Dr. Iryna Khodos

Bilingual Language Behaviours in Multicultural Australia: An Insight into the Nature of Bilingualism and Directions for Sustainable Community Language Practices

Dr. Iryna Khodos

University of Newcastle, Newcastle, Australia

iryna.khodos[at]newcastle.edu.au

Abstract: In today’s pluralistic world, bilingualism is a growing reality, which has furthermore been suggested to contribute to the social and cognitive health of people. Given the inter- and intra-individual variability in bilingual experience, investigating variations in bilingual language practices and further identifying which of them are more efficient in sustaining languages and potentially improving cognitive health of the communities across the world are an important research priority. In the present study, we considered such aspects of bilingual experience as typological proximity/distance, onset age of active bilingualism, language proficiency and language entropy, and explored the way they interplay with (meta)linguistic skills in bilingual adults. Using a background questionnaire and a sentence-judgement task, demographic and language data were collected from 60 linguistically diverse bilingual adults residing in Australia. The results of multiple regression analyses revealed that three of the language variables considered – language proficiency, typological proximity/distance and onset age of active bilingualism – accounted for the variance in metalinguistic data. Specifically, higher levels of language proficiency, use of typologically closer languages and earlier onset age were related to higher metalinguistic scores. These findings reinforce the view on bilingualism as a multidimensional experience, whose consequences depend on a number of distinct but interrelated language learning and use variables. Taken together with a high correlation between language use and language proficiency, the results also suggest that bilinguals who have equally used two languages in the same contexts but with different speakers are more likely to obtain and maintain higher levels of proficiency in both languages, which together with an earlier active use of two typologically closer languages may allow bilinguals to experience advantages in metalinguistic skills. These findings, therefore, underscore the importance of educational programmes and community language practices, which allow people to learn and equally use each of their languages, for maintaining and further developing two/multiple languages.

Keywords: bilingual experience, metalinguistic awareness, language proficiency, language behaviours

1. Introduction

The ability to express the same thought in two different languages is considered to lead to an increased awareness of formal and substantive properties of language (i.e. enhanced metalinguistic awareness; Bialystok, 2001; Galambos and Goldin-Meadow, 1990; Jessner, 2008; Lambert, 1990; Vygotsky, 1962). Given that metalinguistic advantages may, in turn, generalise to other areas of cognitive abilities and thereby contribute to cognitive health, investigating the mechanism underlying metalinguistic benefits in bilinguals is an urgent research task. This is especially true in multicultural Australia, where one in five people speaks a non-English language in addition to English (Australian Bureau of Statistics, 2016). Exploring (meta)linguistic dimensions of bilingualism in this context is, therefore, also crucial for understanding matters concerning integration and socialisation of people as well as for political and educational decision-making. In the present study, we extended the previous metalinguistic research on bilingualism by exploring the interplay between specific dimensions of language experience and metalinguistic skills in bilingual adults in multicultural Australia.

The idea of bilingualism boosting metalinguistic awareness was first expressed by Vygotsky (1962). The psychologist was, moreover, the first to point to the possibility of metalinguistic advantages to contribute to cognitive benefits across the cognitive domains, with the effects depending largely on the metalinguistic skills induced by the use of more than one language. To investigate this enticing hypothesis, Bialystok and Ryan (1985) suggested and innovatively implemented the dual component model. Conceptualising metalinguistic awareness as a form of language processing, the researchers developed the tasks targeting its two skill components: (1) the analysis of linguistic knowledge into structured categories, and (2) the control of attentional procedures to select and process specific linguistic information. For instance, on the basis of existing word awareness tests, Bialystok et al. (2003) developed a sound-meaning task, which required participants to select which of two words matched a target for either the sound (rhyme) or meaning (synonym). Also, Bialystok (1986) manipulated the characteristics of a sentence-judgement task by constructing sentences that were grammatically correct, grammatically incorrect but meaningful or semantically anomalous but grammatical.

