Abstract
This paper provides a dynamic perspective on the linguistic development of adult bilinguals in an L2 environment, and an empirical test for the principles formulated within a dynamic systems approach to L1 attrition. It presents a qualitative analysis of the personal narratives of the post-migratory linguistic development of 27 adult German–English bilinguals residing in Ireland, with particular focus on two participants’ stories. Participants responded to a comprehensive sociolinguistic questionnaire, providing self-ratings of current and past language proficiency in all their languages, and reflecting on the processes of and factors impacting on L2 acquisition and L1 attrition in the L2 environment. These data are complemented by proficiency data elicited with a test battery in L1 German and L2 English. Analyses revealed that most participants have succeeded both in acquiring English to a high level, and in maintaining German. However, other L2s as a rule show a pattern of decline. The results are discussed with recourse to dynamic conceptions of multilingual development.
Keywords
Introduction
Many people have experienced both growth and loss in their languages during their bilingual 1 lives. Perhaps they added a new language to their repertoire of existing ones (language learning/acquisition), started to forget a language that was not practised (language attrition) or, conversely, became more proficient again in a language that had become ‘rusty’ due to lack of practice (language relearning).
A helpful recent perspective for holistically capturing language acquisition and attrition in bilinguals may be found in dynamic approaches to language development (de Bot, 2007; de Bot, Lowie, & Verspoor, 2007; Herdina & Jessner, 2002; Larsen-Freeman & Cameron, 2008). Within such approaches, language is conceived as a complex dynamic system, and language development as being shaped by ‘interrelated patterns of experience, social interaction, and cognitive processes’ (‘Five Graces Group’ et al., 2009, p. 2). As a result, ‘seemingly unrelated linguistic phenomena’ can properly be accounted for within the same theoretical approach, as different ‘facets of the same system’ (‘Five Graces Group’ et al., 2009, p. 2) that are subject to the same forces and constraints.
L1 attrition, the focus of this special issue, has to date been studied as a process largely divorced from other linguistic developmental processes, although it is widely recognised that it usually proceeds in the context of broader changes to the linguistic system, for example as a result of the simultaneous acquisition of another language (see pertinent contributions from Cook, 2003; Ecke, 2004; Fase, Jaspaert, & Kroon, 1992; Köpke & Schmid, 2004; Seliger & Vago, 1991). Thus, Ecke (2004) observes that ‘[t]he attrition of a language is normally related to the gradual acquisition and increasing strength of one or several competing languages’ and argues that ‘[e]nvironmental (including social) changes seem crucial forces behind linguistic competition and change’ (Ecke, 2004, p. 341).
Dynamic systems theory (DST), on the other hand, conceptualises L1 attrition as a feature of the development of bi-/multilingual systems to be studied within that context (Herdina & Jessner, 2002, pp. 91, 93). This paper presents a holistic view of quantitative and qualitative changes in the proficiency profiles of late bilinguals in an L2 environment across the range of languages known by these individuals. By applying a dynamic perspective to these changes, it provides an empirical test for the principles formulated within a DST approach to L1 attrition.
Principles of dynamic approaches to language development
A basic tenet of DST approaches is the interconnectedness and non-linear interaction of all elements within the system, and with its environment. Although bilinguals clearly are able to separate their languages in use, cross-linguistic effects, such as interference, mean that the languages must be connected at some level. The non-linearity of language acquisition processes has been demonstrated widely, and the available data on attrition can also be interpreted in this way (e.g. de Bot & Clyne, 1994).
Dynamic systems are further characterised by complexity and perpetual flux. The latter as it relates to language development has already been referred to in the opening paragraph, while the complexity of the language system, its elements and their interactions with one another and forces outside the system may be deduced from the immense range of potential predictor variables proposed for attrition, and difficulties in establishing clear, unanimous effects for these variables.
Dynamic systems are adaptive, such that change in one element or part of the system leads to changes in the overall system (Larsen-Freeman & Cameron, 2008, Chapter 2). They also show a ‘sensitive dependence on initial conditions’ (de Bot et al., 2007, p. 7). For language development, these principles imply that the initial proficiency level in a given languages matters for the speed and developmental trajectory of the particular language, and the language system overall.
