Russia and the Kostanay region of Kazakhstan (predominantly Russian-speaking region). All immigrants spoke Russian as L1 and completed secondary ...
L1 Attrition without L2 Acquisition: A case of Russian-speaking Immigrants in Israel
Liubov Baladzhaeva
THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE MASTER DEGREE
University of Haifa Faculty of Humanities Department of English Language and Literature
July 2013
L1 Attrition without L2 Acquisition: A case of Russian-speaking Immigrants in Israel
By: Liubov Baladzhaeva Supervised by: Prof. Batia Laufer
THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE MASTER DEGREE
University of Haifa Faculty of Humanities Department of English Language and Literature
July 2013
Approved by:_______________________Date:_____________ (Supervisor) Approved by:_______________________Date:_____________ (Chairperson of M.A. Committee)
Acknowledgments First and foremost, I want to thank my advisor Prof. Batia Laufer for giving me her help and guidance and for inspiring me. I am grateful to my mother who always supported me and did everything to make my education possible. I also want to thank my husband, who made sure I had enough time and resources to work on my thesis, and who gave me his love and support during the hardest times.
ii
Table of Contents Abstract List of Tables 1. Introduction 2. Literature Review 2.1 Definition of first language attrition 2.2 Attrition versus language loss, language shift and incomplete acquisition 2.3 Theoretical Frameworks of L1 Attrition 2.3.1 Regression 2.3.2 Simplification 2.3.3 Interlanguage and cross-linguistic influence 2.3.4 Universal grammar 2.3.5 Sociocultural theory 2.4 Areas of L1 attrition 2.4.1 Lexicon 2.4.2 Grammar 2.4.3 Pragmatics 2.4.4 Phonology/phonetics 2.4.5 Competence versus performance 2.5 Factors affecting L1 attrition 2.5.1 Age at the onset of attrition 2.5.2 Length of Residence 2.5.3 L2 level 2.5.4 L1 and L2 exposure and use 2.5.5 Attitudes 2.5.6 Educational level 2.6 Attrition of Russian as L1 2.6.1 Attrition in lexicon 2.6.2 Attrition in grammar 2.6.3 Attrition in pragmatics 2.6.4 L1 attrition of Russian in Israel 2.7 Chapter summary 3. Pilot Study 3.1 Aim of the Study 3.2 Research Questions 3.3 Method 3.4 Participants 3.4.1 The first immigrant group – with knowledge of Hebrew 3.4.2 The second immigrant group – without knowledge of Hebrew 3.4.3 The control group 3.5 Materials 3.5.1 Questionnaire 3.5.2 Picture description 3.5.3 Grammaticality judgment test 3.6 Results 3.6.1 Research question 1 3.6.2 Research question 2 iii
IV VI 1 3 3 4 5 5 6 7 8 9 10 10 12 13 14 15 15 16 16 17 18 19 20 21 21 23 24 25 28 30 30 30 30 31 32 32 32 33 33 35 36 38 38 39
3.6.3 Research question 3 3.6.4 Research question 4 4. Main Study 4.1 Aim of the Study 4.2 Research Questions 4.3 Method 4.4 Participants 4.4.1 The first immigrant group – with knowledge of Hebrew 4.4.2 The second immigrant group – without knowledge of Hebrew 4.4.3 The control group 4.5 Materials 4.5.1 Questionnaire 4.5.2 Lexical retrieval tasks 4.5.3 Irregular verb formation task 4.5.4 Future tense formation task 4.5.5 Collocations task 4.6 Results 4.6.1 Research question 1 4.6.2 Research question 2 4.6.3 Research question 3 4.6.4 Research question 4 4.6.5 Research question 5 4.6.6 Research question 6 4.6.7 Research questions 7 5. Discussion 5.1 Attrition of vocabulary 5.2 Attrition of grammar 5.3 Attrition of collocations 5.4 Time on task and level of confidence 5.5 Factors affecting attrition 5.6 Language attrition versus language change 5.7 Summary 6. Conclusion and suggestions for further research 7. Bibliography Appendix A – Pictures used in the picture description task in the pilot study Appendix B – Sentences used in the grammaticality judgment task in the Pilot Study Appendix C – Sentences used in the production of future tense task in the Main Study Appendix D – Sentences used in the recognition and production of collocations task in the Main Study
iv
40 40 42 42 42 42 45 46 46 46 46 46 50 51 52 53 54 54 55 55 56 57 57 59 62 62 65 68 69 71 74 75 76 78 90 95 97 98
L1 Attrition without L2 Acquisition: A case of Russian-speaking Immigrants in Israel
Liubov Baladzhaeva
Abstract First language (L1) attrition in late bilinguals has been attributed to the influence of the second language (L2), insufficient maintenance of L1, length of residence in the L2 environment, and the age of attrition onset (Altenberg & Vago, 2004; Gurel, 2008; Pavlenko, 2010; Schmid, 2007, 2011). There is little research showing that a limited knowledge of L2 may also contribute to some attrition of L1 (Yagmur et al., 1999), but no research on whether L1 attrition is possible when immigrants have no knowledge of L2. My study investigates whether Russian immigrants in Israel with little or no knowledge of Hebrew (L2) experience attrition of Russian (L1). It consists of two parts. In the first part, in the pilot study, three groups of adult Russian L1 speakers (16 immigrants with no knowledge of Hebrew, 44 immigrants with knowledge of Hebrew, and 21 monolingual controls from Russia and Kazakhstan) were compared on lexical retrieval, correctness judgment of collocations and correctness judgment of complex grammatical constructions. In the second part, in the main study, three groups of adult Russian as L1 speakers matched in age and education (30 immigrants with no knowledge of Hebrew, 37 immigrants with knowledge of Hebrew, and 21 monolingual controls, residents of Russia and Kazakhstan) were compared on lexical retrieval, production of irregular verbs, production of future tense of regular verbs and correctness judgment of collocations. ANOVAs and Scheffe‘s post-hoc tests were used for each of the comparisons. I also collected sociolinguistic variables of the participants and performed correlational analyses between the attrition test scores and each one of the variables. Results of lexical retrieval tasks in the main study and production of irregular verbs showed no attrition. On correctness judgment of collocations in both parts, and on the lexical retrieval in the pilot study, participants with no knowledge of Hebrew performed no differently than participants with knowledge of Hebrew, and both performed significantly worse than the controls. Moreover, on the production of future tense of regular verbs, and on the correctness judgment of complex grammatical constructions, immigrants without knowledge of Hebrew performed even worse than immigrants with knowledge of Hebrew. On the v
production of future tense of regular verbs, both groups performed significantly worse than the controls. Test results correlated positively with Hebrew proficiency and with the amount of usage of Hebrew. The age of the onset of attrition correlated negatively with some results in the main study. Correlations with the length of residence in Israel, maintenance of Russian and attitudes towards Russian and Hebrew were not significant. The conclusion is that immigrants with no L2 knowledge can experience just as much, or even more, attrition of L1 as immigrants with L2 knowledge. These results are explained in light of cross-linguistic influence theory. Bilingual immigrants experience L1 attrition due to simplification, diminished use of L1 Russian, and influence from L2 Hebrew. Monolingual immigrants experience L1 attrition due to the influence of attrited language of the bilinguals. Lack of metalinguistic awareness prevents monolingual immigrants from recognizing mistakes in their L1. Paradoxically, knowledge of L2 and higher L2 proficiency may have a positive effect on the maintenance of L1. Several basic notions in the attrition research are challenged by the results of the study particularly, the notion that only bilinguals can experience L2-induced L1 attrition. It was also found that higher education, higher age at time of immigration, positive attitudes towards L1 and maintenance of L1 do not prevent L1 attrition.
vi
List of Tables
Table 1 Table 2 Table 3 Table 4 Table 5 Table 6 Table 7 Table 8 Table 9 Table 10 Table 11 Table 12 Table 13 Table 14 Table 15 Table 16 Table 17 Table 18
Differences between the three groups on the vocabulary production test (ANOVA results) Differences between the three groups on the collocation correctness judgment test (ANOVA results) Differences between the three groups on the complex grammatical construction correctness judgment test (ANOVA results) Correlations of sociolinguistic factors with the test results on all sample (Spearman‘s test results) Hebrew knowledge self-evaluation scale The list of verbs with the irregularities Differences between the three groups on the lexical retrieval tasks (ANOVA results) Differences between the three groups on the irregular verb production task (ANOVA results) Differences between the three groups on the future tense production task (ANOVA and Scheffe‘s results) Differences between the three groups on the collocations task (ANOVA and Scheffe‘s results) Differences between the three groups on time on task (ANOVA results) Differences between the three groups in their degree of confidence (ANOVA and Scheffe‘s results) Correlation of the degrees of confidence with the results. The entire sample. (Spearman‘s results) Correlation of the degrees of confidence with the results. The control group. (Spearman‘s results) Correlation of the degrees of confidence with the results. +Hebrew group. (Spearman‘s results) Correlation of the degrees of confidence with the results. -Hebrew group. (Spearman‘s results) The effect of sociolinguistic variables on the results. Two immigrant groups. (Spearman‘s results) The effect of sociolinguistic variables on language attitudes. Two immigrant groups. (Spearman‘s results)
vii
39 39 40 41 47 51 55 55
56 57 58 58 58 59 59 60 61
1. Introduction This study was inspired by personal experience. Four years ago I immigrated to Israel from Yekaterinburg, Russia. I was always proud (sometimes on the verge of showing-off) of my knowledge of the Russian language. I could deliver speeches, write long texts, and switch between styles and registers. I was immensely proud of my large vocabulary and impeccable spelling and punctuation. This pride might seem strange to some people, but I felt that my language was always my main (and sometimes only) asset to make a living in this world. Then, a couple of years ago, I noticed that something happened to me. I started to make grammar and spelling mistakes, I was not always able to remember a word I needed, and sometimes I simply ―invented‖ words that do not exist in Russian. My friends from Russia started telling me in our Skype talks that I had a foreign accent, and some Russian tourists assumed that I was actually a heritage speaker of Russian, that is, someone who is born to immigrant parents, but is raised in a second-language environment. What was happening to me, I wondered? Is it even possible to forget one‘s own language? Is it possible to become a foreigner in a native language, and in such a short time, too? Then I came across the term ―first language attrition‖. First language (L1) attrition is defined as ―the loss of, or changes to, grammatical and other features of a language as a result of declining use by speakers who have changed their linguistic environment and language habits‖ (Schmid, 2011, p. i). It has been attributed to the influence of the second language (L2), insufficient maintenance of L1, length of residence in the L2 environment, and the age of attrition onset (Altenberg & Vago, 2004; Gurel, 2008; Pavlenko, 2010; Schmid, 2007, 2011). It is not yet a well-researched phenomenon, and the reasons for attrition and processes associated with it are not yet entirely clear. While some people might assume that one needs to live for a long time in L2 environment in order to experience L1 attrition, research (as well as the idiosyncratic evidence from my and my family‘s experience) shows that L1 attrition may start in the early years after the immigration (Frenck-Mestre, 1993; Kopke &Schmid, 2004; Pavlenko & Jarvis, 2002). However, the common assumption is that for the attrition process to start, one needs to have a good L2 knowledge (Seliger & Vago, 1991), or, at least, some L2 knowledge (Beganovic, 2006; Yagmur, 1999, 2010). Here, my personal experience did not support the existing theories of L1 attrition. The Russian-speaking community constitutes 1/6 of Israel‘s population, and reaches 40% in some cities (Remennick, 2004). The Russian social network is well developed and practically all the services that one might need are offered in Russian. This makes it possible for an immigrant to exist in a monolingual Russian-speaking world, without any contact with Hebrew and, therefore, without 1
any need to learn it. As a bilingual speaker, involved in the Russian community‘s life, I often had contact with monolingual Russian-speaking immigrants when I volunteered to translate or offer information about bus schedules, official documents and so on. I noticed that not only bilinguals experienced difficulties in their native language, but monolinguals did too. Similarly to the bilingual immigrants, they made grammatical mistakes, had trouble retrieving certain words, and sometimes even developed a Hebrew accent. However, these were only individual observations, and I decided to embark on a study in order to examine whether these observations would be supported experimentally. The objective of this thesis is to find out whether monolingual immigrants can experience first language attrition, and whether they experience it to a different degree than bilingual immigrants. My immigrant sample consists of Russian-speaking immigrants in Israel, both monolingual and bilingual, which were compared to each other and to a control group of monolingual residents of Russia and Kazakhstan. I decided to focus on attrition in lexicon and grammar and to investigate whether attrition is related to various sociolinguistic factors. I used the methodology from first and second language attrition research, and developed specific tasks for the study. The thesis is structured as follows: in the second chapter, I examine current theories of attrition, research on attrition in general, and on Russian in particular. I also describe the linguistically unique Russian-speaking community in Israel. In the third chapter, I present the pilot study, and in the fourth chapter, the main study. In the fifth chapter, I discuss the results from both studies and offer an explanation in light of cross-linguistic influence theory. In the final chapter, I draw conclusions of the study in light of the current view of first language attrition, describe the limitations of the study, and offer possibilities for further research.