The results of Bialystok’s empirical investigations showed that bilingualism does not have a direct effect on metalinguistic awareness rather it influences the two underlying skill components and, what is more, in a different way. In her studies, the bilingual advantage was seen primarily in tasks demanding a high level of control of linguistic processing. In these tasks, the specific skills of the participants in L1 and L2 were not shown to affect their performance. Thus, a superior performance of bilinguals was seen to be due to the early bilingual experience of dual language systems and frequent attention to formal aspects of language. On the other hand, in tasks requiring high levels of analysis, findings were somewhat mixed and depended on the combined proficiency of bilinguals in both languages. Bilinguals were shown to outperform monolinguals only if they had high levels of proficiency.

Bialystok’s findings are in line with a number of other studies on metalinguistic awareness (Davidson et al., 2010; Galambos and Goldin-Meadow, 1990; Hakuta and Diaz, 1985; Ricciardelli, 1992). Similar to Bialystok, these researchers point to a positive influence of mastering two languages on person’s ability to control the processing of linguistic information. These findings are also consistent with the ‘threshold hypothesis’ proposed by Cummins (1977), according to which an overall bilingual superiority in terms of metalinguistic abilities is found only for those who have attained a high degree of bilingualism.

Given the multidimensional nature of bilingualism (de Bruin, 2019; Khodos and Moskovsky, 2020; Laine and Lehtonen, 2018), we expected metalinguistic skills to be affected by different language experiences differently. However, most of the previous metalinguistic studies have not considered the inter-individual variability in their participants’ language experience. Instead, they treated bilingualism and monolingualism as categorical constructs and compared bilinguals and monolinguals as two distinct groups, with their members categorised either as a homogenous whole or in terms of binary oppositions (e.g., early vs late, simultaneous vs sequential, more proficient vs less proficient, L1-dominant vs L2-dominant, balanced vs unbalanced). Therefore, they were not able to uncover the mechanism underlying metalinguistic advantages in bilinguals.

In order to address the limitations of previous metalinguistic research, we examined metalinguistic skills in linguistically diverse bilingual adults, with the inter-individual variability in their bilingual experience being analytically taken into account. Specifically, we explored the following research questions: 1) which combination of bilingual experience accounts for the most variance in bilinguals’ metalinguistic performance, and (2) the extent to which each variable in the combination contributes to explaining the variance in participants’ metalinguistic scores. Given the peculiarity of the bilingual sample in the present study, we were able to consider the following dimensions of bilingual experience: (1) typological proximity/distance between two languages (Germanic languages/smaller distance and non-Germanic languages/larger distance), (2) onset age of active bilingualism (age at which they began using both languages actively on a regular basis), language proficiency (average language proficiency in two languages) and language entropy (the frequency of use of two language).

2. Procedure

Bilingual adults (aged 20-40 years) speaking English as a second language were recruited from the research sites located in the Newcastle/Hunter area, NSW, Australia. The participants were asked to fill in the Language and Social Background Questionnaire (Anderson et al., 2018) to get data on key demographic and language variables. Following that, the bilinguals completed the sentence-judgement task (Bialystok and Barac, 2012; Moreno et al., 2020) for their metalinguistic skills to be assessed.

2.1 Study setting

All the participants resided in Australia, which has a heterogeneous population compared with other countries in the world. There are Indigenous people, descendants of the original UK settlers and a diverse group of immigrants, who either come to Australia as bilinguals or develop bilingual knowledge in the years following their arrival. Considering this, Australia is a multilingual and multicultural country, with the official language, English, coexisting with Aboriginal and immigrant languages.

According to the 2016 Census data (Australian Bureau of Statistics, 2016), one in five Australians now speaks a language other than English at home. Among them, the most commonly spoken ones are Mandarin, Arabic, Cantonese, Vietnamese and Italian, as compared to Italian, Greek, Cantonese, Arabic, Mandarin and Vietnamese in 2006. This points to two opposite tendencies in Australian society: a substantial decrease in the home use of a number of European languages (in particular German, Italian and Greek) and a great increase in Asian languages, especially Mandarin. In other words, the linguistic diversity of Australia is shifting away from the European languages of the post-war period to languages of new migration waves, mainly from Asia and the Middle East (Clyne et al., 2008).