These characteristics of dynamic systems lead to ‘variation both in and among individuals’ (de Bot et al., 2007) and ‘at all levels of linguistic organization’ (‘Five Graces Group’ et al., 2009, p. 2). Inter- and intra-individual variability are indeed a robust finding in both acquisition and attrition studies, arguably necessarily implying a shift to qualitative analyses at the individual level, as realised here. Ultimately, these characteristics also lead to an inherent unpredictability of the outcomes of change processes. Dynamic approaches must therefore concern themselves with retrodiction rather than prediction (Larsen-Freeman & Cameron, 2008; Verspoor, de Bot, & Lowie, 2011).
Nevertheless, language development is postulated to reveal patterns, such as growth curves akin to biological growth, with stage-like transitions, in interaction with internal and environmental variables. Amongst these, language use/amount of contact has been argued to be a crucial variable. This reflects the theoretical orientation adopted by several leading researchers in the field (e.g. de Bot et al., 2007; Ellis, 2008; Herdina & Jessner, 2002) who argue that the nature of language development is probabilistic and sensitive to usage frequencies.
Thus, changes in both L1 and L2 proficiency should correspond to changes in use, such as in opportunities to practice and natural input. For the participants investigated in this paper, these changes were brought about by relocation to a different linguistic environment where use of L1 is potentially reduced. At the same time, L2 has taken on a more important role since most participants, having emigrated of their own volition, aim to be active members of the new society.
In this paper, I chart the overall linguistic development of adult bilinguals in an L2 environment. In line with the dynamic principles just presented, the focus of the study is on variability within and between participants, on identifying possible patterns of development and on the interaction with reported variables, in particular language use. Relating the findings to the Dynamic Model of Multilingualism (DMM; Herdina & Jessner, 2002), I argue that, contra conceptualisations that view L1 development as fundamentally different to L2 development, the same forces and constraints operate on L1 as on second languages.
Method
In a cross-sectional investigation, 27 German–English bilinguals participated in a semi-structured interview based on an in-depth sociolinguistic questionnaire. The bilinguals were of German (N = 25), Austrian (N = 1) or Swiss (N = 1) extraction who had grown up monolingually in German and moved to Ireland in their adult years (average age at arrival 26.7 years, range 22–42 years; average length of residence 19.5 years, range 8–34 years).
The participants had learned English as a school language, sometimes augmented by short periods spent in an English-speaking country during holidays. Their first significant longer-term exposure to English started with emigration. Participants were required to have at least a good level of proficiency in English as assessed on a five-point self-rating scale. In response to a personal background questionnaire returned prior to the study (N = 24), two participants (8%) rated their level of English as ‘good’, eight (33%) as ‘very good’ and 14 (58%) as ‘excellent’. Although self-ratings need to be interpreted cautiously, they do provide a sensible screening tool for recruitment purposes. Following comparison of the self-ratings with the objective L2 proficiency measures described below, the participants can justifiably be described as proficient or highly proficient bilinguals.
The sociolinguistic questionnaire was administered as part of a wider-ranging investigation, which also entailed the elicitation of proficiency data in English and German on a variety of language tasks, as well as the administration of further self-report instruments and a linguistic aptitude task (see Table 1). Matched bilingual groups of 20 German native speakers and 18 Irish native speakers of English served as controls for tasks in the relevant language.
Test battery deployed in the wider-ranging investigation.
In this paper, I analyse the participants’ responses to the sociolinguistic questionnaire to explore how their language proficiencies have changed since migration and what they attribute these changes to. For converging evidence, I draw on proficiency self-ratings elicited with the can-do scales, test scores achieved on the German and English language tasks, principally C-test, 2 verbal fluency task (fluency in controlled associations, FiCA) and film retelling, and external ratings for English.
The sociolinguistic questionnaire was administered as a semi-structured interview. It contained a total of 65 questions across two parts. Part 1, the language-use questionnaire (36 questions), elicited primarily factual, quantifiable information relating to the frequency and modes of language use and language choice across various domains; to primary interlocutors and social networks; and to participants’ views of themselves as language learners/bilinguals.