2
2. Literature Review 2.1 Definition of first language attrition The phenomenon of attrition has been researched now for a few decades (Kopke and Schmid, 2004), but it still offers more questions than answers. It is not yet absolutely clear what factors affect the process of attrition, what language components are more prone to attrition, and in which order they attrite. There are several definitions of L1 attrition that address the different perspectives of the phenomena. For example, Monica Schmid, one of the most prominent researchers of L1 attrition, emphasizes the lack of L1 use as a cause of attrition: ―‗Language attrition‘ describes the loss of, or changes to, grammatical and other features of a language as a result of declining use by speakers who have changed their linguistic environment and language habits‖ (Schmid, 2011, p. I; my italics). Waas (1996) suggests that language attrition is a natural process that happens in ―language contact situations where one language is not maintained by its speakers‖ (pp. 29-30). Yagmur (1997) points out that the definition might depend on the focus of the researcher. The focus can be on the causes of attrition (sociological aspects), on what is lost (linguistic aspects) or how it was lost (psycholinguistic aspects). Unlike Schmid, who accepts not only loss of, but also changes in language structures, Altenberg and Vago employ a more strict definition: a ―loss of language abilities of non-disordered individuals in a second language (L2) environment‖ (Altenberg and Vago, 2004, p. 105). Gurel (2008), on the other hand, does not mention language loss and restricts attrition to L2-induced L1 change or restructuring in an individual speaker‘s grammar, not taking into account the changes in lexis. Some researchers also attribute lexical changes not to L1 attrition, but to borrowing and/or code-switching, not considering them a manifestation of L1 loss (see, for example Pavlenko, 2004). Kopke (2004a), on the other hand, accepts changes in different areas of L1, but emphasizes four necessary features of L1 attrition: it is non-pathological, intragenerational, individual, and affecting linguistic competence and/or performance. In this study, the definition of L1 attrition is based on Kopke‘s and Schmid‘s definitions mentioned above, except for the declining use of the language, since I will specifically address the question whether L1 attrition is possible in a healthy individual without any decline in language use. Thus, by first language attrition throughout the thesis I will mean Intragenerational non-pathological loss of, or changes to features of a native language of immigrant speakers which result in changed linguistic competence and/or performance.
3
2.2 Attrition versus language loss, language shift and incomplete acquisition It is important to distinguish between language loss and language attrition. While in attrition, some language components might be lost, overall the language remains. Language loss refers either to a loss of a language by a community or to a loss resulting from pathological states in individuals such as aphasia. Non-pathological language loss can be associated with language shift – a whole community shifts from its original L1 to another language in subsequent generations. Such a shift can happen in immigrants as well as in indigenous minorities (Boot & Weltens, 1991). In a community, language shift and language loss are intergenerational phenomena, that is, the change is gradual and continual over generations, finally resulting in a switch to a different language than L1. First language attrition can only be intragenerational (Kopke & Schmid, 2004, de Bot, 2001); L1 in this case stays the same but may lose its dominance. While a complete language loss does not equal L1 attrition, the claim about the loss of some parts of a language as constituting attrition can be challenged as well: as it often happens with immigrants who return to their native country, the lost and forgotten may come back. However, there is evidence that adopted children may forget their L1 completely and even lose the ability to distinguish the sound of L1 from other languages (Ventureyra &Pallier, 2004). Nevertheless, Kopke (2004a) claims that ―nothing allows us to state that a language can actually be lost in the psychological sense of the term‖. She refers to the study in which a long-lost language was recovered under hypnosis (Fromm, 1970). While language loss and language attrition are certainly different phenomena, they might have some common features. Maher (1991) describes the communal L1 attrition, and, according to her, languages in communities isolated from the main L1 community undergo the following changes: reduction in the number of allomorphs resulting in more invariable forms, fewer contextsensitive rules, and increased paradigmatic regularity, replacement of synthetic forms by analytic ones or by periphrastic constructions, progressive reduction in inflectional morphology, entailing less flexible word order, preference for coordinate rather than embedded constructions, and distinctive aspectual constructions in verbal systems. These changes are common for the gradual language loss of both Indo-European and nonIndo-European languages, for example, for such different languages as Finnish and Dyirbal (an aboriginal language of Australia). Some of these phenomena occur in first language attrition as well (see section 1.4).
4
Most of the attrition studies focus on individuals whose attrition started when they were adults. In the case of children who moved to an L2 environment before the onset of puberty we should speak about incomplete acquisition rather than attrition (Kopke, 2004a; Kopke & Schmid, 2004, Kaufman, 2001). This topic will be discussed in more detail when talking about the age of the onset of attrition (see section 1.5.1).
2.3 Theoretical Frameworks of L1 Attrition So far, theoretical explanatory frameworks have scarcely been used in applied research of L1 attrition. Schmid (2002) has identified four linguistic models that could be used in the interpretation of L1 attrition data in some individual studies: regression, simplification, interlanguage, and universal grammar. Later Jimenez Jimenez (2004) made an attempt to interpret L1 attrition from the point of view of the sociocultural theory.
2.3.1 Regression Regression hypothesis for language loss was first proposed by Roman Jakobson (1968). He suggested that language loss, specifically in aphasia, might mirror language acquisition. Actually, it was later found that unlike language acquisition, aphasia is not progressive, but affects linguistic competence only partially, and depends on the type of brain injury, not on the external linguistic factors (Kopke & Schnid, 2004). At the same time, there is evidence that older people do exhibit L1 loss, sometimes mirroring L1 language acquisition, or that they can exhibit L2 language loss and revert to their L1 after years of using L2 (de Boot & Weltens, 1991). Silva-Corvalan (1991) notes that such language loss mirrors not only first and second language acquisition, but also development in creolization. Unlike aphasia, attrition is a gradual process, and it might actually be more similar to language acquisition than aphasia (de Bot & Weltens, 1991), especially since evidence was found in SLA research that L2 attrition mirrored L2 acquisition (Hansen, 1999). Seliger and Vago (1991) claim that ―the phenomenon of first language attrition can be seen as a natural continuation of a general language acquisition‖ (p. 12). Two versions of regression in first language attrition have been proposed: chronological what is learned last would be lost first, and reinforcement - what is learned better or repeated more often would be preserved the longest (Kopke & Schmid, 2004).
5
So far, evidence for regression has been found in L1 attrition of Turkish complex syntactic constructions in immigrant L1 speakers of Turkish in Australia (Yagmur, 2004). Complex syntactic constructions, namely relative clauses, are acquired relatively late in Turkish, only at school. Yagmur (2004) found that such constructions can attrite even in the language of educated TurkishEnglish bilinguals, who received higher education in Turkish and had plenty of practice with relative clauses. Yagmur was able to establish the order of acquisition in syntax and to support the idea of L1 attrition as regression, but in some cases L1 acquisition order is not that clear. De Boot and Weltens (1991) point out that applying regression hypothesis to the studies of L1 attrition of lexicon might be problematic, because lexicon acquisition is not gradual, rule-governed and in a fixed order. Overall, the regression hypothesis has not been thoroughly tested in L1 attrition studies, therefore we do not have enough evidence to make conclusions whether it can be used as a theoretical framework for first language attrition.
2.3.2 Simplification Simplification is considered to be an internally induced attrition, a natural tendency in a language. Marked forms, which are more ―complex‖, attrite faster than unmarked ones, a language becomes more regular, and its norms – more general and uniform (Seliger & Vago, 1991). One of the approaches to simplification suggests that L1 attrition follows the principles of language change. For example, in a study of L1 attrition of Dutch in South Africa, Raidt (1997) found that the intragenerational L1 changes were similar to the early stages of Afrikaans‘ development. Schmid (2002) notes that changes in a language that underwent simplification are more likely to happen in the following areas: reduction in registers due to reduction in functions, reduction of the lowfrequency items in the lexicon, and reduction in the morphological complexity leading to a more analytical language structure. However, Kopke and Schmid (2004) claim that at the moment ―‗simplification‘ or ‗reduction‘ is merely a label with no explanatory power‖ (p. 17). According to them, there are no clear, uniform and well-developed definitions of what is ―complex‖, and there is not enough research of the simplification process. Schmitt (2004) also suggests that more in-depth models will have to be developed and applied to attrition data in order to investigate the simplification hypothesis. Kopke and Schmid (2004) also point out that simplification theory does not take into account L2 influence, and all the changes that are considered to be manifestations of simplification are language-internal and induced by lack of use. Also, it is not certain whether simplification is 6
possible when there is still contact with L1. In the era of modern media it is highly unlikely to lose all contact with L1. Sharwood Smith & Van Buren (1991) even claim that the simplification processes might not be induced by the lack of L1 use, but rather by the changes in the learner‘s perception of the basic structure of his or her L1 grammar. While it is hard to believe that simplification alone can be a manifestation of attrition, it may be one of the processes that happen in L1 attrition. Kaufman (2001) claims that L1 attrition is the interplay of two different processes: an internal L1 change and an external influence of L2.