The substantial number of non-English languages notwithstanding, Australia remains a strongly Anglocentric country, where the dominance of English is largely unchallenged (Rubino, 2010). As pointed out by Clyne (2005), the ‘monolingual mindset’ is still one of the key challenges of modern Australian society. The majority of its native English speakers do not speak any other language. Moreover, they show limited interest in languages and/or language study. In addition to cultural and social attitudes, the limited availability and accessibility of language programs in Australian institutions is another possible barrier to cultivating Australian bilingualism/multilingualism (Rubino, 2010).

In this light, bilingualism and multilingualism in Australia appear to be represented mainly by Aboriginal people and immigrants from non-English speaking backgrounds (individual bilingualism). However, even they tend to abandon their native languages relatively quickly as a consequence of lack of opportunities to apply their native language in broader social contexts (mostly single-language contexts) and lack of institutional support.

2.2 Participants

The sample consisted of 60 bilingual adults (20-40 years old), including 22 males and 38 females. All of them held a higher university degree – either Bachelor’s or Master’s degree (M = 4.00, SD = .00). Thus, education was not considered in further analysis. Descriptive statistics are provided in Table 1.

The bilinguals were from varied non-English speaking backgrounds. Their first language belonged to one of the following language branches: Germanic (11); Romanic (13); Slavic (7); Iranic (9); Indo-Aryan (5); Sinic and Tibeto-Burman (6) and other (9), including Vietnamese (2), Greek (1), Cambodian (1), Azerbaijani (1), Malay (1), Filipino (1), Malayalam (1) and Shona (1). Most of them had started acquiring English in childhood (M = 9.35, SD = 4.64) in a single first language-oriented environment and had begun using both languages on a regular basis (in the same or different contexts) shortly before or upon arriving in Australia (i.e. onset age of active bilingualism; M = 21.33, SD = 7.83).

Table 1: Descriptive statistics for demographic and language variables

Variables N Mean SD

Demographic variables

Gender male – 22

female – 38 - -

Age 60 31.92 4.45

Education 60 4.00 .00

Language variables

Onset age of active bilingualism 60 21.33 7.83

L1 [non-English] proficiency 60 9.33 .76

L2 [English] proficiency 60 8.16 1.00

Language use in close social context 60 3.53 .50

Language use in broad social context 60 1.98 .13

Note. Age and onset age in years. Education on a 4-point scale (1 = upper secondary, 2 = post-secondary non-tertiary, 3 = short-cycle tertiary, 4 = tertiary education). Language proficiency on a 10-point scale (0 = no proficiency, 10 = high proficiency). Language use on a 5-point scale (1 = all English; 3 = half English, half the other language; 5 = only the other language).

In Australia, the bilinguals were immersed in a mostly single second language-oriented environment: on average, they indicated the use of mostly English in terms of broad social contexts (M = 1.98 on a 5-point Likert scale, SD = .13). Nevertheless, the participants varied in the way and extent to which each of the two languages was used in close social contexts (M = 3.53, SD = .50) and language proficiency (L1 [non-English] proficiency: M = 9.33 on a 10-point Likert scale, SD = .76; L2 [English] proficiency: M = 8.16, SD = 1.00).

2.3 Instruments

The Language and Social Background Questionnaire was based on the well-established research tool by Anderson et al. (2018). In line with the original questionnaire, the one used for the purposes of the current study consisted of three sections. The Social Background Section captured demographic information, including age, gender, highest level of education, immigration status. The Language Background Section assessed the number of languages spoken and proficiency for speaking, listening, reading and writing the indicated language(s). Finally, the Community Language Use Behaviour Section elicited information on the language usage at different life stages and in different social contexts.