Part 2, the language proficiency and attitudes questionnaire (29 questions), focused on participants’ current level of proficiency and changes therein in relation to all languages known by them; on their attitudes to language learning and language maintenance and their experience of acquiring a language in an L2 environment, yielding mainly open-ended, qualitative data. Participants were prompted to reflect on the processes of L2 acquisition and L1 attrition with questions aiming to elicit factors promoting acquisition/maintenance, the time course of acquisition/attrition and areas of change within each language (see Table 2 for an overview of relevant questions taken from Part 2 of the sociolinguistic questionnaire). A caveat associated with such retrospective sampling is that it is likely to lead to an overemphasis on general patterns at the expense of detail. However, in a cross-sectional study, eliciting retrospective data is one of the few options open to study development.
Subset of questions from sociolinguistic questionnaire.
Data
Changes in language proficiency levels since migration
As reported in greater detail elsewhere (Opitz, 2011, submitted), by the time of the study the bilingual participants had overall attained a high level of proficiency in their L2 English, while displaying only a moderate amount of first-language attrition, as shown by the results of the various language tasks and self-assessment tools.
In English, the bilingual participants showed significant differences on several of the objective measures compared to the Irish controls; thus, their level of attainment as a group was below that of the native speakers. However, a non-trivial number of them (40–50% across various tests) performed well within the native-speaker range on individual and composite measures (averaged z-scores based on six measures across C-test, FiCA and film retelling). External ratings of participants’ global proficiency on the basis of the film retelling confirm the impression of very high L2 attainment overall.
These results tally with the participants’ self-ratings of their current level of English language proficiency (see Table 3). As a group, the bilinguals also outperformed the German control group by a large margin. The comparison of the self-ratings of their level of English language proficiency prior to migration and at the time of testing (see Table 2, Questions 37 and 38) points to a perception of large positive change (Wilcoxon Signed Rank test: T = 0, z = −4.53, p = .000 one-tailed, r = −.62).
Significant correlations between proficiency scores and self-ratings.
Correlation is significant at the 0.05 level (two-tailed).
Correlation is significant at the 0.01 level (two-tailed).
For German, the initial impression based on the test results is that of first-language maintenance, since there were no significant differences between the bilingual group and the German control group on most of the objective measures taken individually. However, the difference did become significant on the composite z-score measure, which is in line with the participants’ perception of a degree of first-language attrition. The C-test measures correlated significantly with participants’ self-ratings of their productive skills in German, as reported on the can-do scales (see Table 3).
In response to a question about whether their L1 proficiency had changed since migration (Table 2, Question 39), 48.1% stated it was still the same, and 51.9% said it had disimproved. Again, the difference between the participants’ self-ratings of their level of language proficiency prior to migration and at the time of testing (Table 2, Questions 40 and 41) was significant, this time indicating negative change (attrition) in the participants’ L1 (Wilcoxon Signed Rank test: T = 1, z = −2.11, p = .031 one-tailed, r = −.29).
Nevertheless, in their self-ratings, several participants emphasised that their L1 was just ‘a little bit worse’. 3 Others did not seem to judge the characteristics of their current speech to constitute a disimprovement: having stated that their L1 proficiency had not changed, they subsequently listed areas of difficulty in German language use. Interestingly, one participant, S18, argued that her German might even be considered to be better than before, despite the fact that she describes lexical retrieval difficulties (‘the words are not always present’), the need to translate phrases from English to German or to find paraphrases when speaking. She cited speaking with increased language awareness as a result of having to make an effort as an explanation for her assessment of improved L1 proficiency.
Such conflicting views probably reflect individually differing conceptions of proficiency, but also the relatively small scale of the reported changes in L1 proficiency that may lie within the bounds of nativeness. This is borne out in the individual test results, which yield a significant difference between German controls and bilinguals only after being combined.
When asked whether participants saw themselves as bilingual (Table 2, Question 33), only 22.2% felt that German was their stronger language, while 14.8% claimed it was English, and the majority (63.0%) felt the languages were on a par. This constitutes a substantial shift in dominance pattern, since emigration occurred when all participants would have been L1-dominant, confirming the impression of some negative change in German language proficiency, and a large positive change in English language proficiency.
