affecting language in the elderly – physical, social and cognitive – can be observed across the lifespan. In different age ranges one type of change may be more prevalent, e.g. in young adulthood, when decisions relating to career choices, partnership and family life are generally the most significant events, changes may be primarily social. In early childhood and in the fourth age mentioned above, changes may be mostly physical, but they always interact with changes on the other levels. In that sense the position taken here is oriented towards continuity rather than discontinuity.
4. Major life events and language attrition It should be obvious from the discussion of lifespan developmental psychology that this perspective is highly relevant for the study of language attrition, and that language attrition research should be a component of the study of language development over the lifespan. In lifespan developmental psychology, the concept of ‘Major Life Events’ has been propagated (Baltes et al. 1977; Braet & VerhofstadtDenève 1998). This concept refers to events that have a significant impact on the course of life, and includes events like going to school, getting a job, getting married, and migration; but also accidents, bereavement, losing one’s job and so on. What makes an event a major life event is highly dependent on individual settings and characteristics: getting divorced may be the worst possible nightmare for one individual and have a very negative effect on the course of that person’s life, and the best thing that could have happened for someone else. For the study of language development (and thus for the study of language attrition) specific language-related major life events may be relevant. Such languagerelated events may be insignificant on the larger scale of life, but they may be very significant for the development of the language system. Going to a bilingual kindergarten, having a pen friend abroad, choosing a school profile which includes
57
58
Kees de Bot
foreign languages, study abroad, international school exchanges, migration, may all be relevant. Again the patterns will be highly individual. From a language attrition perspective all events that lead to a reduction of the use of one language can be viewed as major life events. First language attrition is basically the interaction between a decline of the availability of one language and the development of another, competing one. Migration to an environment in which another language is spoken and accordingly a reduction in the amount of use of and contact with the L1 is the most obvious major life event for L1 attrition. But other events, such as changes in social networks, and political decisions such as the discontinuation of minority language teaching, may also be relevant. To summarize: the argument here is that the study of language attrition should benefit from developing links with lifespan developmental psychology. This branch of psychology provides a framework that takes into account changes that are not limited to certain age groups but may have an impact over the lifespan and in this way connect different stages in life. A core concept is that of ‘major life events’, events that appear to have an invasive effect on the course of life of individuals. Different types of language attrition are typically linked to major life events such as migration or language-related social changes. Both the theories and the research tools and techniques that have been developed in the lifespan tradition may prove to be useful for the study of language attrition.
5. The how of language attrition: A Dynamic Systems Theory approach Dynamic Systems Theory (DST) is a relatively recent approach to the study of development in complex systems. A dynamic system can be defined as: “a set of variables that mutually affect each other’s changes over time” (van Geert 1994: 50). The main components are therefore sets of variables on the one hand and change over time on the other. Some researchers have made the link between lifespan developmental psychology and DST, e.g. Demetriou (1998), who refers to the work by van Geert and others who have developed some of Piaget’s theories further into a dynamic approach. In recent years approaches to language and language development based on Dynamic Systems Theory (DST) have also emerged. A detailed discussion of DST is beyond the scope of the present contribution, and there are a number of publications that provide more detailed accounts of language as a dynamic system (van Geert 1994, 1998; Larsen-Freeman 1997; Herdina & Jessner 2002; de Bot, Lowie & Verspoor 2005, 2007). The difference from more static approaches to development is that in DST, the variables affecting development are studied in their interaction over time. The argument is that different variables (in the case of language these might be motiva-
Dynamic systems and attrition
tion to learn a language, success in learning a language, contact with a language) do not have a fixed effect, but that they interact and that this interaction itself changes over time: not only do motivation and success interact, but this interaction itself is subject to change. The main characteristics of a dynamic systems based approach are the following: Sensitive dependence on initial conditions: in some systems, small differences in initial conditions may have a big impact in the long run, while in other systems large differences in initial conditions have hardly any effect over time and level out. We have to keep in mind that most systems have no real beginning and that we are only looking at particular windows in time. The best-known example of a dynamic system is the weather system. It is impossible to locate one point of origin for the development of this system, as it has been evolving as long as our planet has existed. For language development this means that small differences between learners may have a large impact on the outcome of the learning process and that the same treatment (educational approach) does not necessarily lead to convergence. The same may be true in language attrition: small differences at the beginning of the attrition period may lead to big differences over time. Clyne’s work on Dutch and German migrants in Australia shows numerous examples indicating such a process: some individuals appear to be losing their first language very quickly, while other individuals that seem to start from the same position maintain it over decades with little observable change. Complete interconnectedness: dynamic systems are typically nested: large systems consist of subsystems and these in turn consist of sub-subsystems and so on. Where language is concerned, different subsystems of the cognitive system such as memory, attention and perception play a role, and in the learning process these and several other subsystems, such as motivation and language aptitude, contribute. All of these subsystems are interconnected, which means that change in one subsystem leads to changes in all others. With respect to language attrition this means that we cannot study factors in isolation, we need to look at how the interaction with other factors plays a role. This assumption is corroborated by findings which indicate that ‘time since emigration’ is only a relevant factor when the interaction with ‘amount of contact with the L1’ is taken into account (de Bot, Gommans & Rossing 1991; Soesman 1997). Non-linearity in development: the relation between cause and effect in a developing system may be disproportionate in the sense that some small change in the system may lead to large effects while another change of similar size may be absorbed. Also, a large change may have hardly any effect on the system, while
59
60 Kees de Bot
a much smaller change may lead to major perturbations. For language attrition there may be a disproportionate relation between time of non-use and decline in the system. Such an effect is observable in the findings from Meara’s (2004) study of lexical attrition (for a more detailed discussion see below). Change and development through internal reorganization and interaction with the environment: the classic example of self-reorganization is a pile of sand (Goles, Morvan & Duong Phan 2002): when grains of sand fall on a flat surface they will build a pile. As grains continue to fall, the pile of sand will change shape and settle for a while and then change again. It will reorganize itself constantly, based on the internal forces in the system. The particular way in which the system reorganizes itself cannot be predicted because the total of forces and variables cannot be calculated. Systems move from one stable state (attractor state) to the next. Attractor states are states the system prefers to settle in. A well-known example of attractor states are different ways in which horses run: they either gallop or trot, but there is no intermediate state. Language is also a self-reorganizing system with attractor states: input will affect the system, which will sometimes simply adopt a new element without internal reorganization, and sometimes adapt itself to the new input; but there may also be internal forces that lead to reorganization. Accordingly, language attrition is not just the loss of single elements or patterns: when elements are lost, the system may reorganize itself to find a new attractor state. Systems are constantly changing and only temporarily settle in so-called ‘attractor states’: the example of the pile of sand also serves to illustrate that systems are constantly changing as long as there are forces working on or in them. Even within a pile of sand there are forces at work: gravity will pull grains down, humidity changes and changes in temperature will lead to changes in adhesion between grains, the surface the pile is on will never be absolutely motionless, and so on. Due to these forces the system will change, but not continually: layers of sand will move and then settle for some time. When the system will change and what the next attractor state will be cannot be predicted, as it is the result of all the forces interacting over time. In language learning fossilization and stages of development can be seen as attractor states, and though there is hardly any research on this, it is likely that there will be similar stages in decline. Complete acquisition of a pattern or a word can also be an attractor state. As research on attrition has shown, complete mastery is not really an end-state: patterns that have been acquired completely can also be subject to loss or change (Schmitt & Meara 1997). Dependency on internal and external resources: the forces at work in a pile of sand could be seen as resources for change, but in human learning other types of resources play a role. Internal resources include motivation to learn, previ-
Dynamic systems and attrition
ous knowledge and aptitude. External resources include opportunities to learn, time invested by the environment, availability of tools for learning and material conditions (food, shelter) that make learning possible. The carrying capacity of a system is the sum of the resources that can be used for optimal development and performance. The carrying capacity of a highly motivated, talented and well educated learner in optimal learning conditions is markedly larger than that of a poorly motivated, untalented learner with little time and opportunity to learn. This all has to do with allocation of resources. Some minimal amount of resources is also needed to keep a system going and to maintain elements within it. For language attrition this means that at least some resources (time, contact) must be spent on a language in order to maintain it, though we simply do not know how much time and contact, and what type of contact, is needed for maintenance (see also Schmid, this volume). Complete non-use eventually means decline: the carrying capacity of the system is so low that the system cannot be maintained in its optimal form. Systems may show chaotic variation over time: over time a dynamic system will move from one attractor state to the next, but what the next attractor state will be is unpredictable. Because of the complex interaction of variables, the same system with almost identical initial conditions may show different attractor states. The term ‘chaos’ has been brought to bear upon this unpredictability, but the meaning of ‘chaos’ as a technical term only partly overlaps with the everyday usage. The bedroom of a 14-year-old may be chaos from the parents’ perspective but it is not really unpredictable: chaos is in fact the normal state in that setting. For language learning, on the other hand, chaos is not the most likely state: the developmental tract in L1 and L2 acquisition shows considerable inter-individual consistency, variability between individual learners notwithstanding. This partial uniformity in development may reflect either system-internal constraints on development (which is basically what proponents of the role of universal grammar argue) or a constraining effect of the language input. Smith, Kirby & Brighton (2003) argue that the poverty of input a child receives in language development is in itself the most important force shaping language: language is the way it is because it can be acquired on the basis of a restricted amount of input. Development is conceived of as an iterative process: this means that the present level of development depends critically on the previous level of development. Language evolves gradually and every use of the language is an iterative step in its development. Similarly language attrition proceeds in steps in which the present state of the system is crucial for the next phase.
61
62
Kees de Bot
The main point is that language can be seen as a dynamic system and that language development is a dynamic process in which the language system may grow or decline depending on resources and interactions between input and internal forces. There is no fundamental difference between the processes of growth and decline, they are both governed by the same principles. All language users are constantly going through phases of growth and decline, as their first and foreign language systems are constantly changing, depending on internal reorganization and input from the environment. Another somewhat problematic characteristic of a DST approach to language development, and particularly to language attrition, is that language development cannot really be explained: the factors playing a role in development may all be known, but how their interaction will change in the course of development cannot be predicted and may in that sense be chaotic. A system develops in a certain way because that is how this system develops. On the one hand this is not a very satisfactory situation since research has traditionally been about hypothesis generation and hypothesis testing. The full interconnectedness of systems means that isolating single variables as explanatory may be unproductive because the impact of that variable on all the other variables is not taken into account. For the moment it seems wise to accept that we cannot explain development and to focus on adequate descriptions of developmental processes. In investigations which use DST as their starting point, simulations have been used to develop models in which complexity can be implemented up to a certain point and therefore allow the exploration of developmental paths. As in simulations many more variables can be included, this approach allows for more complete models than we can ever test in humans. But simulations can never be the final answer. We need to link simulations with empirical data even though that may not always be easy or even feasible in interconnected systems. In the last section of this contribution an example of the application of a simulation is presented that shows some of the potential and some of the problems of the use of DST and simulations in attrition research.
6. An example of research on dynamic systems and attrition: Meara (2004) Meara (2004) is one of the first studies on attrition in lexical networks. He argues that research on language attrition has failed to recognize that the lexicon is not a set of single items, but a network of interrelated elements. Attrition in his view is not the reduction of access to lexical items, but a complete reorganization of a network. Since it is rather difficult to show this in human beings, and because
Dynamic systems and attrition
the human cognitive system, even the part we call the mental lexicon, is far too large and complex to approximate through the simulations of networks even the most advanced computers can do, he developed fairly simple networks and investigated how they behave when elements are lost. He set up networks of 2,500 items, in which each word is connected to two other words, and all elements have two states of activation, on or off (‘K = 2 random autonomous Boolean network’). Activation of a particular word depends on the level of activation of the words it is connected to. So changes in the activation status of one word will lead to a cascade of activation/deactivation until the network settles in a new attractor state. By varying the number of words that are deactivated (which are called ‘attrition events’ in his approach) the changes in the whole network can be studied. The iterations with increasing numbers of switched-off words lead to remarkable differences in attrition patterns. While in some cases there was a clear attrition effect after a fairly small number of attrition events, in other cases the network remained stable despite the large number of attrition events. Meara’s networks are much more rudimentary than a real lexicon. There is considerable evidence that words are not simply switched on or off but show degrees of activation, which are reflected in latencies in reaction time experiments (Hakuta & d’Andrea 1992; Hulsen, de Bot & Weltens 2002). Investigations using the savings paradigm, which tests residual knowledge through relearning, have shown that word knowledge can be retained at a very low level (de Bot & Stoessel 2000). Therefore, a more realistic model would be one in which the lexicon consists of elements with varying degrees of levels of activity. A general deactivation effect of network connections due to lack of use leads to diminished activity for elements once they have reached a certain critical threshold. As Meara repeatedly emphasizes, the mechanics of artificial networks cannot be compared to what happens in a real lexicon. However, some of the findings are suggestive of how a dynamic systems approach may help us to interpret empirical data in language attrition. Firstly, the simulations show that attrition in such networks is clearly nonlinear: the effects of steadily increasing numbers of attrition events lead to very different patterns of attrition (see Fig. 1), with some iterations leading to stable networks without much attrition and others showing heavy decline rapidly. Secondly, the data underline the earlier point of a need for a different approach to variation: while the average patterns show a gradual decline (Fig. 2), none of the individual cases show this pattern. Interestingly, the findings in Meara’s rudimentary attrition network mirror some of the crucial findings in attrition research. The first is that there is considerable individual variation among individuals in more or less similar attrition settings. The work on Dutch and German in Australia (Ammerlaan 1997; de Bot
63
64 Kees de Bot
Figure 1. Ten examples of attrition in a network (Meara 2004: 143, his Fig. 1)
Dynamic systems and attrition
Figure 2. Pooled data abstracted from the ten cases in Figure 1 (Meara 2004: 145, his Fig. 2)
& Clyne 1994; Waas 1996) has shown that while some individuals already show considerable attrition after a few years in the L2 environment, others show remarkable stability of language skills even after more than 25 years. The second is that there seems to be a part of the language system that, once acquired, is highly resistant to loss. This finding first emerged in work on the attrition of Spanish in the USA (Bahrick & Phelps 1987), which led Neisser (1984) to suggest a ‘permastore’ of knowledge that is stable over time despite limited or no contact with the language. This finding has been supported by a host of literature by various researchers (Hansen 2001; Hedgcock 1991; Schmid 2002).
7.
Concluding remarks
In several publications Schmid (2002, 2004) and Köpke & Schmid (2004) have called for a more theoretical approach to language attrition research. There is obviously a need for an overarching theory that links findings from attrition research with findings from other fields, such as sociolinguistics, neurolinguistics and psycholinguistics. There are two main questions; why does language attrition occur? and how does it occur? In this contribution it is argued that two theoretical approaches can be combined to further our understanding of language attrition and to contribute to more solid theory building. Notions from lifespan developmental psychology and in particular what have been labelled language-related major life events can help us to explain how changes over the lifespan on different levels lead to a change in the linguistic setting that may have an impact on language development.
65
66 Kees de Bot
For the how of language attrition, Dynamic Systems Theory may provide useful contributions. Other complex systems have been studied extensively using the concepts and tools of DST. Though the complexity of language in use is such that we may never be able to model language as a complex system in all its details, the formalizations of DST will force us to become much more explicit about how variables interact and what role they are assumed to play in language development. Through the application of this general theory on development, language attrition will lose some if its splendid isolation and independence, but it may gain in significance through its contribution to our understanding of the process of language development.
References Ammerlaan, T. (1997). “‘Corrosion’ or ‘loss’ of immigrant Dutch in Australia: An experiment on first language attrition”. In J. Klatter-Folmer & S. Kroon (Eds.), Dutch Overseas: Studies in maintenance and loss of Dutch as an immigrant language (pp. 69–97). Tilburg: Tilburg University Press. Bahrick, H. & Phelps, E. (1987). “Retention of Spanish vocabulary over 8 years”. Journal of Experimental Psychology: Learning, Memory and Cognition, 13, 344–349. Baltes, P. B. (2003). Die Zukunft des Alterns. Berlin: Max Planck Institut. Baltes, P. B., Reese, H. W., & Nesselroade, J. R. (1977). Life-Span Developmental Psychology: Introduction to research methods. Hillsdale, NJ: Lawrence Erlbaum Associates. Bernicot, J. (1998). “Communication and language development”. In A. Demetriou, W. Doise & C. F. M. van Lieshout (Eds.), Life-Span Developmental Psychology (pp. 137–178). Chichester: Wiley. Birdsong, D. (2005). “Interpreting age effects in second language acquisition”. In J. Kroll & A. M. B de Groot (Eds.), Handbook of Bilingualism: Psycholinguistic approaches (pp. 109– 127). New York, NY: OUP. Braet, C. & Verhofstadt-Denève, L. (1998). “Developmental psychopathology”. In A. Demetriou, W. Doise & C. F. M. van Lieshout (Eds.), Life-Span Developmental Psychology (pp. 447– 500). Chichester: Wiley. Brebion, G. (2003). “Working memory, language comprehension, and aging: Four experiments to understand the deficit”. Experimental Aging Research, 29, 269–301. de Bot, K. & Clyne, M. (1994). “A 16-year longitudinal study of language attrition in Dutch immigrants in Australia”. Journal of Multilingual and Multicultural Development, 15, 17– 28. de Bot, K., Gommans, P. & Rossing, C. (1991). “L1-loss in an L2-environment: Dutch immigrants in France”. In H. W. Seliger & R. M. Vago (Eds.), First Language Attrition: Structural and theoretical perspectives (pp. 31–52). Cambridge: CUP. de Bot, K., Lowie, W. & Verspoor, M. (2005). Second Language Acquisition: An advanced resource book. London: Routledge. de Bot, K., Lowie, W. & Verspoor, M. (2007). “A Dynamic Systems Theory approach to Second Language Acquisition”. Bilingualism: Language and Cognition, 10, 7–21.
Dynamic systems and attrition
de Bot, K. & Makoni, S. (2005). Language and Aging in Multilingual Societies: Dynamic perspectives. Clevedon: Multilingual Matters. de Bot, K. & Stoessel, S. (2000). “In search of yesterday’s words: Reactivating a long forgotten language”. Applied Linguistics, 21, 333–253. Demetriou, A. (1998). “Cognitive development”. In A. Demetriou, W. Doise & C. F. M. van Lieshout (Eds.), Life-Span Developmental Psychology (pp. 179–270). Chichester: Wiley. Demetriou, A., Doise, W. & van Lieshout, C. F. M. (Eds.) (1998). Life-Span Developmental Psychology. Chichester: Wiley. Goles, E., Morvan, M. & Duong Phan, H. (2002). “Sand piles and order structure of integer partitions”. Discrete Applied Mathematics, 117, 51–64. Hakuta, K. & D’Andrea, D. (1992). “Some properties of bilingual maintenance and loss in Mexican background high-school students”. Applied Linguistics, 13, 72–99. Hansen, L. (2001). “Language Attrition: The fate of the start”. Annual Review of Applied Linguistics, 21, 60–73. Hedgcock, J. (1991). “Foreign language retention and attrition: A study of regression models”. Foreign Language Annals, 24, 43–55. Herdina, P. & Jessner, U. (2002). A Dynamic Model of Multilingualism: Perspectives of change in psycholinguistics. Clevedon: Multilingual Matters. Hulsen, M., de Bot, K. & Weltens, B. (2002). “Between two worlds: Social networks, language shift and language processing in three generations of Dutch migrants in New Zealand”. International Journal of the Sociology of Language, 153, 27–52. Irigaray, L. (1973). Le langage des déments. The Hague: Mouton. Kemper, S. & Harden, T. (1999). “Experimentally disentangling what’s beneficial about elderspeak from what’s not”. Psychology and Aging, 14, 656–670. Klein, W. (1996). “Language acquisition at different ages”. In D. Magnusson (Ed.), The Lifespan Development of Individuals: Behavioral, neurobiological and psychosocial perspectives (pp. 244–264). Cambridge: CUP. Köpke, B. & Schmid, M. S. (2004). “Language attrition: The next phase”. In M. S. Schmid, B. Köpke, M. Keijzer & L. Weilemar (Eds.), First Language Attrition: Interdisciplinary perspectives on methodological issues (pp. 1–43). Amsterdam: John Benjamins. Larsen-Freeman, D. (1997). “Chaos/complexity science and second language acquisition”. Applied Linguistics, 18 (2), 141–165. Magnusson, D. (Ed.) (1996). The Lifespan Development of Individuals: Behavioral, neurobiological, and psychosocial perspectives. Cambridge: CUP. Meara, P. M. (2004). “Modeling vocabulary loss”. Applied Linguistics, 25 (2), 137–155. Miyake, A. & Friedman, N. R. (1998). “Individual differences in second language proficiency: Working memory as language aptitude”. In A. Healy & L. Bourne (Eds.), Foreign Language Learning: Psycholinguistic experiments on training and retention (pp. 339–364). Mahwah, NJ: Lawrence Erlbaum Associates. Neisser, U. (1984). “Interpreting Harry Bahrick’s discovery: What confers immunity against forgetting?” Journal of Experimental Psychology: General, 113, 32–35. Radvansky, G. A. (1999). “The fan effect: A tale of two effects”. Journal of Experimental Psychology: General, 128 (2), 198–206. Ryan, E. B., Kennaley, D. E., Pratt, M. W. & Shumovich, M. A. (2000). “Evaluations by staff, residents, and community seniors of patronizing speech in the nursing home: Impact of passive, assertive, or humorous responses”. Psychology and Aging, 15, 272–285.
67
68 Kees de Bot
Schmid, M. S. (2002). First Language Attrition, Use and Maintenance: The case of German Jews in Anglophone countries. Amsterdam: John Benjamins. Schmid, M. S. (2004). “First language attrition: The methodology revised”. International Journal of Bilingualism, 8 (3), 239–256. Schmitt, N. & Meara, P. M. (1997). “Researching vocabulary through a word knowledge framework: Word associations and verbal suffixes”. Studies in Second Language Acquisition, 19, 17–36. Smith, K., Kirby, S. & Brighton, H. (2003). “Iterated learning: A framework for the emergence of language”. Artificial Life, 9 (4), 371–386. Soesman, A. (1997). “An experimental study on native language attrition in Dutch adult immigrants in Israel”. In J. Klatter-Folmer & S. Kroon (Eds.), Dutch Overseas: Studies in maintenance and loss of Dutch as an immigrant language (pp. 181–194). Tilburg: Tilburg University Press. van Geert, P. L. C. (1994). “Vygotskian dynamics of development”. Human Development, 37 (6), 346–365. van Geert, P. L. C. (1998). “A dynamic systems model of basic developmental mechanisms: Piaget, Vygotsky and beyond”. Psychological Review 105 (4), 634–677. Waas, M. (1996). Language Attrition Downunder. Frankfurt: Peter Lang. Weinert, F. E. & Perner, J. (1996). “Cognitive development”. In D. Magnusson (Ed.), The Lifespan Development of Individuals: Behavioral, neurobiological, and psychosocial perspectives (pp. 207–222). Cambridge: CUP.
The grammatical profile of L1 speakers on the stairs of potential language shift Carol Myers-Scotton
Michigan State University, USA
This chapter tests the hypothesis that a shift from an L1 to a community-dominant language as speakers’ main public language is abrupt, not gradual (i.e., from one grammatical system to another, without implying L1 grammatical attrition). When the situation biased urban Xhosa-English bilinguals in South Africa to speak only Xhosa, many speakers showed Xhosa-English codeswitching (English content words in Xhosa grammatical frames). Such codeswitching does not necessarily predict a shift. Rather, in such a situation the frequency of L2 monolingual clauses (English here) does point to a shift and its abruptness. A cluster analysis divided the sample (N = 48) into three groups based on use of English elements; the ‘front runners’ group showed numerous English clauses. Keywords: language shift; codeswitching; South Africa; Xhosa
Introduction All over sub-Saharan Africa, local languages are often in contact with a community-dominant, international language that is often either the sole official language or a co-official language (i.e., English, French, or Portuguese, depending on the nation). If that international language is the main language used in government and business – no matter which languages are designated as official – it inevitably captures prestige as the language necessary to speak for socio-economic advancement, no matter how attached many ethnic groups are to their L1 for the solidarity capital it offers. The emerging picture is that those Africans who want to be on the stairs of upward mobility become bilinguals if that option is open to them through advanced education or other means. They retain their L1 for some ingroup interactions, but speak the invading non-local language in more and more situations. Will they lose their L1? For psychological and sociological reasons, that seems unlikely to happen soon if at all, but that discussion is beyond the
70 Carol Myers-Scotton
scope of this study. The question studied here is, even if they retain their L1, are these bilinguals in the progress of shifting to the international language in their community as their preferred public language? Quite possibly, many of these bilinguals have already made this shift. This paper examines the grammatical steps in how such a shift may happen, reporting on a study of mainly young adult Xhosa-English bilinguals in South Africa. In the corpus studied, their speech includes three types of clauses: monolingual clauses in Xhosa, clauses showing Xhosa-English codeswitching, and monolingual clauses in English. The paper tests a set of hypotheses that include two main propositions: 1. Within a group of bilinguals whose psycho-sociolinguistic profiles seem to make them vulnerable to shift, differences in their use of the target language of shift divides them into distinguishable groupings; that is, not everyone is on the same “bilingual step”. Thus, in the case of the Xhosa-English bilinguals, the prediction is that in an interaction potentially including both English and Xhosa, their use of English can distinguish one sub-group from another. 2. When shift happens, it is abrupt from the grammatical point of view. That is, speakers shift from one grammatical system to another without going through a phase of incorporating grammatical elements from one language into the other. In the case of Xhosa-English bilinguals, even those speakers who show the greatest use of English, who may be on the brink of shifting to English as their main language, do not use critical grammatical morphemes from English when they are producing Xhosa-framed phrases or clauses. Critical grammatical morphemes are those inflections and functional elements that are defined as the two types of “late” system morpheme under the 4-M model (cf. Myers-Scotton & Jake 2001; Myers-Scotton 2002). They serve either (a) to join together elements within a phrase in order to meet a language’s wellformedness conditions for that phrase (e.g. English of in ‘book of Laurie’), or (b) to indicate grammatical relationships between phrases (e.g. subject-verb agreement affixes or clitics, as well as case markers). Under the 4-M model, the hypothesis is that, in language production, these morphemes are structurally-assigned; that is, they only become salient at the level of the formulator when the larger constituents are assembled. Thus, their production is later than that of the other two types of morpheme in the 4-M model that mainly convey content (cf. Myers-Scotton 2005). These other types are assumed to be salient at the level of the mental lexicon. These are content morphemes (e.g. nouns and verbs) and “early” system morphemes (e.g. determiners and other morphemes that add meaning to their content morpheme heads and depend on these content morphemes for information about their form).
Grammatical stairs in language shift
This second proposition is motivated by existing studies of adult bilingual speakers whose main language is no longer their L1. In general, contrary to the view that ‘loss’ and ‘simplification’ mark all aspects of a speaker’s L1 when another language becomes his/her main language, empirical evidence supports the prediction that critical grammatical morphemes are maintained intact in a speaker’s L1 at a high frequency. However, unfortunately, most attrition studies that consider morphology group together their results under one rubric, “morphology”. Still, a number of studies give results specifically for one type of critical morpheme, the morpheme marking case, and those results are instructive. For example, Hlavac (2003) studied young adults who are second generation Croatians raised in Australia; English had become their main language, but they still speak Croatian frequently with in-group members, especially older relatives. Or, Gross (2000) studied German L1 speaking adults who had lived in the United States for about forty years and who speak German only infrequently. In both studies, the researchers found that affixes marking case were retained at a high rate. Hlavac found that 91 per cent of all Croatian NPs (8,444/9,318) receive target-like case assignments for the five Croatian cases. Still, this percentage includes ‘zero’ as an accurate assignment for masculine singular nominatives and masculine accusative inanimates; zero is the realization expected in standard, monolingual Croatian. Yet, if these forms are not considered, then 75 per cent of overt case marking is still target-like. Gross found that in his German data, pre-nominal adjectives, which require case marking, had non-target realizations in only 9.3 per cent of the cases. German determiners were inaccurate in only 2.5 per cent of their occurrences (54/1,917). It is important to note that these numbers for grammatical morphemes contrast sharply with results for content morphemes (e.g. nouns and verbs, etc.). For these, 29.7 per cent shows some modification from their meanings in the home variety of the German dialect studied. Other studies of L1 attrition among adults include the well-known study of East Sutherland Gaelic speakers in Scotland who were bilingual in English, but either also remained fluent speakers of their L1 at the time they were studied, or were labeled as semi-speakers (Dorian 1981, inter alia). Also, de Bot & Clyne (1994) report on a longitudinal study of Dutch migrants to Australia. A more recent study by Schmid (2002, 2004) considered pre-World War II Jewish migrants from Germany 60 years after their migration to Anglophone countries. All of these studies showed a good deal of retention of grammatical morphemes, especially case, although results varied across groups of speakers and the details given. For example, although Dorian found the case system in East Southern Gaelic was weakening, the nominative-dative distinction was strongly maintained in older fluent speakers (98 per cent in two villages, 77 per cent in another). It was even 56 per cent by semi-speakers in two villages and 32 per cent in another
71
72
Carol Myers-Scotton
village (Dorian 1981: 147 cited in Myers-Scotton 2002: 215–216). Schmid’s findings are reported in different terms, but in discussing so-called mistakes in case assignment in her sample, she states that results “... would point to the conclusion that there is no overall reduction of the case system where the dative is lost first and a binary system emerges, as the regression hypothesis predicts ...” (Schmid 2002: 95). And in her 2004 article, Schmid comments on the differences between the emigration groups, stating “Where morphology is concerned, the repertoire of emigra1 [one group] seems not to have been affected a great deal ...” (p. 251) although she goes on to imply that the case system for the other two emigration groups showed some reduction.
1. Language choice among Xhosa-English bilinguals The present study reports on interview data from 48 Xhosa-English bilinguals who are in-migrants to the Gauteng Province, the main industrial area of South Africa; Pretoria, the national capital, and Johannesburg are located there. Most of the interviews took place in Pretoria or Soweto or its environs. The speakers all originally lived in the Eastern Cape Province, an area that is dominated by XhosaL1 speakers. The speakers studied all have been living in Gauteng Province for at least two years; their average age is 25. The motivation for selecting such speakers to study is that, as in-migrants living far from their home area, they are vulnerable to shift from their L1 for many interactions. English is the main lingua franca in Gauteng Province, and in most other parts of South Africa. However, Afrikaans is also regularly used in Gauteng, especially in Pretoria. (Interestingly enough, the speakers, as a group, used very few Afrikaans words.) Informal interviews were the source of data in which the bias was for the interviewee to speak mainly or entirely in Xhosa. Speakers were asked open-ended questions calling for their opinions about current issues that are relevant to their lives. A 33-year-old woman living in Pretoria was the main interviewer. The interviewers were L1 Xhosa-English bilinguals and introduced the interview by speaking Xhosa and then spoke largely Xhosa, with only a few switches to English for words or short phrases. Thus, the matter of language choice was left up to the participants, but Xhosa was the obvious unmarked choice. Still, all speakers used some English and a number used a good deal of English. The degree of a speakers’ proficiency in English was not tested formally; speaker selection was based on the interviewer’s assessment of the speaker’s degree of bilingualism and the speaker’s own self-reports on their daily language use. The interviewers themselves used both Xhosa and English in their everyday lives; also, they held jobs that required English use and they had advanced degrees that required study in English.
Grammatical stairs in language shift
To give some idea of what the interview data look like, short extracts from three interviews are reproduced here. English in the examples is in bold and translations are in italics. Example (1) comes from Nw, a young woman in her early thirties, who has her own business, but is also attending a technical college. Here is her response to a question about social services in South Africa: I see South Africa as a charity work ‘I see South Africa as a charity work.’ they are giving so much to people who are actually aba-na-khathali-yo u-ku-bon-e they-neg-care about-rel you-inf-see-subjunctive ‘they are giving so much to people who are actually – people who do not care about you see–’ eziny’izinto a-zi-kho necessary ja South Africa should give other-c.10-thing neg-it-exist yes ku-bantu to-people ‘Some other things are not necessary – yes – South Africa should give to people’ aba-need-a-yo they-need-indic-rel ‘who are needy’ but eziny’izi-nto a-ba-zi-deserv-i other c.10-thing neg-they-c.10.obj-deserve-neg ‘but they do not deserve other things’. (1)
Example (2) comes from a male in his thirties who has a graduate degree (MA) and works in the government. He is comparing living in the city with living in his home area. (2) eh- kwi- big cities i-challenge e ma-ndi-qal-e ndi-thi loc- cl.9- let-1s-start-subjunct 1s-say ku-qal-a inf-start-indic ‘In the big cities the challenge – eh- let me start – I say to start’ yi-ku-mnandi apha e because u-fuman-a i-experience it-is-nice here 2s-get-indic c.9 ‘It is nice here eh because you get experience’ yo-m-sebenzi o-nge-na wu-yi-fuman-a xa assoc-cl.3-work rel-neg-with 2s-it-get-indic when
73
74
Carol Myers-Scotton
u-kwi-homeland 2s-loc ‘Of work which you do not get when you are in [the] homeland.’
Example (3) comes from S, a young woman in her twenties who lived in Capetown at one time, but now lives in Gauteng Province. (3) Aph’-e-Rhawutini i-life it’s easy here-loc-Gauteng c9-life it’s easy ‘Here in Gauteng, life is easy’ all you need i-mali. It’s money makes the world go round and c.9-money e-Koloni i-mali loc-Cape c.9-money ‘All you need [is] money. It’s money makes the world go round and in Capetown money i-nqabile and into ezi-ninzi zi-nqabile and c.9-difficult thing c.10-many c.10-difficult and ku-ya-sokol-w-a. loc-pres struggle-pass-indic was difficult and many things were difficult and were struggled...’
2. Theoretical framework The study was based on three theoretical premises that refer to the nature of incursions of a sociolinguistically dominant language into the speech of an L1 speaker of a language with less socio-economic prestige in the community. First, a premise is that in a milieu where a language other than the speaker’s L1 is sociolinguistically dominant, speakers of any one L1 who have access to proficiency in the dominant language will show indications of shift to that language, at least for their public interactions. Second, progression to language shift is predicted to consist of separable grammatical steps that are identified by the relative use of elements from the community-dominant language (English in this case); a speaker’s position in terms of these steps is a predictor of that speaker’s likely shift to the L2 as his or her main public language. Third, the notion that there are steps implies that shift is not gradual, but abrupt, in regard to the grammars of languages but is shift from one grammatical system to another. . The assumption in the sociolinguistic literature seems to be that language shift is gradual in regard to changes from one domain to another; the evidence about the grammatical nature of
Grammatical stairs in language shift
General hypotheses were stated in the introduction. Here they are made more specific. The Identifiable Group Hypothesis. Use and frequency of different types of English elements divides 48 Xhosa-English bilinguals into discrete groups (i.e. groups do not overlap in terms of their use of English). A sub-hypothesis is that frequency of use of monolingual English clauses is the feature that most differentiates groups. This leads to the prediction that speakers on the brink of shift will use a high percentage of English-based clauses. This study does not document the actual shift; it only predicts shifters show a high-percentage of English-based clauses relative to Xhosa-framed clauses. (A clausal frame is abstract, but with surface consequences. To say that a clause has a specific language frame means that its surface structure shows whatever range of morpho-syntactic patterns are well-formed in the framing language. In this data set, there are three types of clause: monolingual Xhosa-framed clauses, monolingual English-framed clauses, and Xhosa-framed clauses that include English content morphemes that typically receive Xhosa affixes, i.e., codeswitching). The Implicational Scale Hypothesis. Evidence will point toward an implicational scale such that there is a positive relation between the use of various English elements. For example, a speaker who uses many English-based clauses and English phrases will also use many singly-occurring English content morphemes (nouns and verbs) with Xhosa affixes. The Abruptness Hypothesis. Critical grammatical morphemes from the target language for shift (the L2) will not appear in the L1. Critical morphemes (see above) include two types. One type is the “bridge” late system morpheme; such morphemes serve to make certain language-specific structures well-formed. In English, these include the morpheme of that appears in associative constructions in English (e.g. ‘cover of the book’) and also expletive it which appears in weather constructions in English (e.g. ‘it’s raining’). Certain elements in Comp positions in some languages also seem to be bridge system morphemes; for example that in English seems to be a bridge; its designation needs more study in this corpus shift that is reported here calls into question whether shift from any point of view is ever really gradual. . It is not quite true to say these clauses are monolingual in English. There are occasional instances of Xhosa prefixes for noun class nine (e.g. i-budget) that appear on English nouns in clauses that are otherwise entirely in English. The same class nine prefixes also more typically appear on English nouns when they are in Xhosa-framed clauses. Why these prefixes occur sometimes, but not always, and how they function requires further analysis. They may be Xhosa ‘identity markers’ to remind the listener – in the midst of all that English – that the speaker is a Xhosa person, after all.
75
76
Carol Myers-Scotton
(cf. Myers-Scotton & Jake 2007). The second type of critical morpheme is the “outsider” late system morpheme; it is called an outsider because its form depends on information outside the phrase in which it appears. These morphemes include subject-verb agreement, so they include the English suffix -s, marking third person singular in the present tense. They also include markers for English past tense with regular verbs (-ed and the past tense construction did + unmarked main verb). The Convergence Hypothesis. Xhosa data will show some convergence to English in regard to at least minor syntactic structures, such as adjunct placement. The Demographic Hypothesis. The demographic features of age, education, and gender may correlate with specific patterns of the use of English elements.
3. Study results Testing The Identifiable Group Hypothesis. A two-step Cluster Analysis shows that the 48 L1 Xhosa speakers form three, discrete groups that do not overlap in their use of the English elements studied. The groups are most discrete in regard to the use of English phrases or full clauses. These findings support the premise that sociolinguistically similar Xhosa-English bilinguals are not all on the same ‘step’ in their use of English. The groups are these: the Slow Starters (58% N = 28/48), the Middle Pack (25% N = 12/48), and the Front Runners (17% N = 8/48). The variables studied included these categories: overall total of English words used, English nouns with Xhosa prefixes, English verbs with Xhosa affixes (prefixes or suffixes), English adjectives with Xhosa prefixes, English phrases, and Englishframed clauses. Their production of English-framed clauses most graphically differentiates the groups. The Slow Starters have a mean of only 3.21 for such clauses; the Middle Pack has a mean of 25.33, and the Front Runners have a mean of 50.25. Looked at another way, the separations are equally clear: there were 20/48 speakers with ten or more full English clauses, and of these 12 had 29 or more such clauses; in addition, 22/48 speakers had 11 or more English phrases. They contrast with 13/48 speakers with no English clauses at all, and 13/48 speakers who had three or fewer English phrases. Testing the Implicational Scale Hypothesis. The data did not lend themselves to setting up an actual implicational scale. However, preliminary analyses of the data by means of scattergrams appeared to indicate a close relationship between speakers’ use of English-framed clauses and English phrases in otherwise Xhosa clauses, as well as between the frequency of use of monolingual English clauses on the one hand and English nouns or verbs with Xhosa affixes on the other. A scattergram that locates individual speakers on two axes (two of the English
Grammatical stairs in language shift
English verbs with Xhosa affix
English clauses
Figure 1. Scatterplot of use of English verbs with Xhosa affix and English clauses per speaker
variables just described) shows that those speakers who use many monolingual English clauses are also the ones who use more English verbs with Xhosa affixes (see Fig. 1). Similar relationships were found between the use of English clauses and English nouns with Xhosa affixes, and between English-framed clauses and English phrases (not in English clauses). These impressions were tested by means of a Pearson correlation test, which shows a correlation between frequency of English phrases and English clauses at .724 (p < .01). The correlation between English clauses and English elements that are more integrated into Xhosa frames is less; with English verbs with Xhosa affixes in Xhosa frames it is .463 (p < .01) and it is .370 (p < .01) with English nouns with Xhosa affixes in Xhosa frames. Still, the indication from the scattergrams is that the Front Runners, who are defined as members of group three, the speakers who use the most English-framed clauses, are the speakers who also use the most English verbs with Xhosa affixes. They are the speakers who are the clear candidates for shift to English, at least as their main public language. Testing the Abruptness Hypothesis. This hypothesis is supported in the XhosaEnglish corpus. There are no examples at all of the most critical English grammatical morphemes (i.e. outsider system morphemes) in Xhosa-framed phrases. . There are occasional instances of the English plural -s suffix with English nouns when they appear in codeswitching patterns (Xhosa-framed clauses). However, this element is an early
77
78
Carol Myers-Scotton
Even those speakers who use a great deal of English keep these critical morphemes of English separate from Xhosa contexts where they presumably could occur. For example, one speaker used an associative construction with two English NPs, but a Xhosa associative element (i-rate y-e crime ‘the rate of crime’). This associative marker is multi-morphemic, because in addition to the marker itself (a “bridge” under the 4-M model), it includes the noun class agreement for associative constructions, which is the agreement affix for class nine, the default class for foreign words. This agreement prefix is an outsider system morpheme because it must look for its form for information outside the maximal projection in which it occurs (i.e., it looks to a preceding noun). There are other ample opportunities for English critical morphemes to spill over into Xhosa because some speakers use a Xhosa-framed clause next to an English-framed clause in the same conversational turn (cf. examples (1) through (3) above). Testing the Demographic Hypothesis. The hypothesis is not supported; there is no indication that a speaker’s sociolinguistic profile correlates with his or her use of English elements in the study’s context. That is, Pearson correlations show no significant relations at all between frequency of any of the English variables and either gender, age, or education. The only indication that any of these factors is important within the terms of this study is that a post hoc ANOVA finds that gender is significant at the .05 level as a factor differentiating the 48 speakers into three groups. The reason for this reading of significance surely is the fact that six of the eight speakers in the Front Runners group are women. That the demographic factors are not predictive of frequency of English elements implies that other factors, possibly the psychology of what bilingualism means to each individual, are important. That is, an expectation that the degree of one’s own possibility of socio-economic mobility may be critical to predicting English use. Testing the Convergence Hypothesis. This hypothesis is very weakly supported, but the only support is a small handful of examples. These indicate Xhosa L1 speakers are beginning to use English prepositions in locative constructions so that, in effect, the English preposition doubles a Xhosa locative marker. Thus, there are examples such as (4). The English preposition to seems to be functioning in place of a Xhosa locative form. True, the prefix on town is a locative marker, but indications are that the form e-town has been lexicalized as a single unit. Note that the verb (even though it comes from English) receives the derivational suffix el that typically appears with a locative predicate.
system morpheme under the 4-M model, not one of those morphemes defined here as critical morphemes (late system morphemes under the 4-M model).
Grammatical stairs in language shift
(4) si-mov-el-e to e-town. 1pl-move-appl-subjunct loc-town ‘[that] we should move to town’
Otherwise, there are no indications that the Xhosa sections of this corpus show any convergence to English. In some ways, this lack of convergence is a surprise, and it may be that further analysis will uncover some ways in which there is more subtle convergence. It is likely that these speakers did not learn Xhosa and English at all simultaneously. Given what is known about primary education when these speakers were in the lower grades (MacDonald 1990), they may not have had sufficient exposure to English to have learned it in primary school. Many secondary schools also had poor instruction in English and that state of affairs still persists in some areas. Thus, many in this sample probably developed any real fluency in English on the job. That is, the two languages may be separate phenomena both from the psycholinguistic (age of acquisition) and the sociolinguistic standpoint. Given the lack of obvious convergence to English structures, it is interesting that extensive codeswitching occurs in this corpus. This suggests that codeswitching is not a “mixing” of languages, as is a popular belief. There are many examples of clauses that contain morphemes from both English and Xhosa, but always with Xhosa supplying the grammatical frame; further analysis of these clauses is planned. An informal examination of the corpus shows that these speakers display a limited range of the Xhosa lexicon; for example, the word into ‘thing’ appears over and over, covering a number of different meanings. But this lack of lexical variety may be a function of the interview context. Still, the speakers construct complex sentences in Xhosa; for example, embedded clauses with relative pronouns or complementizers are very frequent. If anything, many of their English clauses are fairly simple and formulaic, but conclusions are premature and more analysis is needed.
4. Conclusions This study makes claims and reports on findings about the grammatical nature of steps that lead to language shift. The findings have potential general theoretical interest, especially if they stimulate research in other venues to see if the results reported here can be replicated elsewhere. In this conclusion, only two findings are highlighted. First, a two-step statistically-based Cluster Analysis shows that the 48 Xhosa-English bilinguals studied divide into three separate groupings based on their use of a set of English elements in informal interviews. That is, bilinguals who are relatively similar in their sociolinguistic profiles fall on different
79
80 Carol Myers-Scotton
‘bilingual steps’ in terms of their use of English. To get a picture of the differences across speakers, consider these findings: The Front Runners (8/48) have a minimum of 21 and a maximum of 100 English-framed clauses. This contrasts with the Middle Pack 12/48 which has a range of 1 to 53 English-framed clauses, and the Slow Starters (28/48) with a range of 0 to 12 English-framed clauses. Producing full clauses in English in a context favoring Xhosa means that these speakers are able to move from using only occasional English elements (i.e. English loan words or codeswitches) in well-formed clauses in only one language (Xhosa) to producing other clauses in another language (English) as well. Still, among the Front Runners, who lead the pack in using English, Xhosa clearly dominates in their speech, at least in this informal interview context. A comparison shows that while these speakers have many monolingual English clauses, they have many more that are framed by Xhosa (either monolingual Xhosa clauses or Xhosa-framed codeswitching clauses). Three of the Front Runners (two females and one male) have English as the framing language for 30% of their total clauses. Three others (two females and one male again) show a lower percentage of English-framed clauses; the percentages range from 12 to 21. How do we relate these findings about the dominance of Xhosa and the expectation that these bilinguals ‘ought to’ be shifting to English if they are going to move up the socio-economic stairs? Some explanations are presented, but exploring them fully is beyond the scope of this paper. First, recall that the interview was biased for speakers to use Xhosa, not English because the interviewer spoke mainly Xhosa and asked questions about the speakers’ views on everyday life. Second, as already indicated, expectations about L1 attrition of grammatical morphemes that are often found in the literature are not supported by data from adults; that is, it seems speakers can maintain an L1 in its full, or near full, form even though they largely speak a socio-economically prestigious L2 acquired much later than early childhood. Third, there is a sociological explanation. In the South African society, that ethnic group membership has instrumental as well as sentimental value should not be overlooked. In “the new South Africa” (Apartheid rule was only overthrown in 1995), the major re-apportioning of political and economic power is still going on. Thus, the prudent citizen ‘knows’ his or her ethnic group because of the competition between groups for the instrumental assets of not just political office, but for the asset that matters most to individuals: jobs. Also, even though these speakers work in cosmopolitan cities where they inevitably must speak a good deal of English during the work day, they associate largely with ethnic brethren and travel to their home areas at times, at least for funerals. Finally, a cautionary comment is in order: the study only documents the speakers’ language choices in one context, an informal interview with a fellow Xhosa-English bilingual, who was known to the speakers before the interview, even though he/she
Grammatical stairs in language shift
was not a personal friend. It is entirely possible that some of the speakers studied have already shifted to English as their main truly public language – as the language they use all day long in their workplaces. Whatever the case, the goal of this study was not to determine who had shifted to English and who had not; rather it was to test hypotheses about the grammatical nature of bilingual language use by potential shifters and differences in the types of English elements used and their frequency across speakers. Another question, of course, is why are sociolinguistically-similar bilinguals on different steps in their use of a dominant L2, given the presumption that heavy users are on their way to making the L2 their dominant public language? A costbenefit analysis provides one answer. That is, heavy users of the L2 are often persons who perceive that the L2 can offer them more benefits than costs; further, they are people whose expectations have an objective basis, given the available evidence (e.g., their education level, their job prospects). In addition, speakers’ personal goals matter, too. Second, the study supports the abruptness hypothesis that is related to the predictive importance of producing L2 elements when the L1 is favored. The sprinkling of many English elements in the speaker’s Xhosa might be taken as evidence that a gradual, not an abrupt, shift is going on. But looked at from the grammatical point of view, the evidence supports the hypothesis that any ongoing shift is abrupt, not gradual. Yes, Xhosa-English bilinguals in this study produce many clauses that show codeswitching between the two languages. But the point to notice is that these clauses never contain any critical grammatical morphemes from English (with the exception of those in embedded English phrases, EL islands). Further, there is no attrition of Xhosa inflectional morphemes. That is, no matter how many English nouns or verbs or discourse markers speakers employ, Xhosa continues to supply the critical morphemes of the morpho-syntactic frames of these numerous clauses. Codeswitching may be a step on the way to shift, but codeswitching itself does not signal that shift is inevitable. Viewed in terms of the data studied here, the conclusion is that it is only when speakers shift from Xhosa-framed clauses to producing many English-framed clauses – clauses framed by critical morphemes from English – that we can speak of the real onset of shift to English as the speaker’s primary public language. The data here are only suggestive of numbers, but a conservative estimate is that if the L2 frames 30% of clauses in an informal context where either the L1 or L1 framed codeswitching frequently is used, then shift is on the way, or has already happened.
81
82
Carol Myers-Scotton
Acknowledgement Research was conducted under a National Science Foundation grant BCS0424829 to Carol Myers-Scotton as the principal investigator. Support is gratefully acknowledged. Thanks go to Laurie Sanders for assisting in data analysis, to Roumen Vasselinov for statistical analyses, and to Nozibele Nomdebevana, the primary interviewer.
References de Bot, K. & Clyne, M. (1994). “A 16 year longitudinal study of language attrition in Dutch immigrants in Australia”. Journal of Multilingual and Multicultural Development, 15, 17– 28. Dorian, N. (1981). The Life Cycle of a Scottish Gaelic Dialect. Philadelphia, PA: University of Pennsylvania Press. Gross, S. (2000). The Role of Abstract Lexical Structure in First Language Attrition: Germans in America. Unpublished Doctoral Dissertation, Columbia, SC: University of South Carolina. Hlavac, J. (2003). Second Generation Speech, Lexicon, Code-Switching and Morpho-Syntax of Croatian-English Bilinguals. Frankfurt: Peter Lang. MacDonald, C. A. (1990). Crossing the Threshold into Standard Three: Main report of the threshold project. Pretoria: Human Resources Research Council. Myers-Scotton, C. M. (2002). Contact Linguistics: Bilingual encounters and grammatical outcomes. Oxford: OUP. Myers-Scotton, C. M. (2005). “Supporting a differential access hypothesis: Code switching and other contact data”. In J. F. Kroll & A. M. B. de Groot (Eds.), Handbook of Bilingualism: Psycholinguistic approaches (pp. 236–348). New York, NY: OUP. Myers-Scotton, C. M. & Jake, J. L. (2001). “Explaining aspects of codeswitching and their implications”. In J. Nichol (Ed.), One Mind, Two Languages: Bilingual language processing (pp. 84–116).Oxford: Blackwell. Myers-Scotton, C. M. & Jake, J. L. (2007). “Uniform structure in code switching, principles and processes”. In B. Bullock & A. J. Toribio (Eds.), Handbook of Codeswitching. Cambridge: CUP. Schmid, M. S. (2002). First Language Attrition, Use and Maintenance: The case of German Jews in Anglophone countries. Amsterdam: John Benjamins. Schmid, M. S. (2004). “First language attrition: The methodology revised”. International Journal of Bilingualism, 8 (3), 239–255.
First language attrition from a minimalist perspective Interface vulnerability and processing effects Ianthi Maria Tsimpli
Aristotle University of Thessaloniki, Greece
The present paper deals with first language attrition under the influence of a second language. Two sets of data are presented: the first is offline data from a variety of tasks aiming to address the question of selective vulnerability of interfaces as opposed to the computational system underlying syntactic competence. This data is drawn from the study of Tsimpli, Sorace, Heycock and Filiaci (2004). The second set of data is drawn from a pilot study (Kaltsa 2006), involving an online grammaticality judgment task testing Case on determiners. This data is argued to address the possibility of syntactic attrition of a formal feature, i.e. case, in the domain of syntactic processing alone indicating that performance rather than competence problems are also involved. Keywords: selectivity; minimalism; interfaces; interpretability; Greek
Introduction It has been repeatedly observed that attrition is a phenomenon that selectively affects language, indicating certain areas of linguistic knowledge as more vulnerable than others. Although L1 attrition studies presuppose the attainment of a mature state of L1 knowledge, it is important to point out the significance of accounting for incomplete L1 learners in attrition studies. Incomplete learning is directly associated with the onset of bilingualism and the onset of attrition (Sorace 2004). In the present paper, the offline data from Tsimpli, Sorace, Heycock & Filiaci’s (2004) study comes from L1 speakers of Greek or Italian, born and raised in Greece and Italy respectively, who acquired English after childhood. These groups of speakers are clearly excluded from the set of incomplete L1 learners since the onset of L2 learning is after childhood. On the other hand, the online data presented
84
Ianthi Maria Tsimpli
from Kaltsa’s (2006) study comes from L1 speakers of Greek, born and raised in Sweden and Germany. L2 acquisition of Swedish or German began in childhood; thus, these cases involve early bilinguals whose attrition effects in their L1, Greek, may also involve age-related factors as well as incomplete L1 acquisition from the point where L2 exposure began. As far as selective vulnerability of language areas in attrition is concerned, the lexicon has been the most frequently reported candidate, in terms of lexical retrieval and access (Isurin 2000; Kohnert, Bates & Hernández 1999; Köpke 2002, Olshtain & Barzilay 1991, a.o.). Pragmatic transfer of L2 concepts to L1 has also been documented (Pavlenko 2000), as well as self-reported claims of general loss of pragmatic abilities (Canagarajah 2001). More recently, research on potentially affected areas within the domain of morphosyntax has been presented. L2 influence on L1 morphosyntax has been found, although not as extensive as in the lexical domain, in relation to case-, numbermarking and gender-assignment (Anderson 1999; Håkansson 1995; Schmid 2000), preposition choice (Schmid 2002), pronominal use (Armon Lotem 2000; Lipski 1996; Satterfield 1999), verbal morphology (Montrul 2002), along with word-order rules and subcategorization (Boyd & Andersson 1991; Waas 1996). Some of the above phenomena, such as pronominal use or word-order flexibility, are areas of syntax which interface with semantic and pragmatic properties. In this respect, they could be distinguished from syntax proper which does not seem to be affected by context or discourse factors. Gürel (2004) examined the selective nature of L1 attrition through data collected from Turkish speakers who have been long-term residents in an English speaking community. The phenomenon investigated is pronominal use and binding, and the findings suggest that the L2 grammatical options reconfigure the L1 options and, hence, affect the level of syntactic competence. Notice, however, that pronominal use and, in particular, the use of the appropriate pronominal form in the appropriate discourse context involves the combination of morphosyntactic and discourse constraints; as such it qualifies as a (syntax-discourse) interface phenomenon. Montrul (2004) examines L1 attrition of Spanish in the data of 24 Spanish heritage speakers involving subject and (direct and indirect) object use which is determined by syntactic, along with semantic and pragmatic features. It was hypothesized that the heritage speakers “should display robust knowledge of null subjects as well as object clitics, but variable behaviour in the pragmatic distribution of null vs. overt subjects […] and cases of semantically-based dative cliticdoubling” (p. 125). The results confirmed the hypothesis while adding that no severe loss was observed, and, thus, offered further evidence for the permeability of interfaces between syntax and the discourse-pragmatic field.