The sentence-judgement task was developed in accordance with the cross-validated dual component model of metalinguistic awareness (Bialystok and Ryan, 1985). It consisted of 24 sentences presented in context, as part of three short passages. In line with the previous studies using this type of task (e.g., Bialystok and Barac, 2012; Moreno et al., 2020), the sentences were constructed along two linguistic dimensions: a semantic one and a grammatical one. This resulted in four sentence frames: grammatical, meaningful (GM; 6 items), grammatical anomalous (Gm; 6 items), ungrammatical, meaningful (gM; 6 items) and ungrammatical anomalous (gm; 6 items). This way it was possible to target analysis and control components separately: the highest level of analysis was required to deal with gM sentences, while the highest level of control was needed to judge Gm sentences (Bialystok, 2001).The participants were given 20 minutes and were asked to judge whether the given sentences were grammatical or ungrammatical irrespective of their meaning. The key point was that judgements had to be made on the basis of how each of the sentences was used in the given text. In case the ungrammatical option was selected, correction was required.

3. Results

The obtained background and metalinguistic data were subjected to multiple regression analyses conducted in R (version 3.6.1). The participants’ sentence-judgement task scores were used as dependent variables. The 24 items were combined according to the sentence frame (GM, Gm, gM and gm), which resulted in four factors consisting of six relevant components. Descriptive statistics are given in Table 2.

Table 2: Descriptive statistics for the dependent variables

Dependent variables N Mean SD

GM 60 3.99 1.27

Gm 60 4.00 1.40

gM 60 3.45 1.37

Gm 60 3.57 1.50

Note. Number of correct sentence-judgement task items out of six.

Demographic and language variables were entered as predictors. In particular, each regression included two demographic variables: gender (1 = male, 0 = female) and age in years; and a set of language variables: typological proximity/distance between two languages, onset age of active bilingualism, language proficiency and language entropy.

Typological proximity/distance was extracted from the data on bilinguals’ L1 and used as a dummy variable: 1 = Germanic languages, 0 = non-Germanic languages. Onset age of active bilingualism was based on the age at which the bilinguals began using their two languages actively on a daily basis and included as a continuous variable in years. Language proficiency was computed on the basis of the average proficiency score for each language by using the calculation as in Vaughn and Hernandez (2018):

Language proficiency was treated as a continuous variable (0 = no proficiency in each of the languages, 20 = high proficiency in both languages). Language entropy, i.e. a continuous measure of how often languages are used (0 = only one language is used, 1 = each language is used equally), was calculated using the equation as in Gullifer et al. (2018):

Here, n represents the total possible languages (two in the present study) and P_i represents the proportion associated with the use of a given language. The proportion of L1 and L2 use for each bilingual was quantified on the basis of the self-reported language use data. Means and standard deviations for the predictors are provided in the Table 3.

Table 3: Descriptive statistics for the predictors

Predictors N Mean SD

Demographic

Gender male – 22

female – 38 – –

Age 60 31.92 4.45

Language

Typological proximity/distance Germanic – 11

non-Germanic – 49

Onset age of active bilingualism 60 21.33 7.83

Language proficiency 60 17.50 1.46

Language entropy 60 .64 .35

Note. Gender: 1 = male, 0 = female. Age and onset age in years. Typological proximity/distance: 1 = Germanic languages, 0 = non-Germanic languages. Language proficiency on a 20-point scale (0 = no proficiency in each of the languages, 20 = high proficiency in both languages). Language entropy on a 1-point scale (0 = only one language is used, 1 = each language is used equally).

Given a statistically significant correlation between language proficiency and language entropy, r = .50, p < .001, we created two base-line models. Both contained gender, age, typological proximity/distance and onset age of active bilingualism. However, one had language proficiency and the other included language entropy. Following that, we performed multiple regressions with backward elimination using the regsubsets function. The results of the analyses showed that the best-fitting model among all for the sentence-judgement task items was the one with language proficiency among the predictors (see Table 4).

Table 4: The best-fitting models showing the capacity of language variables to predict the sentence-judgement task items

Variables B SEB t Sig.