Most of the instruments and questions contained in the sociolinguistic questionnaire focused on German and English. I also collected self-ratings for changes in participants’ proficiency in other languages (Table 2, Questions 42 and 43), since I was interested in gaining an insight into the overall linguistic development of participants. Here, a different picture emerges.
Most bilingual participants had studied or picked up other languages, usually one or two, before migration. At the time of testing, the majority of these (74%) had disimproved, productive skills more so than receptive skills. Three participants reported having lost a previously studied language entirely. A few participants had begun to study a new language post-migration (10 cases), but had not achieved a very high level of proficiency in it, except for one case. The participants have apparently had limited success in learning and maintaining languages in the new language environment, other than their L1 and L2 English.
Given these findings, one would expect participants to be quite content with their current level of German and English, and possibly less so with that of their other languages (Table 2, Question 46). Indeed, a large majority of 88.9% reported being satisfied with the level of English they have achieved, while the remainder felt that there was still scope for some improvement. The same number was content with their level of German language maintenance, while a few individuals felt their level of German could or should be better.
However, only 24.1% are satisfied with their current level of proficiency in their other languages. Thus, there is a clear division between the level of contentment with English and German on the one hand, and the other languages on the other, in line with the relative success in maintenance/attainment. Interestingly, 18.5% state explicitly that they do not care about a particular language, a category not found with either German or English, which therefore must matter (more).
The findings above in many respects corroborate and refine our understanding of adult bilinguals’ potential linguistic development. They provide converging evidence for the growing literature showing that a mature L1 is not immune to a degree of attrition, and that adults are capable of developing a sophisticated level of L2 proficiency within the baselines provided by matched native speakers. The latter is in contrast with the contention that ultimate attainment in a second language is exclusively constrained by age of acquisition, and that native-like L2 attainment by adults is reserved for exceptionally talented individuals.
In this study, there was no significant difference between the bilinguals and the German controls on the three tasks of the linguistic aptitude test. The bilinguals also seem fairly typical in the extent to which they variably took up and enjoyed studying foreign languages before migration, as well as in their pre-migratory proficiency. Their reasons for migrating were not related to linguistic matters, such as a desire to improve their English, but instead to a partner or job, or life-style choices in general. Nevertheless, the bilinguals differ from the German controls both in their degree of L1 maintenance and L2 English acquisition. These differences must therefore be related to the environment the respective groups find themselves in.
Language development in the L2 environment
If the L2 environment is special in some way, how did the participants experience the changes in their languages after relocating to the new environment, and how long did it take for these changes to take place? How did they act in their role of language learners? What has brought about the changes they have noticed, and what could be done to influence their linguistic development?
These are some of the questions to which participants responded during the sociolinguistic interview (see Table 2, Questions 34–36 and 48–50). Although participants were not asked specifically about the course of development of their English or other L2 proficiency, since we may assume that it followed the usual acquisition routes described in the literature for L2 learners, participants did provide relevant insights in responding to this complex of questions. In this and the following subsections, I provide a qualitative analysis of the participants’ responses.
According to the participants’ testimonies, the most intensive phase of English language acquisition takes place during the first one to three years after migration; the actual length probably depends partially on participants’ initial proficiency level. This phase is often experienced as something challenging and potentially stressful: ‘I was utterly shattered during the first one to two years because of the language’, remarked one participant (S4), adding that the combination of language and cultural learning created a cognitive and emotional overload.
One participant, S1, related that she went through something of a ‘silent period’ during which she consigned herself to the role of an observer during conversations, writing down expressions she noticed. Another, S14, talked about realising with excitement how her range of language domains expanded, and the opportunities and challenges associated with that.
Some of the important early challenges and consequently areas of language expansion are, not surprisingly, aural comprehension and colloquial language use. Several participants shared that it took them from two to six months until they felt they were coming to terms with the everyday language used in Ireland, and specifically the type of humour referred to as ‘slagging’ (verbal teasing). The shift in language dominance in some participants took approximately as long: ‘After six months I started to think in English’ (S25).
More specialised and sophisticated language took longer to develop, a process which is never entirely complete. Although participants reported that after a certain time they had more or less achieved the proficiency level they still had at the time of testing, that, in other words, acquisition had slowed down and levelled off, most still occasionally, albeit much less frequently, notice and integrate new expressions into their linguistic repertoire.