2.3.3 Interlanguage and cross-linguistic influence Cross-linguistic influence theory has been accepted by many researchers in the field as an explanation of language attrition (see for example Altenberg, 1991; Pavlenko, 2004; Kaufman and Aronoff, 1991; Isurin, 2000; Gurel, 2002; Leisio, 2001, Pelc, 2001, Seliger, 1991). The assumption is that in the absence of L1 input, a speaker can take L2 as a source of positive evidence, and as a result of that, L2 rules could replace that of L1, especially if L2 rules are less complex or have a wider application (Kopke & Schmid, 2004; Seliger, 1991). According to the same researchers, such language transfer might be the most prominent source of L1 attrition (Sharwood Smith, 1983). Altenberg (1991) suggests that L1 and L2 should be similar in order for L2 to L1 transfer to happen. There is some evidence that this is indeed the case (Altenberg, 1991; Kopke, 2004a; Pelc, 2001). For example, Altenberg (1991) found that in L1 attrition of German in an L2 English environment, gender was less affected by attrition, and attributed it to the fact that English has no gender for nouns, therefore there was no competition between the two languages in this area. It is suggested that there would be more transfer effects if L2 formed a superset of the L1, for example, when in L1 some construction is prohibited, and in L2 it is optional (Gurel, 2008; Ribbert & Kuinen, 2010). In such situations, positive evidence from L2 could make a bilingual speaker assume that the construction is not prohibited in L1 as well. For example, Ribbert and Kuinen (2010) found that L1 speakers of German living in the Netherlands made mistakes in the cases that were ungrammatical in German, but grammatical in Dutch. They concluded that areas of L1 that are similar to L2, but have a slight difference, are more prone to attrition. Schmid (2007) suggested that frequency of retrieval might play a role in the attrition of L1 lexicon: the most frequently used L2 elements will replace their less used L1 counterparts. However, similarity or frequency of use may not be the only factors influencing linguistic domains such as lexicon, which may be more prone to attrition than the other domains, for example, morphology (Kopke & Schmid, 2004). Kopke and Schmid (2004), 7
though, warn us that this conclusion could simply be due to the fact that lexical attrition has been investigated more often than the other kinds. Recently, a new hypothesis has emerged. It suggests that bilinguals do not have two separate linguistic systems, in which one language is ‗normal‘ and the other one is ‗deviant‘, but can simultaneously process both languages as an interlanguage (Schmid and Kopke, 2007, Cook, 2003). Thus, the use of the term ―attrition‖ itself is questioned because the changes in L1 might not be a simple loss of language, but a sign of transition to a bilingual language system. Schmid and Kopke (2009) even suggest that L1 attrition may be a phenomenon that is experienced by all L2 users, from the earliest stages of L2 development. It is possible that attrition is merely the emerging part of the iceberg of bilingualism (Schmid & Kopke, 2009). Cook (2001) proposed the idea of multi-competence. Dewaele & Pavlenko describe multicompetence as ―a constant state of flux both within and between individuals‖ (p.137). They compare the two languages of a bilingual mind to two liquid colors that blend unevenly. Bilinguals might differ from native speakers in both of their languages. For example, bilinguals are found to be less fluent than monolinguals both in semantic and formal fluency and slower in lexical retrieval tasks (Schmid & Kopke, 2009). It is unclear, then, to what degree poorer performance on verbal fluency tasks in L1 can be ascribed to general bilingualism effect, and to what degree it is an outcome of attrition. Seliger and Vago (1991) claim that there is a need to distinguish between L1 attrition and code-mixing or code-switching, which can be an additional communication strategy of bilinguals and not an inability to retrieve words of L1. In sum, L1 of a speaker may undergo some sort of attrition under the influence of L2 in order to produce a bilingual or multi-competent system. Laufer (2003) claims that such attrition is a small price to pay for becoming a multilingual.
2.3.4 Universal grammar The models of Universal Grammar that are applied in L1 attrition are usually based on L2 acquisition research (Kopke & Schmid, 2004). These models have been applied to L1 attrition only recently, but some evidence supporting Universal Grammar as an explanatory framework of attrition has been documented (Gurel, 2002, 2004; McCormack, 2004, Sorace, 2004; Tsimpli, 2007; Tsimpli et al, 2004). According to the minimalist program (Chosmky 1995, 2004), language includes the computational system and two interface levels, Phonetic Form and Logical Form. While the computational system can include non-interpretable features, in the interfaces only interpretable features can be found. In the process of L1 acquisition, the setting of language parameters in the 8
mind happens, which involves the selection of features and their values in a way that leads to syntactic language variation (Chomsky 1986, 1995). Computational system, with its noninterpretable features, is assumed to be immune to L1 attrition since it is not based on evidence from language input. Phonetic Form and Logical Form, on the other hand, can be influenced by L2 input when it is conflicting with L1 settings (Tsimpli, 2007). L1 attrition may manifest itself as indeterminacy or optionality in the use of interpretable morphosyntactic features in the contexts where interface conditions in L1 and L2 differ (Sorace, 2004; Tsimpli, 2007; Tsimpli et al, 2004). For example, L1 attrition of null versus overt pronouns has been documented in Italian, Greek, Turkish, and Japanese in L2 English settings (Gurel, 2002, 2004; McCormack, 2004, Tsimpli et al., 2004).
2.3.5 Sociocultural theory Sociocultural theory applied to bilingualism and language attrition connects social participation, linguistic production, and psychological functioning (Jimenez Jimenez, 2004). The theory can be applied to different levels of language: macro level, micro level, and individual level (Kopke, 2004a). Macro level is concerned with the impact of the political and institutional factors on language behavior. Micro level links language choices with social organization and discourse practices. Individual level deals with bilingual conversation and interprets it according to an individual speaker‘s actions and intentions (Kopke, 2004a). In sociocultural theory, language acquisition is perceived as attaining control over language. Until this control is gained, learners remain other-regulated, that is, they rely on the help of others and/or turn to the culture artifacts such as dictionaries, books, computers, etc., which makes them object-regulated. The ultimate goal is to achieve self-regulation in a language through language acquisition. Language attrition in the sociocultural theory is expressed as a loss of the once attained level of self-regulation in a language (L1 or L2) and a return to a previous stage of object- and/or other-regulation (Jimenez Jimenez, 2004). While there are theoretical attempts to explain attrition from the point of view of the sociocultural theory, at the moment there are virtually no L1 attrition studies that use it. Kopke (2002) suggested that the differences between two groups of German immigrants in France and Canada in her study may be attributed to sociocultural differences. However, the design of the study did not allow her to validate this suggestion.
9
2.4 Areas of L1 attrition L1 attrition may manifest itself in practically all language areas: lexicon, grammar, pragmatics, and phonetics. In the next few subsections I will talk about each area in detail. There are some claims regarding which language areas attrite more and faster than others. For example, Sharwood and Smith (1991) speculate that ―a certain type of knowledge might be more liable to attrition than another. Massive lexical loss might be accompanied by minor syntactic loss, surface morphology might change faster than more basic grammatical features, pragmatically conditioned aspect of word order may change before purely syntactic aspects‖ (p.19). Perhaps the degree of vulnerability to attrition can be established in grammar. There is evidence, for example, that content morphemes are more vulnerable than early-system morphemes, and they, in turn, are more vulnerable than late-system morphemes (Gross, 2004). On the other hand, there were studies that found no attrition of grammar in speakers who resided in an L2 environment for a long time (Gurel, 2007; Porte, 2003). The researchers attributed these results to extensive and frequent L1 use. While attempts to establish the order of attrition in grammar have been made, at the present state of knowledge, and without longitudinal studies, it is almost impossible to make comparisons across different linguistic areas with respect to the speed and degree with which they will be affected by attrition (Schmid & Kopke, 2009).
2.4.1 Lexicon L1 attrition in lexicon and semantics has been widely researched and documented (see, for example, Boyd, 1993; Grabois, 2000; Jaspaert & Kroon, 1992; Latomaa, 1998; Olshtain & Barzilay, 1991; Pavlenko & Jarvis, 2002; Van Hell & Dijkstra, 2002). There is a consensus that lexicon is a vulnerable and sensitive part of the linguistic system (Kopke, 2002) and that it is more affected by L1 attrition than other language areas or that attrition affects lexicon faster and more noticeably than other areas (Schmid, 2011). Schmid (2011) attributes a higher degree of attrition in lexicon to the fact that the lexicon is huge and not all its items are frequently used. Lexicon changes all the time, even in a monolingual community, and different languages have cognates or words borrowed from other languages. Because of its natural ability to borrow and assimilate foreign words, L1 lexicon might be more prone to L2 influence. On the other hand, even a quite high degree of L1 attrition in lexicon might not be so devastating. Since lexicon is much larger than other areas of language knowledge, and the items in it 10
are far less densely connected and less interindependent than in grammar and phonetics, it can tolerate certain amount of changes (Schmid & Kopke, 2009). As I have already mentioned, it is, nevertheless, not absolutely clear whether lexicon is indeed more prone to L1 attrition. It might be that the disproportionately high number of studies on L1 attrition of lexicon, compared to the few studies on other language areas, makes us jump to this conclusion. The most common manifestation of L1 attrition in the lexicon is the difficulty in lexical retrieval – slower retrieval, inaccurate responses, or inability to retrieve a word. Almost every study that addresses this question finds evidence for this type of attrition (Amerlaan, 1996; Latomaa, 1998; Olshtain & Barzilay, 1991; Stoessel, 2000; Waas, 1996, Yagmur, 1997). Picture-naming appears to be a valid measure for detection of lexical retrieval difficulties. Lexical retrieval difficulties can also induce overuse of pauses and fillers, repetitions, and retractions (de Leew, 2007; Schmid, 2007; Schmid & Beers Fägersten, 2010). Lexical recognition tasks, on the other hand, show much less attrition (Amerlaan, 1996; Hulsen, 2000; Schmid &Kopke, 2009). Another manifestation of lexical attrition is reduced lexical diversity, which is usually measured by type-token ratio and lexical frequency profiles (Schmid, 2002; Laufer, 2003). Semantic changes can be a sign of L1 attrition in lexicon as well. For example, attrition of lexical semantics can manifest itself in semantic extension, when a particular meaning of a polysemous L2 word is assigned to an L1 word that shares some, but not all, meanings with an L2 word (Pavlenko & Jarvis, 2002). A different relationship between signifier and signified in L1 might appear as a result of L2 influence. For instance, a signifier might incorporate several signified objects in addition to its original one signified. This can happen if in L2 these different objects are considered to be representations of one more general object, while in L1 these objects are considered to be separate. An example of this was found in L1 Russian under the influence of L2 English, when bilinguals changed categorization of drinking vessels in Russian to the system more similar to English (Pavlenko & Malt, 2011). The networks of associations between words can change as well under the influence of L2, and new associations can develop or existing associations can become more similar to L2 (Grabois, 2000, Yoshida, 1990). These phenomena constitute convergence of L1 and L2 values, distinctions, and boundaries. Such changes may result in L1 becoming more like L2, or in a creation of concepts, domains, or systems that are distinct from both L1 and L2. While there are many studies that documented L1 attrition of lexicon, some researchers are concerned whether current methodology allows capturing real L1 attrition. Schmid (2011) notes that 11
while such popular tasks for researching L1 lexicon attrition, such as picture-naming, are easy to administer, since they do not resemble real-life speech and their results might not accurately reflect the real state of a speaker‘s L1. She suggests that using more time-consuming procedures, such as type-token ratios and detailed lexical analysis of free speech, which would give a more realistic picture of attrition in lexicon (Schmid, 2011).