Gm: R² = 37.5%, p < .001

ΔR²= 37.6%, p < .001

Typological proximity/distance -1.52 .39 -3.86 .001

Onset age of active bilingualism -.08 .02 -2.64 .05

Language proficiency .45 .14 3.13 .01

gM: R² = 29.3%, p < .001

ΔR²= 26.8%, p < .001

Typological proximity/distance -.71 .39 -1.81 .05

Language proficiency .62 .14 4.29 .001

gm: R² = 15.8%, p < .001

ΔR²= 14.3%, p < .001

Language proficiency .56 .17 3.3 .001

Note. Onset age of active bilingualism in years. Typological proximity/distance: 1 = Germanic languages, 0 = non-Germanic languages. Language proficiency on a 20-point scale (0 = no proficiency in each of the languages, 20 = high proficiency in both languages).

In the case of gm, language proficiency was the only predictor in the model (R² = 15.8%, p < .001). As the bilinguals’ language proficiency increased by one point on a 10-point scale, their gm scores increased by .56 points. In the case of gM and Gm, the best model contained more than one predictor. In addition to language proficiency, the best model for gM included typological proximity/distance (R² = 29.3%, p < .001); whereas the best model for Gm contained language proficiency, typological proximity/distance and onset age of active bilingualism (R² = 37.5%, p < .001). In both cases, the participants whose L1 belonged to the Germanic language family performed better (gM: B = .71, p < .05; Gm: B = 1.52, p < .001). The bilinguals also obtained higher scores as their language proficiency increased by one point on a 10-point scale (gM: B = .62, p < .001; Gm: B = .45, p < .01). In the case of Gm, the participants’ scores furthermore increased by .08 points as their onset age of active bilingualism decreased by one year. As for the GM items, none of the models explained the variance in the scores, p > .05.

4. Discussion and conclusion

The study tested linguistically diverse bilingual adults residing in Australia on the sentence-judgement task measuring two metalinguistic skills, i.e. the analysis of linguistic knowledge and the control of attentional procedures. This was done to investigate: (1) which combination of bilingual experience (if any) – typological proximity/distance, onset age of active bilingualism, language proficiency and language entropy – accounts for the variance in bilinguals’ metalinguistic performance, and 2) the extent to which each variable in the combination contributes to explaining the variance in the participants’ metalinguistic skills.

The results of the study showed that variance in participants’ metalinguistic performance was related to differences in their bilingual experience. Specifically, the model with with language proficiency among the predictors accounted for the most variance in metalinguistic scores. According to the data, language proficiency was predictive of the participants’ performance on all the sentence-judgement task items: higher levels of language proficiency were related to higher scores. The use of two typologically closer languages further contributed to better performance on the task items requiring the highest level of analysis (gM), and together with an earlier onset of active bilingualism, it was also related to higher scores on the task items placing the greatest burden on control (Gm).

Consistent with recent studies, our research suggests that particular dimensions of bilingual experience rather than bilingualism per se are linked to bilingual advantages (Bialystok and Barac, 2012; Gullifer et al., 2018; Khodos and Moskovsky, 2020). However, our study is unique in that it provides insight into dimensions of bilingual experience which may boost and further maintain enhanced (meta)linguistic skills in adults. In particular, the results of the multiple regression and correlation analyses suggest that bilinguals who have equally used two languages in the same contexts but with different speakers are more likely to obtain/maintain higher levels of proficiency in both languages, which, in turn, may be related to enhanced analysis skills. When combined with the use of two typologically closer languages and an earlier onset of active bilingualism, higher language proficiency may furthermore allow bilinguals to experience advantages in control skills.

Given that most of the previous studies have not considered the inter-individual variability in bilingual experience while interpreting their participants’ metalinguistic performance, it is quite difficult to reconcile the present findings with the wider literature on metalinguistic awareness. Among the variables explored in the present study, the role of the typological proximity/distance between L1 and L2 appears to have received the least attention. This might stem from the fact that most previous research has worked with participants that were linguistically homogeneous – same L1 and same L2. However, even when the participants varied in their language pairs, the researchers did not control for the typological proximity/distance variable in their studies. Our findings, therefore, extend the previous metalinguistic studies and call for a need to focus more research attention on the individual features of bilingual experience and the ways they interplay with language and other cognitive domains (for related ideas, see de Bruin, 2019; Khodos et al., 2020; Laine and Lehtonen, 2018).