For several participants, the levelling-off/stabilisation of English language proficiency seems to have coincided or been slightly preceded by them first noticing ‘problems’ with their German, somewhere around the 18-month to two-year mark. Usually the problems were encountered during a return trip to Germany, a long telephone conversation and similar more intensive or demanding engagements with German.
Most of these problems relate to the lexical and, as a knock-on effect, the pragmatic areas, primarily due to difficulties retrieving or insecurities about the correct/appropriate form of words and certain types of expressions, for example humorous or complex language. Instead, some participants resort to borrowing words and expressions (‘translating from English’, S18) to counter the resulting disfluencies. A related problem is the non-acquisition of vocabulary specific to domains that participants have not encountered in German (work, research, childcare), or that has become current since participants left their home country: ‘My German is that of an 18-year-old’ (S9).
Although shown to be relatively stable (e.g. Köpke, 2002), morpho-syntax and phonetic features are also affected to an extent. Participants reported syntactic transfer, grammatical mistakes and conscious rather than intuitive rule application, and one participant (S10) claimed to have acquired a ‘foreign’ accent in German. Overall, the self-reported changes to the L1 suggest a pattern similar to the one observed in the proficiency data, with the fluency dimension of L1 proficiency being most affected.
Similar to the eventual slowing down of the growth of English language proficiency, the decrease in German proficiency levelled off after a few years. However, many participants subsequently experienced a temporary or more permanent improvement in their L1. Thus, several participants stated that their German was now ‘back to normal’ (e.g. S16), since it had become more affordable and easier to keep telephone and other contact, often via the internet. Trips to Germany also helped to ‘relearn’ (e.g. S4) the language that otherwise had got ‘rusty’ (e.g. S17) and required ‘warming-up’ (S3), like ‘muscles’ (S11). Other reasons mentioned were starting a new job or training as a translator/interpreter, or having children.
Individual trajectories of change
Although developmental patterns are discernible for the group overall, individuals’ experiences and trajectories differ in interaction with many personal and environmental variables. To illustrate individually changing patterns of proficiency over time, and their non-linear interaction with these variables, I provide a summary of the migration stories from two of the participants in the following case studies.
Case study 1
S17 is a 40-year-old female. Her husband is English, and they have three young children. S17 moved to England at 19 years of age. She moved into a house with a German housemate who had an English boyfriend. As a result, the language spoken at home was almost exclusively English. The language at work was also English. She had no other acquaintances in the beginning but soon met her future husband and his circle of friends. S17 reports that her German declined noticeably during the first five years of her time in England and eventually plateaued, while her English improved rapidly during the first year and a half, and continued to grow more steadily thereafter (see).
Developmental trajectories of languages in S17’s repertoire over time.
Her other L2, French, also declined during that period, since she had no occasion to use it, except for a one-month stint in France for some work experience, which saw her recover the language very quickly, but on her return the intermittent improvement was reverted, and she began to forget French again.
Ten years later the family moved to Ireland. S17 took up work that required her to use German. This prompted a gradual reactivation of the L1, further fostered by having children and her decision to speak to them in German. The year prior to testing, she began to work as a German teacher, following which she felt her level of German returned to its ‘normal’ state. Her French is almost forgotten, and her English is very proficient (rated C2).
Case study 2
S1 is a 48-year-old female who also migrated initially to England in her early 20s. Her main motivation for migration was her wish to leave Germany; she did not particularly care where she went. She soon met her future husband, an Irishman, and like for S17 their common language was English. They also have three children who were grown up by the time of testing.
The couple moved twice between Ireland, Germany and another country. The first child was born in Germany, where S1 mother provided German input during the first 2.5 years of her grandchild’s life, while the family language, and the language S1 spoke with her daughter, was English. The young family then moved to Ireland, where the second child was raised exclusively in English.
By the time the third child was born, S1 had started feeling guilty over not raising her children bilingually. It is clear from her narrative that she herself did not see a need for the children to speak German but that she adopted the opinion of others. While the third child did indeed learn to speak German, the other children were not happy with their mother’s decision. The daughter in particular felt embarrassed when her mother spoke German in shops, etc., even though she understood it well enough. Although S1 kept up her efforts to speak German with the third child, the boy stopped using it at the age of 4. The second child, on the other hand, who had not been raised through German, decided to learn the language in school. During a school trip to Germany, he became thoroughly enthused, since he felt drawn towards and at home in the culture and ways of the people around him. He now studies German at third level.