2.4.2 Grammar Since grammatical elements are categorical, there is far less room for optionality in the grammar repertoire than in the sound system or lexicon (Schmid, 2011). This makes the analysis of L1 attrition of grammar more straightforward than that of other language areas. There is a consensus that L2-induced change in grammar can happen only if L1 and L2 have competing equivalent forms (Altenberg, 1991; Kopke, 2002, 2007). In a situation of disproportion between L1 and L2 input, when L2 is used much more frequently than L1, L2 constructions can replace those in L1 partially or completely. Evidence of such replacement was found in numerous studies (Altenberg, 1991; Kopke, 2002, 2007; Ben-Rafael, 2004; Gurel, 2004, 2007, 2008, Paradis, 2007). Some researchers claim that this replacement can only happen when L2 constructions are simpler than those of L1 (Andersen, 1982; Seliger, 1989, 1996; Schmid, 2002). Sharwood Smith and Van Buren (1991), on the other hand, claim that even more complex L2 structures can replace the competing structures of L1 because it is easier to process one type of structure for two languages than a distinct type for each language. The mind of a bilingual speaker, according to the researchers, tends to economize and use the structure that is more frequently seen in language input for both languages. Insufficient L1 input and output can also lead to difficulties in processing complex L1 forms that do not correspond directly to L2 and are not supposed to compete with anything in L2 (Gurel, 2008). Gurel (2002, 2008) elaborates on competition in grammar of L1 and L2 in a bilingual mind and suggests that when L2 grammar is more strict and conservative than that of L1, no attrition is expected. For example, if L1 allows two optional constructions, but in L2 only one of them is acceptable, the construction that is allowed in both languages will not attrite. Gurel does not think that it is a sign of attrition that the construction that exists only in L1, and is not a valid option in L2, might disappear. However, reducing the options in L1 or overcorrecting can be considered attrition as well, and attriters can judge perfectly grammatical optional constructions of L1 as ungrammatical simply because L2 grammar is stricter. For example, Balcom (2003) found that French L1 speakers, 12
while choosing between two optional constructions in French, tended to judge grammatical sentences as ungrammatical because their L2 English allowed only one option out of these two. The most frequent types of L1 attrition in morphosyntax are simplification or reduction of morphologically-marked distinctions or categories (Altenberg, 1991, de Bot, Gommans & Rossing, 1991; Gurel, 2008; Pavlenko & Jarvis, 2002; Waas, 1996). Pavlenko (2003) found reduced case system, disappearance of verbal aspect and incorrect use of the verbs of motion in L1 Russian attrition (Pavlenko, 2003). L1 Finnish attriters tended to change different word orders of highly inflectional Finnish to standard English SVO order (Jarvis, 2003). Ben-Rafael (2004) found attrition of conditional mood in L1 French in Israel (it became Hebrew-like), loss of subjunctive mood, and omission of indefinite articles that do not exist in Hebrew. Gurel (2004) found loss of binding features for the overt pronoun in Turkish that had a corresponding form in English. For example, a pronoun ―o‖ [s/he] in Turkish placed after subject in a sentence refers not to this subject, but to someone else. Bilinguals in the study misinterpreted sentences because they attributed features of English pronouns to Turkish pronouns (Gurel, 2004). Another set of pronouns in Turkish was also examined, and these pronouns had no corresponding forms in English. No attrition was found in their interpretation (Gurel, 2004).
2.4.3 Pragmatics Research on L1 attrition of pragmatics is only at the very beginning. L1 attrition in pragmatics may manifest itself in the loss of pragmatic competence, that is, in inability to interpret particular intentions of a speaker. For example, Kyoko Mori, a Japanese immigrant to the U.S., describes that she can no longer distinguish a real invitation from the symbolic ones in Japanese and to choose a correct level of politeness in a conversation with other Japanese people (Mori, 1997). Waas (1996) found a lack of idiomatic expressions and proverbs in the speech of L1 German attriters in Australia. Latoma (1998) and Tao and Thompson (1991) found L1 attrition in listener‘s responses and strategies. For example, the attriters used backchannels signals such as ―aha‖, ―uhu‖, ―yeah‖, etc., to indicate that they were listening a hundred times more often than their monolingual counterparts (Tao & Tompson, 1991). Particular pragmatic distinctions can attrite if they are not a part of the L2 community norms (Priven, 2002; Valdes & Pino, 1981). L1 attrition of pragmatics is, therefore, a result of crosslinguistic differences in pragmatic norms, assimilation into L2 culture, and extensive daily exposure to L2 pragmatic norms.
13
2.4.4 Phonology/phonetics While achieving native-like pronunciation is probably the hardest part of second language acquisition for adult learners, and even most proficient L2 speakers often retain some L1 accent, it does not mean that L1 phonology is immune to attrition. Studies in L1 attrition of phonology provide evidence that the human perceptual system remains relatively flexible throughout life and can undergo modifications in response to sensory input (Pavlenko, 2000). L1 attrition was found in such areas of phonology as intonational patterns, allophonic realization of phonemes, and diphtongization of vowels (Andrews, 1999; Seliger & Vago, 1991). Speakers who are proficient in two languages can merge certain sounds together if they are similar but not identical in their L1 and L2 (Flege, 1987; Major, 1992; Mennen, 2004). Even not very proficient L2 speakers can demonstrate some phonological attrition. For example, Bond Chang (2010) studied L1 English learners of Korean and found that the English sounds that were similar to Korean changed during the intensive elementary Korean classes and became more Korean-like. These changes were usually measured by special equipment and software, but they could also be heard and recognized by a ―naked‖ ear. For instance, immigrant speakers of L1 German were perceived as non-native speakers by German listeners (Schmid, 2002; De Leeuv, Schmid & Mennen, 2010). On the other hand, in a similar study, no significant difference was found between the L1 German attriters and a control group (Hopp & Schmid, 2011). Some studies also investigated the changes in phonological perception of L1 attriters. Ventureyra et al. (2004) and Cancila et al. (2005) found that attrition might affect the ability to distinguish certain L1 sounds. Major and Baptista (2007) found that L1 attrition can affect how a foreign accent in L1 is perceived by attriters – speakers of Portuguese residing in the US found nonnative speakers of Portuguese with American English accents sounding less foreign than the native speakers that still live in Brazil did.
2.4.5 Competence versus performance An important question in attrition research is whether L1 attrition entails a change in the knowledge of L1 (competence) or a change in the control of this knowledge (performance). L1 attrition may occur only in performance, or in both performance and competence, but not to the same degree. In performance, attrition may lead to simplification or impoverishment, retrieval and processing difficulties, and insecurity of the speaker manifested by frequent hesitations, self-repair, or hedging strategies (Schmid & Kopke, 2009). While performance is the easiest to test, certain 14
elicitation techniques can also make competence relatively accessible to language attrition research (Sharwood Smith & Van Buren, 1991). Sharwood Smith and Van Buren (1991) suggest that there are two stages of L1 attrition. In the first stage there is a progressive loss of control over an unchanged L1 competence, and during the second stage a new L1 competence is developed in which the previous L1 knowledge is lost or changed. In L1 attriters, performance is usually affected more than competence. For example, lexical recognition tasks that test lexical competence show much less attrition than lexical retrieval tasks that test performance (Amerlaan, 1996; Hulsen, 2000; Schmid &Kopke, 2009).
2.5 Factors affecting L1 attrition Different extralinguistic factors may affect the degree of L1 attrition. Schmid (2011) divides such factors into three categories:
factors related directly to language development (level of L2, exposure to L2, use of
factors related to a speaker‘s overall development (age, gender, education, length of
L1);
residence in a new country);
internal and psychological factors (identity, attitudes and emotions).
Some of these factors are well-studied, while others have not been investigated and we can only make assumptions about their influence on L1 attrition. In the subsections that follow, I am going to discuss the factors that have been examined in studies of L1 attrition, namely age at the onset of attrition, the length of residence in L2 environment, the level of L2, L1 and L2 exposure and use, the attitudes towards L1 and L2 and the educational level of attriters.
2.5.1 Age at the onset of attrition There is evidence that until a certain age a language is not yet fully formed (see, for example Nicoladis & Grabois, 2002; Pallier et al., 2003; Ventureyra & Pallier, 2004), and so most researchers claim that in studies of L1 attrition there should be a clear difference between speakers who experienced L1 attrition as children and whose language system has not yet stabilized, and those who experienced L1 attrition as adults (Kopke, 2004a; Kopke & Schmid, 2004, Kaufman, 2001). Those two categories of speakers should never be treated as one group. Several studies demonstrate that in children, first language can attrite dramatically (Isurin, 2000; Schmitt, 2004; Nicoladis & Grabois, 2002), and the younger the child is at the onset of attrition, the more and faster 15
L1 will attrite (Kopke & Schmid, 2004). Bylund (2009), in the study of L1 attrition of Spanish, examined speakers who immigrated to Sweden between the ages of 1 and 19. He concluded that if migration happened before the age of L2, L1 could deteriorate dramatically or could even be lost completely, while after age L2, L1 became immune to attrition and it did not matter how old someone was at the moment of migration. Similarly, Olshtain (1986) suggests that literacy plays a role in resistance to attrition. The most severe attrition in children takes place before the age of 8-9 years (Kopke & Schmid, 2004), which is often also the age when literacy skills are acquired. Literacy skills, including computer literacy, can protect L1 from some L2 influence and add means of contact with L1 in the immigration setting. The age of an attriter as a factor affecting attrition is sometimes investigated in relation to elder immigrants. While there are claims, or rather ‗myths,‘ about language reversion in older immigrants from L2 to L1, there is no real evidence to support these claims (Schmid, 2011). The only large-scale study that tried to investigate this phenomenon did not find any deterioration of L2 or reversion to L1 (Crezee, 2008).
2.5.2 Length of Residence There is a consensus that attrition is a gradual process that should worsen with time. However, it is not quite clear how many years in L2 environment are needed for the process of L1 attrition to start. It is also not clear what the stages of L1 attrition are. Most of the studies investigating attriters that resided in L2 environment for 10 or more years found little or no effect of length of residence on L1 attrition (Ammerlaan, 1996; Tsimpli et al., 2004; Brown, 2001; de Bot & Clyne, 1994; Gurel, 2002). Therefore, it was suggested that, perhaps, L1 attrition sets in during the first ten years, or between the first five and ten years; after that the language system stabilizes and does not deteriorate anymore (De Bot & Clyne, 1994; Schmid, 2011). Some studies offer evidence that attrition may set in even earlier than after 5 years in L2 environment. They say that switch in language dominance may occur after a relatively short period of time, even after 3 years in L2 environment (Frenck-Mestre, 1993; Kopke &Schmid, 2004; Pavlenko & Jarvis, 2002). Ribbert and Kuiken (2010) found that at least in the case of cognate languages, L2-induced changes in L1 can appear very soon after immigration. L1 attrition in their study group was especially striking, since these students did not intend to stay in the Netherlands for a long time and had plenty of opportunities for contact with German, including frequent visits to Germany. The researchers 16
postulate that if L1 and L2 are closely related, as in the case of German and Dutch, L1 attrition is very likely to happen and can happen very fast (Ribbert & Kuiken, 2010). Nevertheless, a few studies did find significant effect of the length of residence on L1 attrition (de Bot, Gommans & Rossing, 1991; Kopke, 1999; Laufer, 2003). For example, Laufer found significant negative correlation between length of residence and L1 attrition of collocation knowledge, and a big difference in lexical richness between new immigrants and those who spent 20 years in L2 environment. There were also studies that reported that L1 system may remain surprisingly stable even after decades in L2 environment (Hutz, 2004; Jaspaert & Kroon, 1992; Schmid, 2002). Furthermore, it was suggested that the length of residence might have an effect on L1 attrition only when there is severely limited or no contact with L1 (de Bot at al., 1991; Kopke & Schmid, 2004; Soesman, 1997). Clearly, such a situation of complete isolation from L1 is hardly possible at the age of globalization.