First language attrition from a minimalist perspective
In bilingualism too, there has been recent evidence that interface phenomena are more vulnerable than syntax proper. Serratrice, Sorace & Paoli (2004) investigated the distribution of subjects and objects in the Italian and English of a bilingual child (1;10–4;6). The research hypothesis was that a crosslinguistic one-way influence would occur subsequent to the instantiation of the C-system resulting in the asymmetrical use of null subjects and objects. The findings corroborate the prediction. Given that a selective pattern of affected language areas has been found in attrition studies but also in bilingualism, the question is whether this selective vulnerability can be accounted for by the architecture of the language system and the resulting susceptibility of interfaces, as opposed to other areas, in cases where ‘language contact’ in the mind of an individual is involved. Minimalism (Chomsky 1995, 2004) offers the theoretical assumptions to evaluate the possibility of L1-L2 contact and crosslinguistic influence on interface and non-interface phenomena.
1. Minimalism According to the minimalist program (Chomsky 1995, 2004), the language faculty includes the computational system where UG-based operations such as Merge, Agree or Move apply, depending on language-specific properties of formal features such as Inflection, and two interface levels, namely PF (Phonetic Form) and LF (Logical Form). The PF representation provides information visible to the articulatory-perceptual systems (A-P), while the LF representation includes the semantic information contained by grammatical features of lexical items and is visible to the conceptual-intentional systems (C-I). Interpretability is a property that applies to features at each interface: PF-interpretable features are those visible to the A-P systems; they include a phonetic matrix and are subject to constraints of the relevant type. LF-interpretable features are those with some semantic content that is relevant to the LF interface. At the interfaces, only interpretable features can be found. Uninterpretable features play an important role in grammatical derivations but should be deleted prior to the derivation reaching the interface in order for the representation to converge. L1 acquisition is a procedure of parameter setting which involves the selection of features and their values in a way that leads to syntactic language variation (Chomsky 1986, 1995). Assuming that the computational system of language is impenetrable by other modules, unlike PF and LF which, by definition, can interface with external systems and be read by them, we could regard attrition as a process that cannot affect syntax proper and parametric choices of the computational domain. On the other hand, features which are interpretable at either
85
86 Ianthi Maria Tsimpli
PF, LF or both, are vulnerable in the mind of a speaker of two languages. In such cases, attrition may surface as indeterminacy or optionality in the use of interpretable morphosyntactic features in contexts where interface conditions in L1 and L2 differ (Sorace 2004; Tsimpli et al. 2004). The use of overt subject pronouns in null and non-null subject languages is an appropriate example of attrition effects at the syntax-discourse interface: although overt subject pronouns are found in both types of languages, the distribution in null subject languages is conditioned by discourse features such as [focus] or [topic-shift]. In other words, their use is based on the integration of syntactic and pragmatic/discourse information (Platzack 2001; Rizzi 2002). Accordingly, investigating features relevant to interfaces as compared to features relevant to the syntactic derivation alone can be used to examine selective attrition patterns in bilingual populations. Since the production and interpretation of a sentence involves grammatical features which differ in interpretability and, hence, processes affecting interface or non-interface levels, we can concentrate on features with opposite values in L1 and L2 and examine whether grammaticality judgments, interpretations, and syntactic processing of the L1 may be influenced by the L2 (cf. Sanz & Bever 2001). Crucially, online syntactic processing is different from offline comprehension, where time-constraints and automaticity are not necessarily involved, with respect to the effects that feature identification and parsing have. It has been shown that online data can reveal differences between native-like grammatical competence vs. performance (see Juffs & Harrington 1995, 1996). In this respect, there have been studies of very advanced L2 learners who exhibit native-like performance in offline grammaticality judgment tasks as well as spontaneous use, but differ from natives in online syntactic processing (Clahsen & Felser 2006; Felser, Roberts, Marinis & Gross 2003; Papadopoulou & Clahsen 2003; Papadopoulou 2006; Papadopoulou & Tsimpli in press). This contrast indicates that the online tasks do not just require knowledge of grammar but a certain degree of automaticity at the morpho-phonological interface (PF) which is the basic input level for the language processor. The data from advanced L2 learners indicates that lack of automaticity in acquired grammatical knowledge may signal L1 interference effects which lead to non-native-like performance. If this line of reasoning is correct, investigating L1 attrition effects in L1 syntactic processing may reveal that interference is found in cases of bilingual or L2 learners in more than one direction, e.g. L1 to L2 or L2 to L1. Note, furthermore, that if such effects are found in online processing of purely syntactic features (i.e. not just interface phenom-
First language attrition from a minimalist perspective
ena) such as case, then attrition also affects the way in which the syntactic parser ‘reads’ the morphological interface level, i.e. PF.
2. Interface and syntactic features studied The offline data comes from an L1 attrition study which includes Greek and Italian speakers of English with near-native competence in the L2 (the data has been published in Tsimpli et al. 2004). In this paper, production and interpretation data from Greek near-native speakers of English is presented. In the above study, we concentrated on subjects in production and interpretation. In particular, the investigation included (a) null and overt subject pronouns and (b) definite and indefinite preverbal and postverbal DP subjects. The motivation was the parametric difference between the L1s (Greek or Italian) and the L2 English with respect to the null subject property. As is well-known, null subject languages like Greek and Italian use null subject pronouns as the unmarked option whereas overt subject pronouns are used only when some discourse feature, such as [focus] or [topic-shift] motivates them. Thus, the syntactic availability of null subjects interacts with a syntax-discourse interface property, namely pronominal interpretation in particular discourse contexts. In the present paper we concentrate on data involving the second phenomenon, i.e. (in)definite preverbal and postverbal DP subjects. Null subject languages allow both preverbal and postverbal subjects (Rizzi 1982) whereas English, a nonnull subject language, does not (e.g. Cardinaletti 1997, for Italian; Philippaki- Warburton 1987 for Greek, among many others). The syntactic availability of postverbal subjects interacts with interface interpretations as well. Specifically, indefinite subjects appearing in postverbal position are supposed to be ambiguous between ‘old’ and ‘new’ reference, whereby ‘old’ implies reference to an antecedent already available in discourse, and ‘new’ implies reference to some entity not mentioned in the discourse as yet. Thus, in the scenario presented in (1) below, the preverbal subject is expected to refer to one of the twins, whereas the postverbal subject is neutral between reference to one of the twins or some other baby:
. In the generative framework, morphological features have a dual status: an abstract representation in the Lexicon which is fed into the derivation and motivates operations such as Agree or Merge, and an interface realization which involves constraints on spell-out which are also subject to phonological and phonetic language-specific properties. . The study was supported by ESRC grant R000239158.
87
88
Ianthi Maria Tsimpli
(1) I ghitonisa mu ston trito orofo apektise dhidhima. The neighbour my in-the third floor had twins “My neighbour on the third floor had twins.” a. b.
Xtes vradhi ena moro ekleje. Yesterday evening a baby was-crying Xtes vradhi ekleje ena moro. Yesterday evening was-crying a baby
Moreover, unaccusative predicates favour the postverbal position especially when the subject is indefinite, hence the contrast in acceptability between sentences such as (2a) and (2b) below (cf. Belletti 1988; Pinto 1997): (2) a. Irthan kapji fitites. arrived some students b. #Kapji fitites irthan. some students arrived
The preference for postverbal indefinites with unaccusatives is associated with the standard assumption that unaccusative predicates have a single internal argument (Burzio 1986). Thus, the postverbal position is the underlying, non-derived position of the verb’s argument. Moreover, indefiniteness of the argument strengthens the preference for the postverbal position given that indefinite DPs rarely make good topics. Recall that topichood, or ‘old’ reference is linked to the choice of preverbal subjects in null subject languages, hence the contrast between (2a) and (2b). Overall, then, the syntactic availability of postverbal subjects is affected by (a) the definiteness of the subject DP, (b) by discourse reference (e.g. (1)) and (c) by thematic properties of the predicate (e.g. (2)). On the assumption that L1 attrition affects interface constraints rather than L1 syntactic options, in the above study we expected Greek attrited speakers to still have postverbal subjects as a syntactic option in their Greek grammar and to interpret them accordingly, i.e. denoting either ‘old’ or ‘new’ reference. However, we also expect these speakers’ grammar to ‘relax’ the interface conditions that regulate the interpretation of an indefinite preverbal subject as a topic denoting ‘old’ reference, fluctuating between both referential options for preverbal subjects. As far as production of indefinite subjects with unaccusative predicates is concerned, we expected attrited Greek grammars to be inconsistent, using indefinite subjects either pre- or post-verbally. This expectation is based on the assumption that, although unaccusativity is a feature shared by both languages, the interaction of definiteness, unaccusativity and postverbal subject position is an L1 property. Thus, attrited Greek speakers were expected to produce both preverbal
First language attrition from a minimalist perspective
and postverbal indefinite subjects as ‘new’ information, in the same way that they were expected not to distinguish between the interpretation of indefinite subjects in pre- or postverbal position.
2.1 The offline data Starting with the offline data, we will concentrate on the results from two tasks: a production task and an interpretation task from Greek speakers of English with near-native competence in the L2 (originally published in Tsimpli et al. 2004). The participants form two groups: a control group (n = 20) of native speakers of Greek living in Greece with minimum or no knowledge of English, and an experimental group (n = 19) of Greek near-native speakers of English living in Britain. The near-native speakers have been living in Britain for a minimum of six years and were using both the L1 Greek and the L2 English on an everyday basis. They were tested for near-nativeness using White & Genesee’s (1996) criteria (for details see Tsimpli et al. 2004). The Headlines task was the production task used for the elicitation of sentences which included a verb, a definite or indefinite subject DP and an adverbial expression. The aim was to investigate production of preverbal and postverbal subjects in relation to the definiteness condition. The constituents were presented scattered on the computer screen and the participant was asked to form a sentence with them starting with the expression Did you hear that…. The phrases were presented together with a picture depicting the sentence formed. The idea was that the participant would imagine being on the phone to a friend and announcing the ‘headlines’ by putting together the phrases presented on the computer screen. The lack of context or discourse in this task formed an all-focus context where all constituents would be presented as ‘new’ information. Thus, controls, unlike near-natives, were predicted to produce overall more postverbal subjects. Half of the sentences (total n = 30) included a definite and half an indefinite subject DP. An example is presented in (3) below: (3) Akuses oti katerefse ena ktirio eksetias ekriksis? Heard-2s that collapsed a building because explosion-gen “Did you hear that a building collapsed because of an explosion?”
The results from the production of preverbal definite and indefinite subjects in this task are presented in Figure 1 below (from Tsimpli et al. 2004). The two groups differed significantly in the production of preverbal definite subjects (F(1,37) = 9.848, p = .003). An interaction between Definiteness and Group was also found (F(1,37) = 4.018, p < .05). These results point to the attrited
89
90 Ianthi Maria Tsimpli
Figure 1. Production of preverbal definite and indefinite subjects per group
group’s change in the direction of preference for postverbal subjects in the all-focus context, thus supporting our initial prediction. Note, however, that postverbal subjects are produced by the attrited group too, indicating that the syntactic choice is still active in their L1 grammar. The second task was an interpretation task which involved presentation of a context and a test sentence such as (1a) or (1b) above. The sentence was accompanied by a set of pictures from which the participant had to pick the one that matched the sentence best. Participants also had the choice of picking two pictures as being good matches for the sentence. Choosing the third, irrelevant picture was classified as a ‘non-valid’ answer. In Figures 2 and 3, the results from the interpretation of preverbal and postverbal indefinite subjects respectively, are presented. The results show a significant difference between the two groups in the choice of ‘new’ referent (F(1,37) = 12.14, p < .001) and in the choice of ‘both’, i.e. ‘old’ and ‘new’ referents (F(1,37) = 8.612, p < .006). There is also an effect of Reference (F(3,37) = 21.62, p < .0001), an interaction between Reference and Group (F(3,37) = 5.512, p < .001) and an interaction between subject position and Reference (F(3,111) = 9.47, p < .0001). These results indirectly show indeterminacy in the interpretations of the near-native group in that the choice of ‘both’ referents being possible with preverbal subjects is in the right direction, as far as our prediction regarding attrition effects is concerned. . According to an anonymous reviewer the controls’ choice of ‘new’ referent in the preverbal condition is not what we expected initially on the assumption of the topic status of preverbal subjects. Still, the difference between ‘old’ and ‘new’ referent in the choice of the control group (shown also by the interaction between subject position and reference) is found. Furthermore, although the total number of responses of ‘both’ and ‘new’ referents seems to be similar in the
First language attrition from a minimalist perspective
Figure 2. Referent interpretation of preverbal subjects per group
Figure 3. Referent interpretation of postverbal subjects per group
The indeterminacy in the interpretation of postverbal subjects by the attrited group is shown in Figure 3 where the two groups show a significant difference in the preference of ‘both’ (F(1,37) = 7.243, p < .01). Although this difference was not part of our predictions since postverbal subjects are not associated with one or the other interpretation in the monolingual grammar – which is also indicated by the similar degree of preference for ‘old’ and ‘new’ referent by the controls – the high number of ‘both’ in the attrited group can be viewed as a side-effect of the indeterminacy associated with preverbal subjects in Figure 2. A difference between control and the near-native groups, it is important that indeterminacy characterizes the ‘attrited’ group alone.
91
92
Ianthi Maria Tsimpli
the two groups is also found in the preference for the ‘old’ reference (F(1,37) = 6.962, p < .02). Overall then, the data from the offline production task seem to show stronger results in the expected direction: more preverbal subjects are produced in contexts where the non-attrited grammar strongly favours postverbal subjects. The interpretation task, however, shows a weaker tendency towards indeterminate interpretations at the level of discourse reference. The contrast between preverbal as opposed to postverbal subjects and reference is not as clearly shown as was initially expected, possibly because of the complexity of the task (participants had to match a sentence with pictures, two of which differed only minimally) and also because more than one preference was allowed to be indicated for each sentence.
2.2 The online data The processing data comes from four adult native speakers of Greek two of whom were born in Sweden and two in Germany, all born to Greek parents who were immigrants in the above countries (see Kaltsa (2006) for details). All participants were raised in Sweden and Germany respectively, and were early bilingual Greek/ Swedish or Greek/German speakers, which sets them apart from the group of subjects who participated in the study reported in the previous section. As shown in Table 1, all participants in the online study had been exposed to the L2 either early in childhood or from birth, despite the fact that both parents spoke only Greek to them. Exposure to L2 was primarily due to the nursery environment or, in the case of GK and EL who were exposed to German from birth, this was due to German-speaking care-takers. As for the differences between participants with respect to the home language(s) used, EV and GK also heard Swedish and German, respectively, at home because of older bilingual siblings. At the time of testing all participants had moved permanently to Greece in the last 2–4 years but considered Swedish or German to be their dominant language. The task used was a speeded online grammaticality judgment task which aimed at testing case on definite and indefinite DP arguments (Papadopoulou & Tsimpli, in prep.). The definiteness and indefiniteness of the article and the grammatical and ungrammatical use of it comprise the four conditions:
. Kaltsa’s study included ten participants. Two of them are children and were excluded for this reason, and the remaining four showed low accuracy in the fillers of the test presented here. Kaltsa also tested all subjects on another online experiment (Papadopoulou & Tsimpli 2004) involving subject-object ambiguity resolution through morphological cues.
First language attrition from a minimalist perspective
Table 1. Information on participants Participant
AK
EV
GK
EL
Age L2 Age of exp. to L2 Home Language Language of Schooling
23 Swedish 4;0 Gr Swed
25 Swedish 3;0 Gr/Swed Swed
21 German from birth Gr/German Ger: kindergarten Gr: school 3
21 German from birth Gr Ger/Gr
Yrs of residence in Greece 4
3
2
(4) a. Tin kopela agapa o giatros the-acc girl loves the-nom doctor “The doctor loves the girl”
Definite-Grammatical
b. *Tin kopela agapa ton giatro the-acc girl loves the-acc doctor “*The doctor loves the girl”
Definite-Ungrammatical
c. Tin kopela agapa enas giatros The-acc girl loves a-nom doctor “A doctor loves the girl”
Indefinite-Grammatical
d. *Tin kopela agapa enan giatro the-acc girl loves a-acc doctor “*A doctor loves a girl”
Indefinite-Ungrammatical
All sentences were in the OVS order, thus beginning with an accusative-marked definite DP. In the ungrammatical conditions, the first offending element was the fourth segment, the determiner (definite or indefinite) which was also marked accusative. The noun always agreed with the determiner in case. Instructions were in Greek and a training session preceded the actual experiment. There were 24 test sentences and 72 fillers, half of which were grammatical and half ungrammatical. Each word remained on the screen of the computer for 500 ms and immediately after the next word appeared. At the end of each sentence the participants were asked to press a button in order to indicate whether the sentence was grammatical or ungrammatical within 2000 ms. Responses that took longer than 5000 ms were not recorded. All four subjects were over 80% accurate in judging filler sentences as grammatical or ungrammatical. The results from their accuracy in the critical items are presented in Table 2. A control group consisting of twenty native speakers of Greek was also included. As shown in Table 2, all monolingual native controls performed equally well in all conditions. Similarly, AK showed no significant differences between grammatical or ungrammatical items in the definite or the indefinite condition. Of
93
94 Ianthi Maria Tsimpli
Table 2. Accuracy (%) in each condition per subject Def-Gram Def-Ungram Indef-Gram Indef-Ungram
AK
EV
GK
EL
Controls
100 100 88 88
100 100 50 100
75 100 63 100
88 88 75 100
92 95 92 95
the remaining three participants, EV’s data showed a significant interaction between determiner type and grammaticality (x2 = 9.722, p < 01) and a significant difference between Indef-Gram and Indef-Ungram (x2 = 16,667, p < 001). GK’s data reveals a significant difference between Indef-Gram and Indef-Ungram (x2 = 8,399, p < 01) but also between Def-Gram and Def-Ungram (x2 = 3,571, p = 059). Finally, EL showed a significant difference between Indef-Gram and Indef-Ungram (x2 = 3,571, p = 059). Given that ungrammatical items in this experiment exhibit robust ungrammaticality and that all test sentences include few segments, being simple objectverb-subject structures, it is of no surprise that accuracy in the judgments of the controls is high in all conditions. At the same time, it is interesting that three out of four participants show problems in judging the grammatical sentences, with more obvious effects in the indefinite condition. These results point to a number of possibilities that lead to poorer performance. First, it is possible that these speakers have native competence in both L1 Greek and L2 Swedish or German, but ‘dominance’ of the L2 interferes with L1 online processing. Note that these speakers had no problems judging ungrammatical sentences as such, a characteristic which is not shared by typical L2 learners. The latter typically consider more ungrammatical sentences as grammatical, as shown in a number of studies. In the four speakers’ data presented here it seems that L2 interference gives rise to attrition in the ability to judge a formal feature on a functional category, the determiner. Assuming that the PF-interface, which includes morpho-phonological feature realisations, is the input to morphological and syntactic processing, we can argue that language dominance can be translated into attrition effects at interface levels.
. As an anonymous reviewer points out the problem in judging grammatical sentences could be attributed to attrition of a syntax-discourse interface phenomenon. In particular, the OVS sentence with an indefinite object and a definite subject may be considered problematic under the influence of Swedish or German given that the ‘new’ information expressed by the indefinite object precedes ‘old’ information. Although this sounds a possible alternative explanation, I believe that it is also possible in Swedish or German to prepose a ‘focused’ object rather than only a topicalised one. In the case of focus, the object could be new, exhaustive or contrastive
First language attrition from a minimalist perspective
It is important to note that, in spontaneous production, these speakers use case morphology in Greek like natives; their own perception of difficulties in the use of Greek concerns primarily word-finding and communicative rather than grammatical aspects of language. It is on these grounds that they justified their claim that they are dominant in the L2. This implies that L2 interference affects offline processes that involve ‘higher’ levels of language use as well, such as encyclopaedic and lexical knowledge, pragmatic use and complexity in language expressivity. Moreover, grammaticality judgments also involve meta-linguistic abilities, which, in the case of adult speakers, are expected to reflect to a large extent language competence, especially in the present study where a grammatical feature such as case is involved. It is also possible, however, that L2 dominance and subsequent attrition effects on L1 judgments is not the only possible explanation for the attested performance in the online task. It could be argued that the Greek input these speakers received in early childhood was not sufficient in quantity or quality (primarily due to code-mixing) to attain native competence in Greek. If this is the case, then L1 acquisition is incomplete in these learners and, as a result, L2 dominance and subsequent interference effects are expected on an L1 grammar, which has not attained a mature state. In any case, evaluation of the two possible accounts for the attested performance, namely L1 attrition due to L2 dominance or incomplete L1 acquisition affecting online morpho-syntactic processing, requires further extensive testing of these participants.
3. Concluding remarks L1 attrition remains, among other issues, an important psycholinguistic question, with respect to (a) the linguistic domains which seem to be affected by it and (b) the degree of attainable generalisation of attrition effects to other language contact situations in the mind of a speaker with two languages, regardless of language direction and type of bilingualism. The present paper addresses the first question starting from the theoretical position that L1 attrition may only affect interface properties and not pure syntactic L1 options (Sorace 2004; Tsimpli et al. 2004). This claim is primarily based on the minimalist assumptions regarding the architecture of the language faculty which allows the computational system alone to be modular, while interface levels are interactive in nature and, as such, vulnerable to L1/L2 language interference. information and in this case, it could be indefinite. As such, it should not be unacceptable in first position.
95
96 Ianthi Maria Tsimpli
Interference, a case of which is attrition, leads to indeterminacy or optionality in the use of interface properties of the L1 under the influence of the L2. The offline data from Tsimpli et al.’s (2004) study indicates that this claim is in the right direction. The online data from Kaltsa’s (2006) study, on the other hand, indicates that even grammatical features such as case may be subject to attrition effects in speeded online judgment tasks. The implication is that, apart from the syntaxdiscourse interface, other interface levels which serve as input to performance systems, such as the parser, may be subject to attrition effects on grammatical features.
References Anderson, R. T. (1999). “Loss of gender agreement in L1 attrition: Preliminary results”. Bilingual Research Journal, 23 (4), 319–338. Armon Lotem, S. (2000). “Language attrition: Why are resumptive pronouns so susceptible?” In S. C. Howell, S. Fish, & K. Thea (Eds.), Proceedings of the 24th Annual Boston University Conference on Language Development (pp. 58–67). Somerville, MA: Cascadilla. Belletti, A. (1988). “The case of unaccusatives”. Linguistic Inquiry, 19, 1–34. Boyd, S. & Andersson, P. (1991) “Linguistic change among bilingual speakers of Finnish and American English in Sweden: Background and some tentative Findings.” International Journal of the Sociology of Language, 90, 13–35. Burzio, L. (1986). Italian Syntax: A Government-Binding approach. Dordredt: Reidel. Canagarajah, S. (2001). “The fortunate traveller: Shuttling between communities and literacies by economy class”. In D. Belcher & U. Connor (Eds.), Reflections on Multiliterate Lives (pp. 23–37). Clevedon: Multilingual Matters. Cardinaletti, A. (1997). “Subjects and clause structure”. In L. Haegeman (Ed.), New Comparative Syntax (pp. 33–63). London: Longman. Chomsky, N. (1986). Knowledge of Language: Its nature, origin and use. New York, NY: Praeger. Chomsky, N. (1995). The Minimalist Program. Cambridge, MA: MIT Press. Chomsky, N. (2004). “Beyond explanatory adequacy”. In A. Belletti (Ed.), Structures and Beyond: The cartography of syntactic structures, vol. III (pp. 104–131). Oxford: OUP. Clahsen, H. & Felser, C. (2006). “Grammatical processing in language learners”. Applied Psycholinguistics, 27, 3–42. Felser, C., Roberts, L. Marinis, T., & Gross, R. (2003). “The processing of ambiguous sentences by first and second language learners of English”. Applied Psycholinguistics, 24, 453–489. Gürel, A. (2004). “Selectivity in L2-induced L1 attrition: A psycholinguistic account”. Journal of Neurolinguistics, 17, 53–78. Håkansson, G. (1995). “Syntax and morphology in language attrition: A study of five bilingual expatriate Swedes”. International Journal of Applied Linguistics, 5 (2), 153–171. Isurin, L. (2000). “Deserted islands or a child’s first language forgetting”. Bilingualism: Language and Cognition, 3 (2), 151–166.
First language attrition from a minimalist perspective
Juffs, A. & Harrington, M. (1995). “Parsing effects in L2 sentence processing: Subject and object asymmetriesin wh-extraction”. Language Learning, 46, 286–324. Juffs, A. & Harrington, M. (1996). “Garden path sentences and error data in second language sentence processing research”. Studies in Second Language Acquisition, 17, 483–516. Kaltsa, M. (2006). First Language Attrition in Greek. Unpublished BA thesis. Thessaloniki: Aristotle University. Köpke, B. (2002). “Activation thresholds and non-pathological L1 attrition.” In F. Fabbro (Ed.), Advances in the Neurolinguistics of Bilingualism (pp. 119–142). Udine: Forum. Kohnert, K., Bates, E., & Hernández, A. (1999). “Balancing bilinguals: Lexical-Semantic production and cognitive processing in children learning Spanish and English”. Journal of Speech, Learning and Hearing Research, 42, 1400–1413. Lipski, J. (1996). “Patterns of Pronominal evolution in Cuban-American bilinguals”. In A. Roca & J. Jensen (Eds.), Spanish in the United States (pp. 155–173). Berlin: Mouton. Montrul, S. (2002). “Incomplete acquisition and attrition of Spanish tense/aspect distinctions in adult bilinguals”. Bilingualism: Language and Cognition, 5 (1), 39–68. Montrul, S. (2004). “Subject and object expression in Spanish Heritage speakers: A case of morphosyntactic convergence”. Bilingualism: Language and Cognition, 7 (2), 125–142. Olshtain, E. & Barzilay, M. (1991). “Lexical retrieval difficulties in adult language attrition”. In H. W. Seliger & R. M. Vago (Eds.), First Language Attrition (pp. 139–150). Cambridge: CUP. Papadopoulou, D. (2006). Cross-Linguistic Variation in Sentence Processing. Dordrecht: Springer. Papadopoulou, 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. Papadopoulou, D. & Tsimpli, I. M. (2004). “Morphological cues in children’s processing of ambiguous sentences: A study of Subject/object ambiguities in Greek”. In A. Brugos, L. Micciula & C. E. Smith (Eds.), Proceedings of the 28th Boston Conference on Language Development (pp. 471–481). Somerville, MA: Cascadilla Press. Papadopoulou, D. & Tsimpli, I. M. (in press). “Morphological cues in L2 sentence processing: Evidence from subject/object ambiguities in Greek as L2”. To appear in A. Belletti, E. Bennati, C. Chesi, E. Di Domenico & I. Ferrari (Eds.), Language Acquisition and Development: Proceedings of GALA2005. Newcastle: Cambridge Scholars Publishers. Papadopoulou, D. & Tsimpli, I. M. (in prep.). Case on Greek determiners: An online study. Pavlenko, A. (2000). “New approaches to concepts in bilingual memory”. Bilingualism: Language and Cognition, 3, 1–5. Philippaki-Warburton, I. (1987). “The theory of empty categories and the pro-drop parameter in Modern Greek”. Journal of Linguistics, 23, 289–318. Pinto, M. (1997). Licensing and interpretation of inverted subjects in Italian. Unpublished Doctoral Dissertation, University of Utrecht. Platzack, C. (2001). “The vulnerable C-Domain”. Brain and Language, 77, 364–377. Rizzi, L. (1982). Issues in Italian syntax. Dordrecht: Foris. Rizzi, L. (2002). On the Grammatical Basis of Language Development. Ms. University of Siena. Sanz, M. & Bever, T. G. (2001). “A theory of syntactic interference in a bilingual”. In J. L. Nicol (Ed.), One Mind, Two Languages: Bilingual language processing (pp. 134–158). Malden, MA: Blackwell Publishers.
97
98 Ianthi Maria Tsimpli
Satterfield, T. (1999). Bilingual Selection of Syntactic Knowledge: Extending the principles and parameters approach. Dordrecht: Kluwer Academic Publihers. Schmid, M. S. (2000). First Language Attrition, Use, and Maintenance: The case of German Jews in Anglophone countries. Unpublished Doctoral Dissertation, Düsseldof: Heinrich-HeineUniversität. Schmid, M. S. (2002). First Language Attrition, Use and Maintenance: The case of German Jews in Anglophone countries. Amsterdam: John Benjamins. Serratrice, L., Sorace, A., & Paoli, S. (2004). “Crosslinguistic influence at the syntax pragmatics interface: Subjects and objects in English-Italian bilingual and monolingual acquisition”. Bilingualism: Language and Cognition, 7 (3), 183–205. Sorace, A. (2004). “Native language attrition and developmental instability at the syntaxdiscourse interface: Data, interpretations and methods”. Bilingualism: Language and Cognition, 7 (2), 143–145. Tsimpli, I., Sorace, A., Heycock, C., & Filiaci, F. (2004). “First language attrition and syntactic subjects: A study of Greek and Italian near-native speakers of English”. International Journal of Bilingualism, 8 (3), 257–277. Waas, M. (1996). Language Attrition Downunder. Frankfurt: Peter Lang. White, L. & Genesee, F. (1996). “How native is near-native? The issue of ultimate attainment in adult second language acquisition”. Second Language Research, 12, 233–265.
(Psycho)linguistic determinants of L1 attrition Ayşe Gürel
Boğaziçi University, Istanbul, Turkey
This paper addresses the issue of selectivity in first language (L1) attrition of syntax by investigating the effects of L2 Turkish binding properties of reflexives and pronouns in the L1 English grammar. The L2 Turkish-induced L1 English attrition data is discussed in comparison with L2 English-induced L1 Turkish attrition data reported in Gürel (2002). The aim is to explore linguistic and psycholinguistic factors that affect interference-dependent syntactic attrition in the L1. The paper argues that it might be possible to predict L1 syntactic forms that are more vulnerable to restructuring through an attrition model that incorporates learnability-based linguistic factors as well as psycholinguistic factors involved in the retention and forgetting of linguistic forms among bilinguals. Keywords: syntactic attrition; binding theory; activation threshold hypothesis; Turkish
Introduction It has been observed that prolonged exposure to a second language (L2), accompanied by long-term disuse of a first language (L1), might induce a restructuring or change in the syntactic module of the L1 grammar, albeit slowly and selectively (Gürel 2002, 2004a, 2004b; Larmouth 1974; Polinsky 1997; Schaufeli 1996; Seliger 1989, 1996; Sorace 2000; Tsimpli, Sorace, Heycock & Filiaci 2004). Syntactic attrition does not occur as quickly as processes such as lexical attrition, nor are all aspects of syntactic knowledge subject to the same degree of change under the influence of the L2. Understanding the selective nature of L2-induced changes in the L1 syntax has been one of the primary concerns of L1 attrition studies (see contributions in Seliger & Vago 1991; Schmid, Köpke, Keijzer & Weilemar 2004); such studies try to
100 Ayşe Gürel
characterize L2 syntactic forms and structures that are likely to be transferred into the L1 system. Selective transfer is one such issue that has also been widely investigated in L2 acquisition research (see for example White 1989, 2000). In linguistically-oriented L1 attrition and L2 acquisition research, predictions on language transfer are formulated solely in linguistic terms, while psycholinguistic factors determining grammar change in the L1 and L2 have largely been ignored. Therefore, this paper aims to fill this gap by investigating both linguistic and psycholinguistic factors that might affect L2-dependent syntactic attrition in the L1. The syntactic phenomena under investigation here are the binding properties of reflexives and pronouns – a central topic in generative syntax, studied extensively in language acquisition research. This paper argues that it might be possible to predict L1 (English) syntactic forms that are more vulnerable to L2 (Turkish)-induced restructuring by using an attrition model that incorporates both learnability-based linguistic issues (e.g. the subset relations between the L1 and the L2) and psycholinguistic factors involved in the retention and forgetting of linguistic forms among bilinguals. Using L1 and L2 syntactic features, linguistically-oriented attrition theories have predicted the susceptibility of different L1 syntactic properties to L2-induced change. This paper argues that these theories are insufficient in themselves, as research suggests that L1 linguistic properties predicted to undergo attrition do so only when there are also suitable psycholinguistic conditions for attrition. Therefore, in this paper, it is asked to what extent psychological factors such as frequency of use may play a role in the retention of L1 syntactic features. Investigating attrition in the L1 English grammar is of particular relevance to test the impact of frequency of L1 use. Due to the de facto status of English as the lingua franca, native English speakers living and working in Turkey (as in many other countries) continue to use their L1 dominantly over the years. Therefore, this is a group that is not expected to demonstrate much attrition in the L1 grammar. Given this, in this paper the results of the L2 Turkishinduced L1 English attrition study are presented in comparison with the results of the L2 English-induced L1 Turkish attrition study reported in Gürel (2002). We believe that comparison of these two groups can help us check learnability-based linguistic models and the frequency-based psycholinguistic models against the predictions they make for the selectivity in syntactic attrition.
1. Accounts for selectivity in L1 attrition The selective nature of L1 attrition has long been documented in academic literature (Gürel 2002, 2004b; Seliger 1989, 1996; Sorace 2000; Tsimpli et al. 2004), though accounts for this selectivity vary. For example, Seliger (1989, 1996) argues
(Psycho)linguistic determinants of L1 attrition 101
that selective attrition is linked to the markedness of L1 and L2 features. In this account, complex grammatical forms with narrow linguistic distribution are considered to be marked. In language contact situations, marked forms in the L1 are predicted to be replaced by less marked forms in the L2, whereas less marked L1 forms do not easily undergo attrition. In the more recent accounts of Sorace (2000) and Tsimpli et al. (2004), selective attrition in the L1 syntax is again explained in purely linguistic terms, suggesting that the changes in the L1 syntax are restricted to the interface between syntax and the conceptual/intentional cognitive systems (Tsimpli et al. 2004: 257). More specifically, selectivity is considered to be a manifestation of the formal distinction between two types of linguistic features, namely [+interpretable] and [–interpretable] features (Chomsky 1995). In their investigation of attrition in L1 Italian and L1 Greek – two null-subject languages – Tsimpli et al. (2004) observe that null subjects, which result from the specification of [–interpretable] features (e.g. case and agreement on verbs), are not affected by attrition. However, the distribution of null and overt subjects, which is determined by [+interpretable] (or syntactic/pragmatic) features (e.g. [+Topic Shift]), is subject to attrition. In another study, Gürel (2002) analyzes selective attrition within the context of the set-theoretic transfer model (Berwick 1985; Manzini & Wexler 1987; White 1989) by investigating the attrition of binding properties of L1 Turkish overt and null subject pronouns under the influence of L2 English. Turkish and English differ with respect to binding domains and the type of pronominals present in two languages. As a null-subject language, Turkish allows null subject pronouns in main and embedded clauses. It also has a special type of nominative reflexive pronominal, kendisi, for which English has no corresponding form. The following examples illustrate the differences between Turkish and English with respect to pronominal binding in the context of Binding Principle B (Chomsky 1986; Chomsky & Lasnik 1995). (1) a. b.
Elifi [o-nun*i/j kendi-si-nini/j / proi/j çok inatçı Elif s/he-gen self-3sg-gen very stubborn ol-duğ-u]-nu bil-iyor be-nom-3sgposs-acc know-prg Elifi knows that [shei /j /*herself /*pro is very stubborn]
The example in (1a) illustrates that the embedded subject position of a Turkish sentence can be occupied by the pronouns o, kendisi or pro. As the above co-in. The following abbreviations are used in this paper: acc = Accusative; dat = Dative; gen = Genitive; neg = Negation; nom = Nominalization; poss = Possessive; prg = Progressive; pst = Past; sg = Singular.
102 Ayşe Gürel
dexations indicate, the overt pronoun o cannot be bound by the matrix subject, Elif in Turkish;, it only allows a disjoint reading. The overt pronoun kendisi and pro, however, can be bound by either the matrix subject or have disjoint readings. In contrast, English only allows the overt pronoun s/he in the embedded subject position, and the binding properties of this overt pronoun are different from that of its Turkish counterpart o. The difference between Turkish and English overt pronoun binding is attributed to differences in the binding domains of the two languages (Gürel 2002, 2004a). It has been suggested that Determiner Phrases (DPs) do not constitute binding domains in Turkish. Most complement clauses in Turkish are nominalised and they behave like lexical DPs with respect to their internal morphological form and case-marking (George & Kornfilt 1981) (see examples (2a) and (3a)), indicating that Turkish embedded clauses are not finite clauses but DPs and they do not form binding domains. In contrast, both finite clauses and DPs count as binding domains in English. (2) a. b.
Elifi DP[o-nun*i/j çok inatçı ol-duğ-u]-nu Elif s/he-gen very stubborn be-nom-3sgposs-acc bil-iyor know-prg Elifi knows that IP[shei/j is very stubborn]
(3) a. Elifi DP[o-nun*i/j yüz-ü-nü Elif s/he-gen face-3sgposs-acc b. Elifi washed DP[heri/j face]
yıka-dı] wash-pst
In the Turkish sentences in (2a) and (3a), the binding domain is the tensed matrix clause that includes the pronoun and a finite inflection (I0). Therefore, in line with Binding Principle B, the pronoun o has to be free in its binding domain, preventing co-indexation between the overt pronoun and the matrix subject Elif. However, in English, both finite Inflectional Phrase (IP) and DP constitute a binding domain. Therefore, co-indexation between the pronouns she/her and the matrix subject Elif is possible in the English sentences shown in (2b) and (3b). By allowing both finite IPs and DPs as binding domains, the L2 English is a less restrictive language than the L1 Turkish. Therefore, the L2 English constitutes
. In the standard assumptions of Binding Theory, ‘bound’ means co-indexed with a c-commanding DP (e.g. Chomsky 1981). . C-command is defined as: α c-commands β if α does not dominate β and every γ that dominates α dominates β. Where γ is restricted to maximal projections, α m-commands β (Chomsky & Lasnik 1995).
(Psycho)linguistic determinants of L1 attrition 103
the superset grammar in this construction, while the L1 Turkish constitutes the subset. Three different tasks (written interpretation, truth-value judgment and picture-identification) were used in Gürel’s (2002) study, in which native Turkishspeakers living in North America (late L2 learners who are at the L2 end-state) were tested as participants. The study’s results revealed that the binding properties of English overt pronouns (e.g. him/her) are transferred onto the overt Turkish pronoun o due to long-term L2 English influence. However, the properties of the Turkish null pronoun and the overt nominative reflexive form kendisi remain unaffected by the L2 English. In other words, while native Turkish speakers maintain the binding properties of pro and kendisi, their L1 knowledge of the binding properties of the overt pronoun o is restructured in keeping with the L2 English grammar. This suggests that restrictions in the L1 grammar (i.e. a more restrictive definition of the binding domain) were loosened due to L2 English. This finding implies that in cases where the L2 grammar has broader options (the superset) than the L1 grammar, the grammatical restrictions in the L1 are neutralized. That is, due to frequent positive evidence from the L2, the grammatical options of the L1 are expanded on the model of the L2, ultimately leading to a grammar that is divergent from native norms. In contrast, the binding conditions of the null pronoun and the overt pronominal kendisi were maintained at the native competence level. These findings can also be explained using the set-theoretic transfer model. In the case of the null pronoun, the influencing language (i.e. the non-null-subject L2 English), forms the subset of the L1, and neither interferes with the L1 grammar nor leads to its restructuring. In this configuration, the L2 English does not provide any data that is inconsistent with the L1; everything included in the L2 already exists in the L1. Thus, the binding properties of null subjects remain relatively well-preserved in the L1. As for the form kendisi, Turkish and English are distinct from each other with respect to the presence of an anaphoric pronominal; the L2 English does not have a nominative reflexive pronoun with binding properties similar to those of the L1 Turkish form kendisi. The preservation of L1 knowledge of the binding properties of kendisi suggests that so long as the L2 does not have an analogous form, L2 interference does not occur. Gürel (2002) interprets these results in the context of the subset-superset relations of the L1 and L2, stating that in the context of L1 attrition where the ‘influencing language’ (i.e. L2) is the superset of the ‘affected language’ (i.e. L1), more signs of L2 transfer into the L1 grammar are noticeable, resulting in more attrition effects. In contrast, in configurations in which the L2 offers a more restrictive grammar (i.e. as the subset of the L1), it does not lead to any restructuring in the L1 grammar. This account therefore argues that whether or not a particular L1
104 Ayşe Gürel
linguistic structure will undergo attrition can be determined by the subset relation between the L1 and the L2 with respect to that structure. More recently, language attrition in general and selective attrition in particular have been analyzed within a psycholinguistic perspective with reference to constructs such as frequency-based language activation or inhibition. Following Paradis (2000) and Köpke (2002), Gürel (2004b), suggests that selectivity in L1 attrition can be predicted under the Activation Threshold Hypothesis, which essentially suggests that the more frequently a linguistic item is activated, the lower its activation threshold (see Paradis, this volume). In the context of L1 attrition, this hypothesis predicts that when the L2 is frequently used (or activated), the L1 (its competitor) is simultaneously inhibited (Green 1986; Köpke, this volume; MacWhinney 1987; Paradis 1989). The activation threshold of the L1 thereby becomes higher than that of the L2. A heightened activation threshold makes an L1 linguistic form less accessible and thus potentially more susceptible to attrition or change under L2 influence. However, this does not mean that the L1 linguistic system will be inhibited altogether; only when the L1 and the L2 have corresponding forms will there be a competition between them. In these cases, if the L1 form is not stimulated frequently, the actively-used L2 form will win the competition. When the L1 and the L2 have no corresponding forms, the competition between them will not be relevant; L1 forms will not be inhibited due to L2 forms and these L1 forms will be preserved even after long-term disuse. As summarized above, selective attrition can be described adequately using both linguistic and psycholinguistic models. Gürel’s (2002, 2004b) analyses suggest that in some cases, both perspectives make similar predictions as to which syntactic structures will undergo attrition. The study on L1 English attrition detailed in the present paper tests the predictions of these two perspectives, the Subset Model and the Activation Threshold Model, by examining the effects of L2 Turkish on L1 English. Specifically, the L2 Turkish influence on the binding properties of L1 English pronouns and reflexives was examined in the English of a group of native English speakers who have been living in Turkey for an extended period of time. In order to facilitate comparisons between the findings of the two studies a design similar to that used by Gürel (2002) was employed.
2. Issue: Attrition effects in pronominal binding in L1 English 2.1 Pronouns As noted earlier, binding properties of pronouns are subject to Principle B of Binding Theory. The previous section presented some examples involving sub-
(Psycho)linguistic determinants of L1 attrition 105
ject pronouns. Consider now the following examples, which illustrate the binding properties of the English and Turkish object pronouns. (4) a. Georgei likes him*i/j b. Georgei o-nu*i/j sev-iyor George s/he-acc love-prg (5) a. b.
Briani says [Georgej likes himi/*j/k] Briani [George’unj o-nui/*j/k Brian George-gen s/he-acc söyl-üyor say-prg
sev-diğ-i]-ni like-nom-3sgposs-acc
Pronouns cannot be bound by a c-commanding antecedent in their local domains. The pronouns him (4a) and o in (4b) cannot be coreferential with the sentential subject (George) in the local binding domain; they must refer to someone else in the discourse. In biclausal sentences in (5a) and (5b), the clause boundary intervenes between the pronouns and their antecedents. In both the English and the Turkish examples, the binding domain for the pronoun is the embedded clause since it includes the pronoun and its governor (the embedded verb) as well as a potential binder. The same binding conditions also apply in nonfinite clauses, as illustrated below: (6) a. b.
Briani told [Georgej not to criticize himi/*j/k] Briani [George’aj o-nui/*j/k eleştir-me-me-si]-ni Brian George-dat s/he-acc criticize-nom-neg-3sgposs-acc söyle-di tell-pst
In monoclausal sentences as in (7), where there are two potential antecedents, the English pronoun can neither be bound by the subject nor by the object antecedent in the local domain. The pronoun him can only take a disjoint reference. However, in Turkish, the pronoun o can be co-indexed with the dative case-marked indirect object George, or it can have a disjoint reading. Whether native English speakers maintain this L1 property despite L2 Turkish influence is one of the questions examined in this paper. (7) a. Briani showed Georgej a picture of him*i/*j/k
. As noted earlier, embedded clauses in Turkish sentences are nominalised constructions hence nonfinite. Therefore, embedded clauses in (5b) and (6b) are both nonfinite.
106 Ayşe Gürel
b.
Briani George’aj Brian George-dat göster-di show-pst
o-nun*i/j/k s/he-gen
bir resm-i-ni a picture-3sgposs-acc
2.2 Reflexives Binding properties of reflexives (and other anaphors) are subject to Principle A of Binding Theory, according to which an anaphor is bound in its binding domain by a proper antecedent (Chomsky 1981, 1986). Accordingly, as illustrated below in (8a), the English reflexive himself must be bound within the minimal category that contains the reflexive, a potential binder and a governor. The corresponding Turkish form kendi expresses similar reflexive relations (8b). However, the form kendisi is not constrained by Binding Principle A. (8) a. Georgei likes himselfi/*j b. Georgei kendi-nii/*j /kendi-si-nii/j George self-acc self-3sg-acc (9) a. b.
Briani says [Georgej likes himself *i/j/*k] Briani [George’unj kendi-ni*i/j/*k Brian George-gen self-acc sev-diğ-i]-ni söyl-üyor like-nom-3sgposs-acc say-prg
sev-iyor love-prg /kendi-si-nii/j/k self-3sg-acc
Both the English reflexive himself and the Turkish form kendi are bound by the subject of their own clause (8a and 8b). In (9a) and (9b), however, they cannot be bound by the subject of the matrix clause across the subject of the embedded clause (their binding domain). When the intended antecedent of the reflexive is outside the local domain in which the reflexive must be bound, the sentence is ungrammatical. Unlike the reflexives himself and kendi, the form kendisi does not need a sentence-internal antecedent (Enç 1989). It has been suggested that the
. The form kendisi can be used as a pronoun (i) or as a reflexive (ii):
(i) Kendi-si gel-di
(ii)
Self-3sg come-pst ‘S/he came’ Elifi kendi-si-nii/j beğen-iyor Elif self-3sg-acc like-prg ‘Elifi likes herselfi/*j’
(Psycho)linguistic determinants of L1 attrition 107
reflexive kendisi is a special pronoun that is not constrained in any way by Binding Theory (Enç 1989; Gürel 2002). In nonfinite clauses, the case appears to be similar. While the reflexives himself and kendi can only be bound by the local antecedent, the form kendisi has no such constraint. (10) a. b.
Briani told [Georgej not to criticize himself*i/j/*k] Briani [George’aj kendi-ni*i/j/*k /kendi-si-nii/j/k Brian George-dat self-acc self-3sg-acc eleştir-me-me-si]-ni söyle-di criticize-nom-neg-3sgposs-acc tell-pst
In monoclausal sentences, where there are two potential local antecedents for the reflexive, the English sentence is ambiguous as both the subject and the object can serve as the antecedent for the reflexive (11a). In Turkish, the reflexive kendi is subject-oriented; it can only be co-indexed with the subject antecedent. However, the form kendisi leads to an ambiguous interpretation of the sentence. Like the English reflexive himself, the Turkish form kendisi can take both subject and object DPs as antecedents. However, kendisi is different from himself in that it also allows a non-local antecedent. (11) a. b.
Briani showed Georgej a picture of himselfi/j/*k Briani [George’aj kendii/*j /kendi-si-nini/j/k Brian George-dat self self-3sg-gen resm-i]-ni göster-di picture-3sgposs-acc show-pst
The differences between English and Turkish pronominals and the set-theoretic predictions can be summarized as follows: unlike English, Turkish nominalised embedded clauses and DPs do not constitute binding domains. Thus, different binding properties of the Turkish pronoun o and the English pronoun s/he in the subject position are due to the difference in what counts as the binding domain in two languages. In these configurations, the L2 Turkish, with its more restrictive grammar, constitutes a subset of the L1 English. Hence no attrition effects are expected in these constructions. In an object position governed by the verb, L1 and L2 pronouns demonstrate similar binding properties in line with Binding Principle B (see examples (4) and (5)). Thus, L2 transfer effects are not relevant in these constructions. In monoclausal sentences with two potential antecedents, the English pronoun can only take disjoint reference. However, in Turkish, the object pronoun o can both be coreferential with the indirect object antecedent and have disjoint reference. In this configuration, the L2 Turkish grammar therefore constitutes the
108 Ayşe Gürel
superset grammar with more grammatical options; it is in this context that more attrition effects are expected since, under attrition, the L1 grammar is expected to move toward a less restricted L2 grammar. English does not have a nominative reflexive similar to the Turkish form kendisi and in Turkish there is also a true reflexive form that corresponds to the English reflexive in binding features. Thus, the L2 Turkish, having two reflexive forms (kendi and kendisi) with different binding properties, has broader grammatical options than the L1 English. The crucial question is whether or not the L1 English grammar would be expanded on the model of L2 Turkish as predicted by the set-theoretic language attrition model (Gürel 2002). That is, would there be any restructuring in the binding properties of the English reflexive form due to the L2 Turkish? This model predicts finding restructuring in the L1 grammar in the form of expansion towards the L2. According to this, under the influence of L2 Turkish, the English reflexive can be attributed less restrictive binding properties on the model of the L2 form kendisi. If binding properties of kendisi are mapped onto the L1 reflexive, then in biclausal finite sentences, attriters might allow non-local antecedents for the reflexive himself/herself. Recall that in monoclausal sentences, both the English reflexive himself and the Turkish kendisi have ambiguous readings as both forms can take subject and object DPs as antecedents. The reflexive kendi, on the other hand, is subject-oriented. Thus, the L2 form kendi is more restrictive than the L1 English reflexive. Therefore, no L2 influence is predicted in this case. However, unconstrained binding behaviours of the form kendisi might still lead to expansions in the grammatical options allowed in the native L2 grammar.
3. The study 3.1 Participants The L1 English attrition group consisted of fifteen native English speakers (mean age: 48.33). A questionnaire given to each participant prior to the study was used to determine their language background. All participants in this study settled in Turkey at an adult age (between 18–32) and had been living in Turkey for at least 10 years at the time of testing. As in Gürel (2002), a 10-year-stay in an L2 country was taken as the main inclusionary criterion in this study. The length for which the participants had been in Turkey ranged from 10 to 35 years (mean length of stay: 18.6 years). All of the participants had their first L2 Turkish exposure as adults when they moved to Turkey. Eight of them had received formal instruction in Turkish, for
(Psycho)linguistic determinants of L1 attrition 109
periods ranging from 2 weeks to 3 years (mean length of formal instruction: 14.25 months). All participants work in English-speaking environments in Turkey. This is a crucial aspect, since the amount of L1-contact is an important factor in determining the extent of L1 attrition. Five of the participants work as English language teachers and four are instructors at an English-medium university; the rest work for international companies where they mostly use the L1 English. Thus, we can safely say that the participants all use the L1 extensively at work. Similar to the L1 Turkish attriters in Gürel’s (2002) study, participants in the present study have been living in an L2 setting for an extended period of time. However, unlike the participants in the 2002 study, the degree of L1 contact is much higher in the present case, perhaps affecting the L2 Turkish influence on their L1 English. As for L2 contact, participants inevitably had daily contact with L2 Turkish both socially and at work (mean reported amount of L2 use: 4.36 hrs per day), and eleven participants have Turkish partners. L2 Turkish contact was reported to mostly occur in the contexts of listening and speaking. In a language background questionnaire, two participants rated their overall Turkish proficiency as near-native, seven rated themselves as advanced and six rated themselves as intermediate-level L2 Turkish speakers. No separate proficiency test was conducted to determine their L2 Turkish proficiency, however, all L2 Turkish speakers who participated in this study were referred to the researcher as ‘people who have good command of Turkish’. While a subjective assessment, it was useful in the initial participant selection phase. This study, however, operated on the assumption that L2 learners do not have to be near-native L2 speakers to be candidates for L1 attrition (cf. Sorace 2000). Fifteen native English speakers participated in the study as a control group. This group included thirteen American and two Canadian participants living in North America. They took part in the study by answering questions for one of the tests (the written interpretation task). Five of the participants were tested during a short visit to Turkey and ten participants took part in the study through e-mail correspondence.