S1 did not comment much on her own linguistic trajectories. She has used English almost exclusively ever since emigration, even with German-speaking friends, partially for practical reasons, such as not to exclude others from a conversation, and partially as a result of what might be called linguistic inertia. She is a proficient speaker of English with a very authentic sounding accent (rated C2), while her German would indicate some degree of attrition, which may be attributed to limited use and lack of interest in maintaining the language.
However, S1 reported that her German had in fact improved over the last while when to her surprise she discovered that she did like the language after all, even though it was not an essential part of her or her family’s identity. A recent journey to Germany had left her feeling exhilarated and in full command of her German. Around the same time, she coincidentally stumbled upon some linguistic childhood treasures that she re-read with pleasure. During testing, she reacted with a certain degree of exasperation to word-finding and other difficulties. Possibly, this will give S1 the impetus to create more opportunities for using and thus maintaining German.
Summary: Case studies 1 and 2
Despite several similarities in their biographies, and the fact that they were fully proficient native speakers of German at the time of emigration, S1 and S17’s languages have developed differently. We find loss and growth in both participants’ languages, but these are brought on by different forces. S1’s case is very interesting in that her conscious efforts to maintain German, even if this was done for the sake of the children (a similar motivation as for S17) led to nothing, while her improvement in German, when it happened, seemed to surprise her more than anybody.
Personal and environmental variables
These two case studies highlight a number of interacting variables related both to the participants and their environment, which I explore further in this section. When asked what had helped participants achieve their current level of proficiency in English, and learn or maintain their languages more generally, they pointed overwhelmingly to their experience of living abroad. Having previously learned English and other languages at school, and variably enjoyed doing so, participants differed in the extent to which they had realised the attendant opportunities for friendships, job opportunities and travelling. However, even some of the language enthusiasts commented on having found certain aspects of instructed language study (such as formal/grammar/rote learning, the absence of practical application and the lack of progress) boring or difficult.
While many agree that having a basis in a language prior to experiencing immersion is helpful and speeds up progress, the case of two essentially self-tutored participants would seem to indicate that it is not a necessary prerequisite. Living in the L2 environment has afforded participants opportunities for speaking and discovering the language in meaningful contexts. Listening to the radio and the television, copying others (the ‘parrot system’, as one participant, S5, called it), watching films and news in the original language, as well as reading books, although in principle possible in the L1 environment, is much more easily accomplished in an immersion context.
Participants listed a wealth of what might be called learner strategies that utilised ‘all channels’ (S18) for learning and developing all language skills in tandem. The most important ones were active participation in conversations and written language use (i.e. focusing on the productive skills), in addition to exploiting and maximising the input present in the environment. More than half highlighted the need to be prepared to take risks by trying out the language and taking the initiative. Several participants also mentioned the role of their partners in providing feedback, input and opportunities for language use.
The bilingual participants, by their own testimony, have benefited from a rich linguistic environment and sufficient input and practice in achieving a relatively high level of English language proficiency, compared with the German controls. This link is confirmed by a significant correlation between the level of English proficiency and the amount of English language use amongst the bilinguals (see Table 4, for this and the following correlations cited). The importance attributed to a partner from the host culture for creating opportunities to use English at home and possibly easing access to an English-speaking circle of friends was also statistically confirmed.
Significant correlations between predictors and z-scores.
Correlation is significant at the 0.05 level (two-tailed).
Correlation is significant at the 0.01 level (two-tailed).
On the other hand, input disruption or reduction since migration may explain the lack of success in maintaining other L2s, and indeed the L1. Less frequent use of German, along with a longer length of residence abroad, or in other words, a longer time since German language use decreased quantitatively and qualitatively as a result of migration, correlate significantly with poorer German maintenance.
In relation to other languages, where no use data were elicited, S17, for example, observed with a certain degree of exasperation that her French used to be her strong language, but that it suffered while her English improved and eventually overtook French. Echoing others, she pointed to a lack of opportunities to practice, given that they are not naturally present, and a lack of time to create such opportunities.