2.5.3 L2 level Seliger and Vago (1991) describe several ways in which advanced L2 proficiency can affect L1: transfer, interference, convergence and cross-linguistic influence. Speakers may start extending L2 grammar rules to L1. They may also extend the meaning of L1 words on analogy with L2, or produce loan translation, when idiomatic L2 phrase or expression is translated literally to L1 (Seliger & Vago, 1991). Most of the L1 attrition studies focus on fluent speakers of L2. While the logical assumption would be that only advanced L2 can cause L1 attrition, some studies suggest that this might not necessarily be the case. Schmid and Kopke suggest that L1 attrition may be a phenomenon that is experienced by all L2 users, from the earliest stages of L2 development (Schmid & Kopke, 2009). Indeed, research of L1 attrition showed that even immigrants with very limited knowledge of L2 can exhibit signs of L1 attrition (Beganovic, 2006; Yagmur, 1999, 2010). For example, Yagmur (1999), in his analysis of Turkish L1 attrition in Australia, found attrition of relative clause formation in speakers with both high and low proficiency in English. Moreover, he reported that on several tests, speakers with limited knowledge of English performed significantly worse than the speakers with advanced knowledge of English. However, this might be attributed to the fact that his group of participants with limited knowledge of English spent more time in Australia than the second group, and was older and less educated. Beganovic (2006) found that participants that resided in the country for less time and had lower L2 proficiency suffered greater L1 attrition than those who spent 17
significantly more time in the new country and spoke fluent L2. The findings of Bylund, Abrahamsson and Hyltenstam (2010) provide a possible explanation for this. They found that higher language aptitude correlated strongly with lower L1 attrition. They suggest that higher language aptitude may have a compensatory function in L1 attrition, helping to maintain high level of L1 even when contact with L1 is significantly reduced. Perhaps, the poor performance of the non-proficient L2 speakers on the tests could be attributed to their lower language aptitude.
2.5.4 L1 and L2 exposure and use It is argued that L1 attrition occurs when two processes happen simultaneously - dominance of L2 and insufficient L1 input (Schmid &Kopke, 2009, Seliger & Vago, 1991). Increased use of L2 not only leads to less time spent on L1 use, but also inhibits L1 (Levy et al., 2007). A neurolinguist, Paradis (2007), offers a biological view of L1 attrition and suggests that it is a result of long-term lack of L1 stimulation. De Bot, Gommans and Rossing (1991) in their study on L1 attrition of Dutch found that proficiency in Dutch did not change over time when there were many contacts with Dutch, but it deteriorated linearly over time when there were only few contacts with Dutch. They also found a significant connection between the length of residence in the L2 environment and lack of L1 contact. Those who stayed in L2 environment longer gradually lost contact with L1. Influence of L1 and L2 input on L1 attrition is a controversial issue. Several studies support the theory that insufficient use of L1 leads to its attrition (de Bot et al., 1991; Jarvis, 2003; Isurin, 2000; de Leeuw, 2009; Nicoladis & Grabois, 2002; Olshtayn & Barzilay, 1991; Ventureyra & Pallier, 2004). Laufer (2003) found that the maintenance of L1 Russian was related to lesser L1 attrition. Other studies reported that there was no link between the lack or insufficient contact with L1 and attrition (Jaspaert & Kroon, 1989). Sometimes contact with L1 had even negative effect on the linguistic performance in native language (Hulsen, 2000; Kipp, 1999). Schmid (2007), in an attempt to resolve this controversy, claims that not all types of L1 use are helpful in maintaining L1. In her study, L1 use did not seem to predict L1 attrition. She suggests that, perhaps, in L1 attrition the frequency and recency of activation play a less prominent part than in other areas of bi- or multilingualism. Her theory is that L1 attrition might depend not on the frequency of L1 use, but on the fact that the monolingual mode demands the inhibition of other languages. She suggests that bilinguals who have experience in inhibiting L2 while in monolingual L1 mode are more likely not to make mistakes in L1. For example, someone who is used to 18
controlling their L1 speech in formal situations will make fewer mistakes in L1 than a person who is not used to any kind of L1 control, or a person who controls their L1 and inhibit L2 only in informal situations, such as communications with monolingual relatives and friends (Schmid, 2007). Schmid (2011) also claims that the quality and the amount of L1 use are not that important to L1 maintenance as long as the speaker wants to maintain it and feels affection towards it.
2.5.5 Attitudes Attitudes are important in second language acquisition and in interaction with other factors, such as intelligence and language aptitude, which may predict the ultimate attainment of a second or a foreign language (Schmid, 2011). As there are many parallels between L2 acquisition and L1 attrition, one would assume that attitudes might also be important in L1 attrition, but there were only few studies that investigated this. Different languages of a bilingual speaker may also serve different emotional functions and sometimes this may influence the process of attrition (Pavlenko, 2002, 2005; Dewaele, 2004). When an L1 is a highly prestigious language in an L2 environment (such as English in Israel, for example), its maintenance is highly valued both by the immigrants and by the L2 society, hence less attrition is to be expected (Olshtain & Barzilay, 1991). According to the ethnolinguistic vitality theory, a group can have a special identity that makes it likely to behave as a distinctive and active collective entity in inter-group situations (Giles et al., 1977). L1 attrition researchers suggested that strong ethnolinguistic vitality would prevent L1 attrition (Hulsen, 2000; Yagmur, 1997, 2004), but there is no evidence supporting this hypothesis. Yagmur (1997, 2004) found no correlation between subjective EV measures and linguistic performance, and neither did Hulsen (2000). While Schmid (2002) found that extraordinary situations, such as persecutions, might cause strong emotional responses that influence the attrition of the language of the persecutors, two studies on relationship between attitudes and L1 attrition did not find any connection (Schmid & Dusseldorp, 2010; Waas, 1996). Waas (1996) found that attitudes did influence self-perception of a speaker, but they did not have an effect on actual L1 proficiency. There is a very important problem with studying the influence of attitudes on L1 attrition concerning the fact that attrition is a process that happens over time. Schmid (2011) assumes that what determines the degree of L1 attrition are the speaker‘s attitudes towards L1 and L2 at the moment of emigration. However, attitudes can change over time, and the first encounter of a researcher with an attriter usually happens years after the emigration. Schmid (2011) thinks that 19
research on attitudes may be less problematic in second language acquisition than in L1 attrition, because less time usually passes from the beginning of L2 study. According to her, opinions of L1 attriters about their native language are a very intimate and sensitive issue related to their identity, family, immigration experience, etc. This may prevent the attriters from discussing their attitudes openly. In L2 learners, on the other hand, language attitudes are much less of a sensitive issue and researchers are able to get easier access to these attitudes. Longitudinal studies of the influence of attitudes on L1 attrition are needed in order to obtain a clearer picture, but such studies have not yet been conducted.
2.5.6 Educational level Educational level may correlate with language aptitude and general level of intelligence, so it is not surprising that it correlates with the degree of L1 attrition. Pelc (2001) defined the level of education as the number of years of education received in L1, and this variable correlated positively with linguistic performance. A higher educational level was found to prevent attrition to some extent (Jaspaert & Kroon, 1989; Kopke, 2004a; Yagmur, 1997; Waas, 1996). Yagmur (1997) also found that less educated speakers performed worse on the language tasks. However, there is a problem of how to determine whether an educational level affects specifically L1 attrition or linguistic performance in general (Kopke & Schmid, 2004).
2.6 Attrition of Russian as L1 There are large Russian-speaking diasporas in many countries. However, so far most of the studies on attrition of L1 Russian were conducted in the USA (e.g. Isurin, 2007; Pavlenko, 2000, 2003, 2005; Shishkin, 2010). These studies mostly focused on attrition in a lexical area. Some research was also conducted in the attrition of grammar and pragmatics. Also, changes in intonations patterns were observed (Lesio, 2001), but no systematic studies on phonological attrition were conducted. Some of the studies were conducted with adult attriters, incomplete acquirers, and heritage speakers as one group, and the conclusions were based on the results of the whole sample. Therefore, sometimes it is not clear whether the results can be attributed to attrition or incomplete acquisition. Both attrition in competence and performance have been observed in speakers of L1 Russian. Most studies usually investigated L1 attrition in educated speakers that were highly proficient in L2. However, higher education and positive attitudes towards L1 did not necessarily 20
provide immunity to L1 attrition. Isurin (2007), in her investigation of L1 Russian speakers who teach Russian as a foreign language in the US, showed that even people who are exposed to L1 daily and have a strong motivation to preserve their L1 may demonstrate attrition in L1 performance. Only a few studies investigated attrition of L1 Russian in Israel (Feldman, 1996; Laufer, 2003; Ben-Rafael and Schmid, 2007). All of the studies conducted in Israel were very diverse in the language area of investigation and methodology, which makes comparison across the studies quite difficult.
2.6.1 Attrition in lexicon Most of the studies on L1 attrition of Russian focused on attrition in the lexicon, similarly to most of attrition studies in other languages. The following changes were found in the lexicon of Russian speakers: word-for-word translation of collocations from L2 (Laufer, 2003, Pavlenko, 2010), reduced lexical diversity (Laufer, 2003) and reduced lexical fluency (Ben-Rafael and Schmid, 2007; Feldman, 1996), confusion of semantically close prepositions (Zemskaya, 2001), substitutions of some verbs by more general verbs such as ―delat‘‖ [to do, to make] and ―brat‘‖ [to take] (Zemskaya, 2001), and changes in lexical semantics (Leisio, 2001; Pavlenko & Malt, 2011; Zemskaya, 2001). Perhaps, one of the most prominent signs of lexical attrition in L1 Russian found in the studies is the use of lexical calques from L2 both as single words and as collocations. There is, for example, evidence that Russian-speaking immigrants tend to produce incorrect verb-noun collocations, where instead of the required verb, the L2 calque is used, and when such collocations are encountered, they are judged as correct (Pavlenko, 2010; Laufer, 2003). Zemskaya (2001) states that semantic and syntactic calques are one of the most prominent kinds of influence of L2 on L1. It is not clear, whether ―calquing‖ in collocations happens only in performance as a result of more frequent use of L2, or the whole semantic mapping of the mental lexicon changes. Pavlenko and Malt (2011), for example, found such changes in mental lexicon of Russian-speaking immigrants in the US. They investigated changes in naming household objects, which reflected narrowing and broadening of the categories, such as distinctions between ―mug‖, ―cup‖ and ―glass‖. Their subjects sometimes were more similar in language choices during naming kitchen items to native speakers of English than to native speakers of Russian. Sometimes it seemed that their whole system was in a confused state and was similar neither to English nor Russian. The researchers
21
conclude that L1 word use is susceptible to L2 influence not only in abstract nouns, as it was suggested before (Cook, 2003; Pavlenko, 2002, 2010), but also in concrete nouns. Leisio (2001) studied ―Old Finland Russians‖ who were born in Finland or came to Finland before World War II and lived almost all their life outside of a Russian-speaking environment. They had almost no contacts with the speakers of Russian who arrived in Finland in the last decades. The researcher found significant reduction in lexical fluency of these speakers, a lot of borrowing and code-switching and L2-induced changes in lexical semantics. Pavlenko (2000, 2003, 2005) studied the relationship between emotions and language in Russian-English bilinguals and observed a pattern of expressing emotional states in L1 Russian not through appropriate one-word verbs, but through change-of-state verbs such as ―stanovit‘sya‖ [to become] and an emotional adjective. For example, attriters used constructions like ―Ya stal grustnym‖ [I became sad], instead of using the appropriate verb ―rasstroit‘sya‖ [to get sad]. In this way the speech of attriters became more English-like, because there is preference for similar constructions for expressing emotional states in English. Some studies found no L1 attrition in the lexicon of people who spent many years in an L2 environment. Shishkin (2010) compared two groups of Russian-speaking immigrants in the U.S.: those who immigrated before the age of 21 and received their higher education in the U.S., and those who arrived in the U.S. aged 21 and older after receiving higher education in Russian. Their lexical fluency was compared in both oral and written tasks. Shishkin found no attrition in any of the groups, and both older and younger groups were not significantly different from each other. However, she did not have a control group to compare them to, which makes her conclusion about lack of attrition questionable. Dewaele and Pavlenko (2003) compared lexical diversity of Russianspeaking immigrants to that of a control group of residents of Russia in a film-retelling task. No difference was found between these two groups. A possible explanation offered by Schmid and Kopke (2009) is that lexical diversity is a feature of language use that may, at first, gradually show some decline in the attritional process, but may also return to the non-attrited level in a later stage.