3.2 Tests Two tests were used in this study: a written interpretation task and a truth-value judgment task.
. A standardized proficiency test is not yet available for L2 Turkish.
110 Ayşe Gürel
3.2.1 Test 1: Written interpretation task The general format of this test was similar to the one used in Gürel (2002) and Kanno (1997). The test included English sentences with pronouns and reflexives; participants were asked to select (a) possible antecedent(s) for the pronouns and reflexives from among the options provided. For example, in a test item given in (12), participants were expected to select both option (a) and (b), indicating that both David and Scott can be the antecedent for the reflexive. (12) David sent Scott a picture of himself. Question: Whose picture could this be? (a) David’s (b) Scott’s (c) Some other person’s
There were eight categories of test items, as illustrated below, and each category contained six items, so the total number of items was 48. a. b. c. d. e. f. g. h.
Possessive DPs (pronouns): The soldiersi carried theiri/j boxes Monoclausal sentences with pronouns: Jamesi sent Henryj a picture of him*i/*j/k Monoclausal sentences with reflexives: Catherinei gave Dianaj good news about herselfi/j/*k Biclausal finite sentences with pronouns: Chrisi knows that Harryj hates himi/*j/k Biclausal nonfinite sentences with pronouns: Nancyi asked Emilyj not to blame heri/*j/k Biclausal finite sentences with reflexives: Juliai thinks that Susanj admires herself*i/j/*k Biclausal nonfinite sentences with reflexives: Maryi wants Janej to forgive herself*i/j/*k Biclausal finite sentences with subject pronouns: Johni said that hei/j lied to the police
3.2.2 Test 2: Truth-value judgement task (Story task) This test was adapted from Gürel (2002), which was originally based on Dekydtspotter, Sprouse & Anderson (1997) and White, Bruhn-Garavito, Kawasaki, Pater & Prévost (1997). In this task, participants were given 48 short Turkish sto. To eliminate any possible facilitative effect of a particular grammatical structure in the story in judging the target sentence, the story was not given in the target language (i.e. English) (for a similar point, see Dekydtspotter , Spronse & Anderson 1997).
(Psycho)linguistic determinants of L1 attrition
ries (2 to 6 sentences long) and asked to decide whether a following English sentence would be conceivably “true” or “false” according to the context given in the story. All of the sentences were grammatical; their appropriateness depended on the context provided by the stories. The stories were used to make as salient as possible in the minds of participants a particular interpretation for the pronoun or the reflexive. The following is one of the test items used in Test 2: (13) Story: Elif bebeği için dadı arıyordu. Gazeteye ilan verdi. Bir gün eve yaşlı bir hanım geldi. Elif kadını tanımak için sorular sordu. (Elif was looking for a baby-sitter for her baby. She placed an advertisement in a newspaper. One day an old lady came to her home. Elif asked questions to get to know the woman.) Target sentence to be judged: Elifi asked the old ladyj about her*i/*j/k
In this monoclausal sentence, there are two potential local antecedents for the pronoun her and the pronoun could also have disjoint reference. The story above requires coreferentiality between the pronoun and the indirect object DP the old lady. However, since the English pronoun cannot be coreferential with either of these local antecedents, the expected answer is “false”. If participants incorrectly allowed binding of the pronoun by the object, then they would say “true” for this sentence. Two other potential interpretations were forced in other stories in the test; the context in these stories forced binding of the pronoun by the subject antecedent or by someone else in the discourse. Overall, the test included 48 items and the types of sentences were similar to those in the first test. Since the story task included stories written in Turkish, the control group answered only Test 1.
4. Results The results of Test 1 and Test 2 were analyzed separately. In Test 1, the accuracy scores of the attrition and the control group were compared. As mentioned earlier, Test 2 was given only to the attrition group as the test included Turkish stories. In the analysis of Test 2, accuracy scores obtained in different sentence types were compared to examine the L2 Turkish influence.
111
112 Ayşe Gürel
Table 1. Results of Test 1* Sentence types
Attrition Mean (%) SD
Control Mean (%) SD
Possessive DPs (pronouns) Monoclausal sentences with pronouns Monoclausal sentences with reflexives Biclausal finite sentences with pronouns Biclausal nonfinite sentences with pronouns Biclausal finite sentences with reflexives Biclausal nonfinite sentences with reflexives Biclausal finite sentences with subject pronouns
84.44 65.56 42.22 82.22 82.22 97.78 98.89 84.44
68.87 54.42 41.07 63.29 61.04 100 96.64 66.67
34.77 36.98 41.24 34.20 34.20 58.64 43.03 35.34
44.03 40.51 37.66 46.78 45.25 0.00 69.47 44.99
* Table 1 presents the mean percentages and the standard deviations. Percentages indicate how many times the participants assigned correct interpretation for a given pronominal.
4.1 Test 1: Written interpretation task Recall that in this test, participants were tested on 8 different categories. The percentage accuracy scores obtained from these sentence types are given in Table 1. For example, in the first category, participants were given a sentence such as The soldiers carried their boxes and were asked to indicate potential antecedents for the possessive pronoun. They were expected to choose both of the options given, namely, the soldiers and some other people’s. The percentage scores given in Table 1 indicate how many times participants correctly indicated that both DPs are potential antecedents for the pronoun their in English. Overall results of MANOVA indicated no significant difference between the attrition and the control group on the combined dependent variables [Wilk’s Lambda 1.156, p = .370]. In other words, there was no difference between the performance of the attrition and the control group, contrary to the predictions made earlier. There was no change or attrition found in native English speakers’ L1 syntactic knowledge of the binding properties of pronouns or reflexives. In a further analysis, the distribution of the responses was examined. Table 2 below gives this distribution. For example, in the first category, the attrition group correctly assigned both the bound and disjoint readings for the possessive pronoun at a rate of 84.44%. For the same pronoun, they assigned bound-only (i.e. bound by the subject only) interpretation at a rate of 11.11% and disjoint-only interpretation at a rate of 4.44%.
. The mean differences did not come out significant because of the small sample size and high standard deviation.
(Psycho)linguistic determinants of L1 attrition 113
Table 2. Distribution of pronominal interpretation in Test 1 Attrition Mean (%)
Control Mean (%)
Possessive DPs (pronouns): Subject antecedent only Disjoint reference only Both subject antecedent & disjoint reference (√)*
11.11 4.44 84.44
27.78 3.33 68.89
Monoclausal sentences with pronouns: Subject antecedent only Object antecedent only Disjoint reference only (√) Both subject antecedent & object antecedent Both subject antecedent & disjoint reference Both object antecedent & disjoint reference All three options (subject-object-disjoint)
2.22 3.33 65.56 1.11 2.22 20 5.56
15.56 5.56 54.44 0.00 3.33 15.56 5.56
Monoclausal sentences with reflexives: Subject antecedent only Object antecedent only Disjoint reference only Both subject antecedent & object antecedent (√) Both subject antecedent & disjoint reference Both object antecedent & disjoint reference All three options (subject-object-disjoint)
51.11 6.67 0.00 42.22 0.00 0.00 0.00
43.33 15.56 0.00 41.11 0.00 0.00 0.00
Biclausal finite sentences with pronouns: Matrix subject antecedent only Embedded subject antecedent only Disjoint reference only Both matrix subject & embedded subject antecedent Both matrix subject antecedent & disjoint reference (√) Both embedded subject antecedent & disjoint reference All three options (matrix & embedded subject-disjoint)
16.67 0.00 0.00 1.11 82.22 0.00 0.00
33.33 0.00 1.11 0.00 63.33 2.22 0.00
Biclausal nonfinite sentences with pronouns: Subject antecedent only Object antecedent only Disjoint reference only Both subject antecedent & object antecedent Both subject antecedent & disjoint reference (√) Both object antecedent & disjoint reference All three options (subject-object-disjoint)
16.67 0.00 1.11 0.00 82.22 0.00 0.00
35.56 1.11 1.11 0.00 61.11 1.11 0.00
114 Ayşe Gürel
Table 2. (continued) Attrition Mean (%)
Control Mean (%)
Biclausal finite sentences with reflexives: Matrix subject antecedent only Embedded subject antecedent only (√) Disjoint reference only Both matrix subject & embedded subject antecedent Both matrix subject antecedent & disjoint reference Both embedded subject antecedent & disjoint reference All three options (matrix & embedded subject-disjoint)
2.22 97.78 0.00 0.00 0.00 0.00 0.00
0.00 100 0.00 0.00 0.00 0.00 0.00
Biclausal nonfinite sentences with reflexives: Subject antecedent only Object antecedent only (√) Disjoint reference only Both subject antecedent & object antecedent Both subject antecedent & disjoint reference Both object antecedent & disjoint reference All three options (subject-object-disjoint)
0.00 98.89 0.00 1.11 0.00 0.00 0.00
3.33 96.67 0.00 0.00 0.00 0.00 0.00
Biclausal finite sentences with subject pronouns: Subject antecedent only Disjoint reference only Both subject antecedent & disjoint reference (√)
13.33 2.22 84.44
32.22 1.11 66.67
*Correct grammatical options in English were indicated by a check mark (√).
Let us first look at the pronouns in Table 2 to examine the preferences of the attrition group and to see if we can observe any tendency that would suggest L2 Turkish influence. In the first and the last category (i.e. possessive DP pronouns and biclausal finite sentences with subject pronouns), the L2 Turkish only allows the disjoint interpretation for the pronoun, whereas the L1 English pronoun can be bound by the subject DP or can take disjoint reference. As mentioned earlier, no attrition effect in the L1 grammar with broader grammatical options is expected; as predicted, there was no increased preference for the disjoint-only reading for the English pronoun in the attrition group. In monoclausal sentences with pronouns, an attrition effect was predicted since the L2 Turkish allows binding of the pronoun with both the object and a third person in the discourse. As can be seen in Table 2, both groups maintained the L1 English option. That is, they allowed the disjoint-only interpretation at a higher rate (65.56% and 54.44%) than other readings. Nevertheless, the attrition
(Psycho)linguistic determinants of L1 attrition 115
group seemed to allow the L2 Turkish option (i.e. both object antecedent and disjoint reference) at a higher rate than the control group (20% vs. 15.56%). This difference, though not statistically significant, might suggest that the possibility of binding of the pronoun with the indirect object, ungrammatical in the L1 English, is more readily available to the attrition group as a grammatical option, possibly due to the L2 Turkish influence. In biclausal finite and nonfinite sentences with pronouns, no attrition effect is relevant as L1 and L2 pronouns demonstrate similar binding options. As for the sentences involving reflexives, in monoclausal sentences with reflexives, the subject-oriented L2 reflexive kendi was not expected to be transferred into the L1 grammar. However, we found that the attrition group preferred coreferentiality between the reflexive and the subject antecedent at a high rate (51.11%). Since this tendency was also observed in the control group, this effect is probably not due to the L2 Turkish. Also, in this category, it was clear that the unrestricted binding properties of the L2 form kendisi were not adopted for the L1 English form. That is, none of the groups allowed unrestricted binding options for the English reflexive. Finally, in biclausal finite and nonfinite sentences with reflexives, the L2 form kendi and the L1 reflexive behave the same way. Thus, as predicted, no attrition was found. As for the influence of kendisi, it was clearly seen that, in contrast to the prediction, the binding options of the L1 reflexive have not been changed on the model of the L2 form. It is important to recall that the L2 form kendi and the L1 English reflexive have similar binding properties to a large extent. Since it is possible that the presence of an L1-like form in the L2 might help maintain the L1 form, these similar properties would prevent changes in the L1 grammar.
4.2 Test 2: Truth-value judgement task (Story task) The results of Test 2 are summarized in Table 3 below. Recall that in this test, binding properties of pronouns and reflexives were tested within a particular context provided by a story. For example, there were 9 stories involving monoclausal sentences with pronouns (e.g. James sent Henry a picture of him). In 3 stories, the context forced disjoint interpretation for the pronoun. As can be seen in Table 3, in these items, participants correctly allowed the disjoint-only interpretation at a rate of 80%. In the other stories, the context forced coreferentiality with subject and object antecedents. For example, in the subject-antecedent context, participants incorrectly allowed subject-binding for the pronoun hence the low accuracy rate of 46.67%.
116 Ayşe Gürel
Table 3. Results of Test 2 Context**
Attrition group Mean (%)* SD
Possessive DP-pronouns Subject antecedent (√) Disjoint reference (√)
100.00 93.33
0.00 25.82
Monoclausal sentences with pronouns Subject antecedent Object antecedent Disjoint reference (√)
46.67 55.56 80.00
32.85 48.25 27.60
Monoclausal sentences with reflexives Subject antecedent (√) Object antecedent (√) Disjoint reference
95.56 66.67 73.33
17.21 37.80 33.81
Biclausal finite sentences with reflexives Matrix subject antecedent Embedded subject antecedent (√) Disjoint reference
77.78 88.89 95.56
34.89 27.22 11.73
Biclausal nonfinite sentences with reflexives Subject antecedent Object antecedent (√) Disjoint reference
86.67 84.44 95.56
24.56 27.79 11.73
Biclausal finite sentences with subject pronouns Subject antecedent (√) Disjoint reference (√)
86.67 62.22
27.60 24.77
* Percentages indicate accuracy rates obtained in assigning a correct interpretation for a given pronominal in a given context. ** Context refers to a particular interpretation forced by the story. Correct grammatical options in English are indicated by a check mark (√).
Paired-samples t-tests were conducted to analyze the results in Test 2. In this analysis, a series of t-tests were conducted among different variables (i.e. sentence types), comparing the individual accuracy scores obtained in each variable. A significant difference was found only in two sentence types. In monoclausal sentences with reflexives, the difference between accuracy rates in subject-binding (M = .9556) and object-binding (M = .6667) of reflexives was significant (T14 = –3.389, p < .05). Recall that monoclausal sentences with reflexives are potentially ambiguous; however, in this study, even if the context favours the object antecedent for the reflexive and this is a grammatical option in English, participants do
(Psycho)linguistic determinants of L1 attrition 117
not seem to consider this option. Since the L2 reflexive kendi only allows the subject antecedent, the participants’ high preference for subjects as antecedents could be interpreted as L2 effects, though the data do not provide conclusive evidence. In addition, in the last category (biclausal finite sentences with subject pronouns), where the embedded subject pronoun can have either a disjoint reference or be bound by the sentential subject, participants seem to have problems in recognizing the possibility of disjoint reference in these constructions, hence a significant difference between the subject antecedent (M = .8667) and disjoint reference (M = .06222), (T14 = 6.205, p < .05). Lower preference for the disjoint reference cannot be due to the L2 Turkish as that is the only option allowed in Turkish. In sum, the results of Test 1 and Test 2 revealed no evidence for L2 Turkish influence in the L1 English grammar. A bigger sample size and multiple tasks of different modalities may have provided more meaningful results.
5. Conclusion The present study was conducted to investigate syntactic attrition in L1 English in comparison to syntactic attrition in L1 Turkish (Gürel 2002). The ultimate aim was to test the predictions of the learnability-based linguistic models and the frequency-based psycholinguistic models of language attrition. The predictions drawn from the Subset Model of language attrition were not supported in the present study, as in fact no conclusive evidence for L2-Turkish-induced attrition was found in the L1 English grammar. Although Gürel’s (2002) study demonstrated L2 English effects in the L1 grammar of Turkish native speakers living in North America, the same L2 effects were not found in the L1 grammar of native English speakers living in Turkey. This difference points to one important factor in the attrition phenomenon: the frequency of L1 use. As the Activation Threshold Model predicts, L1 attrition should only occur when there is little contact with the L1 (but see Schmid, this volume). No matter how frequent the use of L2 is, if it is not accompanied by infrequent L1 contact, L1 attrition is not likely to occur. These findings also suggest that linguistic models of L1 attrition should be considered in association with psycholinguistic models; so long as psycholinguistic conditions for language attrition are not present, patterns of syntactic attrition predicted by linguistic models will not be observed. Although the present study suggests that syntactic attrition in the L1 English may not occur among native English speakers who continue to use their L1 in the Turkish setting, the results should be considered preliminary (due to the small number of people tested). Furthermore, it has long been acknowledged that
118 Ayşe Gürel
change or restructuring in the L1 syntax is more difficult than changes in other aspects of language. Therefore, the present findings which suggest that the group investigated here has not undergone attrition of their L1 English can only be generalized for the domain of L1 syntax.
Acknowledgement I am grateful to Monika Schmid, Barbara Köpke and the anonymous reviewers for their helpful comments and suggestions on an earlier version of this paper. I would also like to thank Yavuz Akbulut and Kyle Marquardt for their help in the statistical analysis and the editorial work of this paper.
References Berwick, R. C. (1985). The Acquisition of Syntactic Knowledge. Cambridge, MA: MIT Press. Chomsky, N. (1981). Lectures in Government and Binding. Dordrecht: Foris. Chomsky, N. (1986). Knowledge of Language. New York, NY: Praeger. Chomsky, N. (1995). The Minimalist Program. Cambridge, MA: MIT Press. Chomsky, N. & Lasnik, H. (1995). “Principles and parameters theory”. In N. Chomsky (Ed.), The Minimalist Program (pp. 13–127). Cambridge, MA: MIT Press. Dekydtspotter, L., Sprouse, R. & Anderson, B. (1997). “The interpretative interface in L2 acquisition: The process-result distinction in English-French interlanguage grammars”. Language Acquisition, 6 (4), 297–332. Enç, M. (1989). “Pronouns, licensing, and binding”. Natural Language and Linguistic Theory, 7, 51–91. George, L. M. & Kornfilt, J. (1981). “Finiteness and boundedness in Turkish”. In F. W. Henry (Ed.), Binding and Filtering (pp. 105–127). Cambridge, MA: MIT Press. Green, D. W. (1986). “Control, activation, and resource: A framework and a model for the control of speech in bilinguals”. Brain and Language, 27, 210–223. Gürel, A. (2002). Linguistic Characteristics of Second Language Acquisition and First Language Attrition: Turkish overt versus null pronouns. Unpublished Doctoral Dissertation, Montreal: McGill University. Gürel, A. (2004a). “Attrition in First Language Competence: The case of Turkish”. In M. S. Schmid, B. Köpke, M. Keijzer, & L. Weilemar (Eds.), First Language Attrition: Interdisciplinary perspectives on methodological issues (pp. 225–242). Amsterdam: John Benjamins. Gürel, A. (2004b). “Selectivity in L2-induced L1 attrition: A psycholinguistic account”. Journal of Neurolinguistics, 17, 53–78. Kanno, K. (1997). “The acquisition of null and overt pronominals in Japanese by English speakers”. Second Language Research, 13 (3), 299–321. Köpke, B. (2002). “Activation thresholds and non-pathological first language attrition”. In F. Fabbro (Ed.), Advances in the Neurolinguistics of Bilingualism (pp. 119–142). Udine: Forum.
(Psycho)linguistic determinants of L1 attrition 119
Larmouth, D. W. (1974). “Differential interference in American Finnish cases”. Language, 50, 356–366. MacWhinney, B. (1987). “Applying the competition model to bilingualism”. Applied Psycholinguistics, 8, 315–327. Manzini, R. M. & Wexler, K. (1987). “Parameters, binding theory, and learnability”. Linguistic Inquiry, 18 (3), 413–444. Paradis, M. (1989). “Bilingual and polyglot aphasia”. In F. Boller & J. Grafman (Eds.), Handbook of neuropsychology (pp. 117–140). Amsterdam: Elsevier. Paradis, M. (2000). “An integrated neurolinguistic theory of bilingualism (1976–2000)”. LACUS, 27, 5–15. Polinsky, M. (1997). “American Russian: Language loss meets language acquisition”. In W. Browne, E. Dornisch, N. Kondrashova, & D. Zec (Eds.), Annual Workshop on Formal Approaches to Slavic Linguistics (pp. 370–406). The Cornell Meeting, 1995. Ann Arbor, MI: Michigan Slavic Publications. Schaufeli, A. (1996). “Word order patterns in contact: Turkish in the Netherlands”. Southwest Journal of Linguistics, 15 (1–2), 153–169. Schmid, M. S., Köpke, B. Keijzer, M., & Weilemar, L. (Eds.) (2004). First Language Attrition: Interdisciplinary perspectives on methodological issues. Amsterdam: John Benjamins. Seliger, H. (1989). “Deterioration and creativity in childhood bilingualism”. In K. Hyltenstam & L. K. Obler (Eds.), Bilingualism Across the Lifespan (pp. 173–184). Cambridge: CUP. Seliger, H. (1996). “Primary language attrition in the context of bilingualism”. In T. J. Bhatia & W. C. Ritchie (Eds.), Handbook of Second Language Acquisition (pp. 605–626). New York, NY: Academic Press. Seliger, H. W. & Vago, R. M. (Eds.) (1991). First Language Attrition. Cambridge: CUP. Sorace, A. (2000). “Differential effects of attrition in the L1 syntax of near-native L2 speakers”. In C. Howell, S. Fish, & T. Keith-Lucas (Eds.), Proceedings of the 24th Boston University Conference on Language Development (pp. 719–725.). Somerville, MA: Cascadilla Press. Tsimpli, I. M., Sorace, A., Heycock, C., & Filiaci, F. (2004). “First language attrition and syntactic subjects: A study of Greek and Italian near-native speakers of English”. International Journal of Bilingualism, 8 (3), 257–277. White, L. (1989). Universal Grammar and Second Language Acquisition. Amsterdam: John Benjamins. White, L. (2000). “Second language acquisition: From initial to final state”. In J. Archibald (Ed.), Second Language Acquisition and Linguistic Theory (pp. 130–155). Oxford: Blackwell. White, L., Bruhn-Garavito, J., Kawasaki, T., Pater, J. & Prévost, P. (1997). “The researcher gave the subject a test about himself: Problems of ambiguity and preference in the investigation of reflexive binding”. Language Learning, 47 (1), 145–172.
L1 attrition features predicted by a neurolinguistic theory of bilingualism Michel Paradis
McGill University & Cognitive Neuroscience Center, UQAM, Montréal, Canada
The constructs from A Neurolinguistic Theory of Bilingualism (Paradis 2004) that have implications for attrition are outlined and predictions are explored: The activation threshold hypothesis predicts that, all other factors being equal, language disuse leads to gradual loss; the most frequently used elements of L2 will replace their (less used) L1 counterparts; comprehension of forms will be retained longer than the ability to produce them. Elements sustained by declarative memory (e.g., vocabulary) are more vulnerable to attrition than those sustained by procedural memory (i.e., phonology, morphosyntax, lexicon). Declarative items are also more susceptible to interference (and hence to attrition by substitution) than implicit items. Pragmatics and conceptual representations are also modified by attrition. Motivation impacts the rate of attrition. Keywords: neurolinguistics; activation threshold; declarative and procedural memory; motivation
Introduction Many hypotheses integrated into a neurolinguistic theory of bilingualism (Paradis 2004) interact in the process of L1 attrition. In this paper, the constructs in a neurolinguistic theory of bilingualism that have implications for attrition will be briefly outlined (i.e., the relevant components of verbal communication, the activation threshold, the language subsystems, and the three-store hypothesis). Then, the ensuing predictions will be explored: (1) The activation threshold hypothesis predicts that, all other factors being equal, (a) language disuse will lead to gradual language loss; (b) the most frequently used elements of L2 will replace their (less used) L1 counterparts; (c) comprehension of forms will be retained longer than the ability to produce them. (2) Elements sustained by declarative memory (e.g.,
122 Michel Paradis
vocabulary) are more vulnerable to attrition than those sustained by procedural memory (i.e., phonology, morphosyntax, lexicon). (3) These declarative items are also more susceptible to interference (and hence to attrition by substitution) than implicit items. (4) Pragmatics and conceptual representations are modified by attrition. (5) Motivation has an impact on the rate of attrition.
1. Neurofunctional components of verbal communication At least four neurofunctional systems are involved in verbal communication, namely implicit linguistic competence, metalinguistic knowledge, pragmatic capacity and affect/motivation. Each of these systems is necessary but not sufficient. The systems are doubly dissociable by pathology. They may receive inputs from other systems and provide outputs, but these do not alter in any way the internal structure and modes of processing of each system. For example, pragmatic information is used to select from among the available grammatical structures the one that best fits the required meaning in a particular situation (both at encoding and decoding), but pragmatics does not change the existing structure of the grammar. One of the main implications of the theory is that there is no qualitative difference between types of bilingual speakers with respect to these neurofunctional systems, only quantitative ones: they all make use of the same systems, albeit to different extents. These quantitative differences have repercussions for attrition in that the effect of a particular factor will be proportional to the extent of the prevalence of one or another system. Implicit linguistic competence is acquired incidentally, stored implicitly, used automatically, subserved by procedural memory, and is task-specific. In contrast, explicit metalinguistic knowledge is learned consciously, stored explicitly, subserved by declarative memory, and consciously controlled when used (even if speeded-up). All explicit knowledge about a language, whether learned through instruction or from observation, i.e., anything of which one is aware (rules, constructions, forms, form-meaning relationships) is represented in declarative memory. Neurofunctional pragmatic capacity is the ability to select the linguistic and paralinguistic features appropriate to suit the context and to produce the required effect (e.g., humor, sarcasm, indirect speech act, metaphorical meaning), taking into consideration the assumed knowledge shared with one’s interlocutors and their inferred mental and affective states. Motivation is a prerequisite for any act of communication. Without motivation, individuals would never talk. Patients with dynamic aphasia have no impairment in language per se, but there is damage to the connections between the
Implications from a neurolinguistic theory of bilingualism 123
cerebral areas subserving motivation and those that subserve language – and they speak only when insistently prompted. On the other hand, individuals with global aphasia, who are unable to put two words together, will nevertheless forcefully bluster out a complex expression of dissatisfaction in the face of a sudden vexing situation. The lack of motivation at the onset of the microgenesis of utterances produced in a classroom situation (such as asking for an aspirin, not because the speaker has a headache but because it is what is required according to the instructions on page 32) may be one reason for the often mediocre results of foreign language learning. Motivation is known to be the best predictor of success in the acquisition/learning of an L2. Kecskés and Papp (2000) report that students of Russian as a foreign language in Hungary at the time of the Soviet occupation had a significantly poorer command of Russian than those who studied English as a foreign language had of English, even though they had been exposed to their L2 for a longer time than the students of English had been to theirs, reflecting the sociopolitical situation of the time – a good example of the effect of counter-motivation. An even more extreme lack of achievement is reported in French as an L2 in Montreal in the sixties, though here for sociocultural and socioeconomic reasons (Paradis 2004). The implicit linguistic competence for each language is hypothesized to be represented as a subsystem of the general language neurofunctional system. This language system (L) is dissociable from other cognitive systems such as music, arithmetic and face recognition, from which it differs based on the nature of the material being processed. The various languages (L1, L2, L3), which are also doubly dissociable from one another, are however similar in nature and represent parametric variations of general language principles; therefore, they do not constitute different systems altogether, but subsystems of the language system. The processes underlying the implicit competence of each language are more similar to one another than they are to the processes underlying other cognitive systems. Now, the neurofunctional language system itself is known to be composed of several modules (e.g., phonology, morphosyntax, lexical semantics). These modules are independent of each other in that their underlying principles are different in nature. Principles of phonology differ in kind from principles of syntax, for instance. English phonology is more similar to Japanese phonology (different parameters of the same phonological principles) than either is to syntax (whether English or Japanese). An individual speaker’s language neurofunctional system (sustaining implicit linguistic competence), schematically represented as a circle, is thus divided vertically into as many language subsystems as there are languages acquired by the speaker, and each such language subsystem is divided horizontally into modules corresponding to the parametrization of each of the modules of language per se. See Figure 1, where language (L) is represented graphically as
124 Michel Paradis
Figure 1. A schematic graphic representation of the language neurofunctional module L showing two language subsystems, E and J, with their own language-specific modules (phonology, syntax, lexical semantics). (Source: Paradis 2004)
a circle, divided into modules (P, S, Lx) by horizontal lines, and into individual languages (E, J) separated by vertical lines, with their respective modules (EP, ES, ELx; JP, JS, JLx), representing language-specific parametric variations of the L modules, P, S, and Lx. It is further proposed that the language subsystems of an individual are connected to a single conceptual system that contains the representations of conceptual features (the three-store hypothesis). These features are momentarily grouped together in accordance with the specific lexical semantic constraints of words in each language and the relevant pragmatic circumstances at the time of their use, thus bringing a conceptual representation to consciousness. Not all of the features constituting a concept are activated whenever a word is used – only those required by the context (H. Damasio, Tranel, Grabowski, Adolpho & A. Damasio 2004). Hence, whereas, most of the time, some different features are activated in addition to the shared features that make two words (quasi) translation equivalents, in some cases, the very same features may be active when either an L1 word or its translation equivalent is used, if their differences are irrelevant in the context. Each subsystem possesses its own lexicon with its own set of semantic constraints and syntactic properties. Finally, the cerebral substrate of any mental representation requires a certain amount of neural impulses in order to reach activation (its activation threshold). Each time a language item (a word or morphosyntactic construction) is used, its activation threshold is lowered, making it easier to activate again, but it slowly
Implications from a neurolinguistic theory of bilingualism 125
rises when inactive (as evidenced in frequency, recency and priming effects). Production of an item (i.e., recall, self-activation) is more difficult than comprehension of the same item because, among other things, the underlying neural substrate must generate the excitatory impulse from within, as opposed to being excited by stimulation from the incoming auditory or visual signal that impinges on the peripheral sensory organs. This means that comprehension can still be achieved with a threshold that would be too high for production. Activation of a particular element (e.g., a word, a syntactic construction) raises the activation threshold of competing candidates, including translation equivalents.
2. Predictions and some supporting evidence 2.1 The activation threshold The activation threshold hypothesis predicts that: (1) language disuse gradually leads to language loss; (2) the most frequently used elements of L2 will tend to replace their (less used) L1 counterparts; (3) comprehension will be retained longer than production because self-activation requires a lower threshold than comprehension. With extensive use of L2, the L1 threshold for certain items is raised, particularly for declarative items, but eventually also for procedures, which leads to dynamic interference (i.e., L2 procedures are used to generate L1 sentences). Thus, over time, attrition first affects words and frozen expressions, then morphosyntax. In a nutshell, attrition is the result of long-term lack of stimulation. . It is a general condition of neurocognitive functions that when they are not being used, they become more difficult to access over time and eventually atrophy. While this is more apparent for declarative memories, it is also true of skills subserved by procedural memory: Concert pianists and professional tennis players cannot skip practice with impunity. It appears that neurocognitive systems require stimulation to remain in good working condition. Thus, the neural substrate of a cognitive representation is rebooted every time the trace is activated, either by impulses triggered by sensory signals impinging on the peripheral nerve sensors, or deliberately self-activated in cases of willful recall. When the trace is not activated, its resources are slowly depleted and its activation threshold keeps rising (i.e., it gradually needs a greater amount of neural impulses to be activated). One may speculate that a principle of economy is at work: Elements of a system that are not being used are eventually discarded in favor of those that are. A similar phenomenon is observed during the pruning of neural connections early in life; likewise, when a function ceases to be used subsequent to loss of the necessary peripheral mechanisms (such as in acquired blindness or deafness), the underlying substrates of the lost function are often colonized to subserve another function that is now being used to a greater extent in order to compensate.
126 Michel Paradis
The activation threshold hypothesis has been explicitly found to be consistent with attrition data (Gürel 2004; Köpke 2002, 2004). Besides, a number of phenomena reported in attrition studies can be accounted for in terms of activation thresholds. For example, the first sign of language attrition is not the loss of certain items but an increase in the length of time needed for their retrieval (Hansen 2001); re-acquiring a language can be dramatically more rapid than acquiring it the first time (Berman 1979; Slobin, Dasinger, Kyntay & Toupin 1993; Yukawa 1998, 1999); for recall, a high level of activation is needed, whereas for recognition, a lower level is sufficient – if the activation drops even further, then the item can no longer be retrieved (Nelson 1978). In Yukawa’s (1998) study, attrition was the result of a processing failure to use knowledge that had remained virtually intact. Comprehension is retained longer than production (Hulsen 2000). Hulsen (2000) also reports raised levels of L1 inhibition through greater use of L2. Ammerlaan (1996) reports that the main factor influencing degree of attrition is the age at which immigrants left their native land. Subsequent contacts with L1 speakers in the country of residence is also reported to have influenced degree of retention. Similarly, Ramírez (2003) reports an evident correlation between the years lived in a foreign country and the degree of L1 change. She further reports that the intense use of L2 in most social contexts (home, work, friends, etc.) results in changes in the use of oral L1 in informal situations. Ramírez concludes that permanent contact with people who speak the language is an important factor in maintaining L1. Attrition is only one of several sources of evidence consistent with the activation threshold hypothesis. It is further supported by experimental and clinical psycholinguistic evidence. For instance, according to Smith (2005), both Dell’s (1986) spreading activation theory and Coltheart, Curtis, Atkins & Haller’s (1993) dual route cascaded model of reading out loud subscribe to the activation threshold concept. Juncos-Rabadán, Pereiro & Rodríguez (2002) consider that the concept of activation threshold has great explanatory and heuristic power and has proven of great value in therapeutic contexts.
2.2 Subsystems Behavioral and electrophysiological studies indicate that learning a second language late in life has different effects on distinct subsystems within language (Sanders & Neville 2003). One may therefore predict that modules within the L1 subsystem will display varying degrees of attrition. De Bot and Weltens (1993) interpret the findings of cross-linguistic experiments (Brown, Sharma & Kirsner 1984; Kerkman 1984; Kirsner, Smith, Lockhart, King & Jain 1984) as providing
Implications from a neurolinguistic theory of bilingualism 127
support for the subset hypothesis. To the extent that L2 competence is acquired, it forms a subsystem that competes with the L1 subsystem for production. In a neurolinguistic theory of bilingualism (Paradis 2004), the cross-linguistic cognates are (quasi) redundantly represented in each subsystem, for at least three reasons: (1) A particular word may be available to a bilingual aphasic patient, but not its cognate in the other language (the same argument that rules out the extended and tripartite systems); (2) lexical meanings (as well as pronunciations – in spite of their graphic representation) are very rarely identical; and (3) lexical items and their cognate translation equivalents have different intralinguistic connections and often possess different syntactic features (e.g., the verb ‘telephone’ requires a direct object in English but an indirect object in French; ‘information’ is a mass noun in English, but a count noun in French; etc.). Dynamic interference (i.e., interference during processing) is made possible by the fact that each language is a subsystem. The two subsystems are part of the same neurofunctional system; thus, the interference does not come from another system or another module, but is between items of the same nature within a single system. (The implicit linguistic competence for the two languages of bilinguals is not processed by different brain areas but by different neural circuits within the same areas.) Interference from one subsystem on the other may occur when the activation threshold of an item is considerably lower in one subsystem than its equivalent’s threshold in the other. However, there is greater opportunity for interference with items stored in declarative memory (items of which we are conscious – e.g., words, but also any surface forms such as suffixes and prefixes, etc.) because unlike procedural memory, declarative memory is not modalityspecific and hence is easily susceptible to interference from other sources. Psycholinguists have used neurolinguistic arguments in support of the incorporation of declarative and procedural components in their models of language processing (Pienemann, Di Biase, Kawaguchi & Håkansson 2005) and of L2 acquisition and learning (N. Ellis 2005; R. Ellis 2005). Because of the modular nature of each language, each component of L1 (phonology, morphology, syntax, lexical semantics) is selectively susceptible to different rates (faster/slower), extent (more/less), and order (first/last) of attrition. Unlike aphasia, which affects selective modules depending on the site rather than the severity of the lesion, attrition is more liable (as Jakobson (1971) inaccurately predicted for aphasia) to follow the reverse of the developmental sequence found in L1 acquisition: prosody is likely to be attrited last, preceded by other aspects of phonology, syntax and morphology, with words attrited first. (Even though children acquire some words before syntax (i.e., before they start putting two words together), they continue to increase their vocabulary until young adulthood and, to a lesser extent, beyond.)
128 Michel Paradis
The order of word attrition will be a function of the degree of structural proximity of L2 and the frequency of use of L2 cognates and “false friends.” One may speculate that, in language pairs that are maximally distant structurally (e.g., English and Japanese), all things being equal, attrition might be slower because one fewer facilitating factor will be involved (i.e., interference from cognates and false friends) – but so many other factors will still be in play (e.g., motivation/affect, frequency of L1 contacts with friends or through the media, including television, etc.) that it is difficult to make strong predictions.
2.3 Motivation Motivation has been shown to have a considerable impact on L2 acquisition/learning (Schumann 1998). For the same reasons, the theory predicts that motivation has an impact on the rate of attrition. Motivation/affect may play an important role by influencing the activation threshold. Thus attrition may be accelerated by a negative emotional attitude toward L1, which will raise the L1 activation threshold. It may be retarded by a positive emotional attitude toward L1, which will lower its activation threshold. Indeed, higher literacy and better L2 maintenance are generally considered to be attributable to the learner/attriter’s language attitudes and motivation (Shewell & Hansen 1999). Pavlenko (2005) and Schmid (2002, 2004) have shown how emotional investments and disinvestments play a prominent role in L1 attrition. Opitz (2003) recognizes that a range of social and attitudinal factors contribute to the possibility of language loss.
2.4 Pragmatics Attrition of language-specific pragmatics results in changing the interpretation of pragmatic cues. In addition, L2 metaphors get translated into L1 instead of using the L1 metaphor that would be appropriate to the particular circumstance. In the process of acculturation, one will also eventually change one’s responses to particular verbal situations. For instance, a French person, unaccustomed to the frequent compliments in American everyday communication, would automatically respond to the statement “I like your dress” by minimizing the value of the dress, stating that it was a very cheap buy at some flea market. Eventually, however, possibly after several years, she will learn to simply say “thank you,” and many years later may end up automatically replying “merci” to a genuine compliment from a French speaker. Japanese students who have spent some years at an American institution will report that they have lost the automatic use of the complex system
Implications from a neurolinguistic theory of bilingualism 129
of politeness markers when, for instance, addressing older Japanese students or professors.
2.5 Semantic and conceptual levels The three-store hypothesis and the activation threshold together predict the attrition of concept boundaries so that L1 words no longer activate their original conceptual meanings, but incorporate some or all conceptual features of their L2 equivalents. Semantic constraints on the use of L1 words become blurred. False friends from L2 will creep into L1 (e.g., an English speaker living in a French environment will call a lecture a conference) or a less colloquial calque will be used (he will tell a friend that his flight was annulled – rather than cancelled). Most psycholinguists today have adopted some form of the three-store hypothesis in that lexical representations are considered to be connected to a common conceptual system (Abel 2003; Brauer 1998; Francis 1999; Grainger 1993; Grosjean 1982, 1998; Hermans, Bongaerts, de Bot & Schreuder 1998; Klosty Beaujour 1989; Laganaro & Overton Venet 2001; Larsen, Schrauf, Fromholt & Rubin 2002; Malmkjær 1991; Paivio & Desrochers 1980; Pavlenko 1999, 2002; Perecman 1984). However, the proposed neurolinguistic theory of bilingualism insists that there is a lexical semantic representational level, part of each language subsystem, that is independent of the corresponding conceptual representation, because aphasia does not necessarily affect the conceptual representations (Alajouanine & Lhermitte 1964; Gurd & Marshall 1993; Lecours & Joanette 1980; Marshall 1984). The distinction between the lexical semantics and conceptual levels is beginning to emerge in psycholinguistics (e.g., Francis 2005; de Groot & Poot 1997; Holowka, Brosseau-Lapré & Petitto 2002; Schriefers 1990) and neurology (Damasio et al. 2004), and at least implicitly in various domains of language research (Haritos & Nelson 2001; Hudon 2004; Koch 2005) and artificial intelligence (Jiang, Zong & Deng 2005; Sona, Avesani & Moskovitch 2005). There is, unfortunately, still some confusion in the literature regarding the use of the terms semantic and conceptual. Often the term conceptual refers to the lexical semantic meaning and semantic refers to the conceptual level (see Francis 2005, for a discussion of the confusion). Because the sound-meaning (and graph-meaning) relationships of words are conscious and hence stored in declarative memory, they are attrited before the rest of language. Ramírez (2003) found that most examples of changes fell into the category of semantic enrichment due to the cross-linguistic influence of L2. In most cases, the native speakers assigned a different meaning to an L1 word on the basis of similarity with a word in L2.
130 Michel Paradis
2.6 Loss or inaccessibility? Is L1 ever lost? In very young immigrants, possibly, if the language is not yet wellestablished at the time of immigration and is never used thereafter (Ventureyra, Pallier & Yoo 2004). In adult immigrants, probably not, but it eventually becomes inaccessible because of its overly high activation threshold: individuals have been reported to regain an attrited language under hypnosis (As 1962; Fromm 1970; Footnick, Pallier & Prieur 2005; cf. Footnick, this volume) or in altered brain states such as high fever, temporary insanity and cerebral trauma (Rush 1812).
3. Conclusions From a neuropsychological point of view, three phenomena conspire to bring about L1 attrition: (1) the characteristics of declarative and procedural memory systems; (2) the activation threshold; (3) the organization of implicit linguistic competence into subsystems. In addition to the various parts of the grammar, L1 pragmatics and the selection of conceptual features corresponding to the meanings of L1 words may be affected by attrition. The phenomenon of attrition, like that of recovery patterns in bilingual aphasia, is a very complex one, subject to influences from different factors whose combination results in the facilitation or hindrance of a particular neurofunctional subsystem (here, L1). There may be a primary cause (i.e., disuse) modulated by a number of other factors (e.g., age, attitude, motivation, structural similarity between the languages, degree of automatization of each, amount of metalinguistic knowledge, etc.) that increase or decrease the rate of attrition.
References Abel, B. (2003). “English idioms in the first language and second language lexicon: A dual representation approach”. Second Language Research, 19, 329–358. Alajouanine, T. & Lhermitte, F. (1964). “Non-verbal communication in aphasia”. In A. V. S. de Reuck & M. O’Connor (Eds.), Disorders of Language (pp. 168–182). London: Churchill. Ammerlaan, A. (1996). “You get a bit wobbly …”: Exploring bilingual lexical retrieval processes in the context of first language attrition. Unpublished Doctoral Dissertation, Nijmegen: Katholieke Universiteit Nijmegen. As, A. (1962). “The recovery of forgotten language knowledge through hypnotic age regression: A case report”. American Journal of Clinical Hypnosis, 5, 24–29.
Implications from a neurolinguistic theory of bilingualism 131
Berman, R. A. (1979). “The re-emergence of a bilingual: A case study of a Hebrew-English speaking child”. Working Papers on Bilingualism, 19, 157–180. Brauer, M. (1998). “Stroop interference in bilinguals: The role of similarity between the two languages”. In A. F. Healy & L. E. Bourne, Jr. (Eds.), Foreign Language Learning: Psycholinguistic studies on training and retention (pp. 317–337). Mahwah, NJ: Lawrence Erlbaum Associates. Brown, H., Sharma, N. & Kirsner, K. (1984). “The role of script and phonology in lexical representation”. Quarterly Journal of Experimental Psychology, 36A, 491–505. Coltheart, M., Curtis, B., Atkins, P. & Haller, M. (1993). “Models of reading aloud: Dual route and parallel distributed processing approaches”. Psychological Review, 100, 589–608. Damasio, H., Tranel. D., Grabowski, T., Adolpho, R. & Damasio, A. (2004). “Neural systems behind word and concept retrieval”. Cognition, 92, 179–229. de Bot, K. & Weltens, B. (1993). “Lexical aspects of language attrition and shift”. In G. Extra & L. Verhoeven (Eds.), Immigrant Languages in Europe (pp. 308–322). Clevedon: Multilingual Matters. de Groot, A. M. B. & Poot, R. (1997). “Word translation at three levels of proficiency in a second language: The ubiquitous involvement of conceptual memory”. Language Learning, 47, 215–264. Dell, G. (1986). “A spreading activation theory of retrieval in sentence production”. Psychological Review, 93, 283–321. Ellis, N. C. (2005). “At the interface: Dynamic interactions of explicit and implicit language knowledge”. Studies in Second Language Acquisition, 27, 305–352. Ellis, R. (2005). “Measuring implicit and explicit knowledge of a second language – A psychometric study”. Studies in Second Language Acquisition, 27, 141–172. Footnick, R., Pallier, C. & Prieur, N. (2005). “A lost language: Recovery is triggered by hypnosis”. Paper presented at the Second International Conference on First Language Attrition. Amsterdam: 17–20 August. Francis, W. S. (1999). “Cognitive interpretation of language and memory in bilinguals – Semantic representations”. Psychological Bulletin, 125, 193–222. Francis, W. S. (2005). “Bilingual semantic and conceptual representation”. In J. F. Kroll & A. M. B. de Groot (Eds.), Handbook of Bilingualism: Psycholinguistic perspectives (pp. 251–267). Oxford: OUP. Fromm, E. (1970). “Age regression with unexpected reappearance of a repressed childhood language”. International Journal of Clinical and Experimental Hypnosis, 18, 79–88. Grainger, J. (1993). “Visual word recognition in bilinguals”. In R. Schreuder & B. Weltens (Eds.), The Bilingual Lexicon (pp. 11–25). Amsterdam: John Benjamins. Grosjean, F. (1982). Life with Two Languages: An introduction to bilingualism. Cambridge, MA: Harvard University Press. Grosjean, F. (1998). “Studying bilinguals: Methodological and conceptual issues”. Bilingualism: Language and Cognition, 1, 131–149. Gurd, J. M. & Marshall, J. C. (1993). “Aphasia and intelligence”. In G. Blanken, J. Dittmann, H. Grimm, J. C. Marshall, & K. Wallesch (Eds.), Linguistic Disorders and Pathologies (pp. 310–317). Berlin: Walter de Gruyter. Gürel, A. (2004). “Selectivity in L2-induced L1 attrition: A psycholinguistic account”. Journal of Neurolinguistics, 17, 53–78. Hansen, L. (2001). “Language attrition: The fate of the start”. Annual Review of Applied Linguistics, 21, 60–73.
132 Michel Paradis
Haritos, C. & Nelson, K. (2001). “Bilingual memory: The interaction of language and thought”. Bilingual Research Journal, 25, 417–439. Hermans, D., Bongaerts, T., de Bot, K. & Schreuder, R. (1998). “Producing words in a foreign language: Can speakers prevent interference from their first language?” Bilingualism: Language and Cognition, 1, 213–229. Holowka, S., Brosseau-Lapré, F. & Petitto, L. A. (2002). “Semantic and conceptual knowledge underlying bilingual babies’ first signs and words”. Language Learning, 52, 205–262. Hudon, M. (2004). “Conceptual and lexical compatibility in thesauri used to describe and access moving image collections”. Canadian Journal of Information and Library Science, 28, 94. Hulsen, M. (2000). Language Loss and Language Processing: Three generations of Dutch migrants in New Zealand. Unpublished Doctoral Dissertation, Nijmegen: Katholieke Universiteit Nijmegen. Jakobson, R. (1971). Studies in Child Language and Aphasia. New York, NY: Mouton. Jiang, M.-G., Zong, C.-Q. & Deng, B.-X. (2005). “Self-organizing map analysis of conceptual and semantic relations for noun”. Lecture Notes in Computer Science, 3498, 977–982. Juncos-Rabadán, O., Pereiro, A. X. & Rodríguez, M. J. (2002). “Treatment of aphasia in bilingual subjects”. In F. Fabbro (Ed.), Advances in the Neurolinguistics of Bilingualism (pp. 275–298). Udine: Forum. Kecskés, I. & Papp, T. (2000). “Metaphorical competence in trilingual language production”. In J. Cenoz & U. Jessner (Eds.), English in Europe (pp. 99–120). Clevedon: Multilingual Matters. Kerkman, H. (1984). “Woordherkenning in twee talen”. In A. Thomassen, L. Noordman & P. Eling (Eds.), Het Leesprocess (pp. 139–152). Lisse: Swetz & Zeitlinger. Kirsner, K., Smith, M., Lockhart, R., King, M. & Jain, M. (1984). “The bilingual lexicon: Language-specific units in an integrated network”. Journal of Verbal Learning and Verbal Behavior, 23, 519–539. Klosty Beaujour, E. (1989). Alien Tongues. Ithaca, NY: Cornell University Press. Koch, P. (2005). “Aspects cognitifs d’une typologie lexicale synchronique. Les hiérarchies conceptuelles en français et dans d’autres langues”. Langue française, 145, 11–37. Köpke, B. (2002). “Activation thresholds and non-pathological L1 attrition”. In F. Fabbro (Ed.), Advances in the Neurolinguistics of Bilingualism (pp. 119–142). Udine: Forum. Köpke, B. (2004). “Neurolinguistic aspects of attrition”. Journal of Neurolinguistics, 17, 3–30. Laganaro, M. & Overton Venet, M. (2001). “Acquired alexia in multilingual aphasia and computer-assisted treatment in both languages: Issues of generalization and transfer”. Folia Phoniatrica et Logopaedica, 53, 135–144. Larsen, S. F., Schrauf, R. W., Fromholt, P. & Rubin, D. C. (2002). “Inner speech and bilingual autobiographical memory: A Polish-Danish cross-cultural study”. Memory, 10, 45–54. Lecours, A. R. & Joanette, Y. (1980). “Linguistic and other psychological aspects of paroxysmal aphasia”. Brain and Language, 10, 1–23. Marshall, J. C. (1984). “Multiple perspectives on modularity”. Cognition, 17, 209–242. Malmkjær, K. (1991). The Linguistics Encyclopedia. London: Routledge. Nelson, T. O. (1978). “Detecting small amounts of information in memory: Savings from nonrecognized items”. Journal of Experimental Psychology: Human Learning and memory, 4, 453–468. Opitz, C. (2003). “L1 attrition and L2 acquisition – The impact of L2 proficiency on L1”. In L. Sayahi (Ed.), Selected Proceedings of the First Workshop on Spanish Sociolinguistics. Somerville, MA: Cascadilla Press.
Implications from a neurolinguistic theory of bilingualism 133
Paradis, M. (2004). A Neurolinguistic Theory of Bilingualism. Amsterdam: John Benjamins. Paivio, A. & Desrochers, A. (1980). “A dual-coding approach to bilingual memory”. Canadian Journal of Psychology, 34, 388–399. Pavlenko, A. (1999). “New approaches to concepts in bilingual memory”. Bilingualism: Language and Cognition, 2, 209–230. Pavlenko, A. (2002). “Conceptual change in bilingual memory: A NeoWhorfian approach”. In F. Fabbro (Ed.), Advances in the Neurolinguistics of Bilingualism (pp. 69–94). Udine: Forum. Pavlenko, A. (2005). Emotions and Multilingualism. Cambridge: CUP. Perecman, E. (1984). “Spontaneous translation and language mixing in a polyglot aphasic”. Brain and Language, 23, 43–53. Pienemann, M., Di Biase, B., Kawaguchi, S. & Håkansson, G. (2005). “Processing constraints on L1 transfer”. In J. F. Kroll & A. M. B. de Groot (Eds.), Handbook of Bilingualism: Psycholinguistic perspectives (pp. 128–153). Oxford: OUP. Ramírez, D. B. (2003). “L1 regression in Spanish-English bilinguals”. In L. Sayahi (Ed.), Selected Proceedings of the First Workshop on Spanish Sociolinguistics (pp. 89–95). Somerville, MA: Cascadilla Press. Rush, B. (1812). Medical Inquiries and Observations upon Diseases of the Mind. Philadelphia,PA: Kimber. Sanders, L. D. & Neville, H. J. (2003). “An ERP study of continuous speech processing II. Segmentation, semantics, and syntax in non-native speakers”. Cognitive Brain Research, 15, 214–227. Schmid, M. S. (2002). First Language Attrition, Use and Maintenance: The case of German Jews in Anglophone countries. Amsterdam: John Benjamins. Schmid, M. S. (2004). “Identity and first language attrition: A historical approach”. Estudios de Sociolingüística, 5, 41–58. Schriefers H. (1990). “Lexical and conceptual factors in the naming of relations”. Cognitive Psychology, 22, 111–142. Schumann, J. H. (1998). “The neurobiology of affect in language”. Language Learning, 48, Supplement 1. Shewell, J. & Hansen, L. (1999). “The role of alphabetic literacy in second language retention: The case of Korean”. Paper presented at the conference on Bilingualism: Acquisition and Attrition. Applied Linguistics Association of Korea, Seoul. [Cited in Hansen, L. (2001). Language attrition: The fate of the start.] Slobin, D. I., Dasinger, L., Kyntay, A. & Toupin. C. (1993). “Native language reacquisition in early childhood”. In E. V. Clark (Ed.), Proceedings of the 24th Annual Child Language Research Forum (pp. 179–196). Stanford: Stanford University. Smith, D. J. (2005). http://www.smithsrisca.demon.co.uk/ psycholinguistics-glossary.html. Sona, D., Avesani, P. & Moskovitch, R. (2005). “Helping physicians to organize guidelines within conceptual hierarchies”. Lecture Notes in Artificial Intelligence, 3581, 141–145. Ventureyra, A. G., Pallier, C. & Yoo, H.-Y. (2004). “The loss of first language phonetic perception in adopted Koreans”. Journal of Neurolinguistics, 17, 79–91. Yukawa, E. (1998). L1 Japanese Attrition and Regaining: Three case studies of two early bilingual children. Tokyo: Kurosio. Yukawa, E. (1999). “L1 Japanese attrition and regaining: The age and pre-attrition proficiency variables”. In P. Robinson (Ed.), Representation and Process: Proceedings of the 3rd Pacific Second Language Research Forum, Vol. 1 (pp. 321–335). Tokyo: Aoyama University.