The perceived importance of input factors also becomes apparent in response to the question of what participants felt could be done in order to counteract the forgetting of languages. Strategies mentioned closely resemble the ones listed in relation to L2 acquisition and predominantly hinge on ways to maximise practising the language. Similarly, when considering what, if anything, participants would change in their linguistic behaviour if they could go back to the time of emigration, some participants wished they had used more German from the start or had not neglected their other languages, while others regretted not having had a better foundation in English, or having isolated themselves from the host community at the beginning.
However, the replies also hint at other variables, such as attitudes towards particular languages, language learning experience and linguistic aptitude, which seem to play a part in both language acquisition and maintenance. It is likely that these variables greatly interact and confound one another, since ease of learning a language may lead to greater enjoyment of language learning, and a more positive attitude towards language(s), perhaps resulting in the resolve to learn more languages and to create opportunities for maintenance.
Thus, higher values placed on bilingualism (rather than on German versus English) correlated significantly with higher German maintenance (Table 4), while many participants commented that they simply did not need their other L2s now, and therefore experienced a lack of motivation to ‘work’ (S13) on and keep up those languages. I did not test the impact of attitudes on participants’ proficiency in those languages, but a positive orientation towards English and towards foreign languages in general correlated significantly with higher levels of English proficiency.
Discussion and conclusions
These findings provide converging evidence for many of the dynamic principles presented at the beginning of this paper. The results show intra-individual variation over time, with languages being forgotten and relearned. Moreover, they point to variability between participants, or different outcomes in the development of the language system despite similar initial and continuing conditions, in the differing bilingual profiles of the participants. As the case studies show, individual developmental trajectories are not linear, nor are the relationships between different variables, such as attitude, and the proficiency levels achieved or maintained. The languages system adapts to and is affected by the environment and individual preferences and needs, but although these co-adaptations are complex in principle, developmental patterns can be discerned, which are supported by statistically significant interactions (see Table 4).
The picture that emerges from the participants’ narratives is that all languages, including the L1, are subject to the same constraints and require looking after, in order to be acquired and maintained. The changes in participants’ language proficiencies are congruent with the linguistic environment participants find themselves in, and the variables referred to relate to language needs and opportunities for language use, which in turn interrelate with language attitudes and practical constraints.
The results echo findings from several recent studies (e.g. Birdsong, 2006; Birdsong & Molis, 2001; Bongaerts, 1999; Flege & MacKay, 2004; White & Genesee, 1996) with regard to L2 attainment. These studies argue that native-like attainment in late bilinguals ‘is not typical, but neither is it exceedingly rare’ under conditions of ‘sufficient L[ength] o[f] R[esidence] and contact with natives’ (Birdsong, 2006, p. 20). Bongaerts, Mennen, and van der Slik (2000, p. 298) similarly conclude from their analysis ‘that a combination of input, motivational, and instructional factors may compensate for the neurological disadvantages of a late start’ for L2 acquisition, possibly facilitated by typological proximity (Birdsong & Molis, 2001).
The findings presented here suggest that the same factors are operative in relation to the L1. In other words, any of the languages in a multilingual system may change, and it does so in interaction with internal and environmental variables that create, though not in an absolute way, the conditions for likely trajectories for language development.
Language development and the Dynamic Model of Multilingualism
These observations, in particular the importance attributed to language use, also tally with the principles espoused in Herdina and Jessner’s Dynamic Model of Multilingualism (DMM) (Herdina & Jessner, 2004). A central notion in the DMM is that of ‘general language effort’, the composite of ‘acquisition effort’ and ‘maintenance effort’. All languages of a multilingual system are postulated to require adequate language effort for acquisition and/or maintenance. If insufficient effort is spent on a particular language, then it will stop growing or even attrite, either in terms of general proficiency, or it will become functionally specialised. For psychological and practical reasons, the overall effort a person is willing or able to expend towards their language system is thought to remain relatively constant over the lifespan.