2.6.2 Attrition in grammar Zemskaya (2001), in an overview of studies (mostly qualitative), based on materials from several countries, analyzed general tendencies in Russian L1 attrition of grammar and language change in heritage speakers. Unfortunately, as in some other studies on L1 attrition of Russian, she did not separate attriters from heritage speakers, so it is not clear to what degree these changes are 22
experienced by L1 attriters. According to this analysis, the changes are possible in morphology, such as changes in noun declensions, the loss of animate/inanimate distinction in nouns, wrong formation of comparative degrees, and the loss of less productive suffixes (Zemskaya, 2001). Reduction of the case system was also observed in late Russian L1 attriters by Pavlenko (2003, 2010). However, Schmitt (2010) found no attrition in the case system of adults who immigrated to the U.S. as children at the age of 8 to 10. She attributed this stability to the lack of similarity between Russian and English case systems. Leisio (2001), in a study of Russian-speaking immigrants in Finland, found attrition of morphosyntax and loss or incorrect use of inflectional and derivational rules. Schmitt (2000) claims that the true indicator of L1 attrition is not simply the loss of rules, but convergence. By convergence she means a situation when L1 root morphemes are used according to the rules of L2 grammar. L2 grammar structures become dominant and replace the structures of L1 (Schmitt, 2000). In her study of Russian-English bilinguals she found almost no such attrition (Schmitt, 2010). She observed more attrition in content morphemes, such as word roots or derivational affixes, than in system morphemes, such as inflectional affixes, that were relatively stable (Schmitt, 2010). Several studies on Russian L1 attrition and heritage Russian in Russian-English bilinguals reported signs of simplification of the L1 verb system (Pavlenko, 2010; Polinsky 2008; Zemskaya, 2002, 2004). The simplification was expressed in a loss or change of such grammatical categories of verbs as aspect, directionality, and manner of motion. Zemskaya (2001) claimed that the verb aspect was not affected by attrition, but Pavlenko (2010) found that the aspect category might attrite even in adults/late attriters. Pavlenko noticed that attriters tended to use the imperfect aspect in situations when the perfect aspect was required. This tendency further manifested itself in the speech of heritage speakers where aspectual distinction almost disappeared, and imperfect aspects replaced the perfect in all situations (Polinsky, 2008). So far, L1 attrition of Russian syntax has been overlooked by researchers. Zemskaya (2001) describes some changes in syntax that she observed, such as the overuse of passive constructions and the loss of particle/preposition ―li‖. These changes have not been formally investigated yet.
2.6.3 Attrition in pragmatics As in studies on L1 attrition of other languages, attrition of pragmatics in Russian was barely researched. Priven (2002) investigated the attrition of Russian formal 2nd person singular pronoun
23
―vy‖1 in the speech of male immigrants in Canada. The use of this pronoun is not simply a matter of lexical or grammatical choice, but is a very important part of politeness in communication. Russian has a complicated system of politeness, in which a person is addressed differently based on age, social rank, degree of familiarity with a speaker and many other factors. The use of pronoun ―vy‖ also requires grammatical agreement with other words in the sentences, and, most often, a special polite version of an addressee‘s name. The study found a clear attrition of the pronoun and all the language features associated with it. However, the researcher himself pointed out the limitations of his method – he collected the data in a very informal situation (men playing soccer). On the one hand, even in this informal situation the pronoun ―vy‖ should have been used since there were significant age differences between the players that required a certain degree of formality in language (age difference almost always requires mandatory use of the polite pronoun ―vy‖). On the other hand, it might be that in a more formal setting, the pronoun ―vy‖ would have been used more often.
2.6.4 L1 attrition of Russian in Israel As mentioned earlier, most of the current research on the attrition of Russian as L1 was conducted with Russian-speaking immigrants to the USA (for example, Pavlenko 2010; Schmitt, 2010). Though Russian-speaking people form a significant minority in Israel, one sixth of the Israeli population, so far only a few studies investigated the attrition of L1 Russian in Israel (Feldman, 1996; Laufer, 2003; Ben-Rafael and Schmid, 2007). One study focused on Russian as a heritage language in children of immigrants to Israel (Kopeliovich, 2009) and one study investigated the attitudes of Russian-speaking immigrants towards their native language (Abu-Rabia, 1999). However, in the latter study the participants‘ L1 performance was not assessed. All the studies on attrition were very different in their subjects and methodologies. Feldman (1997) investigated lexical retrieval in participants who immigrated to Israel as children. She found L1 attrition in lexical retrieval in naming the semantic opposites and retrieving words from an offered definition. Participants had difficulties retrieving the target items and often code-switched to Hebrew, or provided paraphrases or neologisms instead of a target word. They also used many pauses, hesitations and experienced tip-of-the-tongue states. This study, however, does not tell anything about L1 attrition in late attriters, and in the case of child attrition, it is more proper 1
There are two second person singular pronouns “ty” and “vy”, their usage depends on the level of formality of a situation. The singular pronoun "vy" is identical in form to the plural second person pronoun "vy", and the following verb takes the plural form as well (Braun, 1988; Kasatkin, Klobukov and Lekant, 1991; Priven, 2002).
24
to talk about incomplete acquisition. Another methodological problem is that the study did not have a control group of Russian-speaking monolinguals. Laufer (2003) examined the processes of attrition in collocational knowledge and lexical diversity in written expression. Her subjects were late attriters, all of them with academic degrees (received mostly in the country of origin), who showed signs of attrition both in collocations and in lexical diversity. She suggested that L1 collocation knowledge was influenced by L2 as a result of becoming proficient in L2. There was a negative correlation between the results of the collocation test and the length or residence in Israel, that is, the participants who spent more time in Israel achieved worse results on the test. Maintenance of Russian correlated with the results positively. All her subjects were highly proficient in Hebrew. Since then the linguistic situation in Israel changed drastically. More speakers of Russian arrived in the country, and more speakers opted for functioning in Russian while acquiring only limited Hebrew, or not acquiring it at all. Globalization and the development of Internet radically changed and expanded the options for L1 maintenance. Ben-Rafael and Schmid (2007) examined the difference between speakers of L1 French and L1 Russian in Israel. They conducted semi-structured interviews with questions about attitudes towards Israel and the Hebrew language, and then analyzed the interviews for code-switching. Francophones who had a higher level of emotional attachment to Israel and a more positive attitudes towards Hebrew, were more integrated and used Hebrew more, had a significantly higher number of code-switching and difficulties in lexical retrieval than Russian-speaking participants. While emotions and attitudes are now considered to be highly related to the linguistic performance in bilinguals (see, for example Pavlenko 2005), the difference between the two groups might have been related to factors other than attitudes towards Israel. The French-speaking subjects in this study were much older than the Russian-speaking ones, and arrived in Israel about 40 years ago, while the Russian-speaking subjects spent on average about 15 years in Israel. The attitudes toward Israel might also have been affected by the length of residence: since the immigration of French-speaking participants happened a long time ago, they were more adjusted and assimilated in the society and, naturally, had more affection for it, while the recent immigrants from the former USSR might still have been experiencing traumatic effects of the immigration process. Abu-Rabia (1999) corroborates Schmid and Ben-Rafael‘s (2007) assessment of the attitudes of Russian speakers towards Hebrew. He investigated language attitudes of Russian-speaking students in Israel and found that their motivation to learn Hebrew was mostly instrumental; they did
25
not want to assimilate completely. Thus, even though they managed to reach a high proficiency in Hebrew, it almost did not affect their maintenance of Russian. The resistance of Russian-speaking immigrants to a complete assimilation challenged the Israeli ―melting pot‖ ideology and forced the acceptance of a multicultural society ideology, while de facto such a society already existed (Ben-Rafael, 1994; Horowitz, 2001; Remennick, 2004). Many Russian-speaking immigrants also believe Russian culture to be superior to the Israeli one (Kheimets & Epstein, 2001; Remennick, 2004), which contributes to their resistance to assimilate. Remennick (2003) describes Russian immigrants of the first wave (1970s) who initially adopted Hebrew language and culture, but then returned to Russian after a huge influx of the immigrants from the former USSR in the late 1980s and early 1990s opted for maintaining their L1. The two waves of immigrants became very similar in their language use. She found that, even though codeswitching to Hebrew was prominent in the language of Russian-speaking immigrants, the basic structural and grammatical framework of their speech remained Russian. The Hebrew words were used only to refer to some typical Israeli realities. Even in those who learned Hebrew well, the bilingualism was additive and not replacive (Remennick, 2003, 2004). Since the studies were qualitative, it is not clear how prominent the code-switching was and how often the incorrect structures were used compared to the speech of monolinguals. Overall, the population of Russian-speaking immigrants in Israel is very diverse: they may be found in every social stratum, they have different levels of education, religiosity, and adjustment to their new country, plus different attitudes toward Israel, Hebrew and Russian. For example, there are people who are ashamed to speak Russian and try to integrate in the new society as soon as possible, while others take pride in maintaining Russian and prefer to live and function inside the immigrant community. Russian-speaking immigrants constitute a significant minority in Israel (about 1/6 of population of the country). In some cities the population of speakers of Russian reaches 30-40% (Remennick, 2004). Ben-Rafael (1994) claims that the case of Russian community in Israel is completely unique in terms of language and culture maintenance. The community is large, cohesive, and autonomous. Thus, many services are available in Russian, and in most of the offices there are people who speak Russian on some level (either native or heritage speakers). There is an Israeli TV channel broadcasting in Russian; cable companies offer original Russian channels and closed captions in Russian. There are Russian sections in public libraries, Russian bookstores and food stores (even several chains of food stores). Many culture events in Russian are offered in every big city. There are retirement homes for Russian-speaking immigrants, where all the residents 26
and staff speak Russian. It is possible to live in Israel quite comfortably with basic Hebrew or without any knowledge of it, communicating only in Russian, and some immigrants embraced this opportunity. At the same time, formal schooling for immigrants is available only in Hebrew and even if parents did not acquire Hebrew, their children do. The children are usually much more fluent in Hebrew than in Russian, and often act as translators for their older relatives. The community does not remain closed, and there is always some influence of Hebrew. Sometimes the mixed input is coming from Israeli newspapers and TV in Russian, where code-switching or grammatical errors are not infrequent. With such a huge and diverse population we probably cannot talk about a single pattern of L1 attrition of Russian in Israel. It may be affected by many different factors. However, this diversity also offers a possibility to study different aspects of attrition, one of which is a change in L1 of people who did not acquire Hebrew or have only elementary knowledge of it. The immigrants who are not able to find skilled employment are the ones who are most likely to acquire Hebrew minimally or not at all. For such immigrants, then, the existence of a self-contained Russianspeaking community with its own media, consumer services and job market, becomes necessary. Such a community, in turn, produces more and more opportunities for immigrants to survive without Hebrew (Remennick, 2004). The other reason for supporting the Russian-speaking community is transmitting Russian as a native language to a younger generation. It is one of the most important values in the Russian-speaking community (Remennick, 2003), and even the immigrants who are proficient in Hebrew support the dominance of Russian in the community for the sake of children who otherwise would switch to Hebrew completely. Kopke (2004a) suggests that in L1 attrition studies a distinction should be made between the communities of migrant workers and the migrants who are a part of a larger community and are not in contact with other migrants. Russian-speaking immigrants in Israel present a unique case in this sense because they are not migrant workers. Ethnically, and in many aspects culturally, they are a part of the Jewish community. On the other hand, the penetration of Russian language and culture in Israeli society is so noticeable, that even immigrants from the former USSR who wish to avoid any contact with Russian can hardly do that. Russian-speaking community in Israel, thus, presents a unique case and its L1 attrition patterns may differ greatly from the Russian diasporas all over the world. Few studies of L1 Russian attrition in Israel were conducted a while ago and might not be relevant to the current linguistic
27
situation. Also, each of them investigated different areas of language, different kinds of population and used different techniques. Therefore, the results cannot be compared. Another unique feature of Russian-speaking community in Israel is the presence of a great number of Russian speakers who did not acquire Hebrew at any level. The observations I made about such speakers led me to believe that their native language did not remain unchanged. Although there is some research on L1 attrition in people who have limited L2 knowledge, it is scarce, and there is virtually no research on the processes in L1 of the immigrants with no knowledge of L2. The following study is the first attempt to conduct such a research.