The role of L1 use for L1 attrition Monika S. Schmid
Rijksuniversiteit Groningen, The Netherlands
While the factor ‘language contact’ is often named among the most important for the development of individual language attrition, empirical validations of this claim are few and far between. This contribution argues that a bilingual’s use of the first language comprises very diverse situations which cannot be subsumed under one predictor variable. Grosjean’s (2001) framework of language modes is invoked as a useful way of structuring the use of the L1 by immigrants. A statistical investigation of these different types of L1 use on the one hand and language proficiency data on the other demonstrates that the impact of both active and passive exposure to the first language on attrition is anything but straightforward. Keywords: language mode; language use; activation; inhibition; German
Introduction One of the basic predictions of psycholinguistic research with respect to L1 attrition is that language loss can be attributed to language disuse (e.g. Paradis, this volume; Köpke, this volume). According to this prediction, attrition will be most radical among those individuals who rarely or never speak their L1 in daily life, while those speakers who use the L1 regularly, for example within their family or with friends, will to some degree be protected against its deterioration. This assumption is based on the simple fact that rehearsal of information can maintain accessibility. There is ample evidence to demonstrate that in bilinguals, the accessibility of items in either language system depends on what has been called the Activation Threshold of the item, and that this threshold is a function of frequency (how often has the item been called upon?) and time (how long ago was it last activated?) (e.g. Paradis 1993, 2004). Disuse of a language system affects accessibility of lexical items most immediately, but will eventually also impact on grammatical
136 Monika S. Schmid
knowledge (Paradis 2004, this volume; Köpke, this volume). In other words, the less often a bilingual uses one of her languages, the more difficult she will find it to retrieve the correct lexical and grammatical information from memory under the time pressure of normal discourse. With respect to L1 attrition, Paradis (this volume) makes a number of predictions from the point of view of the Activation Threshold Hypothesis (ATH, see Paradis 2004), among which are the following: a. language disuse will lead to gradual language loss; b. the most frequently used elements of L2 will replace their (less used) L1 counterparts. Prediction (a) refers to the behaviour of individual speakers, while prediction (b) hinges on the characteristics of the linguistic systems which interact in the contact situation. Interestingly, most specific applications of the ATH have investigated hypothesis (b) (Gürel 2004; Köpke 2002; Schmitt to appear). The overall degree to which a speaker goes on using her L1, on the other hand, has largely gone uninvestigated. This chapter will present some findings which indicate that the relationship L1 use/L1 attrition is not a straightforward one, and that not all situations of L1 use impact on attrition in the same way. A classification of different types of L1 use, based on Grosjean’s model of language modes (Grosjean 2001), is proposed, and it is hypothesized that frequent L1 use in the different modes may impact differently on the attrition of an L1.
1. The role of L1 contact for attrition The amount of use which a potential attriter makes of her L1 strikes most researchers intuitively as one of the most important factors in determining the attritional process (e.g. Cook 2005; Paradis, this volume). There is, however, little direct evidence that the degree to which a language system will attrite is dependent on the amount to which the language is being used in everyday life. Only two studies report that those subjects who used their L1 on an extremely infrequent basis showed more attrition over time (de Bot, Gommans & Rossing 1991 and Köpke 1999). On the other hand, there is also some evidence for a negative correlation, suggesting that the attriters who used their L1 on a daily basis actually performed worse on some tasks (Jaspaert & Kroon 1989). When findings are contradictory, more often than not it is the methodology which is at fault. In the case at hand the discrepancy might be due to an unwar-
The role of L1 use for L1 attrition 137
ranted simplification of a set of relationships and speech situations. “Language use” refers to a complex pattern of behaviour in everyday interaction, and can therefore probably not easily be reduced to one dichotomous factor, which is what all of the studies cited above attempt to do. De Bot et al. (1991) and Köpke (1999) make a distinction between “more” and “less” use of the L1, while Jaspaert & Kroon (1989) use the L1 of the subject’s partner as a measuring stick. This latter factor also played an important role for Schmid (2002) and proved significant for a number of linguistic variables – in this case, however, the correlation was positive, suggesting that speakers whose partner had a different L1 from themselves made more rather than fewer errors. A further distinction of self-reported L1 use in a number of contexts (with parents, partner, siblings, and children) could not account for the variance among the attrition data found in that study either (Schmid 2002). The only study to date which attempts to assess the impact of L1 use in a more detailed framework is Hulsen’s work on three generations of Dutch speakers in New Zealand (2000). Hulsen investigated L1 attrition within a Social Networks framework, and found that the amount of L1 contact speakers had, particularly in their primary network, correlated strongly with speed and accuracy on a naming task. While this is an important study and an interesting finding, Hulsen does not distinguish first, second and third generation speakers in her analysis. It would therefore be premature to generalise her findings to the one group of speakers which we might call ‘true’ L1 attriters – speakers who completed their L1 acquisition in a monolingual setting before emigration – as opposed to heritage speakers who grew up bilingually and whose acquisition of Dutch will probably have been incomplete to some degree.
2. Language modes The inconsistencies reported above suggest that there are important methodological issues to be addressed where the role of L1 use for L1 attrition is concerned. At the root of the problem may be the simple fact that, among bilinguals, L1 use does not necessarily equal L1 use. As Grosjean (2001) points out, it is crucial for investigations of bilingualism to distinguish between language use in the monolingual and bilingual mode (see Fig. 1). The difference between the monolingual and bilingual modes hinges on the relative degrees of activation of a bilingual’s linguistic systems, symbolically represented in this diagram by the colour of the squares: the darker the colour, the more highly activated the corresponding linguistic system. In the monolingual mode (1), one of the bilingual’s languages is largely deactivated or inhibited
138 Monika S. Schmid
LANGUAGE A (base language)
MONOLINGUAL LANGUAGE MODE
1
2
3
BILINGUAL LANGUAGE MODE
LANGUAGE B
Figure 1. The language mode continuum (Grosjean 2001: 3, his Fig. 1.1)
(Green 1998; Paradis 1993), while the other is highly activated. In the bilingual mode (3), both languages are highly active, and in this type of situation, language mixing, codeswitching and interferences are very frequent. (2) represents the intermediate mode, where Language B is not switched off, but still far less active than Language A. The language mode is dependent on a number of factors in any communicative setting, such as participants, situation, form and content of the message and function of the language act (Grosjean 2001: 5). Interactions with monolingual speakers of either of a bilingual’s languages will typically be in the monolingual mode (unless control of language mode has been impaired, for example due to a pathological condition) and there will be little codeswitching and interference from Language B. Informal interactions with other bilinguals, on the other hand, will often take place in the bilingual mode. If, however, two bilinguals interact in a more formal context, or if the speaker knows that her interlocutor does not like to mix languages, codeswitching and interferences will be reduced, although Language B will still remain active. Such situations, then, will usually take place in the intermediate mode. The language mode continuum has, so far, been most often applied to experimental settings, where bilinguals typically perform differently depending on the language mode within which they complete a given exercise. However, language modes also obtain outside the experimental context, in daily life. Every instance of language use by a bilingual is situated somewhere along the language mode continuum. For potential language attriters, this implies a variety of language
. There is strong experimental evidence (e.g. van Hell & Dijstra 2002) that none of a bilingual’s languages can ever be completely switched off, so the square can never become entirely white (Grosjean 2001).
The role of L1 use for L1 attrition 139
Figure 2. Types of L1 use among emigrants
mode settings: a German-speaking couple, for instance, who emigrated together and still use German frequently among themselves will probably do so in the bilingual mode, codeswitching frequently. They will also often use the L2 with each other, for example when non-German speaking friends are present. On the other hand, when one of them speaks to a friend or relative back in Germany, particularly someone who is unfamiliar with the L2, the conversation will be largely in the monolingual mode. On the basis of the language mode continuum, we can therefore distinguish five types of everyday language use among potential attriters (Fig. 2). At the extreme ends of this continuum, largely monolingual use of L1 (Type I) and L2 (Type V) are situated. Type V, the use of the L2 with native speakers of that language who are unfamiliar with the attriter’s L1, is probably the most common language use situation in daily life for the average subject of attrition studies – well-integrated immigrants with a relatively high proficiency in L2. Type I situations, on the other hand, usually play a quantitatively relatively insignificant role for most of these speakers, as there will be few monolingual speakers of that language with whom they have daily contact. This type of language use only applies to contacts with and in the country of origin – either in distance communication via telephone, letters and/or email or during visits (both from and to that country). The midrange of the continuum contains two situations of bilingual language use, one where L1 is the base language and one where L2 takes on that role. Informal language use among bilinguals, such as language use within the family and among friends, fall into this category. It is not possible to make a classificatory distinction between the two situational types included here, since the base language
140 Monika S. Schmid
may change from exchange to exchange and therefore can only be determined if the actual speech situation has been documented. Both are therefore included under Type III. Type II and Type IV are then the intermediate situations, where both languages are active to some degree, but one of them is used predominantly, and switches and mixing are avoided. Emigrants often report Type IV use – predominant use of the L2 with other bilinguals – when someone is present who is trying to learn the L2 (e.g. other, more recent, emigrants; visitors or Au Pairs), or when both monolinguals and bilinguals participate in the conversation. Type II situations typically include L1 use for professional purposes, for example by language teachers (who try to stick to the L1 as far as possible, but sometimes have to take recourse to the L2 for more complicated explanations), translators (who receive L2 in the input but have to produce target-like, unmixed L1 in the output) or foreign language correspondents. This type of L1 use may also include interactions in heritage language clubs, societies or churches. These exchanges are usually constrained by social pressures to keep the L1 ‘unmixed’, since the L1 and the desire to maintain it is the common denominator shared by the members of the club or church (this function will be even stronger if the L1 has a particular role for the religion). As there may be great ideological variance with respect to code-mixing between individual societies (the reports I had from the members of the German and Austrian Vancouver ‘Alpenclub’, for example, point towards a rather strict policy of non-mixing among its members, while Barbara Köpke (p.c.) reports a great deal of codeswitching among the members of a Montreal German language society), it will have to be established on a case-by-case basis whether such instances of L1 use are to be classified as Type II or Type III. To summarize, I propose a distinction of the following types of L1 use: Type I: monolingual mode L1 use – distance communication with country of origin (telephone, email, letters) – visits to and from country of origin Type II: intermediate mode L1 use – professional L1 use – L1 use in clubs and societies with an (explicit or tacit) non-mixing policy Type III: bilingual mode L1 or L2 use – L1 use within the family – L1 use with friends, acquaintances, colleagues
. Note that predominantly social or informal use of the L1 with other bilingual colleagues does not fall into this category.
The role of L1 use for L1 attrition 141
Type IV: intermediate mode L2 use – L2 use with recent emigrants wishing to acquire L2 – L2 use with native speakers who have a rudimentary knowledge of L1 Type V: monolingual mode L2 use – L2 use with monolingual speakers – L2 use with bilingual native speakers of other languages Among the groups of bilingual speakers typically investigated in language attrition studies (immigrants from cultural background relatively similar to that of the host country, e.g. Western Europeans in North America or Australia/New Zealand), Type V is probably the most frequent speech situation by far, as most of these speakers are well-integrated into the host society and use their L2 both professionally and socially on a daily basis. Type III L1 use will also be relatively frequent for some speakers (but not for all) as will Type II. On the other hand, Type I and Type IV will probably not be used on a very frequent basis by most speakers. How often the L1 is used in these different types of situations in daily life might impact differently upon L1 attrition, since selection of the language mode is achieved by two processes: activation and inhibition. These two processes also determine the activation threshold (AT) of any given linguistic item or language system (Paradis 2004). The level of activation corresponds to the frequency and recency with which any particular item (or the overall language system) has been called upon, so the prediction is that those speakers who make more use of their L1 (and particularly those who have done so recently) will experience less accessing difficulties and other attrition phenomena. Inhibition, on the other hand, in our context refers to the fact that in order to be able to successfully access any item of linguistic knowledge, a speaker has to inhibit all competitors. Inhibition of L1 is therefore a process that speakers operating in an L2 environment routinely have to perform, and by so doing, the activation threshold of L1 items and of the overall L1 system are raised. Since the different types of language mode introduced above may impact on activation levels of L1 and L2 rather differently, they should not be lumped together under the common factor ‘L1 contact’ in attrition studies. Type I use will require the inhibition of L2 to some degree. In this type of monolingual L1 use, effort expended on inhibition will be highest at the beginning of the interaction or visit, and then decrease as the monolingual mode in L1 has been established. The same process, mutatis mutandis, goes for activation and inhibition of L2 and L1, respectively, in Type V situations. The intermediate types, II and IV, on the other hand, will require constant inhibition of L2 or L1, as external stimuli and the knowledge that the interlocutors are also bilingual will make it more difficult to entirely deactivate that system. Lastly, in Type III interactions, little effort will
142 Monika S. Schmid
have to be expended on the inhibition of either L1 or L2, since the use, mixing and switching of both is socially acceptable and contextually appropriate here. From the point of view of the ATH, it can be predicted that frequent use of L1 in the monolingual mode (Type I) may help prevent L1 attrition, while frequent use of monolingual mode L2 (Type V) may help accelerate it. This latter type of language use (interaction with monolingual speakers of the L2) is probably the most frequent for most attriters and used on a daily basis. On the other hand, experience shows that the quantitative differences with respect to situations where L1 is the base language are quite dramatic among potential attriters. It is therefore problematic that the distinction between the different modes of L1 use is not usually made or acknowledged in L1 attrition studies.
3. Predictions The ATH predicts that disuse of L1 and frequent use of L2 will initially lead to a higher activation threshold, that is, reduced accessibility, of lexical knowledge (Paradis, this volume). In other words, the area that will be affected first and most severely by attrition is lexical access. This will lead to a number of well-documented attrition phenomena which will manifest themselves in the following ways: – impaired performance on lexical naming and/or fluency tasks (slower responses, reduced accuracy, e.g. Ammerlaan 1996; Hulsen 2000) – a decrease in lexical richness in free speech (higher type-token ratios, e.g. de Bot & Clyne 1994; Schmid 2005) – an increase in hesitation phenomena in free speech (pauses, filled pauses, repetitions, self-repair, e.g. Nakuma 1997) This paper will attempt to assess the impact of the mode of L1 use as predictor variables on these indicators of attrition.
4. Method 4.1 Subjects The data for this study were collected from three groups of L1 speakers of German, one in an L2 English context (in the area of Greater Vancouver, BC, Canada; n = 53) one in an L2 Dutch context (in the area commonly referred to as the “Randstad”, the densely-populated area of the Netherlands between Rotterdam and Amsterdam; n = 53) and a control group of speakers in Germany (in the
The role of L1 use for L1 attrition 143
Table 1. Age factors across groups* CA (n = 53) mean stdev Age 63.23 26.13 Age at emigration Length of residence 37.09
10.92 7.07 12.25
NL (n = 53) mean stdev
CG (n = 53) mean stdev
Total (n = 159) mean stdev
63.36 29.08 34.28
60.89 – –
62.49 27.60 35.69
9.55 7.46 11.02
11.60 – –
10.71 7.42 11.74
* CA=German speakers in Canada, NL = German speakers in The Netherlands, CG = control group of German speakers in Germany. Group size is not always consistent, since some speakers missed some tests
Table 2. Education (EDU) across groups Level 1 Level 2 Level 3 Level 4
CA (n = 53)
NL (n = 53)
CG (n = 53)
Total (n = 159)
13 22 5 13
9 21 6 17
13 23 6 11
35 66 17 41
Rhineland and the lower Rhine area) who had never lived abroad for an extended period of time (n = 53). Contact with the émigrés (henceforth: attriters) was made through advertisements in both German and English/Dutch newspapers, through German clubs, churches, libraries, schools and TV channels, and through further contacts suggested by the participants themselves. The control group subjects were approached through advertisements in newspapers, through clubs and organizations whose members were assumed to be in the target age group, and through personal contacts. We stipulated as criteria for participation for the attriters that they should have lived in an L2 environment for at least 10 years (one exception was made for a German speaker in Canada who was very eager to participate, but whose emigration was only 9 years ago). All efforts were made to control factors such as age, age at emigration, and length of residence (LOR) across groups (see Table 1). In addition, the groups were controlled for sex (it was not possible to achieve a 50–50 distribution of men and women, the CA and CG groups contained 35 women and 18 men, the NL group contained 34 women and 19 men) and education. For education, a variable with four levels was chosen: Level 1 comprised those subjects who attended school until the end of the German Volksschule or Hauptschule, subjects in level 2 completed the German Realschule or Mittlere Reife, subjects in level 3 had obtained the Abitur or Fachabitur, and subjects in level 4 had received a university degree. The distribution of these levels across groups was as in Table 2. Unfortunately, as is clear from this table, it turned out impossible for practical reasons to achieve a complete match across groups with respect to this factor.
144 Monika S. Schmid
4.2 Materials 4.2.1 Independent variables The aim of this investigation was to assess to what degree frequency of use of the L1 in everyday life impacts on overall performance in that language, and what role the language mode in everyday L1 use plays in this respect. Information on L1 use was elicited by means of a sociolinguistic and personal background interview, based on a catalogue of 78 questions on personal background and lasting between ½ and 1½ hours. The interview was conducted in German. The question catalogue contained a number of binary or yes/no questions (such as gender), a set of ordinal variables (such as education level), a large number of 5-point Likertscale preference or frequency indications (such as L1 use in daily life, with family and friends, cultural affiliation, language preference), and some genuine interval variables such as age and length of residence in the country of emigration. This interview was recorded and later transcribed orthographically (at the time of writing, these data are in the process of being analyzed linguistically). For the purpose of the present analysis, three independent variables pertaining to the frequency of language use were then established: bilmod This variable is an average of a total of 18 Likert-scale questions on the frequency of L1 use with family and friends, i.e. in typically bilingual-mode settings intmod This variable is an average of a total of 4 Likert-scale questions on the frequency of L1 use in clubs, churches and at work, i.e. in typically intermediate-mode settings monmod This variable is an average of a total of 2 Likert-scale questions on the frequency of L1 use with speakers in Germany, i.e. in typically mono- lingual-mode settings The averages per group for these variables are summarized in Table 3.
4.4.2 Dependent variables The experimental design consisted of a number of tests intended to elicit an overall picture of individual L1 proficiency (for a description of the overall design see
. In line with common practice in the social sciences, these Likert-Scales were treated as interval variables. . The term “family” here refers to partner, children and grandchildren in the country of emigration.
The role of L1 use for L1 attrition 145
Table 3. L1 contact and use per emigration group CA (n = 53) BilModTot IntModtot MonModTot
NL (n = 53)
Mean
Max.
Min.
Mean
Max.
Min.
0.28 0.29 0.55
0.66 0.94 1.00
0.00 0.00 0.00
0.28 0.35 0.63
0.75 0.81 1.00
0.00 0.00 0.25
Keijzer & Schmid 2005). For the purpose of the present analysis, the results from two of these tests will be used.
(a) Verbal Fluency (VF) The first of these consisted of two verbal fluency (VF) tasks (Goodglass & Kaplan 1983). In this task, the subject is invited to produce as many items as she can that belong to a particular semantic field (e.g. animals) or fulfil a particular phonological condition (e.g. start with the letter ‘p’). Following Yağmur (1997) two semantic stimuli were used: animals on the one hand, and fruit and vegetables on the other, and a 60-second production period for both. Since it was the purpose of this task to establish lexical access in the L1, all items which were correct German terms for objects in those fields were scored, while L2 items and repetitions were omitted from the count. Items which were German terms, but were used with an incorrect article or plural allomorph were included in the count. (b) Charlie Chaplin film retelling task The second experiment was designed to elicit relatively free speech in a controlled context. This was done by the Charlie Chaplin film retelling task (Perdue 1993). In this task, the subject is asked to watch a 10-minute sequence from the silent Charlie Chaplin movie Modern Times, and retell what happens in her own words afterwards. These retellings were between 3 and 12 minutes long (max. 2296 tokens, min 176 tokens, mean 753 tokens). They were orthographically transcribed and converted to CHAT-format (for a description of CHAT see http://childes.psy. cmu.edu/).
. The sequence commonly used in this experiment starts ca. 33 minutes into the film, at the scene where Chaplin (freshly released from prison) applies for work at a shipbuilding company. It ends ca. 10 minutes later, when Chaplin and Goddard wake up from their daydream outside a little suburban home, to find a policeman is watching them, and walk off together. . I am grateful for the dedicated work of two student assistants, Anja Fislage and Linda Marie Schulhof, who were responsible for most of the transcriptions.
146 Monika S. Schmid
Table 4. Scores on the verbal fluency (VF) tasks per group*
Mean Stdev Max Min
VF 1 (Animals) CA NL CG n = 51 n = 50 n = 53
VF 2 (Fruit & Vegetables) CA NL CG n = 53 n = 53 n = 51
CA n = 53
NL n = 53
CG n = 53
20.98 5.69 38 10
19.72 5.10 29 9
20.44 4.59 33.5 10.5
20.88 4.81 31 10
24.92 4.67 37 15.5
22.84 6.00 36 11
27.34 5.55 46 11
19.06 4.58 32 8
22.59 5.72 38 9
VF total
Table 5. Lexical diversity (as measured by D) per group
Mean Stdev Max Min
CA n = 52
D NL n = 49
CG n = 53
70.45 17.12 135.23 41.21
63.93 15.67 111.15 34.70
75.35 17.90 133.03 44.76
5. Analyses 5.1 Independent variables From the raw data described above, the following measures were established per subject: 1. VF: Performance on the two verbal fluency tasks was averaged to one variable, VF: The results are summarized in Table 4. 2. Charlie Chaplin film retelling task: With respect to the retellings of the Charlie Chaplin film sequence, the following measures were established with the help of the CLAN package (http://childes.psy.cmu.edu/). D: The lexical diversity measure D was established per subject. The results per group are summarized in Table 5. In addition, a number of fluency measures were standardized per subject and 1,000 words of spoken data. The following phenomena were counted: pauses (PAUS), filled pauses (FP), repetitions (REP) and self-corrections (RETR). The results per group are summarized in Table 6. . D (or VOC1) is a measure of lexical diversity that is based on type-token frequencies. Unlike the traditional TTR, which is highly sensitive to sample size, however, D is robust even if applied to text samples of varying length (see MacWhinney 2006: 134ff.; McKee, Malvern & Richards 2000).
The role of L1 use for L1 attrition 147
Table 6. Fluency phenomena per group PAUS/TOK FP/TOK REP/TOK RETR/TOK CA NL CG CA NL CG CA NL CG CA NL CG n = 52 n = 49 n = 53 n = 52 n = 49 n = 53 n = 52 n = 49 n = 53 n = 52 n = 49 n = 53 Mean 14.79
14.82
6.30
35.01 52.75 34.44 12.65
10.04
5.48
15.80
15.54
11.20
Stdev 14.35 Max 81.84 Min 0.00
14.27 58.17 0.00
10.05 45.89 0.00
26.16 35.05 24.66 11.38 134.54 134.70 103.13 56.15 0.00 0.00 0.00 0.00
6.70 34.07 0.00
5.06 23.43 0.00
8.13 41.70 2.00
9.34 52.11 0.00
6.77 37.67 0.00
Table 7. Dependent variables across groups df8
F*
VF D
(2, 149) (2, 152)
14.124*** 5.873**
PAUS/TOK FP/TOK REPTOK RETRTOK
(2, 152) (2, 152) (2, 152) (2, 152)
7.478** 6.607** 10.371*** 5.304**
* (One-way ANOVA, ** = p < .01, *** = p < 0.001)
(a) One-way ANOVA To assess whether there were discernible signs of language attrition among the experimental groups, scores on the 6 dependent variables were first compared across the three groups. A one-way ANOVA established highly significant differences on all of the variables described above (see Table 7). In other words, the three groups did perform differently on all of the measures. For all variables except filled pauses, a Tukey Post Hoc procedure established that the significance was due to a difference between both attriting groups on the one hand and the control group on the other (with respect to filled pauses, the CA and CG groups behaved similarly), filled pauses were therefore dropped from the subsequent analysis. The group effects were established to be stable across age and education by means of a univariate analysis of variance (ANCOVA, see Appendix, Table 2). 5.2 Predictor variables The ANOVAs reported in the previous section established that the attriting groups did indeed perform differently on the linguistic measures than the control . The degrees of freedom vary somewhat for individual measures, since not all participants were able to complete all tests.
148 Monika S. Schmid
Table 8. Linear regression of linguistic and extralinguistic measures VF R2 chg ß
D R2 chg ß
PAUSE R2 chg ß
REP R2 chg ß
RETR R2 chg ß
.038
.016
.005
.000
.001
AGE
–.194* .036
LOR EDU
.083 –.168 .157
.052 BilMod IntMod MonMod
–.128 .014 –.316 .199 .008
–.010 .170 .127 R2 =.125
.073 .037 –.192 .014 .046
.045 –.067 –.061 R2 = .108
–.002 .308 –.023 .034
–.048 –.025 –.194 R2 = .065
.031 .056 .396* .033 .004
.095 .171 –.082 R2 = .071
.032 –.005 –.069 R2 = .061
F (6, 99) = 2.358* F (6, 95) = 1.912 F (6, 94) = 1.096 F (6, 94) = 1.198 F (6, 94) = 1.014 * = p < .05
group: there is evidence for attrition in the sample at hand. However, there is also considerable interpersonal variability within the attriting groups: some speakers perform much better than others (see the overview above, Tables 4–6). The predictions made above were that individual performance would be considerably influenced by the amount to which the L1 is used in daily life, and that the impact of habitual L1 use in different language modes might play different roles in facilitating language maintenance. In order to establish the impact of these predictor variables, a linear regression was carried out on the data from the two attriting groups (n = 106). The predictor variables were entered in three blocks: the first block contained a single predictor variable which has the potential to influence performance in any sample and is therefore not per se attrition linked, namely age at the time of data collection. In the second block, two variables were entered which have previously been shown to play a role in language attrition, namely length of residence (LOR) and education (EDU). The third block contained the variables which pertain to our research question: L1 use in the bilingual mode (BILMOD), in the intermediate mode (INTMOD) and in the monolingual mode (MONMOD). The results from the linear regression are summarized in Table 8. From these results it is evident that the age of the speaker does not impact on performance in free speech. It is only where the VF is concerned that there is a slight interaction (this finding is unsurprising, since elderly speakers have often been found to perform more weakly on fluency tasks). EDU and LOR also appear not to interact with performance, the only measure where there is a significant interaction with LOR is RETR: those speakers who have a longer emigration span
The role of L1 use for L1 attrition 149
self-correct slightly more often than speakers at the lower end of the LOR spectrum. Most astonishing, however, is the fact that there is absolutely no interaction between any of the lexical access, lexical diversity and fluency measures used here on the one hand and frequency of L1 use in daily life in any language mode. In order to assess whether any one language use variable might, after all, have an effect on the performance which in the above tests was masked in the cumulative variables, several more linear regressions were performed. Blocks 1 and 2 were left unchanged (and will not be reported here). The first analysis entered two variables in block 3: L1 use within the family (i.e. with partner, children and grandchildren, BILMOD1) and with friends (BILMOD2). The second set of analyses investigated language use in the interactive mode in more detail: INTMOD1 comprises L1 use in German clubs, churches etc., while INTMOD2 represents the use of the L1 for professional purposes. Finally, the third set of analyses distinguishes L1 use in the monolingual mode in distance communication (MONMOD1) and during visits to Germany (MONMOD2). The results are summarized in Table 9. Table 9. Linear regression of linguistic measures and language modes VF R2 chg ß
D R2 chg ß
PAUSE R2 chg ß
REP R2 chg ß
RETR R2 chg ß
.020
.040
.015
.025
.000
.175 –.086
BilMod1 BilMod2 .074*
.013 .037 .275**
IntMod1 IntMod2 .024 MonMod1 MonMod2
.189 –.209 .043 –.115 .021 .032
.132 .099
–.129 –.019 .035 .049 –.204* .048
.054 –.181
–042 .175 .048 .194 .030 .009
–.199 –.100
.021 –.021 –.152* .161* .004
.063 –.078
–.029 –.059
* = p < .05, ** = p < .01
Again, the results are meagre. Subjects who regularly use their L1 for professional purposes achieve a slightly higher VF score and use slightly less pauses in free speech. None of the other measures has any impact.
150 Monika S. Schmid
6. Discussion Considering how widely accepted the dictum is that L1 use prevents L1 attrition, the results presented above are quite startling. While the initial analyses clearly showed that the attriters differ from the control group speakers, there is little or no correlation with the self-reports on frequency of L1 use. In other words, while there clearly is an attrition effect, the amount of use of the L1 in daily life does not seem to have any predictive power for this effect. In other areas of bilingual investigation, the impact of frequency of activation on accessibility has clearly been established. The findings from this study, however, suggest the possibility that frequency and recency of activation play a less prominent role in L1 attrition than they do in other bilingual or multilingual contexts. This suggests a rather intriguing possibility: it is probably safe to claim that there are no other areas of abstract knowledge that are rehearsed as extensively as a linguistic system that is acquired in a monolingual setting until adulthood. It may thus be possible that there is a kind of saturation point of rehearsal. Such a process of stabilization of knowledge through massive rehearsal has been suggested as one possible explanation for the so-called Critical Period, as an alternative to biologically-based brain plasticity models (Pallier, this volume). I would suggest the possibility that once this point has been reached, frequent activation is no longer necessary to maintain accessibility, and that in such a case, inhibition becomes the process which impacts upon the activation threshold. In other words, attrition might depend less on the mere frequency to which the L1 is continued to be spoken than had previously been assumed, since quality of contact might be more important than quantity, and more and more to the fact that monolingual mode use of the L2 demands that the L1 be inhibited (see also Köpke, this volume). In this context it is interesting to see that, meagre though they are, those predictors that do have a significant impact on lexical access and diversity all belong to L1 use in the intermediate mode. It should therefore also be considered that virtually any experimental setting investigating L1 attrition belongs to that same type of interaction. The participants in the investigation will be aware that the investigator is proficient in their L2, and furthermore the setting is usually the subject’s country of residence, so that L2 can be considered quite active and visible. On the other hand, the relative formality of the situation (a scientific experiment) and the prestige of the investigator (an academic, a scientist) will in all probability signal to the subject that codeswitching is less appropriate in this situation than it would be in an informal interaction with friends. As was pointed out above, this is probably the type of situation in which inhibition of L2 is most difficult. It is therefore possible that the results presented above are to some degree the outcome of more or less ‘practice’ with L1 use and L2 inhibition in this kind
The role of L1 use for L1 attrition 151
of situation. Subjects who routinely use their L1 in this type of interaction may find it easier to inhibit their L2, and this might account for the better results and easier access to L1 which these subjects have shown on some of the dependent variables. If such were to be the case, then the impact of both L1 and L2 use in the situations detailed above for other types of linguistic knowledge should have to be further looked into. The present study has confined itself to issues of lexical accessibility. Further studies should investigate to what degree grammatical knowledge (e.g. with respect to inherent and contextual inflection, such as gender and plural information across the NP, or information on tense inflection of words) is affected. Most of all, however, the findings from this investigation suggest that it is relatively meaningless to study the attrition and use of only one of a bilingual’s languages in isolation, and exclude the development and use of the other. The results above suggest a delicate balance of the two language systems and their activation, inhibition, and accessibility. To what degree actual underlying knowledge is affected by this state of balance, which is probably in continual shift across the lifespan, and almost certainly affected by far more parameters than this study has thought to include, is, and remains, a mystery.
Acknowledgements I am grateful to Marianne Gullberg, Merel Keijzer, Barbara Köpke, Chris McCully as well as two anonymous reviewers for their helpful and valuable comments on earlier versions of this paper. With respect to the statistical analysis, I would like to thank Elise Dusseldorp for her insights and advice. Any remaining mistakes and oversights are mine.
References Ammerlaan, T. (1996). “You get a bit wobbly...” – Exploring bilingual lexical retrieval processes in the context of first language attrition. Unpublished Doctoral Dissertation, Nijmegen: Katholieke Universiteit Nijmegen. Cook, V. (2005). “The changing L1 in the L2 user’s mind”. Paper presented at the 2nd International Conference on First Language Attrition, Amsterdam: 18 August 2005. de Bot, K., Gommans, P. & Rossing, C. (1991). “L1 loss in an L2 environment: Dutch immigrants in France”. In H. W. Seliger & R. M. Vago (Eds.), First Language Attrition (pp. 87–98). Cambridge: CUP.
152 Monika S. Schmid
de Bot, K. & Clyne, M. (1994). “A 16-year longitudinal study of language attrition in Dutch immigrants in Australia”. Journal of Multilingual and Multicultural Development, 15 (1), 17–28. Goodglass, H. & Kaplan, E. (1983). The Assessment of Aphasia and Related Disorders. Philadelphia, PA: Lea & Febiger. Green, D. (1998). “Mental control of the bilingual lexico-semantic system”. Bilingualism: Language and Cognition, 1, 67–81. Grosjean, F. (2001). “The bilingual’s language modes”. In J. Nicol (Ed.), One Mind, Two Languages: Bilingual language processing (pp. 1–22). Oxford: Blackwell. Gürel, A. (2004). “Selectivity in L2-induced L1 attrition: A psycholinguistic account”. Journal of Neurolinguistics, 17 (1), 53–78. Hulsen, M. (2000). Language Loss and Language Processing: Three generations of Dutch migrants in New Zealand. Unpublished Doctoral Dissertation, Nijmegen: Katholieke Universiteit Nijmegen. Jaspaert, K. & Kroon, S. (1989). “Social determinants of language loss”. Review of Applied Linguistics (I.T.L.), 83/84, 75–98. Keijzer, M. C. J. & Schmid, M. S. (2005). “Building language attrition research”. Toegepaste Taalwetenschap in Artikelen, 73, 201–207. Köpke, B. (1999). L’ attrition de la première langue chez le bilingue tardif : Implications pour l’ étude psycholinguistique du bilinguisme. Unpublished Doctoral Dissertation, Toulouse: Université de Toulouse-Le Mirail. Köpke, B. (2002). “Activation thresholds and non-pathological L1 attrition”. In F. Fabbro (Ed.), Advances in the Neurolinguistics of Bilingualism (pp. 119–142). Udine: Forum. MacWhinney, B. (2006). The CHILDES Project. Tools for analyzing talk – Electronic edition. Volume 2: Transcription format and programs. Part 2: The CLAN Programs. http://childes. psy.cmu.edu/manuals/CLAN.pdf. McKee, G., Malvern, D. & Richards, B. (2000). “Measuring vocabulary diversity using dedicated software”. Literary and Linguistic Computing, 15 (3), 323–337. Nakuma, C. (1997). “A method for measuring the attrition of communicative competence: A pilot study with Spanish L3 subjects”. Applied Psycholinguistics (APsy), 18 (2), 219–235. Paradis, M. (1993). “Linguistic, psycholinguistic, and neurolinguistic aspect of ‘Interference’ in bilingual speakers: The Activation Threshold Hypothesis” International Journal of Psycholinguistics, 9 (2), 133–145. Paradis, M. (2004). A Neurolinguistic Theory of Bilingualism. Amsterdam: John Benjamins. Perdue, C. (1993). Adult Language Acquisition: Cross-linguistic perspectives. Cambridge: CUP. Schmid, M. S. (2002). First Language Attrition, Use and Maintenance: The case of German Jews in Anglophone Countries. Amsterdam: John Benjamins. Schmid, M. S. (2005). “First language attrition: The methodology revised.” International Journal of Bilingualism, 8 (3), 239–255. Schmitt, E. (to apear). “When boundarieas are crossed: Evaluating language attrition data from two perspectives”. Bilingualism: Language and Cognition (special issue L1 attrition, 2009). van Hell, J. & Dijkstra, T. (2002). “Foreign language knowledge can influence native language performance in exclusively native contexts”. Psychonomic Bulletin & Review, 9 (4), 780– 789. Yağmur, K. (1997). First language Attrition Among Turkish Speakers in Sydney. Tilburg: Tilburg University Press.
The role of L1 use for L1 attrition 153
Appendix Table 1. ANCOVA dependent variables Variable
Group
Age
Education
R2
VF1 VF2 D PAUS/TOK RETRTOK REPTOK
.000 .006 .002 .010 .009 .000
.095 .012 .178 .678 .297 .952
.022 .706 .026 .511 .606 .629
.244 .121 .129 .067 .080 .103
Critical periods in language acquisition and language attrition Christophe Pallier
Unité de Neuroimagerie Cognitive, INSERM U562, SHFJ & IFR 49, Orsay, France
This paper reviews some of the predictions and assumptions made by different versions of the Critical Period Hypothesis which assumes that language acquisition will in principle be more successful in younger subjects. Findings from investigations of early vs. delayed acquisition of both a first and a second language which support this claim are presented. The question is then addressed whether the overall difference in ultimate attainment is a consequence of a loss of neural plasticity, or of the stabilizing of neural connections through the language learning process itself. The findings from a study of the L1 attrition of Korean adoptees in France are presented in support of the latter explanation. Keywords: critical period; adoptees; neuroimaging; Korean
The issue of the “critical period hypothesis for language acquisition” is the focus of a vast literature (see, among others, Birdsong 1999a; Doupe & Kuhl 1999; Harley & Wang 1997; Lenneberg 1967; Long 1990; Newport, Bavelier & Neville 2001; Singleton & Lengyel 1995). It is important to stress at the outset that the expression “Critical Period Hypothesis” (CPH) is used with two different meanings. The first meaning of the CPH corresponds to an empirical hypothesis according to which humans are more efficient at language learning in the first years of life. In other words, the CPH states that age of acquisition is an important predictor of ultimate proficiency: the older one starts to learn a language, the smaller the odds of reaching native-like proficiency. The expression “critical period hypothesis” is also sometimes used to refer to the concept that an age-related decline in neural plasticity is the cause of increasing difficulties in language learning (Penfield & Roberts 1959). According to this second meaning, the CPH is a potential explanation of age effects on language acquisition. The two meanings must be distinguished because there may
156 Christophe Pallier
be a critical period according to the first meaning, that is, a detrimental effect of age of acquisition on ultimate proficiency in a language, even if the explanation in terms of loss of neural plasticity is wrong. In other words, an adverse effect of age of acquisition (AoA) may have other causes than irreversible neural changes (Birdsong 1999b lists some alternative explanations). Another point is worth emphasizing. In the framework of the Critical Period Hypothesis (with the second meaning), it is often assumed that the putative loss of plasticity is due to maturational factors. Yet, it could also be the outcome of language acquisition itself, as suggested by the following quote from Wilder Penfield, one of the earliest advocates of the CPH: Before the child begins to speak and to perceive, the uncommitted cortex is a blank slate on which nothing has been written. In the ensuing years much is written, and the writing is normally never erased. After the age of ten or twelve, the general functional connexions have been established and fixed for the speech cortex. (Penfield 1965: 792, cited by Dechert 1995)
Another variant is defended by Steve Pinker: Language acquisition circuitry is not needed once it has been used. It should be dismantled if keeping it around incurs any cost [...] Greedy neural tissue lying around beyond its point of usefulness is a good candidate for the recycling bin. (Pinker 1994: 294–295)
The common theme is that once the child has learned the language(s) spoken around him, the neural modifications are irreversible. Before assessing the merits of competing theories, it has to be considered what the empirical evidence for a critical period with the first meaning is. Can language be acquired at any age or are the first years of life indeed critical? Studies of critical periods in animals involve depriving the animals from some relevant stimulation during a given time window. For obvious reasons, this experimental approach cannot be used with humans: it would not be ethically acceptable to deny children exposure to language. Nevertheless, a few observations of abandoned children suggest that their language skills remained severely limited even after language instruction (e.g. Curtiss 1977; Itard 1964). Another source of information comes from studies of groups of congenitally deaf people who learned sign language as an L1 at different ages (Mayberry & Eichen 1991; Newport 1990). These studies show that those who were exposed to sign language in their very first years of life control it better than those who learned it in midchildhood; the latter, in turn, perform better than those who were first exposed to sign language in their teens or later.
Critical periods, language acquisition and attrition 157
While it is sometimes proposed that neural changes critical for language acquisition occur around puberty, it seems that they can take place much earlier. Studies of auditory and language development in deaf children who receive cochlear implants – an auditory prosthesis that stimulates the auditory nerve in order to transmit acoustic information to the central auditory system – reveal that beneficial effects of earlier implementation can be observed in children when they are as young as 1 to 3 years of age (McConkey Robbins, Burton Koch, Osberger, Zimmerman-Philips & Kishon-Rabin 2004; Tomblin, Barker, Spencer, Zhang & Gantz 2005). Are endogenous maturational factors the only cause of the age effect on L1 acquisition? There is evidence against this hypothesis. The fact that the deaf children who did not learn L1 in infancy were deprived from normal linguistic input in their first years of life plays a major role. This has been clearly established by Mayberry, Lock & Kazmi (2001) who compared two groups of deaf adults who had learned American Sign Language (ASL) relatively late, between 9 and 15 years of age. ASL was the L1 for the participants of the first group, who were congenitally deaf. Participants in the second group were born with normal hearing and had started to acquire English before they became deaf. Therefore, ASL was their L2 (note that the grammar of ASL differs markedly from English grammar (Klima & Bellugi 1979)). Mayberry et al. found that the second group largely outperformed the first in ASL. This result shows that experience plays a role because if only maturational factors were at play then the proficiency in ASL should only depend on age of acquisition of ASL and both groups should perform similarly. Moreover, the fact that those who learned an L1 early in life performed better than those who were deprived from any input contradicts Penfield’s idea that exposure to the L1 “fixes” neural connections. On the contrary, learning and using a language in the first years of life maintains the capacity to acquire a new language. This observation corroborates the “exercise hypothesis” according to which the language learning capacity decreases when it is not used (Bever 1981). The studies on the effect of age on the acquisition of an L1 demonstrate that deprivation has rapid detrimental effects on the capacity to learn a language. As deprivation is the usual test applied to assess critical periods in animals, it appears undeniable that there is a critical period for L1 acquisition in humans. The brain must be exposed to language to develop ‘normally’. One possible explanation is that in the absence of linguistic stimulation, the brain areas that normally subserve language processing may be recruited for other functions. This interpretation is supported by data from Lee, Lee, Oh, Kim, Kim, Chung, Lee & Kim (2001) showing that the benefits of cochlear implantation are inversely related to the amount of metabolism in the temporal lobes. In other words, deaf people who have an abnormally low metabolism in the temporal region profit more from
158 Christophe Pallier
Figure 1. Average foreign accent ratings for 240 native speakers of Italian who arrived in English-speaking Canada between the age of 2 and 23 (filled circles) and 24 English controls (squares). (Data are from Flege 1999)
implants than the deaf people who have higher metabolism, presumably because in the latter case, these areas are recruited for extra-linguistic functions. However, more data is needed to confirm this interpretation. Considering the detrimental effect of delays in learning an L1 as the result of deprivation is a hypothesis remote from the ideas expressed in Penfield’s and Pinker’s quotations above. Indeed, they propose that it becomes more difficult to learn a language with age, either because learning the L1 has “fixed” the neural connections (Penfield) or because the language learning systems are disposed of when the L1 is acquired (Pinker). To accommodate the effects described above, it would at least be necessary to modify Pinker’s proposal by adding that the language learning system can also disappear when it is not stimulated, that is, in the absence of linguistic input. How does the CPH fare in the more common situation where people have not been deprived of language exposure in early childhood, but have learned an L2 later? The effect of age of acquisition of an L2 is the focus of a huge literature which cannot be reviewed here (see Birdsong 1999a; Singleton & Lengyel 1995; Strange 1995). In brief, the effect of age of acquisition (AoA) is indisputable: the earlier an L2 is learned, the higher the likelihood of becoming a proficient speaker in L2 (Singleton 1995). For example, AoA has a clear effect on foreign accent in speech production (Flege, Munro & MacKay 1995; Oyama 1976; see Figure 1). Even if other factors like duration of exposure and amount of use of an L2 also influence its final level of proficiency, age of acquisition seems to be the strongest predictor of ultimate achievement.
Critical periods, language acquisition and attrition 159
Figure 2. Self-reported English proficiency for native Chinese immigrants as a function of age of arrival, separated by educational attainment. Data reported in Hakuta et al. (2003)
Having said that proficiency decreases with age of acquisition, two points are still worth mentioning. First, this affirmation is a statistical claim over relatively large samples of subjects. It does not imply that a given person should not try to learn an L2 because he or she is already too old. There is considerable inter-individual variability in L2 attainment and it may be possible that some individuals manage to reach native-like performance even when they have learned their L2 well after childhood (see Birdsong 1999b; Bongaerts 1999; but also Hyltenstam & Abrahamsson 2003). Second, the shape of the curve relating age of acquisition and proficiency does not conform to the notion that critical changes occur around puberty. For L1 acquisition, the effects of delay in age of acquisition on L2 can occur very early. For instance, Weber-Fox & Neville (1996) described Chinese-English bilinguals who were immersed in English before 6 years of age, yet perform significantly worse than native English speakers on some grammatical tests. Moreover, while some versions of the CPH predict a discontinuity around puberty, plots of proficiency in an L2 as a function of age of acquisition display a smooth, gradual decline starting from early childhood and extending throughout life (Birdsong & Molis 2001; Flege et al. 1995; Hakuta, Bialystok & Wiley 2003; see Figure 2). If the effect of age of acquisition is due to a loss of neural plasticity, then the only legitimate conclusion appears to be that the underlying neural changes are quite continuous and are not restricted to a limited time window. . Of course, if one finds individuals who obtain native-like scores on a range of linguistic or psycholinguistics tests, this does not imply that they perform like monolinguals in every respect. Thus, it is impossible to provide definite proof that a given bilingual subject has reached native-like proficiency in his L2.
160 Christophe Pallier
Do delays in age of acquisition have similar effects for L1 and L2 acquisition? This is an important question because the answer could indicate whether the same mechanism is at work in both cases. To my knowledge, data on the effect of age on L1 acquisition are too scarce to know the precise shape of the age effect. Yet, the study by Mayberry et al. (2001) suggests that the effect of delay on L1 acquisition is much greater than for L2 acquisition. That is, the effect of deprivation on language learning seems much more devastating than the effect of having been exposed to an L1. Therefore, I believe it is likely that the effect of age of acquisition for an L2 is due to a different mechanism. One possible explanation appeals to maturational factors that engender a progressive “loss of neural plasticity” in the brain (whether or not such a phenomenon would be specific to language networks is debatable). Another type of explanation postulates that it is the process of learning the L1 itself that stabilises neural connections. The more advanced the learning of the L1, the less the language networks can be modified by exposure to L2. Note that the two hypotheses are not incompatible: both mechanisms could be at play during development. For example, the proactive interference mechanism could be responsible for the slope of the age function observed in the first ten years of life, while the maturational constraints could play a role in the older age range. As an instance of interference of L1 on L2 acquisition, consider the SpanishCatalan bilinguals described in Pallier, Bosch & Sebastian-Gallés (1997). Despite extensive exposure to Catalan from the age of 4 to 6, the native Spanish speakers do not perform like Catalan native speakers on the phonemic contrast between open versus close /e/. A likely explanation is that this phonemic contrast involves speech sounds which are assimilated onto a single phonemic category in Spanish (this situation is called single category assimilation by Best (1995) or perceptual equivalence by Flege, Takagi & Mann 1995). Thus, speakers of Spanish as an L1 hear Catalan through the filter of the Spanish phonetic system, an example of interference from L1 onto L2. Disentangling maturational and interference accounts is not an easy task. Yet, it seems to us that while the maturational hypothesis entails the irreversibility of changes, interference theories need not make this prediction. That is, contrary to Penfield’s assertion, the modifications engendered by the learning processes
. In the title of the paper (Pallier, Bosch & Sebastian-Gallés 1997) we used the expression “lack of behavioral plasticity” to describe this failure to acquire a new phonemic contrast. This is because we deemed it unlikely that this limitation was due to a lack of neural plasticity in the auditory cortex of the Spanish children. In the conclusion of the paper, we questioned whether such an effect was irreversible or not.
Critical periods, language acquisition and attrition 161
could be reversible. This leads to the interesting prediction that the acquisition of an L2 might be facilitated if L1 ceased to be used. What happens when someone learns an L2 after having stopped using L1? International adoption provides an opportunity to address this question. A few years ago we contacted organisations in charge of adoption in France and recruited a small sample of young adults who had been adopted by French-speaking families in their childhood. They were all born in Korea and came to France when they were between 3 and 10 years old. All of them claim to have completely forgotten Korean (though some had memories from their life in Korea.) This is seemingly a very general phenomenon among internationally adopted children (Maury 1995, 1999). French had become the main language of our group of adoptees and they speak it fluently without any detectable accent. We will briefly review here some of the experiments performed on those participants (for more details, see Pallier, Dehaene, Poline, LeBihan, Argenti, Dupoux & Mehler 2003; Sangrigoli, Pallier, Argenti, Ventureyra & de Schonen 2005; Ventureyra 2005; Ventureyra & Pallier 2004; Ventureyra, Pallier & Yoo 2004). Our first three behavioural experiments were designed to assess the adoptees’ residual knowledge of the Korean language. Their performances were compared with that of a control group of native-French speakers who had never been exposed to Korean, nor to any other Asian language (Pallier et al. 2003). The Korean sentence identification experiment involved recognising sentences in Korean among recordings in different languages. In the word recognition experiment, participants heard two Korean words and had to select the one which was the translation of a given French word. Lastly, in the speech segment detection experiment, the task was to decide if specific speech fragments were present in sentences in various languages, including Korean. The results show similar patterns of performance for the adoptees and for the control group of native French speakers (see Figure 3), providing a first validation of the adoptees’ claim that they had forgotten their L1. While the participants performed the speech segment detection task, their brain activity was monitored using functional magnetic resonance imaging (fMRI), a technique that allows to detect changes in patterns of brain activity when the participants process the stimuli. The individual analyses of fMRI data showed no detectable difference in brain activity when comparing the cerebral responses to Korean sentences versus Japanese or Polish sentences, two languages to which the adoptees had never been exposed. Thus, brain imaging data and behavioural data converge on the conclusion that years of exposure to a language in childhood are not sufficient to maintain a solid knowledge of this language. This result can be interpreted in two different ways. First, the Korean language may have been “erased” from the brain of the adoptees. This would con-
162 Christophe Pallier
Figure 3. Results on three behavioural tasks showing similar levels of performance of Korean adoptees and a control group of French native speakers
stitute strong evidence against versions of the CPH that state that some “neural connections” become fixed in the early years of life, as a result of learning and/ or because of maturational factors. Such theories predict that the adoptees (at least those who arrived at older ages) should have displayed some sensitivity to Korean. The metaphor of the imprinted clay tablet used by Penfield therefore has to be rejected. It must be noted, however, that as the adoptees from our studies arrived in France before the age of ten, we cannot exclude the possibility that irreversible changes occur after 10 years of age. A second possible interpretation is that the paradigms used by Pallier et al. (2003) lacked sensitivity and that further testing might uncover effects of the early exposure to Korean. With Valerie Ventureyra, we ran a series of behavioural experiments to test more thoroughly the remnants of Korean in a larger group of twenty adoptees. In brief, we found virtually no significant difference between the adoptees and native French speakers. For example, the adoptees did not perceive the differences between Korean plain, tense and aspirated stop consonants better than French participants (Ventureyra et al. 2005). The only difference between the adoptees and native French speakers came about in an experiment assessing . It would be desirable to know what happens when someone switches to a new language at an age above 10. Studies of language attrition in adults seem to indicate that they show much less attrition than children. Thus, it is possible that plots of language attrition as a function of age show a non-linearity around puberty.
Critical periods, language acquisition and attrition 163
the recognition of the Korean number series: while the adoptees gave the highest scores to Korean recordings (but did not recognise them formally), the control participants who had never been exposed to Korean gave the highest scores to Thaï and Chinese (Ventureyra & Pallier 2004). As these were the only tone languages included in the recordings, this may reflect an implicit knowledge of adoptees that Korean is not a tone language. One important question is whether the adoptees could relearn their native language faster or better than people who have never been exposed to Korean. If the adoptees showed such an advantage, this would provide evidence for remnant traces of early exposure to Korean. From an anecdotal point of view, the adoptees who visited Korea for short stays (from a few days to a few months) did not miraculously recover the ability to speak or comprehend the language, and nor did the few who took Korean courses. Some researchers have reported benefits of exposure to a language in infancy or childhood when relearning it (Au, Knightly, Jun & Oh 2002; Oh, Jun, Knightly & Au 2003; Tees & Werker 1984). For example, Oh et al. (2003) evaluated the perception and production of Korean consonants by three groups enrolled in Korean language classes: one group had spoken Korean regularly for a few years during childhood, another group had heard Korean regularly during childhood but had spoken it minimally, and a last group consisted of novice learners. The first two groups performed better than the novice learners, demonstrating long-term benefits of early childhood experience with Korean. In these studies, however, the participants were not completely severed from the language of interest. In the study by Oh et al. (2003), for example, the participants had been continuously exposed to Korean for four hours a week on average. Therefore, their situation was quite different from that of adoptees who had not been exposed at all to Korean since adoption. Whether or not the adoptees would relearn their L1 faster than novice learners remains an open question. Let us now consider the L2 learned by the adoptees. Data from brain activations comparing the activations elicited by French and foreign languages show similar patterns in the adoptees and the control group of native French speakers (Pallier et al. 2003). Thus, the processing of French recruits the same brain areas . A phonetic training experiment, described in Ventureyra’s thesis, was set up. Unfortunately, most participants did not complete it, probably because the starting level of difficulty was too high. This study is therefore not conclusive. Ideally, the participants in a retraining experiment should be adoptees who enrolled in Korean language classes and have a high degree of motivation. At the 2nd International Conference on First Language Attrition held in 2005 in Amsterdam, Kenneth Hyltenstam mentioned results obtained by Hyeon-Sook Park at the University of Stockholm suggesting that after a few months of courses, Korean adoptees perceived Korean consonants better than novice learners.