Whether language work is in fact experienced as ‘effortful’ probably depends on the context a person finds herself in, and the stage of language development. If the context is congruent with the language to be learned or maintained, like supporting the mother tongue in the L1 environment, no particular initiative on the part of the speaker would usually be required, since opportunities for maintenance are naturally present. On the other hand, the task to acquire the language of the host community in an L2 environment to a sufficient level does appear to require effort, according to participants’ narratives. The linguistic ‘stress’ (S4) experienced by some participants during the first few years after migration is probably due to a high acquisition effort for English, accompanied by increasing maintenance effort for English and high maintenance effort for German and possibly other languages. These demands exceed the overall language effort usually spent by people, who consequently feel linguistically stretched in the attempt to keep up with the increased demands, and may compensate for them by reducing their effort in other parts of the system.
DMM sees the language maintenance effort increase in second-order terms relative to the growth of the language system. In other words, a more complex language system requires proportionally higher language maintenance, leaving fewer resources for further increases in proficiency. This notion seems to be supported by a participant who expressed the view that at that particular stage of her language development, ‘achieving a higher level [in English] would require a lot more effort’ (S24).
The fact that attrition in German became apparent only after a lag time in the new environment is also plausible within this model. DMM models language development – positive and negative – as an S-curve in line with biological growth models, where rapid growth or loss is preceded and followed by flatter inclines/drops (Herdina & Jessner, 2002, p. 89). This general pattern seems to be confirmed by the participants’ narratives about their acquisition of English. Actual changes in proficiency are preceded by ‘increased scatter’ or optionality (Herdina & Jessner, 2002, p. 96). A related concept is that of critical thresholds, below or beyond which proficiency changes are less likely or less noticeable. Before migration, German would have been known at or close to asymptote, the ideal native-speaker proficiency level. In other words, the level of German language proficiency had passed a critical threshold, beyond which language knowledge is relatively resistant to forgetting (de Bot, 2007). Forgetting therefore only registers after a relatively longer time and after more ‘neglect’, all other things being equal.
Similarly, the differential success in maintaining German over L2s other than English may on the one hand be attributed to better maintenance opportunities for German, which continues to be used, and the fact that German was a well-established system at the time of emigration and consequently is better able to absorb the effect of reduced maintenance effort on the other. While English demands a high proportion of language effort, whether experienced as effortful or not, other languages also need support for maintenance in line with their proficiency level, and where not provided, they fall to a proficiency level consistent with that support.
Under the general linguistic conditions presented by the L2 environment, the participants’ language development in the L2 environment appears to have been unidirectional to a degree. However, this obscures a large amount of inter-individual variation and variability over time. In DMM terms, such variation is a direct consequence of the dynamism of the multilingual system, for two reasons: firstly, the non-linear interaction of various factors amongst themselves and with the language system, and secondly, the compensating ‘M-factor’ (multilingual aptitude/metalinguistic abilities) on the other (Herdina & Jessner, 2002, pp. 111, 116).
Concerning the latter, the results show significant correlations between some parts of the linguistic aptitude test and language proficiency (see Table 4 and Opitz, 2011, p. 302). The effect is stronger for English, where both the lexical memory test and the grammatical inference test were correlated with the proficiency scores. English proficiency also correlated with the number of languages known. For German, the lexical memory test produced a significant effect. Thus, the M-factor seems to be relevant both for successful L2 acquisition, and L1 maintenance, in line with the posited interconnection principle.
As reported above, significant effects were also found for some of the other variables under discussion. However, a large part of the variation remains unexplained. This would seem to indicate that the interaction between the various predictors and the language system is neither simple nor straightforward.
To conclude, this paper has discussed the relevance of dynamic approaches and, in particular, the DMM for studying the language development of adult migrants. The concept of ‘language effort’, as a function of language use, proved especially useful, both from the participants’ perspective and in the computed correlations. English demands a large part of the available effort, initially for acquisition and later for maintenance. Languages that are subject to reduced maintenance effort – L1 and other L2s – show evidence of attrition. There is a further multitude of variables that more or less predictably shape participants’ specific linguistic setups and the trajectories of their linguistic development. This paper has provided insights into these trajectories, and the variables with which they interact.
Footnotes
Acknowledgements
I would like to acknowledge the helpful suggestions received from Dorota Lubińska, Lorna Carson and the anonymous reviewers during the preparation of the manuscript.
Funding
This work was supported by Dublin City Council.