2.7 Chapter summary The studies that I discussed in this chapter show that first language attrition can be explained in light of several more general linguistic phenomena, namely: regression of the first language, simplification of the linguistic system, emergence of an interlanguage that consists of L1 and L2 features, cross-linguistic influence from L2 to L1, universal grammar that leads to unification of the two linguistic systems, and loss of language self-regulation. However, currently these explanatory frameworks have little evidence to support them. The various explanations are not necessarily mutually exclusive. It might be that L1 attrition is a complex process that encompasses several different linguistic phenomena, for example, both simplification and cross-linguistic influence. L1 attrition has been observed in practically all language areas, and apparently nothing in a native language is absolutely immune to attrition. Attrition in the lexicon has been investigated most extensively, and this sometimes leads to a conclusion that attrition first and foremost affects the lexicon. However, there is not enough data to confirm it. There are much fewer studies investigating attrition in grammar and phonology, and pragmatics is probably the most under-researched area in L1 attrition. Attrition may affect either a speaker‘s performance, or both performance and competence. It is not absolutely clear which factors affect attrition, because studies often contradict each other. Age at the onset of attrition clearly matters in case of children who arrive in L2 environment before puberty, but for adult immigrants hardly any age effect was found. Length of residence in L2 environment is sometimes reported as an influential factor for L1 attrition development, and sometimes as a factor that has no effect on attrition. Similar conflicting results were found regarding the influence of the amount of L1 and L2 input on L1 attrition. Usually, only proficient L2 speakers were selected for participation in the studies, therefore it is hard to speculate regarding the influence of L2 level on L1 attrition. However, several studies demonstrated that even 28
speakers with limited knowledge of L2 could experience attrition. The higher educational level of a speaker prevented attrition to some extent, while positive attitudes towards native language did not. The process of L1 attrition of Russian is similar to attrition of other languages. Attrition in lexicon, grammar, phonology and pragmatics has been observed. Most of the studies on L1 attrition of Russian were conducted in the US. However, Russian-speaking community in Israel presents a unique case for attrition research. The uniqueness of the community is in its size (one sixth of the country population), its unwillingness of many immigrants to assimilate, and a large population that has acquired very little or no Hebrew. This situation offers a unique opportunity to investigate whether L1 attrition is possible in immigrants with no L2 and whether the level of L2 knowledge affects L1 attrition.
3. Pilot Study 3.1 Aim of the Study The aim of the pilot study was to examine the attrition of Russian L1 under Hebrew L2 influence in immigrants to Israel, namely which factors affect the process of attrition of vocabulary, collocations and complex grammatical constructions. I wanted to examine specifically the group of people who emigrated to Israel a long time ago but did not acquire Hebrew on any level. My wish was to investigate whether lengthy residence in Israel and exposure to attriting Russian of the fellow immigrants affect L1 of the immigrants with no knowledge of Hebrew. A small group of immigrants with no knowledge of Hebrew (―-Hebrew‖) was to be compared to a group of immigrants with knowledge of Hebrew (―+Hebrew‖) and to a control group consisting of current residents of Russia.
3.2 Research Questions 1.
Will the +Hebrew group and the –Hebrew group show signs of attrition of L1
vocabulary compared to the control group? Will the two study groups be different from each other? 2.
Will the +Hebrew group and the –Hebrew group show signs of attrition of L1
collocations compared to the control group? Will the two study groups be different from each other? 3.
Will the +Hebrew group and the –Hebrew group show signs of attrition of L1
complex grammatical constructions compared to the control group? Will the two study groups be different from each other? 4.
What sociolinguistics factors affect L1 attrition (length of residence, knowledge of
Hebrew, exposure to Hebrew, maintenance of Russian)? 29
3.3 Method The collected data included sociolinguistics and test data. The participants filled out a sociolinguistic questionnaire during a semi-formal interview. Following the interview, a test of vocabulary production was administered, where the participants were asked to describe 10 pictures. The next test was a grammaticality judgment test, in which the participants were asked to determine the correctness of collocations and complex grammatical constructions, e.g. subjunctive and conditional mood. Two tasks were chosen rather than one, because the description task was aimed to test production, while the judgment task mainly tested awareness and recognition of correct or incorrect structures (though, some production was involved when participants corrected the sentences). Data collection was intentionally conducted fully in Russian, without code-switching on the part of the interviewer, in order to create a monolingual environment. In a bilingual environment, code-switching and deviations from the standard variety could be assumed as unmarked options and would not be a sure evidence for attrition, while they do appear as marked in a monolingual environment. The assumption was that the speakers, aware of the code, would consciously borrow words and patterns from Hebrew only in the situations when they were not able to find a Russian analogue, and there would also be unconscious borrowing and errors caused by the attriting proficiency.
3.4 Participants Three groups participated in the study: two groups of immigrants to Israel from Russia and other parts of the former Soviet Union and one group consisting of residents of the Ural region in Russia and the Kostanay region of Kazakhstan (predominantly Russian-speaking region). All immigrants spoke Russian as L1 and completed secondary education in Russian in their native countries. They did not live in any other country besides Israel and the country of their origin. The republics of the Soviet Union were considered to be one country for the purpose of the study, since even the participants, who emigrated from countries other than Russia after the break-up of the Soviet Union, lived in a Russian-language environment and most of the time did not know the new official languages of the countries and were not in contact with them. After settling in Israel, the participants did not return to their home country for more than short visits. In order to eliminate the
30
possible influence of languages other than Hebrew, I included only those people who did not have a good knowledge of other languages, according to their report. All participants were volunteers. The Israeli participants were contacted by several methods: a snowball method when the participants indicated the next people who could participate; ads on the advertisement boards in the University of Haifa and the Technion, and in the Neve-Yosef community center in Haifa; online ads in several communities of Russian-speaking Israelis on the Livejournal.com social net; and through an article on the web portal publishing local news in Russian. The interviews were mostly conducted either in the homes of the participants or in the University of Haifa. Some of the participants were interviewed in the Neve-Yosef community center and in the retirement home in Haifa. I conducted all the interviews with the Israeli participants. The participants from Russia and Kazakhstan were found by the snowball method and interviewed in their homes. The interviews were conducted by Alexander Kotlyarov, a volunteering linguistics student from the Chelyabinsk State University. He was also consulted during the process of grading the grammaticality judgments. 3.4.1 The first immigrant group – with a knowledge of Hebrew There were 44 participants, all adults, 20 to 79 years old; they came to Israel from 1 to 21 years ago. Their Hebrew proficiency ranged from elementary to native-like. The participants came from different socioeconomic backgrounds, different levels of education (from the high school graduates to the holders of PhD degrees), and different language environments. For example, there were participants who lived in mostly Russian neighborhoods, had TV, papers and books in Russian, and had jobs that did not require a knowledge of Hebrew (like cleaning or guarding a retirement home for speakers of Russian). On the other hand, there were two participants who lived in a remote kibbutz on the Golan Heights, where they were the only speakers of Russian, they did not have access to Russian TV or books, and their children and spouses were native speakers of Hebrew. 3.4.2 The second immigrant group – without a knowledge of Hebrew The second group consisted of 16 people, 33 to 87 years old, who came to Israel 3 to 21 years ago. They were not able to speak, read or write Hebrew, though they did know some nouns used in daily life and Hebrew names of the state institutions. Their educational level ranged from secondary to higher education. They all lived in a Russian-speaking environment, with access to Russian-language TV, newspapers, libraries, stores, physicians and social workers. They did not 31
have any contact with Hebrew. Those of them, who worked, had jobs that did not require any knowledge of Hebrew.
3.4.3 The control group The control group consisted of 21 adults, 18 to 74 years old, from the Ural region of Russia (Chelyabinsk and Yekaterinburg cities) and the Kostanay region of Kazakhstan (the region is predominantly Russian-speaking). The participants had no knowledge of Hebrew. Though the immigrants were mostly from the regions different than the control group, Russian is relatively homogeneous across the country, dialectal differences are rather small, compared, for example, to German or Italian, and almost entirely phonetic (Auty, Obolensky and Kingsford, 1981, p. 32), which has no effect on the spelling, or on mutual intelligibility. Standard variety is used in schools and universities, in TV translations across the country, and is generally spoken by the educated part of the population. Phonetic issues and dialectal words will not be tested in this study; therefore the possible dialectal differences should not have any effect on the results.