164 Christophe Pallier
in both groups. Note however that this is a comparison between groups and that there is non-negligible interindividual variability. Therefore it cannot be determined if the brain areas used by the adoptees to process French are precisely the same that they used for processing Korean in their childhood (Kim, Relkin, Lee & Hirsch 1997). Only a longitudinal study, where children could be scanned before and after adoption, would settle this question. A most important question regards the level of proficiency attained in French by the adoptees. Pallier et al. (2003) did not provide any formal assessment of this. We remarked informally that the adoptees spoke French fluently without any noticeable accent. Yet, more stringent tests might reveal limitations in their control of French (Cutler, Mehler, Norris & Segui 1989). Two further experiments are relevant. The first assessed the sensitivity to phonotactic rules that differ between French and Korean. Korean speakers report hearing a vowel within certain types of illegal consonant clusters, a phenomenon known as vowel epenthesis (Dupoux, Kakehi, Hirose, Pallier & Mehler 1999). Korean participants who have been speaking French for a few years still experience this illusion. When we tested adoptees, however, we found that they perceived consonant clusters in the same way as native French speakers do. Their speech perception system has become tuned to French phonotactic properties. The second experiment dealt with lexical gender, a feature which is particularly difficult to acquire for native speakers of Korean who learn French as an L2. The participants listened to French sentences, some of which contained either gender agreement mistakes or semantic anomalies. The participants were instructed to press a button as soon as they detected any type of anomaly in a sentence, allowing measurement of both error rates and reaction times. Beside a group of adoptees and a group of native French participants, the experiment also included a third group of native Koreans, speakers of French as an L2 who had been living in France for a few years. The results showed that, as expected, this last group had great difficulties in detecting gender agreement mistakes. By contrast, the adoptees’ performance was similar to that of native speakers of French. This was the case even for the participants who arrived in France when they were around 8 to 10 years of age (and in none of the experiments did the age of adoption correlate with the participants’ scores.) These experiments demonstrate that the adoptees had mastered the phonotactics of French and its lexical gender system. Of course, one cannot exclude the possibility that more demanding tasks might eventually reveal deficiencies in the adoptees’ control of French. The set of data obtained in these series of experiments, as well as Mayberry et al.’s results on post-lingually deaf who learned ASL after 9 years of age, argue against irreversible modifications occurring in the first ten years of life, either be-
Critical periods, language acquisition and attrition 165
cause of maturational constraints or as a by-product of learning the L1. Provided one is exposed to a language in the first years of life, the language systems probably remain highly plastic until at least 10 years of age. If so, one may ask why an L2 acquired in the first 10 years of life is not necessarily perfectly mastered? Our experiments on adoptees were inspired by the interference hypothesis according to which L2 acquisition is hampered by L1. To definitely prove this hypothesis, however, it would be necessary to compare the adoptees with immigrants having a similar background, and who have learned L2 while continuing to use their L1. If the interference account is correct, then the adoptees should have a better proficiency in the L2 than the immigrants. We hope that such a study can be conducted in the future. Finally, it is interesting to know what happens when someone switches to a new language at an age above 10. In fact, studies of language attrition suggest that adults show much less attrition than children (Ammerlaan 1996; Pelc 2001 reviewed by Köpke & Schmid 2004). Thus, language attrition, contrary to L2 acquisition, may show a discontinuity around puberty.
References Ammerlaan, T. (1996). “You get a bit wobbly...” – Exploring bilingual lexical retrieval processes in the context of first language attrition. Unpublished Doctoral Dissertation, Nijmegen: Katholieke Universiteit Nijmegen. Au, T. K., Knightly, L. M., Jun, S.-A. & Oh, J. S. (2002). “Overhearing a language during childhood”. Psychological Science, 13 (3), 238–243. Best, C. (1995). “A direct realist view of cross-language speech perception”. In W. Strange (Ed.), Speech Perception and Linguistic Experience: Theoretical and methodological issues (pp. 171–204). Baltimore, MD: York Press. Bever, T. (1981). “Normal acquisition processes explain the critical period for language learning”. In K. Diller (Ed.), Individual Differences in Language Learning Aptitude (pp. 176–198). Rowley, MA: Newbury House. Birdsong, D. (Ed.) (1999a). Second Language Acquisition and the Critical Period Hypothesis. Mahwah, NJ: Lawrence Erlbaum Associates. Birdsong, D. (1999b). “Whys and why nots of the critical period hypothesis for second language acquisition”. In D. Birdsong (Ed.), Second Language Acquisition and the Critical Period Hypothesis (pp. 1–22). Mahwah, NJ: Lawrence Erlbaum Associates. Birdsong, D. & Molis, M. (2001). “On the evidence for maturational constraints in secondlanguage acquisition”. Journal of Memory and Language, 44, 235–249. Bongaerts, T. (1999). “Ultimate attainment in second language pronunciation: The case of very advanced late second language learners”. In D. Birdsong (Ed.), Second Language Acquisition and the Critical Period Hypothesis (pp. 133–159). Mahwah, NJ: Lawrence Erlbaum Associates.
166 Christophe Pallier
Curtiss, S. (1977). Genie: A psycholinguistic study of a modern day ‘wild child’. New York, NY: Academic Press. Cutler, A., Mehler, J., Norris, D. & Segui, J. (1989). “Limits on bilingualism”. Nature, 340, 229– 230. Dechert, H. W. (1995). “Some critical remarks concerning Penfield’s theory of second language acquisition”. In D. Singleton & Z. Lengyel (Eds.), The Age Factor in Second Language Acquisition (pp. 67–94). Clevedon: Multilingual Matters. Doupe, A. J. & Kuhl, P. K. (1999). “Birdsong and human speech: Common themes and mechanisms”. Annual Review of Neuroscience, 22, 567–631. Dupoux, E., Kakehi, K., Hirose, Y., Pallier, C. & Mehler, J. (1999). “Epenthetic vowels in Japanese: A perceptual illusion?” Journal of Experimental Psychology: Human Perception and Performance, 25, 1568–1578. Flege, J. E. (1999). Age of learning and L2 speech. In D. Birdsong (Ed.), Second Language Acquisition and the Critical Period Hypothesis (pp. 101–131). Mahwah, NJ: Erlbaum. Flege, J. E., Munro, M. J. & MacKay, I. R. A. (1995). “Factors affecting strength of perceived foreign accent in a second language”. Journal of the Acoustical Society of America, 97, 3125– 3134. Hakuta, K., Bialystok, E. & Wiley, E. (2003). “Critical evidence: A test of the critical-period hypothesis for second language acquisition”. Psychological Science, 14, 31–38. Harley, B. & Wang, W. (1997). “The critical period hypothesis: Where are we now?”. In A. M. B. de Groot & J. F. Kroll (Eds.), Tutorials in Bilingualism: Psycholinguistic perspectives (pp. 19–51). Mahwah, NJ: Lawrence Erlbaum Associates. Hyltenstam, K. & Abrahamsson, N. (2003). “Maturational constraints in SLA”. In C. J. Doughty & M.H. Long (Eds.), Handbook of Second Language Acquisition (pp. 539–588). Oxford: Blackwell. Itard, J. (1964). “Mémoire et rapport sur Victor de l’Aveyron”. In L. Malson (Ed.), Les enfants sauvages: Mythes et réalité (pp. 119–246). Paris: Union Général d’éditions 10/18. Kim, K. H. S., Relkin, N. R., Lee, K.-M. & Hirsch, J. (1997). “Distinct cortical areas associated with native and second languages”. Nature, 388, 171–174. Klima, E. & Bellugi, U. (1979). The Signs of Language. Cambridge, MA: Harvard University Press. Köpke, B. & Schmid, M. S. (2004). “Language attrition: The next phase”. In M. S. Schmid, B. Köpke, M. Keijzer & L. Weilemar (Eds.), First Language Attrition: Interdisciplinary perspectives on methodological issues (pp. 1–46). Amsterdam: John Benjamins. Lee, D. S., Lee, J. S., Oh, S. H., Kim, S.-K., Kim, J.-W., Chung, J.-K., Lee, M. C. & Kim, C. S. (2001). “Cross-modal plasticity and cochlear implants”. Nature, 409, 149–150. Lenneberg, E. H. (1967). Biological Foundations of Language. New York, NY: Wiley. Long, M. (1990). “Maturational constraints on language development”. Studies in Second Language Acquisition, 12, 251–285. Maury, F. (1995). Les mécanismes intrapsychiques de l’adoption internationale et interraciale. L’ adoption des enfants coréens en France. Unpublished Doctoral Dissertation, Paris: Université de Paris VIII. Maury, F. (1999). L’ adoption interraciale. Paris: L’Harmattan. Mayberry, R. I. & Eichen, E. B. (1991). “The long-lasting advantage of learning sign language in childhood: Another look at the critical period for language acquisition”. Journal of Memory and Language, 30, 486–512.
Critical periods, language acquisition and attrition 167
Mayberry, R. I., Lock, E. & Kazmi, H. (2001). “Linguistic ability and early language exposure”. Nature, 417, 38. McConkey Robbins, A., Burton Koch, D., Osberger, M. J., Zimmerman-Philips, S. & KishonRabin, L. (2004). “Effect of age at cochlear implantation on auditory skill development in infants and toddlers”. Archives of Otolaryngology Head & Neck Surgery, 130, 570–574. Newport, E. L. (1990). “Maturational constraints on language learning”. Cognitive Science, 14, 11–28. Newport, E. L., Bavelier, D. & Neville, H. J. (2001). “Critical thinking about critical periods: Perspectives on a critical period for language acquisition”. In E. Dupoux (Ed.), Language, Brain and Cognitive Development (pp. 481–502). Cambridge, MA: MIT Press. Oh, J. S., Jun, S.-A., Knightly, L. M. & Au, T. K.-f. (2003). “Holding on to childhood language memory”. Cognition, 86, B53–B64. Oyama, S. (1976). “A sensitive period for the acquisition of a nonnative phonological system”. Journal of Psycholinguistic Research, 5, 261–283. Pallier, C., Bosch, L. & Sebastian-Gallés, N. (1997). “A limit on behavioral plasticity in speech perception”. Cognition, 64, B9–B17. Pallier, C., Dehaene, S., Poline, J.-B., LeBihan, D., Argenti, A.-M., Dupoux, E. & Mehler, J. (2003). “Brain imaging of language plasticity in adopted adults: Can a second language replace the first?” Cerebral Cortex, 13, 155–161. Pelc, L. (2001). L1 Lexical, Morphological and Morphosyntactic Attrition in Greek-English Bilinguals. Unpublished Doctoral Dissertation, New York, NY: City University of New York. Penfield, W. (1965). “Conditioning the uncommitted cortex for language learning”. Brain, 88, 787–798. Penfield, W. & Roberts, L. (1959). Speech and Brain Mechanisms. Princeton, NJ: Princeton University Press. Pinker, S. (1994). The Language Instinct. New York, NY: W. Morrow and Co. Sangrigoli, S., Pallier, C., Argenti, A.-M., Ventureyra, V. & de Schonen, S. (2005). “Reversibility of the other-race effect in face recognition during childhood”. Psychological Science, 16, 440–444. Singleton, D. (1995). “A critical look at the critical period hypothesis in second language acquisition research”. In D. Singleton & Z. Lengyel (Eds.), The Age Factor in Second Language Acquisition: A critical look at the critical period hypothesis (pp. 1–29). Clevedon: Multilingual Matters. Singleton, D. & Lengyel, Z. (Eds.) (1995). The Age Factor in Second Language Acquisition: A critical look at the critical period hypothesis. Clevedon: Multilingual Matters. Strange, W. (Ed.) (1995). Speech Perception and Linguistic Experience: Theoretical and methodological issues. Baltimore, MD: York Press. Tees, R. C. & Werker, J. F. (1984). “Perceptual flexibility: Maintenance or recovery of the ability to discriminate non-native speech sounds”. Canadian Journal of Psychology, 38, 579–590. Tomblin, J., Barker, B., Spencer, L., Zhang, X. & Gantz, B. (2005). “The effect of age at cochlear implants initial stimulation on expressive language growth in infants and toddlers”. Journal of Speech, Language and Hearing Research, 48, 853–867. Ventureyra, V. (2005). A la recherche de la langue perdue: Étude psycholinguistique de l’ attrition de la première langue chez des Coréens adoptés en France. Unpublished Doctoral Dissertation, Paris: Ecole des Hautes Etudes en Sciences Sociales.
168 Christophe Pallier
Ventureyra, V. & Pallier, C. (2004). “In search of the lost language: The case of adopted Koreans in France”. In M. S. Schmid, B. Köpke, M. Keijzer & L. Weilemar (Eds.), First Language Attrition: Interdisciplinary perspectives on methodological issues (pp. 207–221). Amsterdam: John Benjamins. Ventureyra, V., Pallier, C. & Yoo, H.-Y. (2004). “The loss of first language phonetic perception in adopted Koreans”. Journal of Neurolinguistics, 17, 79–91. Weber-Fox, C. M. & Neville, H. J. (1996). “Maturational constraints on functional specializations for language processing: ERP and behavioral evidence in bilingual speakers”. Journal of Cognitive Neuroscience, 8, 231–256.
A hidden language Recovery of a ‘lost’ language is triggered by hypnosis Rosalie Footnick
Unité de Neuroimagerie Cognitive (CEA INSERM U562), Orsay, France
Our case study confirms two previous studies demonstrating that a ‘lost’ language can be recovered by hypnosis. These results point to the existence of another type of attrition. Here a distinction is made between this type of attrition of a ‘hidden language’ (HL), that is, one that has not been forgotten but rather has become inaccessible to the speaker, and the traditionally studied attrition of a ‘forgotten language’ (FL). A conflict hypothesis is proposed to explain HL attrition. An attempt is made to show how these two types of attrition could be distinguished and to explain why brain imagery analyses are important for this distinction. Keywords: hypnosis; neuroimaging; language forgetting; accessibility; Ewe
Introduction Investigations into language attrition can encounter at least three different types of language loss, and for each type, loss can be either partial or total. The first type of language loss results in a language system that can be called Type AL (abnormal language), because the speech produced is clearly abnormal. This type is pathological in origin as a direct result of physiological damage to the brain or other parts of the speech system. This type of language loss is very well documented by the extensive literature on aphasia, agnosia, agrammatism, etc. (e.g. Caplan 1992; de Bleser 2003). A second type of attrited language system, non-pathological in nature, can be designated as Type FL for a forgotten language. This type occurs when a speaker has acquired a native language which, either through lack of use together with the use of another one, or through interference when two or more . Attrition can also affect a second or foreign language, but here we are concerned with L1 attrition.
170 Rosalie Footnick
languages are in constant use, is in a process of forgetting or restructuring. And thirdly, we argue, it is also possible that a language only appears to be forgotten. This last type will be referred to here as HL – a hidden language. Knowledge of a language can attrite partially (e.g. Cook 2003; Lambert & Freed 1982; Schmid 2002; Schmid, Köpke, Keijzer & Weilemar 2004; Seliger & Vago 1991) or totally (Pallier, Dehaene, Poline, LeBihan, Argenti, Dupoux & Mehler 2003; Ventureyra, Pallier & Yoo 2004). An example of total language attrition, which seems to be due to forgetting, is described by Pallier et al. (2003) and Ventureyra et al. (2004) in studies on Korean adoptees (mean age 26.6) who no longer had any contact with their native Korean after their arrival in French families between the ages of 3 and 10. A functional magnetic resonance imaging (fMRI) study showed no specific brain activity when the adopted Korean participants were listening to Korean sentences as opposed to sentences in unknown languages, such as Japanese or Polish. Pallier (this volume) suggests that there is empirical evidence that could put into question the notion (Penfield 1965) that the acquisition of an L1 will always leave long-lasting traces in the brain, since their Korean adoptees appear to have entirely lost their L1. However, Pallier allows the possibility that there may be traces of the adoptees’ early exposure to Korean which their paradigm was not able to uncover. Thus, the question remains – is it possible that an L1 cannot be erased from memory, but rather only becomes more and more inaccessible (Sharwood Smith & van Buren 1991)? If it could be demonstrated that in certain cases an L1 seems to be erased from memory but, in fact, is only inaccessible to the speaker, this would be a case of the third type of attrition proposed above, that which results in a hidden language (HL). Contrary to the first type of language loss, which is attrition resulting in a forgetting process, an HL exists when a language has been acquired as a native (or a second) language and, at some later date, becomes largely inaccessible to the speaker, rather than just being restructured or partly forgotten. It is proposed that when the inaccessible language is an HL, the memory traces are still stored in the brain but blocked from accessibility by some unknown mechanism. There are two previous studies (As 1962; Fromm 1970) which report the recovery of a lost language through the use of hypnosis, giving support to the suggestion that such a mechanism is possible. The present study represents an attempt to replicate and confirm these findings. Since hypnosis is not usually considered an acceptable tool, and since the ‘loss’ of the L1 can only be established on the basis of the participant’s self-report, the initial intention was to obtain objective data to document the difference between the ‘during hypnosis’ and ‘out of hypnosis’ conditions. Such data were to be provided by the monitoring of brain activity before, during and after hypnosis by means of fMRI.
Recovery of a hidden language through hypnosis 171
Unfortunately, it became clear during the experiment that this second step would not be possible, due to our participant’s surprising level of recovery of his lost language, Mina, after several sessions of hypnosis (Sections 3.2 and 3.3). This process, however, cannot be taken to indicate that every forgotten language is also an HL. The conditions between the participant for the present study (CK) and, for example, the Korean adoptees were different: CK had had passive exposure to his lost language for a few hours almost every week for 15 years; whereas the adoptees had had no exposure to Korean. Furthermore, the separation between the use of one language and acquisition of the other was less clear-cut in the present case, since CK continued to speak his L1 (French) during the three years in which he was acquiring his L2 (Mina), whereas the adoptees never spoke Korean after leaving Korea. The present paper will propose a ‘conflict’ hypothesis as a possible explanation of the attrition process which results in an HL. Briefly, it is proposed that after acquiring the HL, the speaker had a conflict with regard to producing it, and this resulted in that language becoming inaccessible to him. This hypothesis will be described in the following terms: (1) a conflict can cause information to become inaccessible in a manner that is not ‘normal’ forgetting; (2) hypnosis can decrease conflict, and (3) a decrease in conflict can allow blocked information to become accessible to the speaker. Previous neuroimaging studies will be presented to support each of these three ideas.
1. Past studies of attrition resulting in an HL 1.1 As (1962) As (1962) conducted an experiment to determine if a ‘lost’ language could be recovered through the use of hypnosis, specifically using age-regression. The ageregression technique is used in hypnosis to enable the participant to concentrate completely on a particular past age period specifically in order to enable him/her to describe images and events that happened during that period very vividly. However, the concern is not the entire age period, but the vividness of the description of specific events during that age period. Whether or not this method gives completely accurate access to childhood memories is controversial (Kihlstrom 2003). The participant was an 18-year-old American who had immigrated to the United States from Sweden at the age of 5. At the time of the experiment he said that he could not recall his native language. He was born to Swedish parents living in Finland. They divorced and he moved to the States with his mother, who
172 Rosalie Footnick
remarried when he was almost 7, and told him that they were not to converse in Swedish, since the new stepfather only spoke English. A distribution of scores for the participant’s knowledge of Swedish before and during hypnosis was made, using 56 questions asked orally in Swedish (e.g., How old are you? Do you have a brother? Can you point to your head?). Under hypnosis with age-regression, the participant was able to answer significantly more questions correctly. The author described definite progress, according to the tests he gave, in the participant’s ability to understand and speak Swedish during hypnosis after four sessions.
1.2 Fromm (1970) Fromm (1970) came across the recovery of a ‘lost’ language accidentally. The author was asked to watch a hypnotic training session, in order to give a young therapist feedback. The participant was a 26-year-old Japanese-American student, who said he did not know Japanese. Fromm had just returned from Japan where she had learned a few sentences of that language. During the hypnosis, when the participant was age-regressed to 7 years old, Fromm used her limited repertoire of Japanese which the participant was unable to understand. Some months later, Fromm used the same participant to demonstrate hypnosis and age-regression. This time the participant was age-regressed to 3 years old and, without provocation, the participant spoke rapid Japanese which lasted 10 minutes. Out of hypnosis, he could not speak Japanese and he had spontaneous posthypnotic amnesia. A taped session confirmed that the participant was able to speak Japanese fluently when age-regressed to 3 years old. When the participant listened to the recording out of hypnosis, he was only able to understand a few isolated words. A translation of the tape revealed that from 1 to 4 years old the participant had been put into a relocation camp with his family during World War II where he had spoken Japanese.
2. Pilot study of an HL A pilot study was conducted in collaboration with C. Pallier and N. Prieur to seek confirmation of the results of As and of Fromm. If the reactivation of an HL under hypnosis could be confirmed, it was intended to add fMRI data which would . Americans of Japanese descent in California had been put into these camps during the war.
Recovery of a hidden language through hypnosis 173
show the brain activity associated with the ‘lost’ language before, during and after hypnosis. The 21-year-old participant for this pilot study, CK, was a university student in Paris. He was born in Paris to Togolese parents who only spoke French in the home. At the age of 2½ years, he, his mother and his older brother went to live with his grandmother in Togo, where they stayed for 3 ½ years and then returned to Paris. During that time, he acquired and spoke fluent Mina (a simplified spoken local variant of Ewe) with his grandmother – whom he adored – and with his playmates. Upon his return to Paris, CK’s mother was told by the head of the primary school that speaking Mina in the home would hinder CK’s scholastic achievement. CK’s mother related this to him and told him not to use Mina any more. At the time that the author interviewed CK, he said that he was unable to understand or speak Mina which he had not spoken since he was 6 years old. CK could only understand and produce about 30 isolated words all of which concerned food, school or childhood games, such as, ‘grandmother’, ‘brother’, ‘milk’, ‘to play’, etc. This loss of Mina had taken place in spite of the fact that CK was present at family gatherings where Mina was spoken almost every Sunday. He said he was never able to participate in any of those conversations. The therapist Nicole Prieur (NP), who uses hypnosis for therapeutic reasons, decided to use the age-regression technique during hypnosis with CK, since he had habitually used Mina between 3 and 6 years of age. A preliminary session established that CK was a highly hypnotizable participant. In that session, CK was tested to see what he could understand and produce in Mina out of hypnosis. The participant served as his own control, since his ability to understand and speak Mina out of hypnosis was compared to when he was under hypnosis. A translator (AW) was present to ask CK questions in Mina. AW communicated with the therapist in written French for guidance for the questions. During each session, CK was put under hypnosis for one hour and all sessions were taped. The number of sessions was to be guided by CK’s ability to understand and to produce Mina during hypnosis. About one month elapsed between the sessions for translation and analysis of data. There were six sessions.
. The participant gave written informed consent. Our participant was not an adoptee. For ethical reasons, it is complicated to do the type of experiment that we were setting up with an adoptee.
174 Rosalie Footnick
Due to the work on priming and hypnosis, NP gave CK an exact posthypnotic suggestion just before ending each session – “Forget everything that has taken place during hypnosis, nothing, not a single word, will remain in your mind”. This was to avoid priming effects, since it was undesirable that CK should recover his Mina out of hypnosis before the planned fMRI part of the experiment, which could only take place when it was clear that he could recover Mina under hypnosis. It was agreed with CK that he would not know anything about the sessions until the experimenters chose to give that information.
2.1 Experiment and results The sessions, with CK under hypnosis and age-regressed to 4 or 5 years old, consisted of a free conversation and sometimes included a questionnaire. These parts were designed to observe CK’s ability: (1) to participate in free spontaneous conversation in Mina with the translator during hypnosis and (2) to answer a series of prearranged questions asked by the translator in Mina before, during and after hypnosis. During hypnosis, the beginning of the free conversation always took place with the therapist in French. CK, age-regressed, was asked to describe a scene of his life in Togo that he could remember from that age period. CK would recall these scenes in French and say what he thought the people in the scenes were saying. Then, while continuing to describe these scenes, CK was asked to talk with the translator in Mina about them. For Session 1, twenty prearranged questions (appropriate for a 5-year-old) were read out by the translator in Mina three times: once before, once during and
. Priming is an increase in the ability to identify or produce an item (e.g., a word, a number, etc.) as a result of previous exposure to that item. There are two main types of priming, one type occurs when the participant is not conscious of the effect (subliminal exposure) and the other occurs with the participant conscious of the previous exposure (Posner 1978). . Kihlstrom’s original study (1980) showed that a posthypnotic suggestion given to participants – to forget everything that took place under hypnosis – resulted in priming in a test for implicit memory and no priming in a test for explicit memory. In the declarative (explicit and conscious) memory system is knowledge of facts (semantic knowledge) and events (episodic knowledge). In the procedural (implicit and non-conscious) memory system is knowledge which is expressed through performance in habits and skills, and simple forms of conditioning. Bike riding and native language are considered to be examples of implicit memory (Squire 1987 & Tulving 1972). . A posthypnotic suggestion is an idea given to the hypnotized person which is to be carried out after the person is no longer under hypnosis. Sometimes the therapist gives a cue for carrying out the suggestion after hypnosis and another cue for ending the suggestion.
Recovery of a hidden language through hypnosis 175
once after hypnosis (Can you point to your arms? What games do you like to play? Do you have a brother? What kind of food do you like?). Example out of hypnosis in Session 1: (1) Translator in Mina: “Do you have a brother?” CK: “Je comprends ‘frère’ mais je ne comprends pas le sens de la phrase. Si je dois répondre par ‘oui’ ou ‘non’, j’imagine que je dirais ‘oui’ à une question concernant mon frère.” (‘I understand ‘brother’ but not the meaning of the sentence. If I had to answer ‘yes’ or ‘no’, I guess I would say ‘yes’ to a question concerning my brother.’)
Example under hypnosis in Session 1 with CK age-regressed to 4 years old: (2) Translator in Mina: “Do you have a brother?” CK: “Oui” (‘Yes’)
CK could understand several isolated words but not complete sentences. Several of the words in Mina that CK knew were deliberately used in the questionnaire to help him to begin to speak Mina, but it became evident that CK could sometimes guess the global meaning of the sentence from knowing one or two words. In addition, the fact that it was possible to answer ‘yes’ or ‘no’ to several of the questions made the results unclear as Example (2) above shows. CK’s answers were divided into three categories: Category 1 refers to responses which indicate that he did not understand the question, or which consist of oui (‘yes’) or non (‘no’). Category 2 contains those responses which indicate that a few words of the question, or the question’s global meaning were understood. Category 3 responses show that CK understood the question completely. Table 1 shows that there is no difference before, during and after hypnosis with respect to the distributions of responses between the three conditions in the first session (numbers are given in percentage of total number of responses). In Session 2, for the first 28 minutes after the beginning of hypnosis, the therapist and translator were trying to coax CK, age-regressed to four years old, into answering questions in Mina concerning the scene he was describing about his life in Togo. CK produced a word here and there in Mina during this time. Then, for the first time, CK said a few complete simple sentences in Mina and without any accent or errors in grammar. Table 1. Percentage of CK’s responses for each category in Session 1 Before hypnosis (n = 20) During hypnosis (n = 20) After hypnosis (n = 20)
Category 1
Category 2
Category 3
Total %
30 25 25
60 65 65
10 10 10
100 100 100
176 Rosalie Footnick
Table 2. Percentage of responses for each category in Sessions 2, 3, 4 and 5 during hypnosis7 Category 1
Session 2 (n=64) Session 3 (n=81) Session 4 (n=58) Session 5 (n=79)
57.8 80.2 8.6 27.8
Category 2
6.2 0 0 0
Category 3 CK answers in French
CK answers in Mina
4.7 3.7 77.6 0
31.2 8.6 13.8 72.1
Total %
99.9 92.5* 100 99.9
*In this session, 7.4% of CK’s responses were sounds that were incomprehensible and when he was asked to be more precise, he would just make another sound. Thus these responses could not fit into any category and were not counted in the results.
An example during hypnosis of an interaction totally in Mina in Session 2: (3)
Translator: “What is your grandmother doing?” CK: “She is preparing a sauce.” Translator: “What dish is she preparing?” CK: “Smoked chicken and gombo” Translator: “What else is she putting into the sauce?” CK: “She’s putting in some salt.”
Table 2 shows that in Session 3 during hypnosis, CK understood less than in Session 2. Example (4) below appears to indicate that CK was blocking or refusing to answer the translator in Mina, because it became clear that he was capable of answering. This interaction occurred when CK was describing a scene (under hypnosis and age-regressed) that he said he had seen happening in Togo. There were three interactions of this type in Session 3. (4) Translator in Mina: “Can you tell me what is happening in the street in Mina?” CK in French: “Non.” (‘No’) Translator: “Qu’est que les adultes dans la rue sont en train de dire en mina?” (‘What are the adults in the street saying in Mina?’) CK in French: “Rien” (‘Nothing’) Therapist (in a tone of voice which is gently pushing CK to respond): “Qu’est que les adultes disent en mina?” (‘What are the adults saying in Mina?) CK in Mina: “Don’t play ball in the street.”
. Before and after hypnosis CK was unable to understand any complete sentence or to produce a complete sentence in Mina in all sessions.
Recovery of a hidden language through hypnosis 177
In Session 4 two new elements were introduced. This was done because there had been a decrease in CK’s understanding in Session 3 compared to Session 2. First, NP put CK into a deeper hypnosis, and second, she suggested to CK, when under hypnosis and age-regressed, that he was not speaking Mina himself, but rather just mouthing what the other people in the scene he was describing were saying. An example of this with CK under hypnosis in Session 4: (5) Therapist: “Qu’est-ce qu’elles disent, simplement répète ce qu’elles sont en train de dire.” (‘What are the girls saying, just repeat what they say.’) CK: “Il a pris l’argent avec lui.” (‘He took the money with him.’) Therapist: “Mais maintenant je veux l’entendre des bouches des filles, ce n’est pas vous qui me le racontez, laissez votre bouche dire leurs mots.” (‘But now I want to hear it from the girls, you are not telling me, let your mouth say their words.’) CK: “J’essaye, mais je n’y arrive pas.” (‘I am trying to, but I can’t seem to.’) Therapist: “Ce n’ est pas vous qui dites quelque chose, ce sont les filles qui sont en train de parler ensemble.” (‘You are not saying anything, it is the girls who are talking to each other.’) CK in Mina: “He took the money with him.”
The results in Table 2 show that in Session 4 CK answered the therapist in French when asked questions in Mina by the translator, instead of answering the translator in Mina. However, CK did produce 12 complete sentences in Mina, all grammatically correct. In Session 5, since it was now clear that CK could understand and speak some Mina, two new elements were added. A forced-choice questionnaire was given during and then again after hypnosis. The second element was that the therapist told CK after he was under hypnosis and age-regressed that he was able to speak Mina. After 6 minutes into the hypnotic state and after the therapist had told CK that he could speak Mina; he then conversed with the translator in Mina for 38 minutes. Then, suddenly, CK stopped, said in Mina that he was tired and his grandmother was putting him to bed. The difference should be noted between Sessions 4 and 5: in Session 4, CK answered the therapist (77% of the total number of possible responses) in French when she asked him to answer the translator’s questions which were said in Mina, but in Session 5, he answered the translator (72% of the total number of possible responses) directly in Mina. The results of the forced-choice questionnaire (40 items appropriate for a 4year-old) with three possible answers were not significantly different during and after hypnosis. It was concluded that it was too easy to guess at the correct answer (global meaning) from knowledge of one word in the sentence. The problem in
178 Rosalie Footnick
creating forced-choice questions for a 4-year-old is that negativity (other than just the word ‘No’) and active/passive voice have not yet been mastered in L1 acquisition by this age. Thus all the questions had to be stated in the affirmative with active voice. These simple questions allowed for too much guessing. It was impossible to ask questions such as ‘Don’t you go to school on Sunday?’ ‘Was the lion eaten by the mouse?’ It is also possible that there was in fact a priming effect. At the end of Session 5, CK, out of hypnosis, heard the recording of the 12 complete sentences that he had produced in Mina during Session 4. CK was unaware until this time that he was able to understand and speak Mina. CK’s comments while listening to himself on tape speaking Mina with the translator were recorded. His comments made it clear that he was astounded to hear himself speaking in Mina in full sentences. He was asked to translate these 12 sentences into French. However, he was only able to understand isolated words but not the meaning of any sentence, as illustrated by examples (6) and (7) below. This result was identical to that of the participant in the Fromm (1970) study who heard himself on tape speaking Japanese that he had produced under hypnosis (Section 2.2). Examples from when the tape was played back to CK in Session 5: (6) CK in Mina on the tape: “He took the money with him.” CK’s comment in French after hearing his sentence: “Je ne peux pas traduire la phrase, je ne comprends que le mot ‘argent’ mais je ne comprends pas le sens de la phrase.” (‘I can’t translate the sentence, I only understand the word ‘money’ but not the meaning of the sentence.’) (7) CK in Mina on the tape: “That man was terribly afraid.” CK’s comment in French after hearing his sentence: “Je comprends ‘peur’ mais je ne comprends pas le sens de la phrase.” (‘I understand ‘afraid’, but not the meaning of the sentence.’)
The results from Session 5 (see Table 2) indicated that CK could understand and speak more Mina under hypnosis than he was capable of out of hypnosis. In Session 6, CK was tested to see what amount of Mina he could now produce without hypnosis. The translator had a typescript of Session 5 and asked CK appropriate questions in Mina to try to reproduce the conversation that had taken place in Session 5. This was done twice in order to separate listening comprehension from oral production abilities: the first time, CK was asked to answer only in Mina, and the second time, he was asked to translate what AW said into French. Lastly, the same forced choice questionnaire given in Session 5 was repeated (see Table 3). Table 3 indicates that CK was more capable of answering in Mina during hypnosis in Session 5 than out of hypnosis in Session 6, but that by Session 6, CK had recovered some ability to understand and to produce Mina out of hypnosis.
Recovery of a hidden language through hypnosis 179
Table 3. Percentage of responses per category in Session 5 compared to Session 6 Category 1 Session 5 (n = 79) During Hypnosis 27.8 CK answers in Mina Session 6 (n = 79) Out of Hypnosis 55.7 CK answers in Mina Session 6 (n = 79) Out of Hypnosis 21.5 CK answers in French
Category 2
Category 3
Total %
0
72.1
99.9
11.4
32.9
100
35.4
43.0
99.9
In Session 6 out of hypnosis, CK seems not to be able to understand many questions (55% of the questions asked) when he was asked to respond only in Mina. However, that number decreased to 21% when he was asked to translate the questions into French. Thus, it was considered, given that CK could respond to 32% of the questions in Mina out of hypnosis and could understand more Mina than he could produce, that the difference in his ability to understand Mina under hypnosis and out of hypnosis was no longer great enough to be able to obtain data for a neuroimaging study. This decision was based on three observations. Firstly, although there was a difference between CK’s answers in Mina under hypnosis in Session 5 and his answers in Mina out of hypnosis in Session 6, it was clear from his ability to translate the questions into French that he had understood much more than he could say in Mina out of hypnosis. Secondly, during Session 6 out of hypnosis, CK was able to answer the forced choice questions as well as he did when he was under hypnosis in Session 5, showing that he had increased comprehension in Mina. Thirdly, given that Session 5 could have had priming effects on Session 6, a last test was given before finishing Session 6. The translator was asked to talk with CK in Mina about CK’s life today (a totally different style of speech and vocabulary than the scenes in Togo). Although there was a lack of vocabulary, CK was nonetheless able to converse on this subject for six minutes in Mina. The only factor that might have influenced CK’s recovery of Mina out of hypnosis was that at the end of Session 5 CK heard the tape of himself speaking Mina during Session 4. Even though he could not understand himself, he was clearly visibly surprised and pleased to hear himself speaking Mina. Perhaps knowing that he could understand and speak Mina decreased his conflict which allowed him access to the HL when he was not under hypnosis.
180 Rosalie Footnick
2.2 Conclusions of the pilot study There were three conclusions from this experiment. 1. This case study confirmed the two previous ones in that a ‘lost’ language can be recovered under hypnosis, which shows that there is attrition resulting in an HL. 2. In spite of precautions taken with a posthypnotic suggestion to forget everything that happened under hypnosis, an unexpected recovery of Mina out of hypnosis did take place. 3. Due to CK’s recovery of the comprehension of Mina out of hypnosis, it was considered likely that no significant differences with CK listening to Mina before, during and after hypnosis would be perceivable in neuroimaging data. The results of this study indicate that it is possible for memory traces of a lost language to remain in the brain although that language can be inaccessible to the speaker. Perhaps for the neural connections to still remain in the brain, a necessary condition is that there is some passive exposure to the HL during the time when that language is not spoken, which was the case for our participant.
3. Preliminary directions for HL research If the neural traces for the language no longer functionally existed, which presumably happens when a language is completely forgotten, then the results of the three studies (our pilot study; As 1962; Fromm 1970) on HL recovery would not have been possible. Because this type of recovery is global, as opposed to partial, and the recovered language is grammatically correct, it might be preferable for HL research not to attempt to describe the attrited language as is done for an FL, but rather to explain the mechanisms that might produce an HL. What type of process might cause the loss of the language resulting in an HL and what mechanism could allow this global type of recovery? How can an HL be distinguished from an FL? A preliminary hypothesis to answer these questions will be proposed based on the assumption that some sort of conflict has caused the HL to become inaccessible. This ‘conflict’ hypothesis will be put forth linking ideas concerning conflict, brain activity and hypnosis.
Recovery of a hidden language through hypnosis 181
3.1 Explaining an HL All three studies on HL attrition could be interpreted in the following way: the participants had had an involuntary conflictual situation associated with speaking the HL when they were young after they had acquired the language. This conflict might have resulted in the language becoming inaccessible. In the As (1962) study, speaking Swedish linked the participant to his mother but would have excluded his stepfather, thus his mother made the request for them not to speak Swedish. For the Fromm (1970) study, speaking Japanese in the family would have been natural, of course, but would have reminded the participant and his parents of the painful time during the war in the relocation camp, and thus it was not spoken. In our pilot study, after the participant returned to Paris, he was told by his mother not to speak Mina, since that could cause him not to do well in school. If an individual has blocked a language and it has thus become inaccessible, then during hypnosis there could be increased or decreased conflict with respect to that language depending on the hypnotic suggestions being given. For example, in Session 4, when the therapist repeatedly asks CK to try to answer in Mina during the hypnosis, this seems to create ‘conscious’ conflict. CK answers (5) “I am trying to [speak in Mina], but I can’t seem to.” However, in Session 5 when NP tells CK during hypnosis that he is able to understand and to speak Mina, this seems to have ‘allowed’ him to do so without the conflict, as he then conversed non-stop in Mina for 38 minutes. Assuming there was conflict involved in the use of the HL when CK returned from Togo, then the suggestion here is that when the conflict arose there was ‘active’ forgetting. Neuroimaging studies have shown that a conflict can cause information to become inaccessible in a manner that is not ‘normal’ forgetting. Experiments were done on the neural underpinnings which control unwanted memories using a retrieval practice paradigm (Anderson & Green 2001; Levy & Anderson 2002). In this paradigm, a habitual response (one that has been previously learned and practiced) has to be overridden (or inhibited) to a target word, the stimulus. To override a response means to actively stop the memory retrieval process for that response. These authors showed that by actively blocking a memory, which is called ‘active’ forgetting, it became more difficult at a later time to recover it. In another experiment (Anderson, Ochsner, Kuhl, Cooper, Robertson, Gabrieli, Glover & Gabrieli 2004), participants formed an association between a target word and a response word before fMRI scanning. During scanning their task was to actively prevent the associated response from coming into memory when the target word was presented. During the suppression of the memory, the dorsolateral and ventrolateral prefrontal cortex, anterior cingulate cortex (ACC) and motor areas increased activation. However, the suppression reduced activa-
182 Rosalie Footnick
tion in the hippocampus which is known to be important for successful memory formation relative to recollection (Squire 1987; Squire & Zola 1993). This would indicate that the memory will be more difficult to retrieve later. Anderson et al. (2004) compared this activation against simple (normal) forgetting and found the patterns of activation to be different. It is well documented that the ACC plays a role in conflict monitoring and attention (Botvinick, Braver, Barch, Carter & Cohen 2001; Botvinick, Cohen & Carter 2004; Botvinick, Nystrom, Fissell, Carter & Cohen 1999; Bush, Luu & Posner 2000; Posner & DiGirolamo 1998; Posner, Petersen, Fox & Raichle 1988), thus the authors postulate that the ACC activation signals a conflict, which was that the associated response is summoned by the target word and needs to have an override response, in order for the associated response to be inhibited. The prefrontal cortex is activated by the signals of the ACC and is known to be involved in stopping prepotent motor responses, that is, inhibiting a motor response (for example, linked to speech) that is about to happen. Thus, the associated response (the response word which has been associated with the stimulus word) is consciously being blocked and this leads to active forgetting which makes the associated response much more difficult to retrieve. If there is conflict and active forgetting of the HL, then how could conflict be reduced so as to eliminate the override response which makes the HL inaccessible? There is some research on conflict and ACC activation, as well as on the reduction of conflict by hypnosis. The Stroop task is classically used to study conflict (Stroop 1935). The participants, who have to be experienced readers, are asked to name the ink colour of a word, the meaning of which is a colour, e.g., the word green displayed in red ink. Reading the word and incorporating its meaning is automatic and leads to the behavioural Stroop Interference Effect (SIE), that is, an involuntary conflict when the meaning of the word and the colour of the ink of the word are incongruent. When the ink colour and the meaning of the word are congruent, the participant responds faster and with fewer errors. Several neuroimaging studies have shown that ACC activation covaries with the levels of response (reaction time and errors) due to conflict, such as that found in the Stroop task (Botvinick et al. 1999, 2001, 2004; Carter, MacDonald, Botvinick, Ross, Stenger & Noll 2000; Casey, Thomas, Welsh, Badgaiyan, Eccard, Jennings & Crone 2000; MacDonald, Cohen, Stenger & Carter 2000; Ullsperger & von Cramon 2001). The levels of the behavioural interference effects, indicating conflict for the participants, have been shown to increase or decrease in correlation with the increase or decrease of ACC activity (Botvinick et al. 1999; Carter et al. 2000; Casey et al. 2000; Kerns, Cohen, MacDonald, Cho, Stenger & Carter 2004). Hypnosis is often not considered to be a real phenomenon. However, its use has been established for example with respect to the control of pain and the de-
Recovery of a hidden language through hypnosis 183
crease of conflict. It has been well documented that hypnosis can decrease both the sensation of pain itself and the unpleasantness of the noxious stimuli (the subjective experience of physical pain), both of which are correlated with a decrease in ACC activity (e.g. Crawford, Gur, Skolnick & Benson 1993; Faymonville, Laureys, Degueldre, DeFliore, Luxen, Franck, Lamy & Maquet 2000; Raij, Numminen, Navanen, Hiltunen & Hari 2005; Rainville, Duncan, Price, Carrier & Bushnell 1997). Thus in many hospitals, patients are conscious when they undergo surgery with local anaesthetics combined with light sedation and hypnosis, instead of general anaesthetics. Recent neuroimaging research has shown that in highly hypnotizable participants a specific posthypnotic suggestion, to perceive written words as nonsense strings during the Stroop colour task, was able to reduce the involuntary conflict (Raz, Shapiro, Fan & Posner 2002) and that with the same posthypnotic suggestion the decrease in the involuntary conflict was associated with decreased ACC activation, as well as decreased activity in visual areas as compared to controls (Raz, Fan & Posner 2005). Since hypnosis can reduce conflict, with the reduction corresponding with ACC activity, it is possible that in the three case studies, hypnosis reduced the conflict associated with the HL which allowed its recovery. To recapitulate: if the idea of a conflict provoking an HL is assumed, then an explanation of the mechanism provoking an HL and its recovery is possible by linking the above ideas on conflict, hypnosis and brain activity to an HL. First, it is hypothesized that the conflict causes the HL to be continuously associated with ‘active’ forgetting and this results in inaccessibility of the HL. Second, since reduced conflict (and reduced pain) has been shown to be correlated with reduced ACC activation, and hypnosis has been shown to reduce conflict (and reduce pain) which is correlated with reduced ACC activation, it is hypothesized that through hypnosis the conflict linked to the HL is reduced allowing the speaker to access the HL. This proposition could be tested by means of neuroimaging studies which could link HL recovery, conflict and ACC activation.
3.2 Possible studies Such neuroimaging studies could be set up to monitor brain activity while the participants are being scanned as they hear the HL before, during and after hypnosis. The prediction is that this type of study would show an increase in ACC
. For example, in France, Belgium, Germany, Finland and the USA (Büchel, Bornhövd, Quante, Glauche, Bromm & Weiller 2002; Crawford et al. 1993; Faymonville et al. 2000; Petrovic, Petersson, Ghatan, Stone-Elander & Ingvar 2000; Ploghaus, Becerra, Borras & Borsook 2003).
184 Rosalie Footnick
activity before and after hypnosis and a decrease in ACC during hypnosis which would be correlated with the recovery of the HL. Presumably, there would also be a difference in activity in the language area correlated with the recovery of the HL; whereas there would be no activity in that area before and after hypnosis when the HL was inaccessible to the speaker. These types of studies would also allow the differentiation between an FL and an HL. If the lost language is an FL, then during hypnosis there will be no global recovery of the language and no decrease in ACC activity. A partially attrited FL will be correlated with activity in the language area unless the FL has been completely forgotten in which case there will be no activity in the language area. Thus, what appears to be a completely attrited FL from linguistic as well as neuroimaging data could be an HL. This could only be revealed by inducing hypnosis while scanning the person who has lost the language. Would the Korean adoptees show no brain activity specific to hearing Korean as opposed to other unknown languages before hypnosis and then show a specific activity in the language area during hypnosis if Korean were to be recovered? If during hypnosis there were to be no recovery, it might show that passive exposure to a lost language is a necessary condition for it to have long-lasting neural traces, although it remains inaccessible to the speaker. It is proposed that losing an FL is a slow process with parts of the language disappearing individually in a systematic way over a long time due to a normal forgetting process, while the loss of an HL is a rapid process in which the entire language, except for a few isolated words, disappears over a short time, due to psychological events involving conflict concerning the use of the HL. This conflict causes the speaker to block access to the language through active forgetting. The HL is inaccessible to the speaker, but the language still exists in the brain. If it is recovered, there is no grammatical loss of that language, it is fluent and spoken without any accent. An FL is either partly inaccessible due to a gradual grammatical and lexical loss or it is totally forgotten with no neural trace of the language in the brain. Here the attempt has been to show (1) how language memory traces could be blocked from consciousness and (2) how this ‘lost’ language could be retrieved back into consciousness. Clearly the ‘conflict’ hypothesis is based on only three studies and needs to be verified.
Acknowledgement I thank Christophe Pallier for very useful comments on a first version on this text.
Recovery of a hidden language through hypnosis 185
References Anderson, M. C. & Green, C. (2001). “Suppressing unwanted memories by executive control”. Nature, 410, 366–369. Anderson, M. C., Ochsner, K. N., Kuhl, B., Cooper, J. Robertson, E., Gabrieli, S. W., Glover, G. H. & Gabrieli, J. D. E. (2004). “Neural systems underlying the suppression of unwanted memories”. Science, 303, 232–235. As, A. (1962). “The recovery of forgotten language knowledge through hypnotic age regression: A case report”. The American Journal of Clinical Hypnosis, 5, 24–29. Botvinick, M. M., Nystrom, L. E., Fissell, K., Carter, C. S. & Cohen, J. D. (1999). “Conflict monitoring versus selection-for-action in anterior cingulate cortex”. Nature, 402, 179– 181. Botvinick, M. M., Braver, T. S., Barch, D. M., Carter, C. S. & Cohen, J. D. (2001). “Conflict monitoring and cognitive control”. Psychological Review, 108 (3), 624–652. Botvinick, M. M., Cohen, J. D. & Carter, C. S. (2004). “Conflict monitoring and anterior cingluate cortex: An update”. Trends in Cognitive Sciences, 8 (12), 539–546. Büchel, C., Bornhövd, K., Quante, M. Glauche, V., Bromm, B. & Weiller, C. (2002). “Dissociable neural responses related to pain intensity, stimulus intensity and stimulus awareness within the anterior cingulate cortex: A parametric single-trial laser functional magnetic resonance imaging study”. The Journal of Neuroscience, 22 (3), 970–976. Bush, G., Luu, P. & Posner, M. I. (2000). “Cognitive and emotional influences in anterior cingulate cortex”. Trends in Cognitive Sciences, 4 (6), 215–222. Caplan, D. (1992). Language: Structure, processing and disorders. Cambridge, MA: MIT Press. Carter, C. S., MacDonald, A. M., Botvinick, M., Ross, L. L., Stenger, V. S. & Noll, D. (2000). “Parsing executive processes: Strategic vs. evaluative functions of the anterior cingulate cortex”. Proceedings of the National Academy of Science, USA, 97, 1944–1948. Casey, B. J., Thomas, K. M., Welsh, T. F., Badgaiyan, R. D., Eccard, C. H., Jennings, J. R. & Crone, E. A. (2000). “Dissociation of response conflict, attentional selection, and expectancy with functional magnetic resonance imaging”. Proceedings of the National Academy of Science, USA, 97, 8728–8733. Cook, V. (Ed.) (2003). Effects of the Second Language on the First. Clevedon: Multilingual Matters. Crawford, H. J., Gur, R. C., Skolnick, B. & Benson, D. M. (1993). “Effects of hypnosis on regional cerebral blood flow during ischemic pain with and without suggested hypnotic analgesia”. International Journal of Psychophysiology, 15, 181–195. de Bleser, R. (Ed.) (2003). The Science of Aphasia: From therapy to theory. Oxford: Pergamon. Faymonville, M. E., Laureys, S., Degueldre, C., DeFliore, G., Luxen, A., Franck, G., Lamy, M. & Maquet, P. (2000). “Neural mechanisms of antinociceptive effects of hypnosis”. Anesthesiology, 92, 1257–1267. Fromm, E. (1970). “Age regression with unexpected reappearance of a repressed childhood language”. The International Journal of Clinical and Experimental Hypnosis, 18, 79–88. Kerns, J. G., Cohen, J. D., MacDonald III, A. W., Cho, R. Y., Stenger, A. & Carter, C. S. (2004). “Anterior cingulate conflict monitoring and adjustments in control”. Science, 303, 1023– 1026. Kihlstrom, J. F. (1980). “Posthypnotic amnesia for recently learned material: Interactions with ‘episodic’ and ‘semantic’ memory”. Cognitive Psychology, 12, 227–251.
186 Rosalie Footnick
Kihlstrom, J. F. (2003). “Hypnosis and memory”. In J. F. Byrne (Ed.), Learning and Memory, 2nd ed. (pp. 240–242). Farmington Hills, MI: Macmillan. Lambert, R. D. & Freed, B. F. (Eds.) (1982). The Loss of Language Skills. Rowley, MA: Newbury House. Levy B. J. & Anderson, M. C. (2002). “Inhibitory processes and the control of memory retrieval”. Trends in Cognitive Science, 6 (7), 299–305. MacDonald III, A. W. Cohen, J. D., Stenger, V. A. & Carter, C. S. (2000). “Dissociating the role of the dorsolateral prefrontal and anterior cingulate cortex in cognitive control”. Science, 288, 1835–1838. Pallier, C., Dehaene, S., Poline, J.-B., LeBihan, D., Argenti, A.-M., Dupoux, E. & Mehler, J. (2003). “Brain imaging of language plasticity in adopted adults: Can a second language replace the first?” Cerebral Cortex, 13, 155–161. Penfield, W. (1965). “Conditioning the uncommitted cortex for language learning”. Brain, 88, 787–798. Petrovic P., Petersson K. M., Ghatan P. H., Stone-Elander S. & Ingvar M. (2000). “Pain-related cerebral activation is altered by a distracting cognitive task”. Pain, 85, 19–30. Ploghaus, A., Becerra, L., Borras, C. & Borsook, D. (2003). “Neural circuitry underlying pain modulation: Expectation, hypnosis, placebo”. Trends in Cognitive Sciences, 7 (5), 197–200. Posner, M. I. (1978). Chronometric Explorations of Mind. Hillsdale, NJ: Lawrence Erlbaum Associates. Posner, M. I., Petersen, S. E., Fox, P. T. & Raichle, M.E. (1988). “Localization of cognitive operations in the human brain”. Science, 240, 1627–1631. Posner, M. I. & DiGirolamo, G. J. (1998). “Executive attention: Conflict, target detection and cognitive control”. In R. Parasuraman (Ed.), The Attentive Brain (pp. 401–23). Cambridge, MA: MIT Press. Raij, T. T., Numminen, J., Navanen, S. Hiltunen, J. & Hari, R. (2005). “Brain correlates of subjective reality of physically and psychologically induced pain”. Proceedings of the National Academy of Science, USA, 102, 2147–2151. Rainville, P., Duncan, G. H., Price, D. D., Carrier, B. & Bushnell, M. C. (1997). “Pain affect encoded in human anterior cingulate but not somatosensory cortex”. Science, 277, 968– 971. Raz, A., Shapiro, T., Fan, J. & Posner, M. I. (2002). “Hypnotic suggestion and modulation of Stroop interference”. Archives of General Psychiatry, 59, 1155–1161. Raz, A., Fan, J. & Posner, M. I. (2005). “Hypnotic suggestion reduces conflict in the human brain”. Proceedings of the National Academy of Science, USA, 102, 9978–9983. Schmid, M. S. (2002). First Language Attrition, Use, and Maintenance: The case of German Jews in Anglophone countries. Amsterdam: John Benjamins. Schmid, M. S., Köpke, B., Keijzer, M. & Weilemar, L. (Eds.) (2004). First Language Attrition: Interdisciplinary perspectives on methodological issues. Amsterdam: John Benjamins. Seliger, H. W. & Vago, R. M. (Eds.) (1991). First Language Attrition. Cambridge: CUP. Sharwood Smith, M. A. & van Buren, P. (1991). “First language attrition and the parameter setting model”. In H. W. Seliger & R. M. Vago (Eds.), First Language Attrition (pp. 17–30). Cambridge: CUP. Squire, L. (1987). Memory and the Brain. New York, NY: OUP. Squire, L. & Zola, S. M. (1993). “Episodic memory, semantic memory and amnesia”. Hippocampus, 8, 205–211.
Recovery of a hidden language through hypnosis 187
Stroop, J. R. (1935). “Studies of interference in serial verbal reactions”. Journal of Experimental Psychology, 18, 643–662. Tulving, E. (1972). “Episodic and semantic memory”. In E. Tulving, W. Donaldson & G. H. Bower (Eds.), Organization of Memory (pp. 381–403). New York, NY: Academic Press. Ullsperger, M. & von Cramon, D. Y. (2001). “Subprocesses of performance monitoring: A dissociation of error processing and response competition revealed by event-related fMRIs and ERPs”. NeuroImage, 14, 1387–1401. Ventureyra, V., Pallier, C. & Yoo, H.-Y. (2004). “The loss of first language phonetic perception in adopted Koreans”. Journal of Neurolinguistics, 17, 79–91.
Identity, immigration and first language attrition Petra Prescher
Rijksuniversiteit Groningen, The Netherlands
Long term bilingual immigrants often report that they perceive first language attrition after years of living in a second language surrounding. This paper focuses on the individual’s own perception concerning first language attrition and identity. It deals with questions such as: To what extent does living in another country and speaking another language influence one’s mother tongue? If individuals experience first language attrition, do they really ‘suffer’ from it? Is there any objective evidence for their supposed decline of language skills? Data from a study of attrition in German migrants in the Netherlands will be discussed, showing the close relation between language (decline) and identity as a result of interacting factors over the lifespan. Keywords: identity theories; autobiographical narratives; sociopsychological approach
Introduction In his novel The Unbearable Lightness of Being, the Czech-French émigré writer Milan Kundera describes emigration as “walking a tightrope high above the ground without the net afforded a person by the country where he has his family, colleagues, and friends, and where he can easily say what he has to say in a language he has known from childhood” (Kundera 1984: 71). When the 53-year old Kundera wrote these lines in 1982 in his mother tongue Czech, he had been living in France for seven years. Apparently, he experienced emigration as a balancing act, feeling detached and out of place. As an immigrant, according to Kundera, one persistently tries to maintain the equilibrium between the life before and the life after migration, between two cultures, a familiar past and an unfamiliar present and, last but not least, between two languages of which the language of childhood seemingly still transmits the feeling of easiness and comfortable safety.
190 Petra Prescher
It is not surprising that émigré writers such as Kundera reflect in their novels on migration-connected loss and longing and on the impact this has on the mother tongue as, for writers, language is the medium par excellence to express themselves. It is the indispensable tool, the artifact in the Vygotskyan sense, through which they interact with their environment. Other émigré writers who describe the experiences and struggles of the late bilingual migrant include Joseph Conrad (1946), Vladimir Nabokov (1967), Joseph Brodsky (1988), Agota Kristof (1995) and Amin Maalouf (2000). In her bestseller Lost in Translation: A Life in a New Language (1989), the Polish-American author Eva Hoffman told the story of her own long and difficult process of becoming bicultural and bilingual. The acquisition of an L2 and the adoption of a new culture in her case apparently coincided with the attrition of the L1 and a kind of release from the original culture and selfhood. To her, the easiness of the mother tongue, which Kundera so longed for, seemed to be treacherous, a temporary state, since it became threatened. Moreover, L1 seemed to have an impact on identity. This raises interesting questions: if, following Vygotsky, we take language to be the “tool of tools” among the artifacts through which human beings mediate their actions in order to interact with their environment, to facilitate, shape and transform mental processes (Cole & Wertsch 2002), and to give sense to their lives – what consequences, then, has the notion of L1 attrition for the individual’s self? To what extent does living in another country and speaking another language influence one’s mother tongue and identity? How does it feel to be involved in language changing? Do individuals really experience L1 attrition and if so, do they really ‘suffer’ from it? Is there any objective evidence for their perceived L1 attrition? The present chapter presents aspects of a larger ongoing study on L1 attrition and identity, in particular on the individual’s own perception concerning bilingual and bicultural identity in relation to L1 attrition. Its predominant focus is the “in-between position” (Walter 2003) of adult immigrants, their sense of belonging or not-belonging and their attempts (or failure) to re-create their self in an L2environment. The study is based on first person accounts of German long-term immigrants in the Netherlands.
1. Theoretical framework 1.1 L1 attrition The study of L1 attrition has become a challenging subfield of applied linguistics. While (second) language acquisition has long been a subject of considerable sci-
Identity, immigration and first language attrition 191
entific interest from various disciplines, language forgetting has been studied systematically for only about two decades, mostly by applied linguists (see Ecke 2004 for a detailed overview). Progress has been made in the area of theoretical interpretation and evaluation of the phenomena related to language attrition. However, only a limited number of studies have investigated (first) language attrition with reference to psychological and philosophical phenomena such as self and identity (e.g. Kouritzin 1999; Mills 2004; Pavlenko & Blackledge 2004; Schmid 2002), whereas psychological, philosophical and sociological studies about identity have paid little or no attention to the relationship between identity and the forgetting of the mother tongue (e.g. Boym 2001; Bruner 2003; Elovitz & Kahn 1997; Holstein & Gubrium 2000; King 2000; McAdams 2003; Neisser & Libby 2000; Nelson 2003). Questions concerning the sociopsychological consequences of language forgetting, such as whether or not it affects the individual’s identity, have been the focus of writers and philosophers (Hoffman 1989; Kristeva 2000) perhaps rather than of researchers.
1.2 Identity Identity is a keyword in many contemporary debates on societal issues: “People use it to make sense of themselves, of their activities, of what they share with others and how they differ from them.” (Misztal 2003: 132). Even though research on identity has a tradition which reaches back to ancient times, in the past few decades various identity-concepts have been taken up more broadly, especially within philosophy, and (social) psychology. In psychology, identity is one of the central and most frequently studied topics of investigation. Much research has been done on ethnic, cultural, national, personal, social, and gender-identity as well as on, for instance, identity and emotions. Several theories on identity and identity formation have been developed, and some of them have become classic, such as the theory of personality of the psychoanalyst Erikson (1963, 1968a, 1968b) who describes eight psychosocial developmental stages from infancy to late adulthood, and the social identity theory (SIT) of Tajfel and Turner, later reformulated as self-categorization theory (SCT) (Oakes, Haslam & Turner 1994; Tajfel & Turner 1979, 1986) which is based on the assumption that there are three levels of category abstraction which can be used to classify the self: personal identity (the self as an individual), social identity (the
. Identity research cannot be detached from research on the self which is a widely and controversially discussed concept within psychology. In recent years, there has been an explosive revival of research on the self (for an overview see Hogg & Vaughan 2005).
192 Petra Prescher
self as a group member) and interspecies (the self as a human being). In the last two decades, more attention has been paid to identity as a complex, dynamic and unpredictable process and to the individual’s interaction with his environment during the process of identity formation. Psychological dynamic models emphasize the interconnectedness, characteristic patterns and self-organizing properties of complex systems (van Geert 1991, 2000; Haviland-Jones & Kalbaugh 2000; Lewis 1997). From a dynamic systems perspective, Haviland-Jones & Kalbaugh describe identity as a “result of (inter)subjective memories, present events, and emotional resonances that change over time and constantly provide new configurations as well as periodic repetitions” (p. 301). A special role in both changing and stabilizing the larger identity system is attributed to emotional events such as motivational systems which may be small at the origin, i.e. the initial state, but can attract more and more resonating elements over time and may have (unexpected) dramatic effects (Haviland-Jones & Kalbaugh 2000). In this respect, a life experience such as migration, connected with the notion of L1 attrition, can be seen as an emotional event that can have a considerable impact on the individual’s life and may result in an alteration of the existing identity system.
1.2.1 Transcultural identity In the late 1990s, Yoshizawa Meaders (1997), a Japanese psychiatrist in New York, developed a theory on identity formation by migrants. Working with mainly immigrant clients and analyzing their personal narratives, she observed recurring patterns. Her observations resulted in a theory on transcultural identity building which contends that transcultural identity formation occurs in three ‘phases of adjustment’ during the process of immigration. The first phase, the immersion phase, is described as the most dynamic and difficult one during which losses are acutely felt. Language is perceived as the most difficult challenge of this period, since language is vitally important in communication, and in defining identity in relationship to others. In the second phase, the phase of bicultural identity, the conscious recognition of aspects of identity is reflected as rooted in both the original and the adopted cultures. In the third, transcultural phase, finally, the migrant has developed his or her ‘transcultural self ’ – biculturalism (including bilingualism) is no longer experienced as a conflict. In addition, Yoshizawa Meaders distinguished three groups of immigrants. Migrants of the first group adjust quickly but superficially to the new environment and culture. Their adjustment, therefore, can be fragile and may contain problematic undercurrents which can lead to disappointment, disillusionment, anger, and depression and in some cases even to a disintegration of the self. Immigrants of the second group are just the opposite: they are described as unwilling and unable to make necessary changes, to be open or exploratory. They tend to hold on to their original culture,
Identity, immigration and first language attrition 193
language and identity. If chronic, this behavior may result in a further retreat into rigidity with minimum adjustment and, finally, into isolation. Migrants of the third group try to take the middle course: while making themselves respectful and open to the new environment, they nevertheless try to retain their own cultural identity and, by doing so, make the most lasting and flexible adaptation. Cross-cultural psychologists have described ‘personality shifts’ in bilinguals in the culture framing hypothesis (Hong, Morris, Chiu, Benet-Martinez 2000) which holds that bilinguals are “bi-culturals who have acquired two internalized cultures with personalities consistent with the two cultural systems” (Smith, Bond & Kağitçibasi 2006: 142). Several researchers suppose that language shifts can generate correspondent personality shifts in either direction (e.g. Harzing & Maznevski 2002; Hong et al. 2000; Ramirez-Esparza, Gosling, Pennebaker & Benet-Martinez 2006). Ward, Bochner & Furnham (2001) distinguished three categories of reactions of members of one cultural group to members of another: ‘psychological adaptation’ which means affective reactions such as pleasure or anxiety, ‘sociocultural adaptation’, i.e. the process of culture learning, and ‘acculturation’, or the process of identification with or separation from the local community. Despite the huge number of publications on self and identity, the role of language in psychological research on identity formation seems to be limited. Although the importance of language for social life and the transmission of cultural knowledge is broadly acknowledged and language is seen as a “primary means to gain access to the contents of other minds” (Krauss & Chiu, quoted in Bavelas 2005), in social psychological research language usually plays a role only as a means by which participants’ responses are elicited and in which they respond. However, cross-cultural psychologists acknowledge that language does more: language can be used to define one’s identity in relation to others, whereas language proficiency determines power relations and influences the degree of mutual sympathy. Language, nevertheless, is just one element of a larger set of cross-cultural skills (Smith et al. 2006). But in the process of socialization which is also the process of individualization, language plays a role which should not be underestimated. It provides a means to encode and access memories (Bauer & Wewerka 1997), a narrative framework for organizing memory (Reese, Haden & Fivush 1993) and therefore a modality for restoring and revaluating a memory that otherwise might be forgotten (Neisser & Libby 2000). From this perspective, language can be seen as an inseparable part of our identity since we use it to construct, tell and retell our life-stories. With respect to bilinguals, studies on bilingual autobiographical memory, emotions and selfconcept have demonstrated that bilinguals tend to connect certain memories and emotions to one of their languages (language-dependent memory hypothesis) (e.g. Marian & Neisser 2000; Marian & Kaushanskaya 2004; Pavlenko 2002;
194 Petra Prescher
Panayiotou 2004; Schrauf & Rubin 1998, 2003). Different languages elicit different associations: what seems to be the same cue appears to be a different one in different languages. Marian & Kaushanskaya suggest, furthermore, that bilinguals use different self-concepts and/or self-presentations depending on the language spoken at any given time. Yet it remains unclear if and to what extent these diverse self-concepts built on different languages affect identity, and how this phenomenon is perceived. Do people undergo ‘cross-cultural accommodation’ and how do they make sense of their special position? With respect to late bilinguals and immigrants, a further set of questions can be asked: how does it feel to use one’s L2 as a dominant language for an extended period of time? Are long-term migrants aware of L1 attrition and does it affect their perception of their identity? How do people fit in their cross-cultural experiences with their personal life-stories, how do they make sense of their world? The present study is an attempt to find an answer to such questions.
2. Personal narratives of long-term late bilingual migrants on identity and L1 attrition Identity is widely seen as a cognitive construct, a life-story which is built on memories (McAdams 2003; Squire & Kandel 2000). These memories can be based on personal as well as transmitted experiences, that is, memories and knowledge of others (e.g. Sacks 2005). Language works as the glue that binds memories in the autobiographical storyline, by and through which people try to give meaning to experiences and important events. It is widely acknowledged that this narrative process is a more or less creative one – although narrated life stories are based on autobiographical facts, they often go beyond the facts as people, depending on the situation, selectively take aspects of their experience and imaginatively construct narratives that integrate their life and make sense to them and to others (King 2000; McAdams 2003; Neisser & Libby 2000; Nelson 2003; Polkinghorne 1988; Squire & Kandel 2000). Remembering is inherently dynamic, determined by the individuals’ attitudes or values at the time. It is anything but permanent and fixed (Edelman 1989; Sacks 1996). Nevertheless, narratives, e.g. the first person accounts elicited in the case of this study, complement abstract and generalized statistical data by animating them and making them concrete. Little stories make the difference; they are indispensable to understand the larger whole.
. Cross-cultural accommodation refers to a kind of personality shift depending on the given language (Bond & Yang 1982; Harzing & Maznevski 2002).
Identity, immigration and first language attrition 195
To learn more about the individual’s own feelings in these respects, the present investigation used a deductive approach which combines qualitative and quantitative methods of data collection. This heuristic research method allows flexibility and provides space for reinterpretation of the collected data. However the most important data and the main starting point for the present investigation are people’s own statements and interpretations. Although narratives contain imaginative elements, narrative knowledge remains an important source for understanding complex processes of alteration concerning human beings and their lives.
2.1 Interviews The participants in this project were ten male and ten female German long-term immigrants in the Netherlands who left Germany as adults (i.e. after the age of 18) and who had been living in a Dutch-speaking environment for a minimum of 10 years. The participants were selected by snowball sampling, a technique often employed in qualitative research (Coolican 2004). All participants grew up monolinguals and became Dutch-German bilingual during their residence in the Netherlands. If the selected participants had partners, these were Dutch. The narratives were elicited through semi-structured in-depth interviews in the L1. In order to elicit memories and at the same time encourage a natural and uninhibited oral production, both loose and structured interview-techniques were used, which stimulate targeted as well as open recall. To trigger targeted recall, the participants were asked to remember a particular past event, such as the day of immigration. For open recall, the cue-word method was applied, which means that the participants simply reported whatever memory a particular cue, such as Freundschaft (friendship) or Sprache (language) brought to mind. The advantage of this approach – which is often applied in psychological research and has been extensively described by Neisser & Libby (2000) – is that the participants may choose the events themselves, only slightly manipulated by the researcher. In general, a relatively natural conversation was attempted, sometimes adapting the interview questions to the context, in order to get richer, fuller, more genuine and more realistic information.
. The study is based on a corpus of spoken data of 20 autobiographical narratives of German immigrants in the Netherlands. The obtained corpus is complemented by a sociolinguistic questionnaire, Can-Do-Scales (Dutch and German), a Dutch C-test and a German C-test. The sociolinguistic questionnaire and the language tests (C-test and Can-Do test) are part of the Language Attrition Test Battery which was developed and put together by the team of the Language Attrition Research Project (see Schmid, Köpke, Keizer & Weilemar 2004).
196 Petra Prescher
Each interview took approximately two hours. The atmosphere during the interviews was open and comfortable. All participants were willing and very interested and became more and more relaxed as the interview proceeded. Only one person was reluctant to speak about personal experiences, feelings and emotions. In most cases, the interviewees seemed to be very free in telling their stories of migration and language attrition and obviously felt the need to talk about them. One informant said: “Endlich wird das mal untersucht.” (‘It is high time that this was investigated.’ BE, m, 42). So far, 16 of the 20 interviews have been transcribed.
3. Some first results 3.1 Perception of L1 attrition Reviewing the accounts of German long-term immigrants, and following Yoshizawa Meaders’ model, two main groups can be distinguished: one group of migrants who wanted to assimilate as quickly as possible, rejecting their linguistic and cultural origin; and a second group who tried to maintain their original culture and identity but were more or less open to the Dutch environment. None of the informants belonged to the third category (no attempt to adapt to the new circumstances). This confirms general impressions on the assimilation of German migrants in the Netherlands. The linguistic and cultural differences between Germany and the Netherlands are generally not perceived as serious enough to lead to isolation or separation. The fragment below illustrates the struggle of a woman from the first group who adjusted very quickly, rejecting her L1 and even experiencing a kind of “personality loss”: (1) Ich hab mich so schnell angepasst, ich wollte sofort dazu gehören hier und wollte sofort alles richtig machen. Ich glaube, dass da schon ‘n gewisser Persönlichkeitsverlust da war. Ich hab überhaupt kein Deutsch mehr gesprochen, nur noch Niederländisch. Weil es – ja, irgendwie konnte ich mich in dieser Sprache gleich sehr gut ausdrücken und es hatte was Befreiendes für mich. Ich hab, ich glaube in den ersten Jahren hab ich mich zu schnell auf die andere Seite geschlagen. Das hat sich später dann gerächt und ich bin dann auch, ich war auch bei der Studentenpsychologin und die hat zu mir gesagt, du musst deine deutsche Hälfte mal abholen, in Deutschland. Also ich hab in zwei Identitäten gedacht. Das hab ich dann gemerkt, wenn ich *zu nach Hause gefahren bin zu meinen Eltern zum Beispiel, dass ich mich in Konflikten – konnte ich mich ganz schlecht äußern. Und hier dann war’s kein Problem. (LV, f, 46, 27 years in NL)
Identity, immigration and first language attrition 197
‘I adapted so quickly, I immediately wanted to be one of them and wanted to do everything right. I think I suffered from a kind of loss of personality. I absolutely didn’t speak German anymore, only Dutch. Because – yes, in some way I could express myself very well in this language; in some way I felt liberated. I think in the first years I switched to the other side too fast. Later on I came to regret it. I went to the psychologist and she told me, pick up your German half in Germany. Thus, I thought in two identities. I realized that when I went home to my parents, for instance, I could hardly express myself in conflict situations. But here it wasn’t a problem.’
In the interviews, participants often reflect on their linguistic and cultural inbetween position. Without any exception, all informants report a decline in their L1 skills. The subjects mostly perceive attrition on the phonological, lexical and morphological/syntactical level: (2) Also wenn ich in Deutschland gerade ankomm, dann sagen meine Freunde und meine meine Eltern und so dann, sagen schon, dass ich einen holländischen Akzent habe in meinem Deutsch. Manche Sachen kann ich viel einfacher und schneller auf holländisch sagen als dass ich sie auf deutsch – dann muss ich wieder puzzeln, wie ging’s nun wieder? Und die Grammatik, die dann etwas unterschiedlich ist, die macht’s dann halt, dass es irgendwie nicht ganz passt. Der Satz geht irgendwie in die falsche Richtung. (GJK, m, 33, 11 years in NL) ‘Thus, when I’ve just arrived in Germany, my friends and my my parents tell me that I have a Dutch accent in my German. Some things I can say easier and more quickly in Dutch than in German – then I have to puzzle, how to do this again? And the differences in grammar, sometimes they cause slight problems. The sentence goes in the wrong direction.’ (3) Also ich weiß, dass ich manchmal ‘n Wort suchen muss. Und dass ich manchmal auch drüber nachdenken muss, so grammatikalische Geschichten. War das jetzt so oder war das so? Also, im Schreiben hab ich mehr *Problemen. (JvD, f, 45, 21 years in NL) ‘So I know that I have to search for a word, sometimes. Sometimes I have to think about grammar. This way or that way? Writing is more difficult.’ (4) Man hört’s schon, mein mein Deutsch, dass ich auf jeden Fall aus’m Ausland komme. *Maar ich denk mir mein Deutsch ist in dem Sinne abscheulicher als als *dass mein Niederländisch. In Kassel, wo ich ja herkomme, bin ich als als Ausländer ausgeschimpft. (LE, m, 33, 12 years in NL) ‘You can hear it, my my German, that, anyhow, I come from abroad. *Maar [‘but’] I think my German is in that sense much more horrible than *that my Dutch. In Kassel, where I come from, they scolded me for being a foreigner.’
198 Petra Prescher
Even though all participants report a decline in their L1 skills, they do not perceive it as a definitive loss, albeit they suspect partial loss: (5) A: Ich hab irgendwie die Befürchtung, dass es [das Deutsch] eh nicht mehr so selbstverständlich vorhanden ist wie früher. Q: Bedauerst du das? A: Ja schon. Q: Warum? A: Weil es einmal so war und und ja, weil es dann eh dann praktisch verloren gegangen is, weil da etwas verloren gegangen ist. (CW, f, 51, 25 years in NL) A: ‘Somehow, I am afraid that it [German] isn’t at my disposal as naturally as before.’ Q: ‘Do you regret that?’ A: ‘Yes.’ Q: ‘Why?’ A: ‘Because once I had it and and, yes, because it is gone, because something was lost there.’
However, while the L2 has become the most prominent one in everyday life, the mother tongue seems to remain the most familiar language: (6) Wenn ich müde bin, dann geht’s [das Niederländisch] schlechter oder ich vergess Worte. Wenn ich richtig müde bin, dann merk ich, dass meine Muttersprache doch Deutsch ist. Da kommen auch deutsche Sätze dann plötzlich raus! (JvD, 45, 21 years in NL) ‘If I am tired, it [Dutch] becomes worse or I forget words. When I am really tired, I do notice that German is my native language, after all. Then sentences suddenly come out in German!’
It is remarkable that a decline in L1 skills is reported in every single interview. The majority of participants want to preserve the L1 skills and therefore make a conscious effort to retain apparently forgotten words, phrases and language rules. After observing indications of attrition in their L1 skills, people try to speak more German (to their children), to visit Germany more frequently, to read more German books and newspapers, to watch German TV, etc.
3.2 Reflections on identity One of the questions that arise in relation to L1 attrition is if and to what extent living in another country and speaking another language influences one’s selfhood. Some participants describe their linguistic and cultural position as “in-between”, “living in a no-man’s-land”, feeling forlorn:
Identity, immigration and first language attrition 199
(7) [Aber das Leben in den Niederlanden] das hat ‘ne ganz bestimmte Einsamkeit mit sich gebracht. Ab und zu. Weil man halt, *was ich selber hatte, du kommst halt einfach in’n Niemandsland *zurecht. Und das Niemandsland ist dann halt, dass du selber nicht mehr weißt, wo du eigentlich herkommst, was du tust und äh, ja, welche Sprache du eigentlich sprichst. Welche Sprache dich – na ja, deine Gefühle ausdrücken, auszudrücken, und das wird dann halt ab und zu schwierig. (LE, m, 33, 12 years in NL) ‘[But living in the Netherlands] has brought with it a certain loneliness. At times. Because, what I myself experienced, you just arrive in no-man’s land. You are in a no-man’s land because you don’t know where you are anymore what you are doing and, yes, which language you are actually speaking. Which language you – well, to express your feelings can be difficult, sometimes.’ (8) Wenn ich in Deutschland bin, sagen die, denken die eher, dass ich Holländer bin *dan dass ich Deutscher bin (GJK, m, 33, 11 years in NL) ‘When I am in Germany, they say, they rather think that I am Dutch *dan [‘than’] that I am German.’
There were differing answers to the question on whether speakers now felt foreigners in Germany. However, a majority does no longer feel at home in their country of origin: (9) Q: Fühlst du dich in Deutschland als Fremde? A: Ich fühl mich da auch nicht mehr zu Hause, nee, da müsste ich ja lügen. Dazu lebe ich zu lange hier. Ich lebe inzwischen länger hier in den Niederlanden als in Deutschland. Sehr viel Dinge, die die sind auch fremd geworden, Probleme, die die Menschen da haben, da hat man natürlich nichts mit zu tun. Q: Und bedauerst du das? A: Ja. (UH, f, 51, 29 years in NL) Q: ‘Do you feel like a foreigner in Germany?’ A: ‘I don’t feel at home there anymore, no, to be honest. I’ve lived here too long. I’ve lived in the Netherlands longer than in Germany. Many things have become unfamiliar to me. I cannot relate to problems that people have, of course.’ Q: ‘Do you regret that?’ A: ‘Yes.’
On the other hand, the personal narratives show the obvious desire of the immigrants to integrate languages, cultures, lives, i.e. the life before and the life after emigration into one self. One speaker attempted to integrate her supposed two identities by means of psychotherapy:
200 Petra Prescher
(10) Dann hab ich ‘ne Gruppentherapie gemacht und also auf Niederländisch, hab ich meine ganze Jugend ungefähr durchgespielt. Also das, *he, also ich hab selber Rollen gespielt, ich habe auch Personen Figuren aus meiner Kindheit spielen lassen und damit gesprochen, Probleme besprochen, die ich hatte, die dann natürlich völlig deutsch waren und die hab ich dann auf Niederländisch, hab ich die dann noch mal wiederholt, sagen wir mal. Durch die Therapie ist das wieder zusammengekommen. Ich glaube, bis zu dem Moment war das schon ‘n bisschen gespalten. Aber ich fühle mich nicht als Ganzes niederländisch. Das ist nicht so. Ich fühle mich mehr als Ganzes an sich und da drin ist deutsch und niederländisch.” (SH, f, 49, 25 years in NL) ‘And then, during group therapy, I re-enacted my whole youth in Dutch. And, *he, I played some of the roles myself, and some of my childhood contacts were played by other persons and I talked to them, and I discussed the problems I had, which were very German, of course, and then I repeated them in Dutch, I repeated them, as it were. And through the therapy it came together, again. Up to that moment, I think, I was somewhat split. But I don’t feel wholly Dutch. No. I feel more as an integrated whole, and within this whole there is German and Dutch.’
One participant, a musician, first regards language as essential for consciousness and mind. But although language seems to be strongly connected with the mind (‘Geist’), he wonders whether different languages (and cultures) bring forth different minds. If so, according to him, there must be something which unifies both minds, and that something is not language but art and music. When writing a composition, he uses elements of both systems and so creates a synthesis which brings together the possible two minds: (11) Aber wenn man zwei Sprachen, Kulturen lebt, dann auch zwei Geister? Weiß ik nicht. Wer hält diese beiden Selbste zusammen? Da muss ja was sein, was diese zusammenhält. Sonst kannste ja nicht mehr differenzieren. Wenn ich jetzt was komponier, dann hat sich dit hier so nicht geändert. Also jetzt in dem Kunstwerk. Man nimmt zwar Aspekte jetzt von der einen Seite noch mit und von der anderen, ich meine jetzt Einflüsse, das schon, aber es wird dann wieder das Eine. Irgendwie Kunst wäre ja dann das Eine und das ist vielleicht das, was beides dann zusammenhält, zusammenbringt. (KG, m, 50, 20 years in NL) ‘But if you live two languages, cultures, does that mean two minds, too? Don’t know. Who holds those two selves together? There has to be something, hasn’t there? Otherwise you are not able to differentiate. If I write a composition, nowadays, nothing has changed. I mean, within the artifact. Sure, you take aspects from both sides, I mean, influences, but in the end, it becomes the One. Somehow or other, art may be that which holds them together, brings them together.’
Identity, immigration and first language attrition 201
4. Considerations The scope of the present study and the examples presented above do not lend themselves to massive generalizations. Qualitative research emphasizes meanings, experiences, descriptions and therefore, in the present case, first person accounts. In the narratives on their immigrant experiences, participants describe their own intraindividual variability, i.e. stability and instability across time and different conditions, during the process of acculturation. This process is characterized by diversity, unpredictability, and non-linearity. Identity and language development across the lifespan appear to be comparable dynamic systems with similar characteristics: open, unstable, fluid, unpredictable, fluctuating, multidimensional, and interconnected. The accounts of the German migrants in the Netherlands seem to illustrate this assumption. The narratives also seem to support Yoshizawa Meaders’ hypothesis of transcultural identity building; after initial quick assimilation during the first years after emigration (“I wanted to adapt so quickly, it was almost a kind of loss of my personality”, LV), immigrants in the ‘phase of bicultural identity’ tend to struggle with their languages (“I didn’t know anymore what my real language was”, CB) and their self (“I don’t know which identity I have anymore”, LE). However, the longer the duration of immigration, the more subjects try to reflect on or to return to their original identity and language. Most of them try to preserve their L1 skills as an essential part of their identity. Interestingly enough, there is apparently no difference in this respect between the two groups of German immigrants distinguished above: the one which tried to adjust quickly but superficially and the one which preferred gradual but flexible adaptation. Across time, long-term migrants of both groups become more confident about their origin, culture and mother tongue and are more critical about the culture, language and mentality of the guest country. The longer the duration of immigration, the stronger the attempt to return to the original identity and language. Moreover, the more participants try to assimilate in Dutch society, the more problems they perceive concerning their identity: wanting to integrate completely, at the end of the day they have to accept they will remain different, resulting in a disinclination to adapt any further. In the end, immigrants are used to being different; they accept their “nationlessness”, their imperfect languages, they have created their “home” in their own transcultural selves.
References Bauer, E. & Wewerka, S. (1997). “Saying is revealing: Verbal expression of event memory in the transition from infancy to early childhood”. In P. van den Broek, P. Bauer & T. Bourg
202 Petra Prescher
(Eds.), Developmental Spans in Event Comprehension and Representation: Bridging fictional and actual events (pp. 139–168). Hillsdale, NJ: Lawrence Erlbaum Associates. Bavelas, J. B. (2005). “The two solitudes: Reconciling social psychology and language and social interaction”. In K. L. Fitch & R. E. Sanders (Eds.), Handbook of Language and Social Interaction (pp. 179–199). Mahwah, NJ: Lawrence Erlbaum Associates. Bond, M. H. & Yang, K. S. (1982). “Ethnic affirmation versus cross-cultural accommodation: The variable impact of questionnaire language on Chinese Bilinguals in Hong Kong”. Journal of Cross-Cultural Psychology, 13, 169–185. Boym, S. (2001). The Future of Nostalgia. New York, NY: Basic Books. Brodsky, J. (1988). “The condition we call exile”. In The New York Review of Books, 34 (21&22), http://www.nybooks.com/articles/4548, Retrieved 05-02-20, 12:09. Bruner, J. (2003), “Self-Making Narratives”. In R. Fivush & C. A. Haden (Eds.), Autobiographical Memory and the Constructions of a Narrative Self (pp. 209–225). Mahwah, NJ: Lawrence Erlbaum Associates. Cole, M. & Wertsch, J. V. (2002). “Beyond the individual-social antimony in discussions of Piaget and Vygotsky”. Math Forum Internet Library. Conrad, J. (1946). A Personal Record: The works of Joseph Conrad. London: Dent Uniform Edition. Coolican, H. (2004). Research Methods and Statistics in Psychology. London: Hodder Arnold. Ecke, P. (2004). “Language attrition and theories of forgetting: A cross-disciplinary review”. International Journal of Bilingualism, 8 (3), 321–354. Edelman, G. M. (1989). The Remembered Present. New York, NY: Basic Books. Elovitz, P. H. & Kahn, Ch. (Ed.) (1997). Immigrant Experiences: Personal narratives and psychological analysis. Madison, WI: Fairleigh Dickinson University Press. Erikson, E. (1963). Childhood and Society. New York, NY: Norton. Erikson, E. (1968a). Identity: Youth and crises. New York, NY: Norton. Erikson, E. (1968b). Life History and the Historical Moment. New York, NY: Norton. Harzing, A.W. & Maznevski, M. (2002). “The interaction between language and culture: A test of the cultural accommodation hypothesis in seven countries”. Language and Intercultural Communication, 2, 120–139. Haviland-Jones, J. M. & Kalbaugh, P. (2000). “Emotion and identity”. In M. Lewis & J. M. Haviland-Jones (Eds.), Handbook of Emotions (pp. 293–306). New York, NY: The Guilford Press. Hoffman, E. (1989). Lost in Translation. London: Vintage. Hogg, M. A. & Vaughan, G. M. (2005). Social Psychology. Harlow: Pearson. Holstein, J. A. & Gubrium, J. B. (2000). The Self We Live By: Narrative identity in a postmodern world. New York, NY: OUP. Hong, Y. Y., Morris, M. W., Chiu, C. U. & Benet-Martinez, V. (2000). “Multicultural minds: A dynamic constructivist approach to culture and cognition”. American Psychologist, 55, 709–720. King, N. (2000). Memory, Identity, Narrative. Edinburgh: Edinburgh University Press. Kouritzin, S. (1999). Face(t)s of First Language Loss. Mahwah, NJ: Lawrence Erlbaum Associates. Kristeva, J. (2000). Crisis of the European Subject. New York, NY: Other Press. Kristof, A. (1995). Hier. Paris: Edition du Seuil. Kundera, M. (1984). The Unbearable Lightness of Being. London: Faber and Faber.
Identity, immigration and first language attrition 203
Lewis, M. (1997). Altering Fate: Why the past does not predict the future. New York, NY: The Guilford Press. Maalouf, A. (2000). In the Name of Identity: Violence and the need to belong. New York, NY: Arcade Publishing. Marian, V. & Kaushanskaya, M. (2004). “Self-construal and emotion in bicultural bilinguals”. Journal of Memory and Language, 51, 190–201. Marian, V. & Neisser, U. (2000). “Language-dependent recall of autobiographical memories”. Journal of Experimental Psychology, 129 (3), 361–368. McAdams, D. P. (2003). “Identity and the Life Story”. In R. Fivush & C. A. Haden (Eds.), Autobiographical Memory, Narrative and Self (pp. 187–207). Mahwah, NJ: Lawrence Erlbaum Associates. Mills, J. (2004). “Mothers and Mother Tongue: Self-construction by mothers of Pakistani heritage”. In A. Pavlenko & A. Blackledge (Eds.), Negotiation of Identities in Multilingual Contexts (pp. 161–192). Clevedon: Multilingual Matters. Misztal, B. A. (2003). Theories of Social Remembering. Philadelphia, PA: Open University Press. Nabokov, V. V. (1967). Speak, Memory: An autobiography revisited. London: Weidenfeld and Nicolson. Neisser, U. & Libby, L. K. (2000). “Remembering life experiences”. In E. Tulving & F. I. M. Craig (Eds.), The Oxford Handbook of Memories (pp. 315–332). New York, NY: OUP. Nelson, K. (2003). “Narrative and self, myth and memory”. In R. Fivush & C. A. Haden (Eds.), Autobiographical Memory, Narrative and Self (pp. 3–28). Mahwah, NJ: Lawrence Erlbaum Associates. Oakes, P. J., Haslam, S. A. & Turner, J. C. (1994). Stereotyping and Social Reality. Oxford: Blackwell. Panayiotou, A. (2004). “Bilingual emotions: The untranslatable self ”. Estudios de Sociolingüística, 5 (1), 1–19. Pavlenko, A. (2002). “Bilingualism and emotions”. Multilingua, 21, 45–78. Pavlenko, A. & Blackledge, A. (Eds.) (2004). Negotiation of Identities in Multilingual Contexts. Clevedon: Multilingual Matters. Polkinghorne, D. E. (1988). Narrative Knowing and the Human Sciences. Albany, NY: State University of New York Press. Ramirez-Esparza, N., Gosling, S., Pennebaker, J. & Benet-Martínez, V. (2006). “Do bilinguals have two personalities? A special case of cultural frame-switching”. Journal of Research in Personality, 40, 99–170. Reese, E., Haden C. & Fivush, R. (1993). “Mother-child conversations about the past: Relationships of style and memory over time”. Cognitive Development, 8, 403–430. Sacks, O. (1996). “Gowers’s Memory”. Neurology, 46, 1467–1469. Sacks, O. (2005). “On Memory”. In The Threepenny Review, Winter 2005, http://www. threepennyreview.com/samples/sacks_w05.html, Retrieved 07-02-19, 18:59. Schmid, M. S. (2002). First Language Attrition, Use and Maintenance: The case of German Jews in Anglophone countries. Amsterdam: John Benjamins. Schmid, M. S., Köpke, B., Keijzer, M. & Weilemar, L. (Eds.) (2004). First Language Attrition: Interdisciplinary perspectives on methodological issues. Amsterdam: John Benjamins. Schrauf, R. W. & Rubin, D. C. (1998). “Bilingual autobiographical memory in older adult immigrants: A test of cognitive explanations of the reminiscence bump and the linguistic encoding of memories”. Journal of Memory and Language, 39, 1–21.
204 Petra Prescher
Schrauf, R. W. & Rubin, D. C. (2003). “On the bilingual’s two sets of memories”. In R. Fivush & C. A. Haden (Eds.), Autobiographical Memory, Narrative and Self (pp. 121–145). Mahwah, NJ: Lawrence Erlbaum Associates. Smith, P. B., Bond, M. H. & Kağıtçıbası, Ç. (2006). Understandig Social Psychology Across Cultures: Living and working in a changing world, London: Sage Publications. Squire, L. R. & Kandel, E. R. (2000). Memory: From mind to molecules. New York, NY: Scientific American Library. Tajfel, H. & Turner, J. C. (1986). “The social identity theory of inter-group behavior”. In S. Worchel & L. W. Austin (Eds.), Psychology of Intergroup Relations. Chigago, IL: NelsonHall. Tajfel, H. & Turner, J. C. (1979). “An integrative theory of intergroup conflict”. In W. G. Austin & S. Worchel (Eds.), The Social Psychology of Intergroup Relations (pp. 33–47). Monterey, CA: Brooks/Cole. van Geert, P. (1991). “A dynamic system model of cognitive and language growth”. Psychological Review, 98, 3–5. van Geert, P. (2000). “The dynamics of developmental mechanisms from Piaget and Vygotsky to Dynamic Systems Models”. Current Directions in Psychological Science, 9, 64–68. Vygotsky, L. (1978). Mind in Society. Cambridge, MA: Harvard University Press. Walter, R. (2003). Narrative Identities: (Inter)cultural in-betweeness in the Americas. New York, NY: Peter Lang. Ward, C., Bochner, S. & Furnham, A. (2001). The Psychology of Culture Shock. East Sussex: Routledge. Yoshizawa Meaders, N. (1997). “The transcultural self ”. In P. H. Elovitz & C. Kahn (Eds.), Immigrant Experiences: Personal narrative and psychological analysis (pp. 47–59). Madison, WI: Fairleigh Dickinson University Press.
Language attrition and ideology Two groups of immigrants in Israel Miriam Ben-Rafael and Monika S. Schmid
Alon College, Israel / Rijksuniversiteit Groningen, The Netherlands
There is a great body of research on the role of attitudes, motivation and emotions for the development of bilingualism, providing compelling evidence that these factors are among the most important for predicting success in the ultimate attainment of second and foreign language learners. Attempts have also been made to establish a similar link between these emotive factors and language attrition. Overall, findings from such investigations have been inconclusive and unsatisfactory. It has been hypothesized that this may in part be due to the fact that there was not enough diversity between the subjects investigated, and that clearer findings might be achieved through investigations which contrast different immigrant populations. The present study presents a comparison of two groups of immigrants in Israel. Keywords: attitude and motivation; identity; integration; Hebrew; French; Russian
1. Theories of language attitudes Investigations on the role of language external influences on the development of bilingualism typically focus on different sets of factors. On the one hand there are the ‘classical’ sociolinguistic variables: personal background factors which are relatively easy to quantify in a sociological or socioeconomic framework, such as age, gender, socioeconomic background, education and so on. Far more problematic for any linguistic investigation, but particularly in the context of bilingualism, are more subjective factors, such as attitude, motivation and emotion. While it is generally accepted that these are important predictor variables for individual and societal bilingualism, it has proven hard to establish practical approaches able to cope with these elusive factors in a quantitative manner.
206 Miriam Ben-Rafael and Monika S. Schmid
Attempts to account for the role of attitudinal factors in bilingual situations are typically situated in one of the following approaches: socioeducational, sociopsychological or neurobiological.
1.1 The socioeducational approach One of the earliest and most widely used approaches to measure the influence of attitude and motivation is the socioeducational model, which links attainment in (instructed) L2 acquisition to attitudes towards the language and its speakers, to the learning situation, and towards integration. This model is most closely associated with the work of Robert Gardner and his colleagues, and was first proposed in 1959 by Gardner & Lambert. Considerable work has since gone into developing generally accepted, quantitative applications of this model, the most widely used of which is the so-called Attitude/Motivation Test Battery (Gardner 1985). The model proposed here distinguishes three groups of factors, motivation, integrativeness and attitudes toward the learning situation (Masgoret & Gardner 2003: 124). The test battery constructed on the basis of this model comprises 63 Likert-Scale items which fall into these categories, plus optionally a number of additional questions to measure factors such as motivational intensity (Gardner 1985). This Attitude/Motivation Test Battery (AMTB) has been applied in a large number of investigations, chiefly on instructed L2 learning, and consistently achieved high correlations with attainment by individual learners. These ideas find another application in the ‘affective filter’ in Krashen’s monitor model (Krashen 1981). Krashen argues that, where the learner has a positive attitude, high motivation, low anxiety etc., linguistic input can be assimilated and converted into knowledge. If, on the other hand, these factors are lacking, an ‘affective filter’ will prevent the input from becoming intake. Problems that have been pointed out with respect to the AMTB (and the affective filter) mainly pertain to its explanatory power: while the predictor factor groups measured here appear to be valid and important ones for success in language learning, they have been shown to be extremely difficult to weigh against each other, and it remains to be established “just which of these variables, in what combinations, and to what degree, are supposed to affect learning, and why?” (Long 1990: 275). Furthermore, the impact of individual factors may be different according to the context: while an integrative vs. instrumental motivation generally influences the outcome of the learning situation positively, there may also be learning environments where an instrumental attitude is better (Pavlenko 2005: 32). It does appear that, while the overall power of the AMTB has been well-established, the role of individual factor groups needs to be looked at more
Language attrition and ideology 207
closely. It may also be beneficial to broaden further investigations to include other environmental factors of the learning situation, and not to confine the perspective to correlations between the AMTB on the one hand and measures of proficiency and achievement on the other.
1.2 The sociopsychological approach For sociopsychological investigations on the role of attitudes and emotions for bilingual development, factors such as individual proficiency, success in language learning, and ultimate attainment play a less central role. The focus is on the individual speaker’s language systems and the development and portrayal of identity and identities. This line of investigation is based on the recognition of the fact that language is one of the most important factors of identity and identification: people speak like the people they want to be like, and use the way they talk to signal aspects of their own personalities (Edwards 1999). A great deal of research has been conducted on migrant multilingualism on the one hand and identity and emotions on the other, based in particular on theories such as Le Page & Tabouret-Keller’s Acts of Identity (1985). Such work typically investigates uninstructed (immersed) language learners and the use they make of their various languages to express (or hide) emotions and to attain the personality they deem most desirable (for a comprehensive review see Pavlenko 2005). A further area of investigation in this context is the influence of sociodemographic and other group issues on language learning and language retention. Within the theory of Ethnolinguistic Vitality, it has been postulated that factors such as group size and prestige, institutional support, endogamy/exogamy and so on can be among the most critical predictor variables within the context of minority language loss and maintenance (e.g. Giles, Bourhis & Taylor 1977). This latter area of investigation indicates a further problem with studies of the impact of attitude on language learning and maintenance among migrants: while attitudes may vary individually across one community of immigrants, they will also to some degree be consistent across the group. More importantly, while groups consider themselves as stratified, they are often viewed as homogenous from the outside (Breakwell 1983: 191). Thus, while an individual might have perceived herself as belonging to a social class with a certain social prestige while still in the country of her origin, after emigration she may find attitudes towards her influenced by stereotyped notions associated with ‘immigrants’ or ‘immigrants of a certain ethnolinguistic background’. It is therefore important that studies of the impact of language attitude for multilingual development attempt to make a systematic comparison of groups
208 Miriam Ben-Rafael and Monika S. Schmid
of different linguistic and ethnic backgrounds, or compare the same groups in different settings. To ensure comparability, both L1 and L2 of the groups under investigation should be identical or as similar as possible (e.g. Turks in Australia and Turks in the United States) in order to assess the impact of varying attitudes, prestige, and institutional support on the development of interlanguage.
1.3 The neurobiological approach More recently, it has been pointed out that, instead of casting phenomena witnessed in second language acquisition (SLA) in abstract terms of rules, principles and strategies, the field might benefit from a view that takes into account the neurocognitive substrate (Schuman 1994). In other words, it is argued here that SLA investigations might profitably adopt an approach that assumes that “brain causes behavior” (Schuman 1994: 231). Schuman argues that the ‘affective filter’ proposed in Krashen’s model (the idea that certain attitudes on the part of the language learner facilitate the ‘taking in’ of information, see above 1.1) could be recast from a neural perspective. Two brain structures are relevant in this context: the amygdala and the orbitofrontal cortex. Both appear to play a role in stimulus appraisal: the amygdala has a crucial role in assessing the emotional significance of stimuli, while the orbitofrontal cortex helps monitor their reward potential (Schuman 1994: 233). Both these brain structures are linked to the emotional memory system, which, according to Schuman, could constitute an alternative explanation to the affective filter. Affect and cognition, in this view, are “distinguishable but inseparable parts of a mutually interacting system” (Schuman 1994: 240).
2. Attitudes and language attrition As motivation has a demonstrated impact on the development of one of a bilinguals’ language systems, language attrition research has long been fascinated with the idea that the converse might also be true: if wanting to learn something (or wanting to be part of a community of speakers) can help with acquisition, can wanting to forget something (or wanting to no longer be part of a speech community) help forget it? The psychological process of suppressing unwanted or traumatic memories is well-known, but can it also apply with respect to a linguistic system, that is, knowledge that is represented in quite a different way? One of the first larger studies that attempted to investigate this phenomenon was Waas’ investigation on the L1 attrition of German in Australia (Waas 1996),
Language attrition and ideology 209
which applied a two-way classification of identity along two criteria: residential status (‘permanent resident’ vs. ‘naturalized citizen’) and ‘ethnic affiliation’, measured by systematic exposure to German in daily life. This classification did not yield significant results when correlated with performance on a linguistic test. A more rigorous framework of ethnic identity is applied in Yağmur (1997) and Hulsen (2000) who apply the theory of Ethnolinguistic Vitality Theory, as developed among others by Giles, Bourhis & Taylor (1977) in their respective investigations of Turkish in Australia and Dutch in New Zealand. Here, too, no significant impact of the predictor variables of Subjective Ethnolinguistic Vitality could be established. The findings from these three studies seem to underscore that socio- and ethnolinguistic frameworks that rely on subjects’ self reports about their use of and attitude towards their L1 seem to be inadequate tools for the prediction of the linguistic aspects of language attrition: subjective evaluation at the current point in time cannot establish a link between attrition and concepts such as identity. This may conceivably be down to the fact that people’s identity, affiliation and self-concept does not remain stable across the lifespan (Breakwell 1986: 18). Obviously, if a connection between attitudes or identities and attrition is to be found, it would have to be looked for in the moment of emigration, i.e. at the possible onset of attrition. This point was recognized in Schmid’s investigation of language attrition among German Jews who had emigrated from Germany during the Nazi regime (Schmid 2002). Based on a historical classification of progressive radicalization of persecution during those years, this study investigates the impact of the period during which emigration had taken place, hypothesizing that longer exposure to increasingly traumatic events would lead to a greater wish among the refugees for disassociation from Germany and the German language. This classification proved to be highly relevant in explaining the outcomes of the linguistic investigation. The latter study illustrates a point that was made above: attitudes, in particular language attitudes, are usually consistent to some degree across a particular community of migrants. In order to shed more light on the impact of motivation and emotion for individual language attrition, it is therefore necessary to compare groups of migrants for whom the context of migration or the migrational setting can be clearly differentiated.
3. Multilingualism in Israel The linguistic situation of Israel is unique in many ways. Firstly, as is often pointed out, the process of reactivating a dead language (ancient Hebrew) by its adoption
210 Miriam Ben-Rafael and Monika S. Schmid
and establishment as the national and community language is the only known successful example of this kind of language engineering and policy. This revival of the language of the Jewish Bible as a living national language is a particularly striking illustration of the close link between language and identity, since it was, from its very beginning, motivated by ideological drives and the desire to concretize a new Jewish and national identity (Myhill 2004). One may see in this Hebrew revival what Castells (1999: 21) calls a national identity project set in motion by individuals through personal and collective events, and it can been argued that modern Hebrew forms “an invisible national border and institutes links between private and public life as well as between past and present” (ibid: 71). The ideology of Hebrew revival was reinforced by assumptions which are common in many Western societies today: immigrants will only truly be able to integrate linguistically if they acquire the legitimate language of the society and weaken their commitment to their L1 (Kheimets & Epstein 2001; Remennick 2003: 431). In the case of Israel, however, this myth was further reinforced by the desire to bring up a first generation of native speakers of Hebrew (E. Ben-Rafael 1994: 63–65). In view of these strong ideological underpinnings of the Hebrew revitalization endeavour, it is striking that Israel has come to formulate a more flexible approach over time by its adoption of a policy of greater tolerance of societal multilingualism. This is a relatively recent development: initially new migrants were strongly encouraged to learn and adopt the national language as a marker of their new Israeli identity, and to discontinue at least public use of their heritage languages (Spolsky 1999: 168). However, in September 1996 a new policy became effective, providing for mother tongue teaching and reinforcing second/foreign language education (Spolsky & Shohamy 1996), and giving, in principle, “full formal recognition to the values of Arabic and of immigrant and community and heritage languages, and the importance of international languages to build national capacity” (Spolsky 1999: 170). This policy marked a developmental process characterized by the strengthening of cultural pluralism, the growing influence of globalization and English in daily life and, most importantly, the recent strong influx of immigrants from the former Soviet Union (Deutch 2005: 282) who today make up 20% of Israel’s Jewish population (E. Ben-Rafael, Lyubansky, Glöckner, Harris, Israel, Jasper & Shoeps 2006: 25–30; Remennick 2003: 432).
4. The study The present study addresses the issue of L1 attrition among two communities of migrants in Israel. Based on the reflections and considerations outlined above, the question is asked to what extent the different prevailing language policies and
Language attrition and ideology 211
ideologies at different points in time find their reflection in the ideologies, attitudes, and processes of language attrition of two different groups of immigrants in Israel. The first group comprises 15 ideologically-motivated French-speaking immigrants who see their immigration to Israel as a major life accomplishment. These participants expressed this idealistic attitude by joining a collective village (kibbutz) in the late 1950s and early 1960s, at a time when the official Israeli language policy was strongly monolinguistic and encouraged all newcomers to quickly acquire Hebrew and use it in all spheres of social and private life. The second group consists of 15 Russian immigrants who came to Israel in the late 1980s and early 1990s. The motivation of immigrants in this period, particularly those from Eastern Europe, was less ideological and more primarily motivated by practical and pragmatic considerations – improving their standard of living, escaping the threat of anti-Semitism, looking for new life opportunities, and providing a better future for their children (E. Ben-Rafael, Olshtain & Geijst 1996; E. Ben-Rafael & Peres 2005; E. Ben-Rafael et al. 2006). These immigrants settled individually in urban areas rather than in kibbutzim; and during a period when language policy was much more tolerant than it had been in the past. This change found its reflection in language attitudes: while Russian immigrants of earlier decades shared a wish to make the transition to Hebrew as quickly as possible, and even experienced their Russian accent as a “stigma” (Remennick 2003: 441), the large influx of new Russians in the 1990s effected a kind of language shift among Israel’s Russian population. The Russian language and culture became an important factor for these speakers (Remennick 2003: 443). The speakers in the sample under investigation here form no exception: while willing to acquire Hebrew, they were typically very keen to keep Russian as a vernacular and cultural language and to transmit it to their children (Rozenbaum-Tamari 2004; Laufer 2003).
4.1 Participants 4.1.1 The Francophone sample The Francophone sample (n = 15) comprises 10 men and 5 women, aged between 55 and 75 (mean: 67, 81), inhabitants of a kibbutz in Galilee which will be referred to as Klila for the purpose of this paper. At the time of the data collection in 2005, the participants had been living in Klila for about 45 years (from 34 to 54 years, mean: 45, 66). All participants had immigrated from the mid-1950s to the early 1960s, when they were between 18 and 20 years old and had completed high school education. Their countries of origin were France (n = 6), Belgium (n = 3), Morocco (n = 2), Algeria (n = 3) and Tunisia (n = 1). The decision to join the kib-
212 Miriam Ben-Rafael and Monika S. Schmid
butz was taken on the basis of ideological convictions, including an eagerness to adopt Hebrew for private and public purposes alike. This attitude was sustained by the strict provision of the kibbutz veteran population – in accordance with Israel’s central language policy – that any newcomer should acquire and use the new national language exclusively.