3.5 Materials
3.5.1 Questionnaire The following demographic and sociolinguistic information was explicitly stated by the participants: name, age, level of education, level of Hebrew knowledge (this was based on self-report and measured on a scale from 0 to 4, in which 0 is no knowledge of Hebrew, 1 is elementary, 2 is intermediate, 3 is advanced and 4 is native-like), and length of residence in Israel. In addition, six multiple-choice questions were used to assess the exposure to Hebrew and the maintenance of Russian. This questionnaire was developed by Laufer (2003) for a study of attrition of Russian in Israel and it was slightly modified for the current study. For example, the question about using the Internet was added. The questionnaire was conducted in a form of semiform interview. The questions were: 1. На каком языке вы читаете книги, журналы и газеты? [In what language do you read books, magazines and newspapers?] (a) На русском [In Russian] 32
(b) На иврите [In Hebrew] (c) На русском и иврите [In both languages] 2. На каком языке вы смотрите телевизор и\или фильмы? [In what language do you watch TV and/or movies?] (a) На русском [In Russian] (b) На иврите [In Hebrew] (c) На русском и иврите [In both languages] 3. На каком языке вы читаете новости в интернете, пользуетесь поиском, заходите на сайты и общаетесь в социальных сетях (Facebook, LiveJournal, Twitter и т. п.)? [In what language do you read news and other types of content on the Internet, use an internet search, and communicate in social networks (Facebook, Livejournal, Twitter etc.)?] (a) На русском [In Russian] (b) На иврите [In Hebrew] (c) На русском и иврите [In both languages] 4. На каком языке вы общаетесь с семьей и друзьями? [In what language do you speak with your family and friends?] (a) На русском [In Russian] (b) На иврите [In Hebrew] (c) На русском и иврите [In both languages] 5. В течение недели вы используете русский язык… [During the week you use Russian language…] (a) Меньше 10 часов [Less than 10 hours] (b) 10-15 часов [10-15 hours] (c) 15-30 часов [15-30 hours] (d) Больше 30 часов [More than 30 hours] 6. В течение недели вы используете иврит… [During the week you use Hebrew language…?] (a) Меньше 10 часов [Less than 10 hours] (b) 10-15 часов [10-15 hours] (c) 15-30 часов [15-30 hours] (d) Больше 30 часов [More than 30 hours] 33
In the first four questions each language received 1 point. In the last two questions each language received 1 to 4 points depending on the answer, (a) to (d) respectively. Then the points were added together in order to calculate the score for each language. Theoretically the score for Russian could range from 1 to 8, and for Hebrew from 0 to 8 (since there were participants who did not know Hebrew). However, in the study the lowest score one participant received for Russian was 3, and most of the scores for Russian were 8 or 7. Generally, the scores for Russian were higher than those for Hebrew. Many people reported, for example, that though they were able to read books in Hebrew, they prefer to do so in Russian. But there were also individuals with equal scores for both languages, and the ones who received a higher score for Hebrew than for Russian were those who also reported native-like knowledge of Hebrew. If the participants reported that they also used a language other than Russian and Hebrew for any of the situations in question, their answer was written down but only the data on Russian and Hebrew was used for the analysis.
3.5.2 Picture description In this test, 10 pictures were used as stimuli in order to induce a speech act in which the key words would be named. The pictures used in the study are provided in the Appendix A. The description task, rather than naming, was chosen because the key words would be produced in context, which would be closer to a normal speech situation, thus the assumption was that those who produced the key words in Hebrew rather than in Russian in a study environment would do so in a non-study environment as well. The participants were not told explicitly in which language they should describe the pictures, but since the explanation of the task was in Russian, and there was no code-switching in the speech of the interviewer, they seemed to be aware of the code of the situation. For example, one participant, who speaks mostly Hebrew or Hebrew-Russian mix in her daily life, was often unable to come up with a Russian word instantly, so she first whispered a Hebrew word, looking embarrassed, and then after a pause in a full voice she said Russian word if she could (sometimes after a long pause she admitted that she did not remember the respective Russian word and repeated the Hebrew word). Most often the participants who used Hebrew words in descriptions did not notice it and did not correct themselves. Occasionally, the search for a Russian word was indicated by a long pause; some participants started to utter a Hebrew word and stopped after a syllable, some explained that they did not know the Russian word for an object in question or thought that there was no appropriate Russian word (although other Israeli participants, as well as the control group, found appropriate Russian words for these objects). 34
There were 10 keywords, one word per picture. Each picture presented some daily life situation and included the object corresponding to a keyword. The participants were instructed to describe what they saw on the picture, what kinds of people and objects were present, and what the situation was. They were not told how detailed the description should be (and the descriptions ranged from one sentence to a full story per picture) and were not stopped or encouraged to continue during the description. Only if they failed to state the keyword, were they once more given the instruction (for example, in a situation where the participant described a picture of children getting on a school bus with two girls in the foreground looking at a cell phone as simply ―Girls‖). The participants got 1 point if they said a keyword in Russian and no points if they said it in Hebrew (even if they then came up with a Russian word after some pause) or did not state it at all. Since all the participants in the control group used all 10 keywords, the assumption was that if the participant did not say the word it was because of forgetting the word in Russian. Synonyms of the keywords in Russian were also accepted. For example, for the keyword ―rynok‖ (market) the word ―bazar‖ was accepted. For the keyword ―shtora‖ (curtain) such synonyms as ―gardina‖, ―zanaveska‖, ―zanaves‘‖, and ―portyera‖ were accepted. The points for each picture were added together in one score for the picture test.
3.5.3 Grammaticality judgment test The test consisted of 32 sentences in Russian, all on daily life topics, in a conversational style. The sentences used in the study are provided in the Appendix B. Eight of them contained errors in collocations, modeled on the collocations existing in Hebrew. Eight contained errors in complex grammatical constructions, modeled on the Hebrew constructions. Sixteen sentences were correct and served as distracters. Each collocation was used only once. An example of a sentence with a collocation error is ―Skol‘ko vremeni beryot dobrat‘sya do raboty?‖, literally, ―What time does it take to get to work‖. The correct verb in this collocation in Russian is ―zanimat‘‖, and not ―brat‘‖ (both are translated into English as ―to take‖). There were four types of errors in complex grammatical constructions, two sentences per each type of the error:
Errors in formation of a dependent clause with indirect question. While in Hebrew the indirect question connected to the main clause just by a conjunction ―im‖ (if/whether), in Russian the conditionality of the clause is marked by the conjunction 35
―li‖ (whether) inserted between predicate and subject and the inversion of the word order. Moreover, the Hebrew conjunction ―im‖ may be translated to Russian both as ―yesli‖ (if) and ―li‖ (whether), but in an indirect question only ―li‖ can be used. The example of the sentence with such an error is ―Sonya sprosila Mishu yesli on khochet yeschio chayu‖ which literally means ―Sonya asked Misha if he wants more tea‖. The correct version would be ―Sonya sprosila Mishu khochet li on yeschio chayu‖, literally ―Sonya asked Misha wants whether he more tea‖.
Errors in formation of dependent clause in subjunctive mood (in sentences like ―He demanded that the car should be repaired‖). In Russian there is only one way of forming the dependent clause subjunctive mood, for which the conjunction ―shto‖ (that), the particle ―by‖ (the conjunction and the particle merge in this case into ―shtoby‖ or shorter from ―shtob‖) and the verb in the past tense are used. In Hebrew, the conjunction ―she‖ (that) is used with the verb in the future tense. The example of sentences with such an error is ―Ya khochu, shto ty nemedlenno polozhish knigu na mesto‖, literally ―I want that you instantly will put the book in its place‖. The correct sentence in Russian would be ―Ya khochu, shtoby ty nemedlenno polozhil knigu na mesto‖, literally ―I want that you instantly put (past tense) the book in its place‖.
Inclusion of third person pronoun between subject expressed by noun and predicate expressed by noun or adjective. Both in Russian and Hebrew the present tense of the lexical verb ―to be‖ is not used in descriptive sentences. However, while Russian simply omits it, not adding anything else (―Olya studentka‖ – ―Olya [is] a student‖), Hebrew sometimes adds a third person pronoun in cases where the subject is a noun (―Miri hi studentit‖ – ―Miri she [is] a student‖), though this construction is optional. An example of a sentence with such an error is ―Marina ona otlichnaya podruga‖, literally ―Marina she a great friend‖, while in the correct usage the pronoun ―ona‖ (she) should be omitted, and this construction is obligatory in Russian. The construction tested in this task is not to be confused with left dislocation which is acceptable in normative Russian. Left dislocation is marked by a pause after the advanced constituent in spoken Russian, and by a comma after it in written Russian. The erroneous construction that was observed in the speech of immigrants did not employ a pause after a constituent, and, therefore, did not constitute a case of left
36
dislocation. Such a construction was unanimously considered incorrect by the nonimmigrant speakers of Russian.
Violation of the pro-drop rule of usage of the pronoun ―eto‖ (it) in impersonal sentences. In Russian, an expletive pronoun is not present in impersonal sentences, while in Hebrew such construction is possible (though the pronoun is optional). An example of a sentence with such an error is ―Kazhdomu chelaveku eto vazhno znat‘ angliyskiy‖, literally ―For everyone it important to know English‖. The correct sentence would omit the pronoun ―eto‖.
In the original design, the distracters were supposed to play a role in the study and the corrected, or judged as incorrect, items were counted as errors, but apparently there was a problem with these sentences that was not spotted during the design stage. Although they were formally correct, some of them were somewhat doubtful, usually stylistically, and many participants, both in the study groups and in the control group made many corrections. It was decided to count only corrections made in the 16 target sentences. The participants were given three pages with sentences, unnumbered, so they did not know the number of sentences. They were told that in this test there are correct and incorrect sentences, and their task was to find the errors and correct them. They were instructed to decide in each case if the sentence was correct. If it was, they were to leave it without any changes. If they spotted an error they were instructed to underline it and to write the correct version if they knew it. They were specifically instructed that there are no errors in orthography and punctuation. The score for the correctness judgment task was calculated as follows: an error recognized and corrected counted as 2 points; an error recognized but not corrected, or corrected wrongly counted as 1 point. The corrections made in distracters were not counted. Thus, for each group of sentences the score could range from 0 (no errors recognized) to 16 (all errors recognized and corrected). In some uncertain cases (for example the wrong collocation ―delat‘ dietu‖(to be on a diet) was corrected as ―derzhat‘ dietu‖ instead of common in Russian ―sadit‘sya na diety‖), the dictionaries of modern usage were consulted.
3.6 Results Three sets of scores were compared – scores for the vocabulary production test, scores for the collocations test and scores for the complex grammatical construction test. Anova test and Scheffe‘s post-hoc test were performed in order to look for differences between the three groups. 37
The Spearman‘s rank correlation coefficient was chosen for examining possible correlations between the test results and such variables as the length of residence in Israel, level of Hebrew, maintenance of Russian and usage of Hebrew. The Pearson‘s product-moment correlation coefficient was rejected after examining the data for linearity: it appeared that none of the pairs of variables were related by a linear function, which is necessary for Pearson‘s coefficient to be calculated.
3.6.1 Research question 1 Will the +Hebrew group and the –Hebrew group show signs of attrition of L1 vocabulary compared to the control group? Will the two study groups be different from each other?
Table 1: Differences between the three groups on the vocabulary production test (ANOVA results)
n
Control Group +Hebrew Immigrants -Hebrew Immigrants
Mean (max. 10) 21 10.00* 44 9.02 16 8.87 F=7.022
Standard Deviation 0.00 1.24 1.25 *p