4.1.2 The Russian sample The Russian sample (n = 15) comprises 8 men and 7 women, aged between 30 and 50 (mean: 39, 46) who had settled individually in urban areas in Israel. At the time of the data collection in 2004/5 they had been living in Israel for 10 to 15 years (mean: 14, 06). Their immigration in the 1990s found a setting that had, ever since the late 1970s, recognized linguistic pluralism as a socially legitimate fact and had become much more open-minded toward cultural diversity than in the past. The immigrants themselves were mostly motivated by utilitarian drives and aspired to remain as close as possible to their Russian culture and language. The immigrants, who form a highly educated group, had mostly completed their professional training in the Former Soviet Union. 4.2 Method Data collection was conducted by means of a semi-structured interview, based mainly on open-ended questions about past events, motives for immigration, linguistic problems encountered, and current attitudes towards French/Russian and Hebrew. The interviews, which were tape recorded, were conducted by native speakers of the immigrants’ L1 who were members of the respective communities. In the case of the French migrants, the interviews took approximately 60 to 80 minutes, while the Russian interviews were somewhat shorter (25 to 35 minutes). The spoken data were subsequently transcribed orthographically and analysed in order to identify instances of codeswitching and characterize language attitudes and motivations of immigration.
. The interviews of the French sample were conducted by co-author Miriam Ben-Rafael and the interviews of the Russian sample by Moshe Koningstein, a professional sociologist who immigrated to Israel from Moscow in 1990. He was also in charge of the translation of the interviews in English. The authors of this article wish to thank Moshe Koningstein for his valuable contribution.
Language attrition and ideology 213
4.3 Attitudes 4.3.1 The Francophone sample The participants in the Francophone sample consistently ascribed their emigration to Israel to ideological and political motivations. Many of them mentioned the Shoa, as well as their individual socialist and communist aspirations. The decision to settle in a kibbutz seemed fitting in this context (see ex. (1)–(6) below). (1) le sionisme et le socialisme # j’étais imprégné de tout de cet idéal [...] le kibouts pour moi était une réponse # venir en Israel pour moi c’était venir au kibouts (Yair, Morroco) ‘Zionism and socialism # I was impregnated with these ideals [...] the kibbutz for me was the answer # to come to Israel for me meant going to the kibbutz.’ (2) je venais d’une famille assimilée # mais la shoa m’a beaucoup touché et quand on parlait de kibouts d’égalité et tout ça c’étaient des choses qui étaient près de mon coeur (Yaakov, Algeria) ‘I came from an assimilated family # but the Shoa has affected me a lot and when one spoke about the kibbutz, equality, and all that were mentioned, it touched my heart.’ (3) à l’époque, je pense j’ai été poussée par l’idéologie c’est-à-dire le sionisme socialisme # on est venu au kibouts # je pense que c’était aussi par rapport à la guerre et l’antisémitisme aussi # nous à la maison on parlait beaucoup de ça # c’est-à-dire maman a été très marquée # par la guerre l’antisémitisme et # je pense que c’est en fait les éléments essentiels (Drora, Paris) ‘At that time I think I was urged by ideology i.e. Zionism socialism # we came to the kibbutz # I think also because of the war and the anti-Semitism # at home we spoke a lot about it # my mother has been very much affected # by the war, the anti-Semitism and # I think these are the essential reasons.’
While the process of immigration and integration into society and work life went relatively smoothly for most participants ((4) through (6)), acquisition of Hebrew initially was problematic for many. Most participants acquired the new language by immersion, since it was the accepted norm in the kibbutz to use Hebrew exclusively. (4) j’ai pas de souvenir d’avoir souffert ou de m’être sentie mal à l’aise [...] j’ai travaillé au gan et au khadar okhèl [...] j’ai travaillé dans les baté jeladim macheu kazè # [...] on habitait au bet margoa # enfin on avait une chambre [...] on allait manger au khadar okhèl ensemble, on allait ensemble au moadon (Sara) . The symbol # indicates a pause.
214 Miriam Ben-Rafael and Monika S. Schmid
‘I don’t remember suffering or feeling bad [...] I worked in the kindergarten, in the dining room, in the children’s homes, # something like that, we lived in the rest house of the kibbutz # okay, we just had a room [...] we used to go together to the dining room, to the club.’ (5) lo zokhér che-haju li beajot, gaaguim non # ça s’est bien passé (Tsviko, Algeria) ‘I don’t remember having problems or feeling nostalgia no # everything was all right’ (6) le fait de travailler en tant que metapèlèt m’a beaucoup appris # les enfants me disaient al tagidi kakha tagidi kakha (Sara) ‘The fact I worked with children as an educator helped me a lot # the children used to tell me don’t say it like that say it like this.’
The general view was that kibbutz reality had changed dramatically over the years, giving more importance to individual life and economic privatisation. However, the acquisition of Hebrew, its exclusive use in daily life, together with the renouncement of the immigrants’ L1 has remained an ideological aspiration of the community. (7) bati kdé lehitnaték mi ha tsarfatit [...] kdé lijot chutafa letkhijat hasafa haivrit (Margalit) ‘I came in order to get rid of my French [...] in order to participate in the revival of Hebrew.’ (8) on voulait parler hébreu on voulait devenir de vrais israéliens (Drora) ‘We wanted to speak Hebrew, we wanted to become real Israelis.’
Hebrew has become the prevalent language in daily life and family interaction, even for couples who are both Francophone. The aspiration to bring up a generation of native speakers of Hebrew also lead to parents not passing on their L1 to their children. Even terms of affection were and generally still are Hebrew ones – khamudi (‘sweet’), pachoch (‘little bird’), ahuvati (‘my love’), although a very limited number of French terms, such as poulette (‘little chick’) or chouchou (‘darling’) appear sometimes. All respondents feel that their L1 proficiency is diminished as a result of the predominance of Hebrew in daily life, and the geographical distance between them and the Francophone countries of origin, which gives them little opportunity to use French. Several speakers indicate that they are aware of mixing French and Hebrew, and during the interview try to discipline their language use. Many express a feeling of loss, of the inability to retrieve French lexical items, and express a sense of linguistic impoverishment and lack of confidence, which frequently makes
Language attrition and ideology 215
them feel ill at ease when speaking French with “real Francophones”. In such situations particularly, many speakers perceive their language attrition to be dramatic. (9) je m’exprime plus en hébreu qu’en français et à la maison et au travail # donc c’est sûr quand on parle pas une langue et tout # finalement tu perds # j’ai pas oublié mais y a des mots qui me manquent parce que je les utilise pas souvent (Yaakov) ‘I speak more Hebrew than French at home and at work # so naturally when you don’t use a language and so on # at the end you lose # I haven’t forgotten it all, but there are words I’m missing because I don’t use them frequently.’ (10) quand ma belle-soeur et mon frère viennent en visite au début de leur séjour on se sent perdu # en huit ou dix jours en pratiquant on se remet un peu à flot # au début on se sent invalide du point de vue de la langue # je cherche mes mots et j’ai du mal à me mettre au diapason (Sara) ‘When my sister-in-law and brother come and visit us, at the beginning of their visit we just feel lost # after eight or ten days of practice it gets better # at the beginning we feel handicapped from a linguistic point of view # I look for my words and it’s hard to get in tune.’
Similar to the sentiments expressed in (10) above, a number of respondents have the feeling of a gradual ‘recovery process’ after some time of continued exposure. Some also indicate that they now have more opportunities to use French, as the amount of time at their disposal has substantially increased since their children left home or they themselves retired from work. They enjoy reading French books or browsing French Internet sites. Sara even asks herself if her adoption of the Hebrew language and culture really was as fundamental as she had previously thought (11). (11) ça montre peut-être qu’on est pas vraiment intégré ni dans la langue ni dans la culture ça pose un siman cheéla, # si disons après 40 ans on a plus de plaisir à lire en français parce que c’est plus facile (Sara) ‘Perhaps this means that we are not really integrated, neither linguistically nor culturally. It raises a question # if after 40 years we still enjoy reading in French more, because it’s easier.’
To the final question, “If you had to do it all over, would you follow the same path in Israel and in the kibbutz?”, the interviewees unanimously answered positively (12); some even started singing Edith Piaf ’s popular song – “rien, rien de rien, je ne regrette rien!” (‘Nothing, nothing, I regret nothing.’)
216 Miriam Ben-Rafael and Monika S. Schmid
(12) si c’était à refaire je le referais [...] les règles du jeu ont changé # mais je ne regrette absolument pas [...] c’est une expérience intéressante et je suis content de l’avoir fait (Tsviko) ‘If I had to do it again I would do it [...] the rules of the game have changed # but I absolutely don’t regret it [...] it is an interesting experience and I am pleased to have done it.’
Interestingly, however, almost all respondents question their policy of Hebrew monolingualism in retrospect – possibly in response to the changing of political strategies and prevailing attitudes towards multilingualism. They still endorse the imperative of speaking Hebrew to assert their collective identity, but regret not having given their children the chance to become bilingual (13)–(15). (13) la réalité d’alors s’est imposée # on pouvait pas tellement gérer les choses différemment # au niveau de l’ identité # du sentiment d’appartenance (Drora) ‘The reality was such that we couldn’t act differently # it was a question of identity # the feeling of belonging.’ (14) si c’était à refaire je parlerais en français à mes enfants # je ferais un drôle d’effort pour qu’ ils acquièrent la langue [...] on voulait une langue nationale et pour ça on était prêts à ne plus parler dans la langue maternelle (Yaakov) ‘If I had to do it all over I would have spoken French with my kids # I would have made efforts to let them acquire the language [...] we wanted a national language and for that we were prepared to no longer use our mother tongue.’ (15) ani mitstaèrèt ki ani roa èt ha-kmija chèl ha-jeladim ve chèl ha-nekhadim chéli ledabér (Margalit) ‘I regret it because I see my children’s and my grand-children’s eagerness to speak French.’
4.3.2 The Russian sample Unlike the Francophones, the Russian-speaking immigrants invoke reasons for their decision to emigrate to Israel which are predominantly instrumental and pragmatic (16). (16) esli chelovek s vysshim obrazovaniem, kvalifitsirovannyi spetsialist, ne mozhet prokormit’ svoju semju, obespechit’ ei dostoinyj uroven’ zhizni, on estestvenno ishchet vykhod iz polozhenia (Oleg) ‘If a person with higher education, a qualified professional, cannot feed his family, ensure them a decent living standard, it’s understandable that he will look for a way out.’
Language attrition and ideology 217
Unlike the Francophones, who never looked back once they had settled in the kibbutz, many report that their initial time in Israel had been fraught with difficulties and homesickness, and feelings of culture shock (17). (17) samoe trudnoe – eto privyknut’ k novoj strane srede, klimatu, jazyku, vse bylo chuzhim. Ja skuchala po Kievu s ego parkami i dozhdem, po Dnepru, ja skuchala po svoim druzjam # po vsemu k chemu ja privykla (Mariana) ‘The most difficult thing – was to get used to the new country, environment, climate, language, everything was foreign. I missed Kiev with its parks and rain, Dnepr [local river], I missed my friends # everything I was used to.’
Anna, who came from Moscow to Rehovot in 1989 at the age of 17, thought at the time that she had arrived in a khor, literally a “hole”, a deserted spot. For several respondents the worst problems were the language issue and finding a job. They often explained they could not find work because they did not speak Hebrew, adding that from the moment they began to master the language it was also easier to find employment: (18) samoe trudnoe – eto otsutstvie jazyka i problemy s rabotoj # my ved’ byli bez jazyka, bez raboty, ne znali elementarnykh veshchej, bank, kupat holim i tak dalee (Alex) ‘The most difficult – was the absence of the language and job problems # we had no language, no job, didn’t know the simplest things, [such as] bank, health insurance, and so on.’ (19) s rabotoj zdes’ ochen’ ploho, trudno ustroit’sia na rabotu, no eto v jazyk i upiraetsia v printsipe – tol’ko jazyk, esli by govorili normal’no problem by nikakih ne bylo, to est’ problemy konechno est’ – finansovye, no uzhe ne chuvstvuesh sebia polnym idiotom (Vladimir) ‘It’s not easy with jobs here, difficult to get a job, but it’s mainly because of the language – only language, if we spoke normally we would not have any problems, that is we would have problems, of course, financial, but you wouldn’t feel like such a complete idiot.’
While the general attitude towards Israel is consistently positive, becoming part of the community is not as important to the Russian participants as it was for the Francophones. Integration takes second place to individual success. Some remain quite pessimistic and still have ambiguous attitudes towards their new country. 17 years after his emigration, for example, Eugene still misses “the thick woods, the rivers, and the waterfalls” of his former home. Yet, strangely enough, when he is asked if it was worth coming to Israel, the answer is ‘yes’:
218 Miriam Ben-Rafael and Monika S. Schmid
(20) ja ne mogu otvetit’ na etot vopros, potomu chto v etot moment ja pochuvstvoval chto ja prosto shaiach # chto eto moja strana, moja zemlia, moj narod (Eugene) ‘I cannot answer this question, because at that moment I felt that I am simply shaiach # that this is my country, my land, my people.’
All interviewees are proficient speakers of Hebrew, but all the same Russian remains important; it is the language of the community and of the family, where it is essential to maintain it and pass it on to the next generation (21). Despite these ambitions, however, many speakers report a great deal of codeswitching, particularly when speaking of typically Israeli topics (22). (21) doma na russkom, hotia rebenok postoianno pytaets’a govorit’ na ivrit, my staraemsia chtoby evo russkij jazyk byl po vozmozhnosti chistym, chtoby sohranit’ jazyk (Alexei) ‘At home [we speak] Russian, though the kid always tries to speak Hebrew, but we try to keep his Russian pure as much as possible, to maintain the language.’ (22) estestvenno v moem slovare mnogo ivritskih slov: ni odin russkij ne skazhet “Institut natsional’nogo strachovania” ili “ipotechnaia ssuda”; oni govoriat “bituach leumi” “mashkanta” (Lena) ‘It’s natural in my vocabulary there are many Hebrew words: no Russian would say “Institute of national insurance” or “mortgage loan”; they say: “bitukh leumi, mashkanta”.
Many Russian-speaking respondents demonstrate a pragmatic attitude towards their languages, as illustrated by the example of Lena, who came to Israel in 1992. Her language use is situationally based: she codeswitches into Hebrew if she has to, although at home she always speaks Russian with her mother and her daughter, and this is also the language that she uses outside Israel when travelling. In Israel, she speaks both Russian and Hebrew. As a rule, the Russian interviewees agree that Russian must be retained; it is their mother tongue and their culture, and they want to preserve both and pass them on to their children. Russian respondents also differ from the Francophones when it comes to the question if, given the chance, they would emigrate to Israel again. Several respondents indicate that they would have preferred to emigrate elsewhere if they had had the opportunity. Oleg, for instance, 12 years after his immigration, is preparing to move again to Australia: (23) ja ne sionist, hotia mozhno skazat’ chto ja patriot Izrailia, ja v razgovorah nikogda ne polivaiu Izrail’ griazju i ne terpliu eto v svoem prisutstvii – ni zdes’ ni tem bolee za granitsej, no posviashchat’ svoju zhizn’ bor’be za svetloe budushchee Izrailia ja
Language attrition and ideology 219
ne gotov, zhizn’ tolko odna i ja hochu blagopoluchnoj, bezopasnoj i obespechennoj zhizni dlia svoej sem’i detej, dlia menia eto glavnoe slikha (Oleg) ‘I’m not a Zionist, though one can say I am an Israeli patriot, in conversations I never pour dirt on Israel nor allow it in my presence – neither here nor abroad, but I am not ready to dedicate my life to the fight for the bright future of Israel [an ironic use of a common Soviet expression], I only live once and I want a prosperous, safe and secure life for my family and children, that’s most important for me, excuse me.’
The previous examples and quotations illustrate a clear distinction between the two samples: while for the Francophones, the motivation to acquire Hebrew was strongly integrational, the Russian speakers mainly acquired it from an instrumental motivation. On the basis of the theories on language attitudes and SLA, it could therefore be predicted that the Francophone sample will be more successful in their ultimate attainment in Hebrew. However, the objective of the present study is to investigate the impact the different perspectives on language and identity have for the attrition or maintenance of the speakers’ L1, the hypothesis being that the L1 of the Francophone sample will show more instances of interferences and language mixing than that of the Russian sample.
5. Borrowing and codeswitching For the purpose of the present investigation, all instances of borrowing and codeswitching in a stretch of 1,000 words from the participants’ free data were analyzed and classified into two categories: single-item codeswitching (SIC) and multipleword sequences in Hebrew (MWS). The results are shown in Figure 1 (for an overview of SIC and MWS per speaker see Appendix, Table 1). The difference in the use of single item codeswitching between the two groups was highly significant (t(28) = 3.011, p < .01, two-tailed). The difference in the use of multi-word sequences between the two groups was marginally significant (t(28) = 1.802, p = .082). There is therefore a clear tendency for the Francophone sample to use more Hebrew, both single elements and phrases, in their L1 than the Russian speakers do. This finding is in accordance with our prediction that the more ideological and integrative attitude towards Hebrew of the French speakers would lead to more language mixing, and possibly also to a higher degree of L1 attrition, than the pragmatic and instrumental attitude of the Russian speakers.
220 Miriam Ben-Rafael and Monika S. Schmid
25,00 20,00 15,00 10,00 5,00 0,00 SIC MWS
Francophone 20,80 3,73
Russian 11,27 1,00
*SIC= single-item codeswitching; MWS=Multi-word sequence
Figure 1. Mean scores for SIC and MWS per 1,000 words*
6. Discussion As was illustrated above, the Francophone sample had more instances of borrowing and code-mixing in their L1 than the Russian sample in a stretch of 1,000 words per speaker. A qualitative investigation of the entire interviews shows that the semantic areas in which borrowings occur are quite similar across the two groups, with the exception that the Francophone speakers frequently use Hebrew expressions to refer to particularities of kibbutz life (24) and their pre-kibbutz training (25). (24)
khavérim aséfot vaadot markolit chkhuna arvut adadit hitnatkut
(25) garin chaliakh tnua
kibbutz members assemblies commissions grocery newcomers’ quarter community help disengagement group emissary youth movement
While the Russian group also uses Hebrew terms when describing their immigration experience, these are more general (26). (26) olim alja sokhnut
immigrants immigration immigration agency
Language attrition and ideology 221
ulpan ulpan alèf ulpan bét khinukh mejukhad mekhina
Hebrew course H. c. number one H. c. number two special education preparatory courses
Terms that refer to (public) institutions are used by both the Francophone (27) and the Russian (28) sample, as are work-related expressions ((29) and (30) respectively). (27)
khaver knèsèt avoda kupat kholim bituakh machlim
parliament member Labour party health insurance complementary insurance
(28) machkanta arnona bituakh léumi
loan payment by instalments insurance
(29 )
pardes falkha mifalim adachot metalizatsja
orangery extensive agriculture factories lenses metallization
(30)
pkida roé khèchbon ganènèt ozèrèt bajit tafkid akharaj atsmai
secretary accountant kindergarten teacher helper role responsible independent entrepreneur
In all of these areas, codeswitching is probably an indication of the fact that these terms are so intimately linked to the reality of life in Israel that French items would not fully convey what the speakers wished to express. Interestingly, however, both groups also use both single-item and multiple-word codeswitchings when describing their emotions (31 contains examples from the Russian sample). (31)
tsarot mesubakh jafe meod chalem im atsmi guch zar
worries complicated very nice I feel good about it outsider
222 Miriam Ben-Rafael and Monika S. Schmid
le-mazalenu by good luck ma che yech it’s what it is al tagidi jije be-seder, don’t say it will be ok, it’s never ok af paam lo be-seder
In the French group, there were indications that codeswitchings in this particular area were often due to the fact that the corresponding French items were either inaccessible (32) or felt to be not entirely appropriate to the feeling the speaker wished to convey (33). (32) c’est plus jail # comment on dit ça? (Ben) ‘It’s more efficient # how do you say that?’ (33) j’ai été memurmèrèt # je saurais pas le dire en français (Sara) ‘I have been disappointed # I wouldn’t know how to say that in French.’
It therefore appears that, while both groups use codeswitching into Hebrew for more or less the same purposes, the Francophone speakers do so more often than the Russian speakers. This impression is confirmed by the fact that, over the entire interviews, the Francophone speakers use a much higher proportion of MWS than SIC: while in the French sample, one MWS occurs for every 3.37 SICs, in the Russian sample that proportion is 11.57 (see Appendix, Table 2). In other words, while for the Russian speakers, Hebrew influence on their L1 asserts itself mainly through the borrowing of isolated lexical items, the Francophone speakers seem to have a higher degree of integration of their linguistic systems when speaking the L1, and language-mixing also extends to grammatical items.
6. Conclusion The present paper has attempted to compare two groups of immigrants in Israel with respect to their attitudes towards Israeli society and the Hebrew language on the one hand and their use of their L1 on the other. The first group consisted of 15 Francophone kibbutzniks who emigrated to Israel in the late 1950s and early 1960s. Overall, their decision to emigrate was ideologically motivated. The society and environment they found, at that time, adhered to a strictly monolingual policy. The Hebrew language was considered an embodiment of the Israel state, and it was the goal of the immigrant generation to contribute to the revival of that language by bringing up a generation of monolingual native speakers. More than that, however, acquiring Hebrew was, for the immigrant, not only a marker but an embodiment of identity. It is likely that the integrative motivation to acquire the language of the country was stronger
Language attrition and ideology 223
for ideologically motivated immigrants to Israel in this period than for any other group of migrants over the past century. The second group comprised 15 Russian immigrants who settled in urban areas in the 1990s. For these participants, the motivation to emigrate had been taken less on ideological and more on pragmatic grounds: they came to Israel not to build a new nation, but to find a better (and safer) home for themselves and their families. While the Hebrew language, to them, represented an important tool for economical reasons, they did not experience the same burning need to become native-like and to raise their children as the first generation of native speakers. On the contrary, the desire to pass on their L1 to their own children was prevalent across this group. The linguistic investigation analyzed the most clearly visible indications of L2 interference in the free spoken L1 data from these two groups: borrowings and codeswitchings. The quantitative analysis established clear differences in the amount of these phenomena: the Francophone sample used significantly more Hebrew elements in their discourse than did the Russian one, and in the discourse from the Francophone speakers, the two linguistic systems were integrated to a higher degree than was the case for the Russians, who used mainly isolated, borrowed Hebrew words. These findings do point towards a link between language attitudes on the one hand and language attrition on the other. More detailed investigations of these and other groups, encompassing more linguistic phenomena and relying on different types of data, will, however, need to be conducted to substantiate this tentative finding.
References Ben-Rafael E. & Peres, Y. (2005). Is Israel One? Religion, nationalism and multiculturalism confounded. Boston/Leiden: Brill Academic Publishers. Ben-Rafael E., Lyubansky, M., Glöckner, O., Harris, P., Israel, Y., Jasper, W. & Shoeps, J. (2006). Building a Diaspora – Russian Jews in Israel, Germany and the USA. Boston/Leiden: Brill Academic Publishers. Ben-Rafael, E., Olshtain, E. & Geijst, I. (1996). “Identity and Language: The social insertion of Soviet Jews in Israel”. In N. Lewin-Epstein, Y. Ro’i & P. Ritterband (Eds.), Russian Jews on Three Continents (pp. 364–388). London: Frank Cass. Ben-Rafael, E. (1994). Language, Identity and Social Division: The case of Israel. Oxford: OUP. Ben-Rafael, M. (2001). Contact de langues: Le français parlé des francophones israéliens. Unpublished Doctoral Dissertation, Tel Aviv: Tel-Aviv University. Breakwell, G. M. (1983). “Identities and conflicts”. In G. M. Breakwell (Ed.), Threatened Identities (pp. 189–213). New York, NY: Wiley. Breakwell, G. M. (1986). Coping with Threatened Identities. London/New York: Methuen.
224 Miriam Ben-Rafael and Monika S. Schmid
Castells, M. (1999). Le pouvoir de l’identité. Paris: Fayard. Deutch, Y. (2005). “Language law in Israel”. Language Policy, 4, 261–285. Edwards, J. (1999). “Refining our understanding of language attitudes”. Journal of Language and Social Psychology, 18 (1), 101–110. Gardner, R. C. (1985). Social Psychology and Second Language Learning: The role of attitudes and motivation. London: Edward Arnold. Gardner, R. C. & Lambert, W. E. (1959). “Motivational variables in second language acquisition”. Canadian Journal of Psychology, 13, 266–272. Giles, H., Bourhis, R. & Taylor, D. (1977). “Towards a theory of language in ethnic group relations”. In H. Giles (Ed.), Language, Ethnicity and Intergroup Relations (pp. 307–348). London: Academic Press. Hulsen, M. (2000). Language Loss and Language Processing: Three generations of Dutch immigrants in New Zealand. Unpublished Doctoral Dissertation, Nijmegen: Katholieke Universiteit Nijmegen. Kheimets, N. G. & Epstein, A. D. (2001). “English as a central component of success in the professional and social integration of scientists from the former Soviet Union in Israel”. Language in Society, 30, 187–215. Krashen, S. (1981). Second Language Acquisition and Second Language Learning. Oxford: Pergamon. Laufer, B. (2003). “The influence of L2 on L1 collocational knowledge and lexical diversity”. In V. Cook (Ed.), Effects of the Second Language on the First (pp. 19–31). Clevedon: Multilingual Matters. Le Page, R. B. & Tabouret-Keller, A. (1985). Acts of Identity: Creole-based approaches to language and ethnicity. Cambridge: CUP. Long, M. H. (1990). “Maturational constraints on language development”. Studies in Second Language Acquisition, 12, 251–285. Masgoret, A.-M. & Gardner, R. C. (2003). “Attitudes, motivation, and second language learning: A meta-analysis of studies conducted by Gardner and associates”. Language Learning, 53 (1), 123–163. Myhill, J. (2004). Languages in Jewish Society: Towards a new understanding. Clevedon: Multilingual Matters. Pavlenko, A. (2005). Emotions and Multilingualism. Cambridge: CUP. Remennick, L. (2003). “From Russian to Hebrew via HebRush: Intergenerational patterns of language use among former Soviet immigrants in Israel”. Journal of Multilingual and Multicultural Development, 24 (5), 431–453. Rozenbaum-Tamari, J. (2004). The olim of the former Soviet Union – Alija, motivations and roles in front of the Israel society. Jerusalem: Ministry of absorption and integration (in Hebrew). Schmid, M. S. (2002). First Language Attrition, Use and Maintenance: The case of German Jews in Anglophone countries. Amsterdam: John Benjamins. Schuman, J. (1994). “Where is cognition? Emotion and cognition in second language acquisition”. Studies in Second Language Acquisition, 16, 231–242. Spolsky, B. & Shohamy, E. (1996). “National profiles of languages in education: Israel. Language Policy”. http://www.biu.ac.il/HU/lprc/lprcprof.htm. Spolsky, B. (1999). “Language in Israel: Policy, practice, and ideology”. In J. E. Alatis & A. H. Tan (Eds.), Georgetown University Round Table on Languages and Linguistics 1999 (pp. 164–174). Washington, DC: Georgetown University Press.
Language attrition and ideology 225
Waas, M. (1996). Language Attrition Downunder. Frankfurt: Peter Lang. Yağmur, K. (1997). First Language Attrition Among Turkish Speakers in Sydney. Tilburg: Tilburg University Press.
Appendix Table 1. Single-item codeswitching (SIC) and multi-word sequences (MWS) per speaker in a stretch of 1,000 words Francophone SIC
MWS
Russian
SIC
MWS
Ben Shula Drora Eli Jenny Yoel Margalit Moshe Nathan Sara Tsviko Yair Yaakov Isra Yoav Mean St.dev.
0 2 0 1 18 0 12 0 0 3 12 3 0 2 3 3.73 5.60
Alex Alexei Alik Anna Ellen Eugene Harry Ilia Irena Lena Marianna Mira Natasha Oleg Vladimir
10 11 9 1 7 22 9 10 28 7 9 10 19 9 8 11.27 6.72
0 0 1 0 0 4 0 0 6 2 0 1 0 1 0 1.00 1.77
10 43 14 33 22 25 16 14 16 38 21 12 15 24 9 20.80 10.26
226 Miriam Ben-Rafael and Monika S. Schmid
Table 2. Single-item codeswitching (SIC) and multi-word sequences (MWS) per speaker in entire interview Francophone SIC
MWS
Proportion Russian SIC/MWS
SIC
Ben Shula Drora Eli Jenny Yoel Margalit Moshe Nathan Sara Tsviko Yair Yaakov Isra Yoav Total
32 49 2 14 14 0 83 0 3 18 15 4 1 3 8 246
1.57 0.36 30.00 15.50 2.13 – 1.22 11.50 15.33 6.00 – 20.00 3.39 9.14 3.53 3.37
10 27 22 3 18 42 12 9 78 17 13 12 19 19 23 324
39 173 31 128 22 49 30 35 60 61 32 46 30 46 48 830
Alex Alexei Alik Anna Ellen Eugene Harry Ilia Irena Lena Marianna Mira Natasha Oleg Vladimir Total
MWS
1
8
12 4 2 1 28
Proportion SIC/MWS – – 22.00 – – 5.25 – – 6.50 4.25 – 6.00 – 19.00 – 11.57
Stimulated recall methodology in language attrition research Antonio F. Jiménez Jiménez
California State University, Channel Islands, USA
The present study explores the value of the Stimulated Recall Protocol for investigations of first language attrition. On the basis of a study of Spanish heritage speakers in the United States it is argued that this method allows a more direct way of tapping into compensatory strategies which speakers apply in situations of communicative breakdown. Insights which speakers provide upon being confronted with their own narratives can provide evidence for changes in the process of regulation which are experienced by attriters, who, faced with the task of engaging in a communicative activity, are forced to retreat to previous stages in this developmental process. Keywords: communicative strategies; stimulated recall protocol; regulation of communicative process; sociocultural theory; heritage speakers; Spanish
Introduction Recent investigations on language attrition are offering new perspectives, methodologies, and findings to a field very much in need of answers (see Schmid, Köpke, Keijzer & Weilemar 2004, for a collection of the latest line of research on language attrition). The present study adheres to this trend of investigation by proposing a methodology that, although extensively used in psychology, education, and language acquisition studies, has never been explored in language attrition research. This methodology, known as stimulated recall protocol (SRP), “is used to explore [speakers’] thought processes (or strategies) at the time of an activity or task. This is achieved by asking [speakers] to report those thoughts after they have completed a task or activity” (Gass & Mackey 2000: xi). The objective of the present article is to analyze the value of using this methodology in language attrition research. Specifically, it focuses on the use of compensatory strategies by a group of attriting speakers of Spanish during an oral narration task. The following section
228 Antonio F. Jiménez Jiménez
will provide an in-depth analysis of the role of compensatory strategies in previous language attrition studies. Then, a detailed description of stimulated recall methodology will follow.
1. Compensatory strategies in language attrition research When faced with a communicative breakdown, as is frequently the case with language attriters, speakers develop compensatory strategies aimed at resolving the problem. Communicative or compensatory strategies have been extensively studied in the field of applied linguistics (Bialystok 1990; Faerch & Kasper 1984; Kasper & Kellerman 1997; Kellerman 1991; Poulisse 1990, 1996, 1999; Tarone 1977, 1983). A number of authors have focused on the analysis of particular strategies. Thus, for instance, Kenny (1996), Nakuma (1997), Christenfeld (1994), Levin, Silverman & Ford (1967), and Taylor (1969) studied the analysis of pauses (filled or silent); Tomiyama (1999) focused on speech rate and repetitions; Russel (1999) on discourse fillers; Channel, Sinclair and Carter (1994) on vague language; Levelt (1989) on self-corrections; Cohen (1986, 1989), and Olshtain and Barzilay (1991) on lexical retrieval strategies; Andersen (1982) on literal translations; and Waas and Ryan (1993) on the use of onomatopoeia. In 1985, Lantolf and Frawley created a functional typology of compensatory strategies based on the idea of regulation. For these authors, communication, on a psychological level, is not regarded as the transfer of information. Instead, communication is considered “as an attempt by speakers to regulate themselves as individuals when verbalizing” (Lantolf & Frawley 1985: 146). Maintaining selfregulation is the ultimate goal for the speaker engaged in communicative activity. At first, the language learner, whether an L1 or L2 learner, is controlled by the language that the language community uses for social interaction (that is, in sociocultural terms, the novice is object-regulated by the language). In the process of gaining control over the language, learners may rely on the help of others (becoming other-regulated) until they become more knowledgeable and independent users of the language (achieving self-regulation). Thus, acquiring a lan-
. It can also be argued that achieving other-regulation becomes fundamental in the communicative activity, not in the sense of trying to control someone but rather of allowing the other person to understand the speaker’s perspective and personal sense about the object of conversation.
Stimulated recall methodology 229
guage is about moving from object- and other-regulation to self-regulation. In this process, language comes under voluntary control of the speaker. This developmental sequence of regulation is not categorical and does not have an end point of permanent self-regulation at all times. The speaker can lose self-regulation at any time depending on the circumstances of the communicative activity. When this happens, “the speaker can reaccess an earlier communicative strategy (i.e., object-regulation, and other-regulation) in order to solve the task of communication, that is, gain self-regulation” (Lantolf & Frawley 1984: 427). In other words, “any speaker has continuous access to all forms of order [i.e., control, regulation] by means of verbalization” and “this availability of ordering strategies is never lost” (ibid.: 427). From this perspective, “language attrition entails the loss of the once attained level of self-regulation of a language in a particular activity and the momentary return to a previous stage of object- and/or other-regulation” (Jiménez Jiménez 2004: 67). That is to say, the speaker, engaged in a communicative activity, needs to reaccess previous levels of development (other- and object-regulation) more frequently than at a previous point in time when the individual was able to maintain a higher degree of self-regulation for the same activity. A communicative breakdown can have many causes, most likely the novelty of a task, a problem with the language of communication, or a conceptual misunderstanding in the course of the communicative activity. Based on the notion of regulation and on the idea that speakers tend to maintain linguistically-based self-regulation at all times, speakers develop strategies that help them continue performing the task at hand. The next sections categorize these strategies depending on their main regulatory function. Even though strategies have been associated with a certain category (self-, other-, or object-regulation), it is important to note here that strategies do not pertain exclusively to one of these categories. Instead, strategies are considered fluid entities which can serve different regulatory functions depending on all the concomitant factors that are involved in a certain communicative activity.
1.1 Self-regulatory strategies According to Lantolf and Frawley’s functional typology of compensatory strategies, self-regulation obtains when choice of strategy is determined by the attempt
. And as Vygotsky argued (1962), this move is optimally achieved in school, because here the learner can attain conceptual (theoretical) understanding of the language and how it can be manipulated for communicative purposes.
230 Antonio F. Jiménez Jiménez
to maintain self communicative activity (Lantolf & Frawley 1985: 153). The use of self-regulatory strategies performs different functions, such as: to buy processing time (with the use of filled and unfilled pauses, and discourse fillers), to focus attention on what is coming next (with self-repetitions), to monitor the uttered speech (self-corrections, self-recasts), to retrieve desired information (self-retrieval), and to guide and plan the course of action (with metacomments). Strategies are manifestations of these higher cognitive functions at work, and they represent attempts by the individual to maintain self-regulation during the communicative activity.
1.2 Object-regulatory strategies According to Lantolf and Frawley (1985), object-regulation manifests itself “when interlanguage as object regulates choice of strategy” (p. 153). When this is the case, the speaker has two options. On the one hand, the speaker can completely avoid the topic, in which case s/he has become object-regulated by his/her lack of control over either the language or the task. On the other hand, the speaker can compensate for his/her own inability to supply the desired outcome by creating an intermediary solution that partially fulfills the communicative requirements and, at the same time, camouflages the speaker’s inability to supply the desired form. The compensatory strategies identified as having a primarily object-regulatory function are circumlocutions, approximations and all-purpose words, word-coinage, literal translations, onomatopoeia, and codeswitching. Message abandonment is considered the most extreme form of object-regulation since it shows a shift from the individual having control over the task/language to the task/language having control over the individual.
1.3 Other-regulatory strategies During speech activity self-regulation cannot always be maintained and objectregulatory strategies are not always available. When this happens, the speaker can rely on the help of the interlocutor. Other-regulation occurs “when presence of other(s) in communicative acts regulates choice of strategy” (Lantolf & Frawley 1985: 153) or when the speaker relies on the help of the interlocutor to solve a given communicative problem. In order to be able to employ other-regulatory strategies, the rules of the activity in which speakers are engaged have to allow cooperative communication. In some situations, appeal to the interlocutor is not readily attainable, such as in testing situations, in which case the speaker is obliged to use other types of strategies to overcome the communicative breakdown.
Stimulated recall methodology 231
SELF-REGULATION ↓ Filled pauses Unfilled pauses Segmental lengthening Slower speech rate Discourse fillers Self-repetitions Self-recasts Self-corrections Self-retrieval Private Speech
↔
OBJECT-REGULATION ↓ Circumlocutions Approximations Word-coinage Literal translations Codeswitching Codemixing Onomatopoeia Message abandonment
OTHER-REGULATION ↓
↔
Repetition requests Clarification requests Confirmation requests Expressions of non- or misunderstanding
Figure 1. A taxonomy of communicative strategies in relation to their main regulatory function
Based on these theoretical grounds, the present chapter considers compensatory strategies as manifestations of the internal processes of a speaker engaged in the task of solving a communicative problem. Some strategies can be recognized indirectly in the speech. This is the case of circumlocutions, in which the researcher can determine that the speaker is struggling to convey the message around the problem of finding the appropriate word or expression. However, in most cases, compensatory strategies are very difficult to identify if the analysis of the data is confined to microlinguistic aspects. For instance, in order to avoid a linguistic problem, speakers can choose to use approximations and vague language, producing linguistically correct propositions. These avoidance strategies, and others, are not reflected in speech and remain hidden within the speaker (Schmid 2004). Thus, if we want to assess the use of avoidance strategies by a certain speaker, there is no other way but to ask the speaker him/herself about the cognitive processes that take place inside his/her head while in the task of language formation.
2. Stimulated recall methodology The study of thought processes in second language research has been approached from very different angles using a wide range of introspective methods (e.g. Cohen 1996, 1998; Faerch & Kasper 1987). Of all these, thinking aloud protocols, where subjects are instructed to verbalize their thoughts while they are engaged in a difficult task (Ericsson & Simon 1993), have been most frequently used in linguistic studies. Participants are asked to say whatever they are looking at, thinking, doing, and feeling as they go about their task. This way, the researcher gains insight into the process involved in solving the task. Although this experimental
232 Antonio F. Jiménez Jiménez
strategy is extensively used, many believe that the externalization of the thought process while doing the activity may change the thought process itself. Kellerman (1974) started applying an early version of this method in order to understand how second languages are learned. He recognized the importance of receiving feedback from the participant, stating that “by bringing the learner into the analysis of his text […], we are making available data that are useful in the refinement of techniques of analysis of language produced by learners beyond the point permitted by the application of experiential insights alone” (p. 97). The proposed methodology in this chapter shares with thinking aloud protocols the fact that participants are asked to externalize their thoughts, with the exception that this is done immediately after the communicative activity has been completed. In order to help participants remember their thought processes, some tangible visual or aural reminder (in the form of a video or audio of the subject while performing the task) is presented. It is assumed that this reminder of the “event will stimulate recall of the mental processes in operation during the event itself ” (Gass & Mackey 2000: 17). This method is superior to a simple post hoc interview in that the participant does not need to heavily rely on memory without any prompts. It also has an advantage over thinking aloud protocols because the participant does not need to go through a process of training in order to be able to perform a task and talk about it simultaneously. This methodology seeks to uncover cognitive processes that are not evident through simple observation. Stimulated recall methodology is useful because it can help determine how knowledge is organized and how cognitive processes, such as retrieval, search or decision making, are being employed (Gass & Mackey 2000: 21). This study seeks to assess the utility of using this methodology in language attrition research. It is hypothesized that stimulated recall will unveil important cognitive processes of attriting speakers, especially as regards their use of compensatory strategies when faced with a communicative problem. More specific information about how this methodology was employed in the framework of this study is presented below.
3. Methodology 3.1 Participants In a globalizing world where migration is steadily increasing, more and more people are putting down roots in other countries. As a consequence, the number of heritage speakers is multiplying. A “heritage speaker” is defined here as someone who is born and raised in a household where a language other than the one main-
Stimulated recall methodology 233
ly used in the wider community is spoken (Valdés 2000). Some heritage speakers merely understand the heritage language, while others can communicate fluently in it. Regardless of where they stand in this continuum, for these speakers the heritage language is normally casual (i.e., they use this language to communicate with family and friends) and is largely influenced by the mainstream language (normally their stronger language). This is language attrition at the inter-generational level (also referred to as language shift), and this linguistic erosion becomes more accentuated each generation down. In general terms, heritage speakers usually have an incomplete acquisition of the heritage language. Therefore, when communicating in this language, especially in more formal contexts, they encounter communicative problems similar to the ones faced by language attriters. Correspondingly, they also make a wider use of communicative strategies in order to compensate for the linguistic difficulties. Since the focus of this chapter is to scrutinize the use of compensatory strategies by speakers who encounter frequent communicative problems (like language attriters and heritage speakers), and not to analyze the language itself that is produced by these very distinct groups, this study used a group of heritage speakers as participants. A total of ten heritage Spanish speakers participated in this study. All of them were students at a small public university in Southern California. They were born and raised in the United States from Mexico-born parents. They were not only the first generation in their lineage to be born to American citizenship, but also the first generation of their families to attend higher education. Participants’ literacy, comprehension, and conversation skills in Spanish varied considerably. However, for the purpose of this article, language proficiency is not considered a determining factor since the main objective is to assess compensatory strategies in language attrition using stimulated recall methodology. The following paragraphs will describe in detail the different phases involved in this methodology as used in this study.
3.2 Phase 1: Information recall This study employed a data elicitation protocol based on the recall of a videoclip. Free oral recall has proven to be one of the most practical and productive means to assess a wide range of diverse linguistic aspects simultaneously, such as vocabulary development (Cohen 1989; Olshtain & Barzilay 1991; Russell 1999; Yoshitomi 1999), syntactic complexity (Yoshitomi 1999), verbal morphology (Reetz-Kurashige 1999), phonology (Tomiyama 1999; Yoshitomi 1999), and communicative competence (Nakuma 1997), as well as error analysis (de Bot & Clyne
234 Antonio F. Jiménez Jiménez
1994; Yoshitomi 1999). For the present study, students had to retell, orally in Spanish (their heritage and weakest language), the story enacted in a five-minute clip of the well-known U.S. TV sitcom, Friends, played in English (their dominant language). For the reader’s reference, a transcript of the clip is included in the Appendix. Immediately after watching the clip, students had to retell the story in the form of a monologue. Monologues are considered here a type of communicative activity, as there is information that is transmitted to a recipient. In fact, based on Vygotsky (1962) and Sokolov (1972), Lantolf and Frawley (1984) claim that “monologue is the highest or most complex form of speech not dialogue” because it can be argued that “a speaker has fully acquired a language when s/he can engage in collective monologue, not in dialogue, since monologue entails the speaker as a fully self-regulated speaker” (p. 426–7). However, as mentioned above, this “fully self-regulated speaker” is an ideal representation rather than a real one, as the speaker relies on other types of speech (other- or object-regulated) when the circumstances require it. It could also be argued that constructing a monologue involves “an ‘I’ that makes choices on what to talk about and a ‘Me’ that interprets and critiques these choices” (Vocate 1994: 12). This conception of monologue as an internal dialogue between an ‘I’ and a ‘Me’ can be corroborated in the examples shown below, as students were able to recall what they were thinking (e.g., hesitations, retrieval difficulties, self-corrections, etc.) during the process of creating a monologue. Participants’ oral production was digitally audio- and video-recorded (with a microphone and a webcam connected to a computer) for later data transcription and analysis. Instructions for the task were given in the target language, that is, Spanish. The language mode was quite informal to make students feel as comfortable as possible in the lab environment. Students did not have any time limit during the oral retelling of the clip. Participants produced narrations that averaged 2 minutes and 45 seconds, containing a mean of 339 words each.
3.3 Phase 2: Transcription and preliminary analysis Once students finished narrating the story, they were given a comprehensive language survey questionnaire to fill out, which took on average 25 minutes to complete. During this time, the main researcher transcribed the oral narration and conducted a preliminary analysis of the text. To expedite the transcription process, the audio and video data were recorded directly to the researcher’s computer, and then immediately transcribed using a specialized transcription software (WAVpedal 4.0), a foot control pedal attached to the computer, and headsets.
Stimulated recall methodology 235
This type of free oral retelling lends itself well to a number of different linguistic analyses. However, for the purpose of this study, the analysis focussed on the use of compensatory strategies. Creating a coherent story in your weakest language is, to say the least, a challenging task. Students engaged in this activity needed to rely on the use of compensatory strategies in order to help them convey the storyline enacted in the video-clip. The researcher examined the texts, identifying and tagging the compensatory strategies as described previously.
3.4 Phase 3: Stimulated recall Once the researcher and the participants finished their respective tasks, they both came together and talked about the participant’s view of her/his own performance. In order to facilitate the task, the researcher showed the video-clip of the participant narrating the story together with the transcription of the retelling. The researcher used the tagged (i.e., analyzed) version of the transcription but, to avoid bias, did not show it to the participant. Following the guidelines described by Gass and Mackey (2000), participants were instructed to stop the video when they wanted to comment on their narration. The researcher was also allowed to stop the video and ask questions about the narration. This was done for every compensatory strategy identified in the preliminary analysis, especially in cases where the preliminary analysis was ambiguous. This recall session was also audioand video-recorded for later analysis. The use of this immediate recall protocol proved to be an excellent instrument to assess avoidance strategies and at the same time to fine-tune the analysis previously conducted by the researcher.
4. Results Most compensatory strategies were easily recognizable during the preliminary analysis of the oral texts. In fact, in 90% of the cases the researcher’s analysis of regulatory strategies was corroborated by the participants. In other words, these strategies can clearly be identified and quantified with a high degree of reliability. However, the use of stimulated recall protocol (SRP) offered a deeper understanding of the participants’ thinking process in their use of these strategies. For instance, while the researcher can clearly analyze a participant’s use of pauses (length, frequency, speech rate, etc.), he or she cannot know for sure what the speaker is thinking in that moment. The speaker could have encountered a linguistic problem that s/he is trying to solve, or maybe the speaker has just momentarily forgotten the story and is trying to retrieve it from memory. SRP allows a
236 Antonio F. Jiménez Jiménez
qualitative analysis of the data that proves to offer invaluable information about the process of language formation. Compensatory strategies are used by all of us, from the novice learner to the emeritus scholar. When the communicative task at hand becomes challenging, speakers access previous stages of other- and object-regulation, making use of strategies to continue the task. The examples provided in this section, uttered by first-generation heritage speakers of Spanish with an incomplete or attriting knowledge of this language, are comparable to the ones found in previous attrition studies. It is theorized that this methodology is also valid for different typologies of attriting speakers and the findings derived from this set of data can also be applicable to language attrition research. For this reason, the examples provided in this section will be discussed as they relate to language attrition. SRP was particularly useful for identifying avoidance strategies, for instance, the use of vague language and approximations, when the speaker uses a vocabulary item or structure which the speaker knows is imprecise or incorrect, but which shares enough semantic features with the desired item to satisfy the speaker’s communicative need (Tarone 1983). It is important to note here that a traditional linguistic analysis would not reveal any trace of attrition since speakers produced grammatically accurate sentences. SRP provides a deeper level of understanding of the attrition process. (1.1) Narration: Translation: Recall:
No quiere trabajar allí y se le olvidó le- um leer su currículum vitae… She doesn’t want to work there and she forgot to re- um read her resume… The speaker used “to read” when she wanted to say “to edit” but she could not remember the word
(1.2) Narration: Translation: Recall:
… le estaban ayudando a… a…poner los el currículum en en el sobre … they were helping her … putting the resumes in in the envelope Empacar meaning “to pack” came to the speaker’s mind, but she knew that it was not the right word so she used a more generic word
. For a better understanding of the context that triggered the examples shown in this section, the Appendix contains the transcript of the video-clip that participants had to narrate. The examples shown here are excerpts of participants’ narrations of the story shown in the videoclip.
Stimulated recall methodology 237
(1.3) Narration: Translation: Recall:
Los dos jóvenes están sentados en el café hablando sobre muchachas The two young guys are sitting in the café talking about girls During the stimulated recall session, the speaker clarified that he really wanted to say “on the sofa” but he did not know how to say it in Spanish, which would be en el sofá
Another avoidance strategy is to recast or rephrase an idea in such a way that the speaker is able to produce a grammatically correct form. The end result is a completely accurate linguistic utterance, but the cognitive path followed would remain invisible without this methodology. (2.1) Narration: Translation: Recall:
… y entonces pues nin- ninguna lo ninguna carta está uh uh bien para ella… … and then so any any any letter was uh uh good for her What the speaker meant to say was that she was not accepted anywhere
(2.2) Narration: Translation: Recall:
… Ross dice que quiere a alguien que um que lo haga sentir algo …Ross says that he wants someone who um who makes him feel something The speaker wanted to say “to stir something in him” but she found she did not know how to express this idea
Pavlenko (2004) noted that when the speaker code-mixes and uses a lexical borrowing from the non-attriting language, we have to examine the reasons for which that particular lexical item has been borrowed. She argues that if the speaker has never learned/used that lexical item in the attriting language, the use of a borrowing should not be considered a sign of language attrition. SRP provides a way to identify these vocabulary items and structures never learned in the attriting language. (3.1) Narration: Translation: Recall:
… de allí hay un una… pausa en el en en la narración en el video and from there there is a a… pause in the in in the narration of the video In the interview the speaker said that she wanted to say “the credits” but she had never learned that word in Spanish, which turns out to be a cognate, namely créditos
(3.2) Narration:
y están echando resumes o ¿cómo se dice? Em… ¡Ay! ¡Bueno! Resumes
238 Antonio F. Jiménez Jiménez
Translation: Recall:
and they are sending resumes or how do you say that? Em… Oh! Well! Resumes In this case the speaker knew that he had learned the Spanish word for “resume” (which is curriculum) but he could not recall it in that moment. When he was told the word, he recognized it
Another manifestation of attrition is the use of circumlocutions, when the speaker describes the characteristics or elements of the objects or action instead of using the appropriate target language item or structure (Tarone 1983). Although sometimes circumlocutions can be easily recognized, some other times are not so clear. (4.1) Narration: Translation: Recall:
Ross se da cuenta que ha escrito algo mal Ross realizes that she has written something wrong The speaker did not know how to say “typo”. An option in Spanish for this word could be una errata
(4.2) Narration: Translation: Recall:
… en cuanto se le arrima la mesera otra vez empieza a como como muy muy este… um… como que está toda- que sí todavía la quiere ¿verdad? … as soon as the waitress comes close again he starts like like very very um… um… it’s like he is still like he still loves her, right? What she really wanted to say is that he had “puppy eyes”. In Spanish hacer ojitos could be said
(4.3) Narration: Translation: Recall:
… mientras que las amigas están uh ab- ab- abriendo las cartas y leyendo lo que dicen uh no encuentran ninguna que que le dé buenas noticias todas que oh gracias por mandar la carta pero no ocupamos su ayuda … while her friends are uh op- op- opening the letters and reading what they say uh they don´t find any that that give her good news all of them say oh thank you for sending the letter but we don´t need your help The speaker later said that he wanted to say that all of the letters “rejected” her. To “reject” is rechazar in Spanish
SRP also helps to identify omitted content in the narration, i.e., when the speaker omits some information from the story because s/he did not know how to say it or felt that it would be complicated to explain. This avoidance strategy is almost impossible to assess with a traditional linguistic analysis, since it cannot account for what is not there.
Stimulated recall methodology 239
(5.1) Narration: Translation: Recall:
... y entonces pues nin- ninguna lo ninguna carta está uh uh bien para ella… … and then so any any any letter is uh uh good for her… The participant wanted to say that the character was not accepted anywhere, which was partially expressed in her utterance but she also wanted to add that all the companies found something negative in her resume
(5.2) Recall:
One of the speakers commented that she left out all the information about the scene of the paper cut because she did not know how to say it
The stimulated recall protocol also provided a better understanding of the reasons for which the speaker uses self-corrections. The use of these strategies implies that the speaker is closely monitoring speech production and correcting for ambiguity and to clarify potential misunderstanding. (6.1) Narration: Translation: Recall:
… y traen consigo un unos um ¿cómo se llama? Un correo, correspondencia… …and they bring with them a some um how do you say that? a mail, correspondence The speaker thought that correo was not correct because she had that word associated with “post office” and not with “mail” and that is why she used a clarifying word afterwards. In Spanish, correo can mean both “post office” and “mail”. However, like in English, the word correo can not be preceded by the undetermined article.
Another example of the usefulness of SRP is that it allows the researcher to distinguish slips-of-the-tongue from improper use of the language. Based on Dell’s (1986) definition of slip of the tongue (“an unintended, nonhabitual deviation from a speech plan”) (p. 284), Poulisse (1999) identifies two criteria: (1) that there must be an error and (2) that the speaker must be able to repair the error (p. 91). SRP allows asking speakers to correct these errors. If unable to do this, we can conclude that it was not a slip of the tongue. (7.1) Narration: Translation: Recall:
Este video es sobre la seria de Friends en la televisión This video is about the series of Friends on television When asked, the speaker was not sure about the word. In Spanish you can use the word serie
(7.2) Narration:
… y le pregunta que si los había leído antes de um emprimirlos para mandarlos
240 Antonio F. Jiménez Jiménez
Translation: Recall:
… and he asks if she had read them before um printing them to send them The researcher asked the speaker to say the word again and she again said *emprimirlos when the right form is imprimirlos
This research methodology also allows clarifying many of the uncertainties that arise in the preliminary analysis. The researcher can ask direct questions (e.g., how do you say this?) to confirm whether the speaker really knows some particular linguistic element or not. (8.1) Narration: Translation: Recall:
… están hablando sobre poker (English pronunciation) … and they are talking about poker The speaker was not sure how to say poker in Spanish (pronounced as /´poker/ or /´poka:/) so she decided to use the English pronunciation (/´p6~k6r/). This comment clarified the analysis since the word poker is a borrowing from English but it was evident then that the speaker was thinking of the English word at that time
Sometimes speakers are unsure about some linguistic element, and they make a guess and eventually produce an utterance that is grammatically correct. A traditional linguistic analysis would not be able to recognize this uncertainty while SRP can. (9.1) Narration: Translation: Recall:
Entonces de ahí se va a otra escena Then from there it goes to a different scene The speaker did not know the word for ‘scene’, so he made it up thinking it could be a cognate. In this case he was successful. This methodology was able to identify this hesitation
Another advantage of using this methodology is that at times, when participants hear themselves, they make corrections to their stories (10.1). SRP offers the opportunity for participants to self-reflect on their own language production, giving them a second chance to refine their narration (10.2). This methodology reminds us of the difference between on-line performance and linguistic competence. In a simple linguistic analysis we could consider improper use of the language as a sign of attrition while, if given the opportunity to reflect on their own linguistic production, speakers could make the necessary corrections.
Stimulated recall methodology 241
(10.1) Narration: Translation: Recall:
… dos amigas de Rachel um entran a la restaurante … two friends of Rachel um come into the (fem.) restaurant The article should agree in gender with its noun (restaurant) in the masculine form. The speaker recognized the mistake and corrected it during the recall session
(10.2) Narration: Translation: Recall:
Y ella quiere saber si ha recibido respuestas de los trabajos que pidió And she wants to know if she has received answers for the jobs that she asked for During the recall session, this speaker said solicitó (to apply), which is a more precise term for the idea that she was expressing
There are other advantages of using this methodology. In many cases, participants made comments that the researcher did not expect or did not think of. Obviously, although many mental processes can be traced by looking at the speaker’s linguistic production, there are many others that remain undetectable. In addition, after the recall exercise, the researcher can ask participants general questions about the task. This open discussion gives speakers the opportunity to make comments about the experience that they consider necessary for the researcher to know. In this study, some participants talked about their perception of their own ability to speak the attriting language in relation to their strongest one. Some of them mentioned that they were nervous or tired, or that they in general have a bad memory. Another speaker said that the task was more difficult for him because he did not know the show and he was not familiar with the characters, the setting, etc. As we see in the examples above, the use of stimulated recall protocol in language attrition research provides numerous advantages. However, the use of this methodology also provides some challenges that the researcher should consider. Some may argue issues of validity and reliability, since this methodology relies on the speakers’ retrospective (hence, subjective) reports of their thought process during the task. Also, it should be noted here that this methodology does not address automatic cognitive processes, which are not accessible to consciousness, as it only informs us about controlled, strategic processing. Nevertheless, the possibility to obtain information about strategic cognitive activity is valuable enough in the framework of this study since the goal is to assess the strategies consciously used by speakers to compensate for their incomplete/attriting knowledge of the language. One important factor to be considered when using this methodology is that recall should be carried out as soon as possible after the actual event, in order to
242 Antonio F. Jiménez Jiménez
avoid interference or contamination during the period between the event and the recall. These “verbal reports, elicited with care and interpreted with full understanding of the circumstances under which they were obtained, are a valuable and thoroughly reliable sort of information about cognitive processes” (Ericsson & Simon 1980: 247). In fact, participants in this study were able to make comments on their own narrations, offer explanations for their linguistic choices, make corrections, identify the elements that they were doubtful about, and talk about their previous linguistic knowledge in relation to the task at hand. In other words, this methodology provided critical information about the cognitive processes of language formation by attriting speakers. That which is hidden became visible before the eyes of the researcher. This is particularly important in attrition studies since this phenomenon is all about invisibility. It is not only about what is said, but is also about what is not said. And that which is not said only resides in the head of the speaker. There is no other way to bring that information to light but by asking the speaker directly.
5. Discussion Despite the numerous efforts dedicated to the study of language attrition in the last two decades, language attrition remains one of the most enigmatic areas in the field of linguistics. Traditionally, the language attrition literature has primarily investigated those linguistic elements that are more vulnerable to attrition in relation to those that are maintained for a longer period of time, while not much attention has been paid to the cognitive processes that the speaker engages in before, during, and after the production of these linguistic elements. This research emphasizes the intrinsic relationship between the linguistic and the cognitive lines of development. One of the most fundamental notions of Sociocultural Theory is that “sociocultural and mental activity are bound together in a dependent, symbolically mediated, relationship” (Lantolf & Pavlenko 1995: 109), language being the main tool that mediates this interaction. Through interaction with the community, social speech moves towards egocentric speech, and, in this process, the individual incorporates the higher mental functions that are present in the community into his/ her own cognitive system. These higher mental functions, which include problem solving, voluntary attention, and the planning, executing, and monitoring of the course of action, are social in origin and have been internalized during speaking activity. Egocentric speech moves underground and becomes silent, representing the essence of conscious mental activity, that is, inner speech. The internalization of higher forms of mental activity allows the speaker to gain self-regulation in the
Stimulated recall methodology 243
different activities s/he engages in. However, when the activity becomes difficult, speakers can reaccess previous stages of development in the attempt to regain control (or self-regulation) over the task. In order to do this, speakers can vocalize their inner speech externally so it becomes visible, material, and at the same time, malleable. This form of speech has been labeled in the theory as private speech and it “represents the externalization of what otherwise would remain as covert mental processes (e.g., planning, remembering, learning, etc.) and emerges in the face of difficult tasks” (Appel & Lantolf 1994: 439). Thus, speakers who encounter some difficulty in the course of a certain activity may rely on private speech to solve the problem. In this case, speakers’ internal thinking activity surfaces externally and verbally in an attempt to gain control over their thinking process. In this view, thinking does not always precede speaking. Instead, thinking may also coincide with or even follow speaking. This is manifested in the use of the compensatory strategies employed by the participants in this study in order to gain control over the diverse communicative problems they were facing during the completion of the narration task. In their narrations, students used the language not only to recall the stories conveyed in the video-clip but also to focus on certain aspects of their linguistic production, to solve communicative problems, and to plan, execute and monitor their speech. For instance, self-repetitions can function as a regulatory device to focus on what is coming next. Pauses, segmental lengthening, and a slower speech rate are symptoms of the internal planning of subsequent linguistic production. Self-corrections and self-rephrasing are manifestations of the continuous monitoring of students’ production, while circumlocutions, approximations, codeswitching, literal translations, word-coinages, etc. are linguistic materializations of communicative-motivated problem solving processes. In other words, (self- and object-) regulatory strategies are considered forms of private speech as they represent external manifestations of internal thinking processes. In addition, other-regulatory strategies could also be regarded as manifestations of private speech in this context since dialogic forms of speech were not expected and most of the students’ attempts to initiate a dialogue can be interpreted as addressed to themselves. This idea is supported in Appel and Lantolf ’s (1994) study on how speaking mediates cognitive activity in L1 and advanced L2 speakers of English in the activity of reading and recalling narrative and expository texts. The authors conclude that “not only can planning occur simultaneously with speech, but that the very activity of speaking can, in fact, be planning, or more precisely, thinking, externalized as self-directed private speech, the goal of which is planning what to say about a particular topic” (p. 440). This view is corroborated in the present study as evidenced in students’ use of regulatory strategies as a means for thinking about their own linguistic production. This finding points to the importance
244 Antonio F. Jiménez Jiménez
of conducting studies on language attrition that take into account both levels of development (linguistic and cognitive), since they form an inseparable unity, the two sides of the same coin.
References Andersen, R. W. (1982). “Determining the linguistic attributes of language attrition”. In R. D. Lambert & B. F. Freed (Eds.), The Loss of Language Skills (pp. 83–118). Rowley, MA: Newbury House. Appel, G. & Lantolf, J. P. (1994). “Speaking as mediation: A study of L1 and L2 text recall tasks”. The Modern Language Journal, 78 (4), 437–452. Bialystok, E. (1990). Communication Strategies. Oxford: Blackwell. Channel, J., Sinclair, J. & Carter, R. (1994). Vague Language (Describing English Language). Oxford: OUP. Christenfeld, N. (1994). “Options and ums”. Journal of Language and Social Psychology, 13, 192–199. Cohen, A. D. (1986). “Forgetting foreign-language vocabulary”. In B. Weltens, K. de Bot & T. van Els (Eds.), Language Attrition in Progress (pp. 143–158). Dordrecht: Foris. Cohen, A. D. (1989). “Attrition in the productive lexicon of two Portuguese third language speakers”. Studies in Second Language Acquisition, 11, 135–149. Cohen, A. D. (1996). “Verbal reports as a source of insights into second language learner strategies”. Applied Language Learning, 7, 5–24. Cohen, A. D. (1998). Strategies in Learning and Using a Second Language. London: Longman. de Bot, K. & Clyne, M. (1994). “A 16-year longitudinal study of language attrition in Dutch immigrants in Australia”. Journal of Multilingual and Multicultural Development, 15 (1), 17–28. Dell, G. S. (1986). “A spreading activation theory of retrieval in sentence production”. Psychological Review, 93, 283–321. Ericsson, K. A. & Simon, H. A. (1980). “Verbal reports as data”. Psychological Review, 87, 215– 251. Ericsson, K. A. & Simon, H. A. (1993). Protocol Analysis: Verbal reports as data. Cambridge, MA: MIT Press. Faerch, C. & Kasper, G. (1984). “Two ways of defining communication strategies”. Language Learning, 34, 45–63. Faerch, C. & Kasper, G. (1987). “From product to process – Introspective methods in second language research”. In C. Faerch & G. Kasper (Eds.), Introspection in Second Language Research (pp. 251–266). Clevedon: Multilingual Matters. Gass, S. M. & Mackey, A. (2000). Stimulated Recall Methodology in Second Language Research. Mahwah, NJ: Lawrence Erlbaum Associates. Jiménez Jiménez, A. F. (2004). “A sociocultural approach to language attrition”. In M. S. Schmid, B. Köpke, M. Keijzer & L. Weilemar (Eds.), First Language Attrition: Interdisciplinary perspectives on methodological issues (pp. 61–80). Amsterdam: John Benjamins. Kasper, G. & Kellerman, E. (1997). Communication Strategies. Harlow: Longman.
Stimulated recall methodology 245
Kellerman, E. (1974). “Elicitation, lateralization and error analysis”. York Papers in Linguistics 4. Reprinted in Interlanguage Studies Bulletin 1, 79–114, 1976. Kellerman, E. (1991). “Compensatory strategies in second language research: A critique, a revision, and some (non-) implications for the classroom”. In R. Phillipson, E. Kellerman, L. Selinker, M. Sharwood Smith & M. Swain (Eds.), Foreign/Second Language Pedagogy Research: A commemorative volume for Claus Faerch (pp. 142–161). Clevedon: Multilingual Matters. Kenny, K. D. (1996). Language Loss and the Crisis of Cognition. Berlin: Mouton de Gruyter. Lantolf, J. P. & Frawley, W. (1984). “Second language performance and Vygotskyan psycholinguistics: Implications for L2 instruction”. In A. Manning, P. Martin & K. McCalla (Eds.), The Eleventh LACUS Forum, 1983, (pp. 425–440). Columbia, SC: Hornbeam Press. Lantolf, J. P. & Frawley, W. (1985). “On communication strategies: A functional perspective”. Rassegna Italiana di Linguistica Applicata, 17, 143–157. Lantolf, J. P. & Pavlenko, A. (1995). “Sociocultural theory and second language acquisition”. Annual Review of Applied Linguistics, 15, 108–124. Levelt, W. J. M. (1989). Speaking: From intention to articulation. Cambridge, MA: MIT Press. Levin, H., Silverman, I. & Ford, B. (1967). “Hesitations in children’s speech during explanation and description”. Journal of Verbal Learning and Verbal Behavior, 6, 560–564. Nakuma, C. (1997). “A method for measuring the attrition of communicative competence: A pilot study with Spanish L3 subjects”. Applied Psycholinguistics, 18, 219–235. Olshtain, E. & Barzilay, M. (1991). “Lexical retrieval difficulties in adult language attrition”. In H. W. Seliger & R. M. Vago (Eds.), First Language Attrition (pp. 139–150). Cambridge: CUP. Pavlenko, A. (2004). “L2 influence and L1 attrition in adult bilingualism”. In M. S. Schmid, B. Köpke, M. Keijzer & L. Weilemar (Eds.), First Language Attrition: Interdisciplinary perspectives on methodological Issues (pp. 47–60). Amsterdam: John Benjamins. Poulisse, N. (1990). The Use of Compensatory Strategies by Dutch Learners of English. Berlin: Mouton de Gruyter. Poulisse, N. (1996). “Strategies”. In P. Jordens & J. Lalleman (Eds.), Investigating Second Language Acquisition (pp. 135–166). Berlin: Mouton de Gruyter. Poulisse, N. (1999). Slips of the Tongue: Speech errors in first and second language production. Amsterdam: John Benjamins. Reetz-Kurashige, A. (1999). “Japanese returnees’ retention of English-speaking skills: Changes in verb usage over time”. In L. Hansen (Ed.), Second Language Attrition in Japanese Contexts (pp. 21–58). New York, NY: OUP. Russell, R. A. (1999). “Lexical maintenance and attrition in Japanese as a second language”. In L. Hansen (Ed.), Second Language Attrition in Japanese Contexts (pp. 114–141). New York, NY: OUP. Schmid, M. S., Köpke, B., Keijzer, M. & Weilemar, L. (Eds.) (2004.). First Language Attrition: Interdisciplinary perspectives on methodological issues. Amsterdam: John Benjamins. Schmid, M. S. (2004). “First language attrition: The methodology revised”. International Journal of Bilingualism, 8 (3), 239–255. Sokolov, A. N. (1972). Inner Speech and Thought. New York, NY: Plenum. Tarone, E. (1983). “Some thoughts on the notion of communication strategy”. In C. Faerch & G. Kasper (Eds.), Strategies in Interlanguage Communication. New York, NY: Longman.
246 Antonio F. Jiménez Jiménez
Taylor, I. (1969). “Content and structure in sentence production”. Journal of Verbal Learning and Verbal Behavior, 8, 170–175. Tomiyama, M. (1999). “The first stage of second language attrition: A case study of a Japanese returnee”. In L. Hansen (Ed.), Second Language Attrition in Japanese Contexts (pp. 59–79). New York, NY: OUP. Valdés, G. (2000). Introduction. Spanish for native speakers, Volume 1. AATSP Professional Development Series Handbook for Teachers K-16. New York, NY: Harcourt College Publishers. Vocate, D. R. (1994). Intrapersonal Communication. Different voices, Different minds. Hillsdale, NJ: Lawrence Erlbaum Associates. Vygotsky, L. S. (1962). Thought and Language. Cambridge, MA: MIT Press. Waas, M. & Ryan, A. (1993). “Onomatopoeia in language attrition”. Linguistische Berichte, 148, 477–482. Yoshitomi, A. (1999). “On the loss of English as a second language by Japanese returnee children”. In L. Hansen (Ed.), Second Language Attrition in Japanese Contexts (pp. 80–111). New York, NY: OUP.
Appendix Video-clip transcript (The group of friends is helping Rachel mail out resumes while whistling the theme from The Bridge on the River Kwai.) Ross: Uh, Rach, we’re running low on resumes over here. Monica: Do you really want a job with Popular Mechanics? Chandler: Well, if you’re gonna work for mechanics, those are the ones to work for. Rachel: Hey, look, you guys, I’m going for anything here, OK? I cannot be a waitress anymore, I mean it. I’m sick of the lousy tips, I’m sick of being called ‘Excuse me...’ Phoebe: Oh! Paper cut! Paper cut! Joey: Here, here. Phoebe: Grapefruit juice! Grapefruit juice! Rachel: Okay, guys, we’re almost done. We only got 20 more to go. Ross: Rach, did you proofread these? Rachel: Uh... yeah, why? Ross: Uh, nothing, I’m sure they’ll be impressed with your excellent compuper skills. Rachel: (upset) Oh my Goood! Oh, do you think it’s on all of them? Joey: Oh no, I’m sure the Xerox machine caught a few. Opening Credits [Scene: Central Park, Ross and Chandler are sitting at a table. Rachel is working. Monica and Phoebe enter.] Chandler: You know? I know they call this a love seat but I’m not really feeling anything special towards you.
Stimulated recall methodology 247
(Ross hits Chandler with the paper) Monica: Hey, guys. Chandler and Ross: Hey. Rachel: Hey... hi, ladies... uh, can I get you anything? (to Monica, quietly) : Did you bring the mail? Monica: Lots of responses. Rachel: (to Monica) : Really? (out loud) : Sure, we have scones left! (to Monica) : OK, read them to me. Phoebe: (reading) : Dear Ms. Green, thank you for your inquiry, however... oh... Rachel: (out loud) : We have apple cinnamon... Monica: (reading) : OK... Dear Ms. Green... yeah... yeah... yeah... No. Phoebe: Wow! Rachel: What? Phoebe: (reading) : Your Visa bill is huge! Rachel: (grabs the bill) Give me that! (Camera cuts to Chandler and Ross at table.) Chandler: You know, I can’t believe you. Linda is so great! Why won’t you go out with her again? Ross: I don’t know. Chandler: Is this still about her whole ‘The Flintstones could’ve really happened’ thing? Ross: No, it’s not just that. It’s just – I want someone who... who does something for me, y’know? Who gets my heart pounding, who... who makes me, uh... (begins to stare lovingly at Rachel) Chandler: ...little playthings with yarn? Ross: What? Chandler: Could you want her more? Ross: Who? Chandler: (sarcastically) Dee, the sarcastic sister from Whats Happening. Ross: Look, I am totally, totally over her, OK, I just... (Rachel comes over, Ross lays head on table) : Hiiii! Rachel: Hi! How are you? Ross: We’re fine, we’re fine. Rachel: OK. (walks away) (Ross keeps staring at her, head on table. Chandler smacks him with a newspaper. Joey enters, Ross and Chandler laugh at him.) Joey: Shut up! Chandler: We’re not – we’re not saying anything. Phoebe: What? Ross: Uhhhh... Joey cried last night. Joey: Thank you. Chandler: (to the girls) We were playing poker, alright... Joey: There was chocolate on the three. It looked like an eight, alright? Ross: Oh, guys, you should’ve seen him. ‘Read ’em and weep.’
248 Antonio F. Jiménez Jiménez
Chandler: And then he did. Rachel: Well, now, how come you guys have never played poker with us? Phoebe: Yeah, what is that? Like, some kind of guy thing? Like, some kind of sexist guy thing? Like it’s poker, so only guys can play? Ross: No, women are welcome to play. Phoebe: Oh, OK, so then what is it? Some kind of... you know, like, like... some kind of, y’know, like... alright, what is it? Chandler: There just don’t happen to be any women in our games. Joey: Yeah, we just don’t happen to know any women that know how to play poker. Girls: Oh, yeah, right. Monica: Oh, please, that is such a lame excuse! Rachel: Really. Monica: I mean, that’s a typical guy response. Ross: Excuse me, do any of you know how to play? Girls: No. Rachel: But you could teach us. Guys: No.
Name index
A Abel 129 Abrahamsson 11, 20, 30, 159 Adolpho 124 Alajouanine 129 Altenberg 18, 21 Ammerlaan 16, 18, 21, 63, 126, 142 Andersen 228 Anderson, B. 110 Anderson, M. C. 19, 181, 182 Anderson, R. T. 84, 96 Andersson 84, 96 Appel 243 Argenti 15, 161, 170 Armon Lotem 84 Aronoff 18 As 17, 130, 170, 171 Atkins 126 Au 163 Avesani 129 Avila 3 B Baddeley 16, 19 Badgaiyan 182 Bahrick 17, 65 Baltes 53, 55, 57 Barch 182 Barker 157 Barzilay 18, 84, 228, 233 Bates 3, 84 Bauer 193 Bavelas 193 Bavelier 155 Becerra 183 Belletti 88
Bellugi 157 Ben-Rafael, E. 210, 211 Ben-Rafael, M. 6, 18, 25, 37, 205, 212 Benet-Martinez 193 Benson 183 Berg 12 Berman 126 Bernicot 54 Berwick 101 Best 160 Bever 86, 157 Bialystok 19, 159, 228 Birdsong 11, 55, 155, 156, 158, 159 Blackledge 191 Bochner 193 Bond 193, 194 Bongaerts 129, 159 Bornhövd 183 Borras 183 Borsook 183 Bosch 160 Botvinick 182 Boujon 12 Bourhis 29, 207, 209 Boyd 84 Boym 191 Braet 57 Brauer 129 Braver 182 Breakwell 207, 209 Brebion 57 Brighton 61 Brodsky 190 Bromm 183 Brosseau-Lapré 129
Brown 126 Bruhn-Garavito 110 Bruner 191 Büchel 183 Burton Koch 157 Burzio 88 Bush 182 Bushnell 183 C Canagarajah 84 Caplan 169 Cardinaletti 87 Carpenter 18 Carrier 183 Carroll 20 Carter, C. S 182 Carter, R 228 Casey 182, 185 Castells 210 Channel 227, 228 Chengappa 13, 17 Chiu 193 Cho 182 Chomsky 22, 43, 85, 101, 102, 106 Christenfeld 228 Chung 157 Clahsen 86 Clyne 59, 65, 71, 142, 233 Cohen, A. D. 228, 231, 233 Cohen, J. D. 182 Cole 190 Collette 16, 19 Coltheart 126 Conrad 190 Cook 1, 3, 4, 20, 136, 170
250 Name index
Coolican 195 Cooper 181 Crawford 183 Crone 182 Crosson 14 Cummins 41 Curtis 126 Curtiss 156 Cutler 3, 164 D D’Andrea 63 Damasio, A. 124, 129 Damasio, H. 124, 129 Daneman 18 Dasinger 126 de Bleser 169 de Bode 15 de Bot 9, 10, 15, 16, 23, 30, 53, 56, 58, 59, 63, 71, 129, 136, 142, 233 Dechert 156 DeFliore 183 de Groot 129 Degueldre 183 Dehaene 15, 161, 170 Dekydtspotter 110 Dell 11, 126, 239 Demetriou 54, 58 Deng 129 de Schonen 161 Desrochers 129 Deutch 210 Di Biase 127 DiGirolamo 182 Dijkstra 3, 11, 48 Doise 54 Dorian 71, 72 Doupe 155 Dressler 10 Duncan 183 Duong Phan 60 Dupoux 15, 161, 164, 170 Dussias 19 E Eccard 182 Ecke 17, 191 Edelman 194
Edwards 207 Eichen 156 Ellis, R. 127 Ellis, N. 10, 127 Elman 11 Elovitz 191 Enç 106, 107 Epstein 210 Ericsson 231, 242 Erikson 191 Escudero 41 F Fabbro 12, 14, 97 Faerch 228, 231 Fan 183 Faymonville 183 Felser 86 Filiaci 83, 98, 99 Fissell 182 Fivush 193 Flege 3, 11, 158–160 Fodor 22 Footnick 6, 17, 130, 169 Ford 228 Fox 182 Francis, N. 11, 28 Francis, W. S. 129 Franck 183 Frawley 228–230, 234 Freed 170 French 48 Friedman 57 Fromholt 129 Fromm 17, 130, 170, 172, 178, 180, 181 Furnham 193 G Gabrieli, J. D. E. 181 Gabrieli, S. W. 181 Gantz 157 Gardner 25, 29, 206 Garrett 22 Gass 227, 232, 235 Geijst 211 Genesee 89 George 102 Ghatan 183
Giles 29, 207, 209 Glauche 183 Glöckner 210 Glover 181 Goles 60 Gommans 59, 136 Goodglass 145 Gosling 193 Grabois 15 Grabowski 124 Grainger 11, 129 Green, C. 181 Green, D. W. 12–14, 19, 24, 49, 104, 138 Gregg 41 Grosjean 4, 11, 19, 25, 48, 49, 129, 135–138 Gross, R. 86 Gross, S. 18, 71 Gubrium 191 Gur 183 Gurd 129 Gürel 5, 10, 12, 84, 99–104, 107–110, 117, 126, 136 H Haden 193 Håkansson 18, 84, 127 Hakuta 63, 159 Haller 126 Hansen 65, 126, 128 Harden 56 Hari 183 Haritos 129 Harley 155 Harrington 86 Harris 6, 210 Harzing 193, 194 Haslam 191 Haviland-Jones 192 Hedgcock 65 Heilman 14 Herdina 58 Hermans 129 Hernández 84 Heycock 83, 98, 99 Hiltunen 183 Hirose 164 Hirsch 164
Hirschfeld 23, 24 Hlavac 71 Hoffman 190, 191 Hogg 191 Holowka 129 Holstein 191 Hong 193 Hudon 129 Hulsen 18, 21, 63, 126, 137, 142, 209 Humphreys 14 Hyltenstam 11, 20, 30, 159, 163 I Ijalba 12, 17, 18 Ingvar 183 Irigaray 55 Israel 210 Isurin 15, 84 Itard 156 J Jackendoff 39, 44–46, 48–50 Jacquet 48 Jain 126 Jake 70, 76 Jakobson 9, 127 Jaspaert 136, 137 Jasper 210 Jennings 182 Jessner 58 Jiang 129 Jiménez Jiménez 6, 227, 229 Joanette 129 Johnson 11 Juffs 86 Jun 163 Juncos-Rabadán 126 Jusczyk 44 Just 18 K Kağitçibasi 193 Kahn 191 Kakehi 164 Kalbaugh 192 Kaltsa 83, 84, 92, 96 Kandel 194 Kanno 110
Name index 251
Kaplan 145 Kasper 228, 231 Kaufman 18, 28 Kaushanskaya 193, 194 Kawaguchi 127 Kawasaki 110 Kazmi 157 Kecskés 123 Keijzer 10, 99, 145, 151, 170, 227 Kellerman 228, 232 Kemper 56 Kennaley 56 Kenny 228 Kerkman 126 Kerns 182 Kheimets 210 Kihlstrom 171, 174 Kim, C. S. 157 Kim, J.-W. 157 Kim, K. H. S. 164 Kim, S.-K. 157 King, M. 126 King, N. 191, 194 Kirby 61 Kirsner 126 Kishon-Rabin 157 Klein 55 Klima 157 Klosty Beaujour 129 Knightly 163 Koch 129, 157 Kohnert 84 Köpke 1, 2, 5, 9–13, 18–21, 23–25, 29, 42, 55, 65, 84, 99, 104, 118, 126, 135–137, 140, 150, 151, 165, 170, 227 Kornfilt 102 Kouritzin 191 Krashen 20, 206, 208 Kristeva 191 Kristof 190 Kroon 136, 137 Kuhl, B. 181 Kuhl, P. K. 155 Kundera 189, 190 Kyntay 126 L Laganaro 129
Lambert, R. D. 170, 206 Lambert, W. E. 25, 29 Lamendella 14 Lamy 183 Lantolf 228–230, 234, 242, 243 Larmouth 99 Larsen 129 Larsen-Freeman 58 Lasnik 101, 102 Laufer 211 Laureys 183 LeBihan 15, 161, 170 Lecours 129 Lee, D. S. 157 Lee, J. S. 157 Lee, K.-M. 164 Lee, M. C. 157 Lengyel 155, 158 Lenneberg 11, 155 LePage 207 Levelt 228 Levin 228 Levine 18 Levy 181 Lewis 192 Lhermitte 129 Libby 191, 193–195 Lipski 84 Liu 11 Lock 157 Lockhart 126 Long 155, 206 Lorenzo-Dus 20 Lowie 58 Luria 11 Luu 182 Luxen 183 Lyubansky 210 M Maalouf 190 MacDonald, A. W. 182 MacDonald, A. M. 182 MacDonald, C. A. 79 MacKay 158 Mackey 227, 232, 235 MacLeod 16 MacWhinney 104, 146 Magnusson 54
252 Name index
Major 4, 21 Makoni 56 Malmkjær 129 Malvern 146 Manzini 101 Maquet 183 Marian 193, 194 Marinis 86 Marshall 129 Masgoret 206 Maury 161 Mayberry 156, 157, 160, 164 Maznevski 193, 194 McAdams 191, 194 McCarthy 18 McClelland 11 McConkey Robbins 157 McKee 146 Meara 20, 60, 62–65 Mehler 3, 15, 161, 164, 170 Meuter 14 Mills 191 Milroy 29 Milton 20 Misztal 191 Miyake 46, 51 Molis 159 Montrul 10, 25, 84 Morris 193 Morvan 60 Moskovitch 129 Munro 158 Myers-Scotton 5, 69, 70, 72, 76 Myhill 210 N Nabokov 190 Nakuma 142, 228, 233 Navanen 183 Neisser 65, 191, 193, 194, 195 Nelson 126, 129, 191, 194 Nespoulous 10, 31 Nesselroade 53 Neville 126, 155, 159 Newport 11, 155, 156 Nicoladis 15 Noll 182 Norman 13, 19 Norris 3, 164
Novack 14 Numminen 183 Nystrom 182 O Oakes 191 Obler 13, 17 Ochsner 181 Oh, J. S. 163 Oh, S. H. 157 Olshtain 18, 21, 84, 211, 228, 233 Opitz 20, 128 Osberger 157 Overton Venet 129 Oyama 158 P Paivio 129 Pallier 6, 10, 11, 15, 55, 130, 150, 155, 160–164, 170, 172, 185 Panayiotou 194 Paoli 85 Papadopoulou 86, 92 Papp 123 Paradis 5, 10–12, 14, 16–18, 21, 46, 48, 56, 104, 121, 123, 124, 127, 135, 136, 138, 141, 142 Pater 110 Pavlenko 14, 15, 18, 26, 84, 128, 129, 191, 193, 206, 207, 237, 242 Pelc 165 Penfield 6, 11, 155–158, 160, 162, 170 Pennebaker 193 Perdue 145 Perecman 129 Pereiro 126 Peres 211 Perner 54 Petersen 182 Petersson 183 Petitto 129 Petrovic 183 Phelps 65 Philippaki-Warburton 87 Pienemann 127 Pinker 42, 156, 158 Pinto 88 Platzack 86
Ploghaus 183 Poline 15, 161, 170 Polinsky 99 Polkinghorne 194 Poncelet 16 Poot 129 Posner 174, 182, 183 Poulisse 228, 239 Pratt 56 Prévost 110 Price 183 Prieur 130, 172, 173 Prince 42 Py 25, 30 Q Quante 183 R Radvansky 57 Raichle 182 Raij 183 Rainville 183 Ramírez 126, 129 Ramirez-Esparza 193 Raz 183, Reese, E. 193 Reese, H. W. 53 Reetz-Kurashige 233 Relkin 164 Remennick 210, 211 Richards 146 Rieussec 10 Rizzi 86, 87 Roberts 11, 86, 155 Robertson 181 Rodríguez 126 Ross 182 Rossing 59, 136 Rozenbaum-Tamari 211 Rubin 129, 194 Rumelhart 11 Rush 130 Russell 233 Ryan, A. 228 Ryan, E. B. 56 S Sacks 194
Sanders, L. 82 Sanders, L. D. 126 Sangrigoli 161 Sanz 86 Satterfield 84 Schade 12 Schaufeli 99 Schmid 1, 2, 5, 10, 11, 13, 15, 19, 20, 23, 25, 29, 31–34, 42, 61, 65, 71, 72, 84, 99, 117, 118, 128, 135, 137, 142, 145, 165, 170, 191, 195, 205, 209, 227, 231 Schmitt, E. 18, 136 Schmitt, N. 60 Schrauf 129, 194 Schreuder 33, 129 Schriefers 129 Schumann 14, 25, 128 Schwartz 43 Sebastian-Gallés 160 Segui 3, 164 Seliger 18, 99, 100, 170 Serratrice 85 Shah 46 Shallice 13, 19 Shapiro 183 Sharma 126 Sharwood Smith 5, 10, 19, 20, 22–24, 39, 41–44, 170 Shewell 128 Shoeps 210 Shohamy 210 Shumovich 56 Silva-Corvalan 25 Silverman 228 Simon 231, 242 Sinclair 228 Singleton 155, 158 Slobin 126 Smith, D. J. 126 Smith, K. 61 Smith, M. 126 Smith, P. B. 193 Soesman 59 Sokolov 234 Sona 129 Sorace 83, 85, 86, 95, 100, 101, 109 Speedie 14
Name index 253 Spencer 157 Sperber 23, 24 Spolsky 210 Sprouse 43, 110 Squire 174, 182, 194 Stoessel 16, 63 Stone-Elander 183 Strange 158 Stenger 182 Stroop 182, 183 Swain 23 T Ta’ir 14 Tabouret-Keller 207 Tajfel 191 Tarone 228, 236, 238 Taylor, D.M. 29, 207, 209 Taylor, I. 228 Tees 163 Thomas 182 Tomblin 157 Tomiyama 228, 233 Toupin 126 Tranel 124 Trenerry 14 Truscott 39, 42, 43 Tsimpli 5, 22, 83, 86, 87, 89, 92, 95, 96, 100, 101 Tulving 174 Turian 18 Turner 191 U Ullman 16–18 Ullsperger 182 V Vago 18, 21, 170 Valdés 233 van Buren 10, 20, 22, 23, 24, 44, 170 van der Linden 16, 19 van Geert 58, 192 van Hell 3, 138 van Heuven 48 van Lieshout 54 Vaughan 191
Ventureyra 11, 15, 130, 161–163, 170 Verhofstadt-Denève 57 Verspoor 58 Vocate 234 von Cramon 182 Vygotsky 190, 229, 234 W Waas 65, 84, 208, 228 Walter 190 Wang 155 Ward 193 Warrington 18 Weber-Fox 159 Weilemar 99, 170, 227 Weiller 183 Weinert 54 Welsh 182 Weltens 10, 63, 126 Werker 163 Wertman 14 Wertsch 190 Wewerka 193 Wexler 101 White 89, 100, 101, 110 Wiley 159 Y Yağmur 18, 145, 209 Yang 194 Yeni-Komshian 11 Yoo 11, 130, 161, 170 Yoshitomi 233, 234 Yoshizawa Meaders 192, 196, 201 Yukawa 126 Z Zhang 157 Zimmerman-Philips 157 Zobl 10 Zola 182 Zong 129
Subject index
4-M model 5, 70, 78 early system morphemes 70 late system morphemes 78 A Access 2, 6, 11, 13, 19–21, 24, 48, 49, 62, 74, 82, 84, 125, 132, 141, 142, 145, 149–151, 171, 179, 183, 184, 193, 229, 236 Activation 9, 11–13, 15, 21, 23, 24, 27, 29, 32, 33, 35, 42, 43, 45, 46, 48–51, 63, 99, 104, 118, 121, 124–131, 135, 137, 141 Adoptees 6, 155, 161–165, 170, 171, 184 Afrikaans 72 Age 11, 15, 18, 20, 21, 26–29, 31, 44, 53–55, 57, 58, 66, 72, 76, 78, 79, 84, 108, 126, 130, 134, 143, 144, 147, 148, 155–160, 162, 164, 165, 167, 170–178, 185, 195, 205, 217 Age-regression 171–173 Aging 55–57, 66 Agrammatism 169 Ambiguous sentences 96 Aphasia 10–13, 17, 18, 34, 119, 122, 123, 127, 129, 130, 132, 169 Aptitude 9, 19, 20, 22, 31, 32, 59, 61, 67 Arabic 210 Aspect 16, 20, 21, 23, 26, 97, 109 Attitude(s) 5, 6, 9, 13, 25, 26, 28, 29, 128, 130, 194, 205–209, 211, 212, 216–219, 222–224 Attractor state(s) 60, 61, 63 Automaticity 86
B Behaviourism 41 Biculturalism 192 Bilinguals early bilinguals 6, 18, 84 late bilinguals 18, 194 Bilingual Interactive Activation Model (BIA) 48–51 Binding 12, 17, 84, 99–108, 111, 112, 114–116, 118, 119 Borrowing 219, 220, 222, 237, 240 Brain imaging 18, 161 Brain plasticity 11, 20, 28, 150 C Case, case-marking 70–72, 74, 84, 87, 92–96, 101–103, 105, 107–108 Central executive 16, 19 Cerebral substrate 124 Chaos theory 5 Children 15, 26–30, 55, 92, 127, 137, 144, 149, 156, 157, 160–162, 164, 165, 198, 211, 214–216, 218, 219, 223 Chinese 159, 163 Circumlocutions 230, 231, 238, 243 Clausal frame 75 Cochlear implants 157, 166, 167 Codeswitching 13, 25, 26, 28, 69, 70, 75, 77, 79–82, 138–140, 150, 212, 218, 219, 221, 222, 225, 226, 230, 243 Competence 3, 9, 11, 19, 22, 24–28, 30, 33, 35, 50, 83, 84,
86, 87, 89, 94, 95, 103, 122, 123, 127, 130, 132, 152, 233, 240, 245 Competition 12, 18, 19, 47, 49, 80, 104, 119, 187 Comprehension 17, 18, 33, 34, 45, 66, 86, 121, 125, 178–180, 233 Connectionism 41, 46, 51 Contact(s) 5, 12, 13, 20, 23–25, 28–30, 32–34, 36, 58, 59, 61, 65, 69, 82, 85, 95, 101, 109, 117, 119, 126, 128, 135, 136, 139, 141, 145, 150, 170 Content morphemes 70, 71, 75 Control mechanism(s) 9, 19 Convergence 59, 76, 79, 97 Critical period (see also sensitive period) 10, 55, 155–157, 165–167 Crosslinguistic influence 85 D Dutch 6, 34, 59, 63, 66–68, 71, 82, 132, 137, 142, 143, 151, 152, 195–201, 209, 224, 244, 245 Dynamic system(s) 17, 53, 58, 59, 61, 62, 63, 68, 192, 201, 204 Dynamic Systems Theory (DST) 53, 58, 62, 66, 67 E Education level 20, 81, 144 Elderly 54–57, 148 Electrophysiological studies 126 Emotion(s) 6, 14, 49, 191, 193, 196, 203, 205, 207, 209, 221 Emotional implication 9, 15
256 Subject index
English 5, 6, 12, 33–36, 49, 69, 70–85, 87, 89, 96–119, 123, 127–133, 142, 143, 157–159, 167, 172, 210, 212, 224, 234, 239, 240, 243–246 Ethnolinguistic Vitality 29, 207, 209 Ewe 6, 169, 173 Explicit knowledge 122, 131 F Feelings 195, 196, 199, 217 Film retelling 145, 146 Finnish 96, 119 fMRI 161, 170, 172, 174, 181 Forgetting 16, 17, 19, 31, 33–35, 67, 96, 99, 100, 169–171, 181, 182–184, 191, 202 Fossilization 60 Free speech 142, 145, 148, 149 French 6, 32, 49–51, 55, 69, 118, 123, 127–129, 161–164, 170, 171, 173, 174, 176–179, 189, 211, 212, 214–216, 219, 221, 222 G Garden path sentences 97 German 6, 36, 59, 63, 68, 71, 82, 84, 92–94, 98, 133, 135, 139, 140, 142–145, 149, 152, 186, 189, 190, 195–201, 203, 208, 209, 224 Grammaticality 20, 21, 31, 83, 86, 92, 94, 95 Grammaticality judgement(s) 20, 21, 31 Greek 5, 83, 84, 87–89, 92–95, 97, 98, 101, 119, 167 H Hebrew 6, 25, 34, 35, 131, 205, 209–224 Heritage speakers 84, 137, 227, 232, 233, 236 Hypnosis 6, 17, 130, 131, 169–186
I Identity 6, 26, 75, 189–194, 196, 198, 201, 202, 204, 205, 207, 209, 210, 216, 219, 222 Implicit knowledge 163 Incomplete learning 83 Inhibition 9, 11–13, 15, 19, 23, 28, 29, 32, 33, 49, 104, 126, 135, 141, 142, 150, 151 Inner speech 242, 243 Input(s) 20, 23–25, 29, 30, 44–48, 50, 56, 60–62, 86, 94–96, 122, 140, 157, 158, 206 Instruction(s) 79, 108, 109, 122, 123, 156, 245 Interface, interface phenomena (see also minimalism) 23, 32, 45, 49, 84–88, 94–96, 98, 101, 118, 131 morpho-phonological interface 86 syntax-discourse interface 86, 87, 94, 96 Interference(s) 2–4, 12, 13, 17–19, 21, 31, 36, 86, 94, 95, 97, 99, 103, 119, 121, 122, 125, 127, 128, 131, 132, 138, 160, 165, 169, 182, 186, 187, 219, 223, 242 Interlanguage 36, 118, 208, 230 Interpretability, interpretable features 83, 85, 86 Italian 5, 83, 85, 87, 96–98, 101, 119, 158 J Japanese 118, 123, 128, 129, 133, 134, 161, 166, 170, 172, 178, 181, 192, 245, 246 K Korean 6, 133, 155, 161–164, 170, 171, 184 L L2 influence 2, 4, 36, 84, 104, 108, 245 Language Acquisition Device (LAD) 43, 50 Language change 41
Language contact 5, 34, 85, 95, 101, 135 Language mode(s) 6, 13, 19, 48, 49, 51, 135, 136, 138, 139, 141, 144, 148, 149, 152, 234 Language shift 32, 67, 69, 74, 79, 211, 233 Language use 5, 9, 10, 12, 13, 15, 23, 25, 27, 28, 48, 56, 72, 81, 95, 135, 137–139, 142, 144, 149, 214, 218, 224 Length of residence (LOR) 143, 144, 148, 149 Lexical borrowing 237 Lexical networks 62 Lexical retrieval 21, 31, 36, 84, 130, 151, 165, 228 Lifespan perspective 53, 54 Limbic system 14, 15 Linguistic knowledge 16, 40, 50, 83, 141, 151, 242 Literacy 9, 20–22, 26–28, 128, 133, 233 Logical Form (LF) 85, 86 M Maturation 1, 6 Memory 6, 9, 11, 15–18, 20–22, 27, 28, 31, 32, 34–36, 45–50, 54, 56, 57, 59, 66, 67, 97, 121, 122, 125, 127, 129–133, 136, 167, 170, 174, 180–182, 184, 186, 187, 193, 195, 201, 203, 204, 208, 232, 235, 241 declarative memory 15–18, 27, 121, 122, 127, 129 episodic memory 35, 49, 50 explicit memory 36, 174 implicit memory 174 procedural memory 17, 18, 27, 121, 122, 125, 127, 130 working memory (WM) 16, 18, 19, 31, 32, 34, 46, 47, 54, 56, 57 Mental grammar 41 Mental lexicon 41, 48, 63, 70 Metalinguistic 21, 28, 43, 51, 122, 130 Metalinguistic ability 43
Migrants 6, 25, 33, 34, 55, 59, 67, 71, 72, 132, 152, 189, 192, 194, 196, 201, 207, 209, 210, 212, 223 Mina 171, 173–181 Minimalism, minimalist program (see also interface) 83, 85 Minority language 58, 207 Modularity 37, 39, 42, 132 MOGUL 5, 39, 42–46, 48–50 Motivation 6, 15, 20, 25–29, 58–60, 72, 87, 121–123, 128, 130, 163, 205, 206, 208, 209, 211, 219, 222–224 instrumental motivation 6, 25, 26, 206, 219 integrative motivation 222 Multicompetence 4, 20, 32 N Null pronoun(s) 33, 118, 103 Null subject(s) 84–88, 101, 103 Null subject languages 86–88 O Object-regulated, objectregulation 228–230, 234, 236 Online processing 16, 21, 30, 86, 94 Online tasks 86 Optionality 22, 86, 96 Other-regulated, otherregulation 228, 229 Overt pronoun(s) 102, 103 Overt subject(s) 84, 86, 87, 101 P Parsing 33, 86 Performance 7, 17, 19, 22, 24, 25, 28, 30, 33, 40, 44, 45, 47, 51, 57, 61, 83, 86, 94–96, 112, 142, 144, 148, 149, 152, 159, 161, 162, 164, 174, 187, 209, 235, 240, 245 Phonetic Form (PF) 85, 86, 87, 94
Subject index 257
Phonetic perception 37, 133, 168, 187 Phonology 27, 32, 41, 121–124, 127, 131, 233 Phonotactics 164 Plasticity 9, 11, 15, 20, 27–29, 36, 150, 155, 156, 159, 160, 166, 167, 186 Plural 77, 145, 151 Polyglot aphasia 11–13, 17, 18, 119 Portuguese 69, 244 Pragmatics 14, 49, 98, 122, 128, 130 Prestige 69, 74, 150, 207, 208 Pro-drop 97 Production task 89, 92 Proficiency 1, 17, 18, 20, 29–31, 67, 72, 74, 109, 131, 133–135, 139, 144, 155–159, 164, 165, 193, 207, 214, 233 Pronominals 101, 107, 118 Puberty 55, 157, 159, 162, 165 R Recall 80, 115, 125, 126, 171, 174, 195, 203, 227, 228, 231–235, 237–239, 241–244 Recast(s) 208, 230, 231, 237 Reduction 1, 54, 56, 58, 62, 72, 182, 183 Reflexives 99, 100, 104, 106, 107, 110, 112–116 Regression 9, 17, 31, 33, 34, 67, 72, 130, 131, 133, 148, 149, 171–173, 185, 186 Regulation 14, 35, 227–230, 236, 242, 243 Relative clause 97 Repetitions 142, 145, 146, 192, 228, 230, 231, 243 Restructuring 4, 25, 28, 30, 99, 100, 103, 108, 118, 170 Retraining 163 Russian 6, 25, 26, 36, 119, 123, 205, 211, 212, 216–226
S Selective, selectivity 48, 49, 83–86, 99–101, 104, 127 Self-corrections 146, 228, 230, 234, 239 Self-regulated, selfregulation 228–230, 234, 242, 243 Sensitive period (see also critical period) 167 Sentence generation 21 Shift 5, 10, 32, 55, 67, 69, 70, 72, 74, 75, 77, 79, 81, 86, 87, 131, 151, 194, 211, 230, 233 Simplification 49, 71, 137 Skills 17, 18, 21, 23, 26, 55–57, 65, 125, 156, 174, 189, 193, 197, 198, 201, 233, 245, 246 comprehension skills 18 conversation skills 233 cross-cultural skills language skills 56, 57, 65, 156, 191 linguistic skills 56, 57 literacy (reading, writing) skills 26, 55 productive skills, production skills 23 oral skills 55 receptive skills 23 Slips-of-the-tongue 239 Social Networks 35, 58, 137 Social Network Theory 29 Sociocultural Theory 6, 37, 227, 242 Spanish 6, 17, 20, 25, 33, 36, 65, 66, 84, 97, 133, 152, 160, 227, 233, 234, 236–240, 245, 246 Specific language impairment 18 Spontaneous speech 21 Strategies avoidance strategy/strategies 231, 235, 236, 238 compensatory strategy/ strategies 227–233, 235, 243 object-regulatory strategies 230
258 Subject index
other-regulatory strategies 230, 243 self-regulatory strategies 230 Subcortical structures 14 Subset Model 104, 117 Subsystems 41, 54, 59, 121, 123, 124, 126, 127, 130 Swedish 20, 84, 92–94, 171, 172, 181 Syntactic interference 97 T Task dependency 9, 21, 22
Tense 76, 97, 151, 162 Transfer 1, 2, 84, 100, 101, 103, 107, 132, 133, 228 Translation equivalents 124, 125, 127 Trauma, traumatic experiences 15, 130 Turkish 5, 12, 33, 37, 84, 99–111, 114, 115, 117–119, 152, 209, 225 U Ultimate attainment 31, 43, 44, 98, 155, 205, 207, 219 Ungrammaticality 94
Universal grammar (UG) 2, 42, 44, 46, 61, 85 V Verbal fluency 145, 146 W Word recognition 50, 131, 161 X Xhosa 5, 69, 70, 72, 75–81
In the series Studies in Bilingualism (SiBil) the following titles have been published thus far or are scheduled for publication: 35 Rocca, Sonia: Child Second Language Acquisition. Expected October 2007 34 Koven, Michèle: Selves in Two Languages. Bilinguals' verbal enactments of identity in French and Portuguese. ix, 329 pp. Expected September 2007 33 Köpke, Barbara, Monika S. Schmid, Merel Keijzer and Susan Dostert (eds.): Language Attrition. Theoretical perspectives. 2007. vii, 258 pp. 32 Kondo-Brown, Kimi (ed.): Heritage Language Development. Focus on East Asian Immigrants. 2006. x, 282 pp. 31 Baptista, Barbara O. and Michael Alan Watkins (eds.): English with a Latin Beat. Studies in Portuguese/Spanish – English Interphonology. 2006. vi, 214 pp. 30 Pienemann, Manfred (ed.): Cross-Linguistic Aspects of Processability Theory. 2005. xiv, 303 pp. 29 Ayoun, Dalila and M. Rafael Salaberry (eds.): Tense and Aspect in Romance Languages. Theoretical and applied perspectives. 2005. x, 318 pp. 28 Schmid, Monika S., Barbara Köpke, Merel Keijzer and Lina Weilemar (eds.): First Language Attrition. Interdisciplinary perspectives on methodological issues. 2004. x, 378 pp. 27 Callahan, Laura: Spanish/English Codeswitching in a Written Corpus. 2004. viii, 183 pp. 26 Dimroth, Christine and Marianne Starren (eds.): Information Structure and the Dynamics of Language Acquisition. 2003. vi, 361 pp. 25 Piller, Ingrid: Bilingual Couples Talk. The discursive construction of hybridity. 2002. xii, 315 pp. 24 Schmid, Monika S.: First Language Attrition, Use and Maintenance. The case of German Jews in anglophone countries. 2002. xiv, 259 pp. (incl. CD-rom). 23 Verhoeven, Ludo and Sven Strömqvist (eds.): Narrative Development in a Multilingual Context. 2001. viii, 431 pp. 22 Salaberry, M. Rafael: The Development of Past Tense Morphology in L2 Spanish. 2001. xii, 211 pp. 21 Döpke, Susanne (ed.): Cross-Linguistic Structures in Simultaneous Bilingualism. 2001. x, 258 pp. 20 Poulisse, Nanda: Slips of the Tongue. Speech errors in first and second language production. 1999. xvi, 257 pp. 19 Amara, Muhammad Hasan: Politics and Sociolinguistic Reflexes. Palestinian border villages. 1999. xx, 261 pp. 18 Paradis, Michel: A Neurolinguistic Theory of Bilingualism. 2004. viii, 299 pp. 17 Ellis, Rod: Learning a Second Language through Interaction. 1999. x, 285 pp. 16 Huebner, Thom and Kathryn A. Davis (eds.): Sociopolitical Perspectives on Language Policy and Planning in the USA. With the assistance of Joseph Lo Bianco. 1999. xvi, 365 pp. 15 Pienemann, Manfred: Language Processing and Second Language Development. Processability theory. 1998. xviii, 367 pp. 14 Young, Richard and Agnes Weiyun He (eds.): Talking and Testing. Discourse approaches to the assessment of oral proficiency. 1998. x, 395 pp. 13 Holloway, Charles E.: Dialect Death. The case of Brule Spanish. 1997. x, 220 pp. 12 Halmari, Helena: Government and Codeswitching. Explaining American Finnish. 1997. xvi, 276 pp. 11 Becker, Angelika and Mary Carroll: The Acquisition of Spatial Relations in a Second Language. In cooperation with Jorge Giacobbe, Clive Perdue and Rémi Porquiez. 1997. xii, 212 pp. 10 Bayley, Robert and Dennis R. Preston (eds.): Second Language Acquisition and Linguistic Variation. 1996. xix, 317 pp. 9 Freed, Barbara F. (ed.): Second Language Acquisition in a Study Abroad Context. 1995. xiv, 345 pp. 8 Davis, Kathryn A.: Language Planning in Multilingual Contexts. Policies, communities, and schools in Luxembourg. 1994. xix, 220 pp. 7 Dietrich, Rainer, Wolfgang Klein and Colette Noyau: The Acquisition of Temporality in a Second Language. In cooperation with Josée Coenen, Beatriz Dorriots, Korrie van Helvert, Henriette Hendriks, Et-Tayeb Houdaïfa, Clive Perdue, Sören Sjöström, Marie-Thérèse Vasseur and Kaarlo Voionmaa. 1995. xii, 288 pp. 6 Schreuder, Robert and Bert Weltens (eds.): The Bilingual Lexicon. 1993. viii, 307 pp. 5 Klein, Wolfgang and Clive Perdue: Utterance Structure. Developing grammars again. In cooperation
4 3 2 1
with Mary Carroll, Josée Coenen, José Deulofeu, Thom Huebner and Anne Trévise. 1992. xvi, 354 pp. Paulston, Christina Bratt: Linguistic Minorities in Multilingual Settings. Implications for language policies. 1994. xi, 136 pp. Döpke, Susanne: One Parent – One Language. An interactional approach. 1992. xviii, 213 pp. Bot, Kees de, Ralph B. Ginsberg and Claire Kramsch (eds.): Foreign Language Research in CrossCultural Perspective. 1991. xii, 275 pp. Fase, Willem, Koen Jaspaert and Sjaak Kroon (eds.): Maintenance and Loss of Minority Languages. 1992. xii, 403 pp.