The need for further research notwithstanding, the present study contributes clearly to our understanding that bilingual experience can offer advantages that extend beyond language. These benefits, as a result, can have socially relevant consequences for educational attainment and future socioeconomic success. In the context of multicultural Australian, this underscores the importance of introducing suitably designed educational, social and political policies encouraging bi-/multilingualism and creating the best possible setting/environment for learning and using two/multiple languages. Establishing language learning programmes and promoting social practices that maintain and further develop Indigenous and community languages seems particularly desirable. Along other bi-/multilingual practices, this may have serendipitous benefit of improving the social and cognitive health of multicultural Australia.

References

Anderson, J.A., L. Mak, A.K. Chahi and E. Bialystok. (2018). The language and social background questionnaire: Assessing degree of bilingualism in a diverse population. Behavior Research Methods, 50, pp. 250–263.

Australian Bureau of Statistics (2016). Census Dictionary: Australia 2016, Catalogue No. 2901.0, ABS, Canberra.

Bialystok, E. (1986). Children's concept of word. Journal of Psycholinguistic Research, 15, pp. 13–32.

Bialystok, E. (2001). Metalinguistic aspects of bilingual processing. Annual Review of Applied Linguistics, 21, pp. 169–181.

Bialystok, E. and R. Barac. (2012). Emerging bilingualism: Dissociating advantages for metalinguistic awareness and executive control. Cognition, 122(1), pp. 67–73.

Bialystok, E., S. Majumder and M. Martin. (2003). Developing phonological awareness: Is there a bilingual advantage? Applied Psycholinguistics, 24, pp. 27–44.

Bialystok, E. and E.B. Ryan. (1985). Toward a definition of metalinguistic skill. Merrill-Palmer Quarterly, 31(3), pp. 229–251.

Clyne, M. (2005). Australia’s language potential. Sydney: University of New South Wales Press.

Clyne, M., J. Hajek and S. Kipp. (2008). Tale of two multilingual cities in a multilingual continent. People and Place, 16(3), pp. 1–8.

Cummins, J. (1977). Cognitive factors associated with the attainment of intermediate levels of bilingual skills. Modern Language Journal, 61, pp. 3–12.

Davidson, D., V.R. Raschke and J. Pervez. (2010). Syntactic awareness in young monolingual and bilingual (Urdu-English) children. Cognitive Development, 25(2), pp. 166–182.

De Bruin, A. (2019). Not all bilinguals are the same: A call for more detailed assessments and descriptions of bilingual experiences. Behavioral Sciences, 9, pp. 1–13.

Galambos, S.J. and S. Goldin-Meadow (1990). The effects of learning two languages on levels of metalinguistic awareness. Cognition, 34(1), pp. 1–56.

Gullifer, J.W., X.J. Chai, V. Whitford, I. Pivneva, S. Baum, D. Klein and D. Titone. (2018). Bilingual Experience and Resting-State Brain Connectivity: Impacts of L2 Age of Acquisition and Social Diversity of Language Use on Control Networks. Neuropsychologia, 117, pp. 123–134.

Hakuta, K. and R.M. Diaz (1985). The relationship between degree of bilingualism and cognitive ability: A critical discussion and some new longitudinal data In K.E. Nelson (Ed.), Children's language (Vol. V, pp. 319–344). Hillsdale, NJ: Lawrence Erlbaum.

Jessner, U. (2008). A DST model of multilingualism and the role of metalinguistic awareness. Modern Language Journal, 92(2), pp. 270–283.

Khodos, I. and C. Moskovsky. (2020). Dimensions of bilingualism promoting cognitive control: Impacts of language context and onset age of active bilingualism on mixing and switching costs. Linguistic Approaches to Bilingualism. Advanced online publication. http://doi.org/10.1075/lab.19064.kho

Khodos, I., C. Moskovsky and S. Paolini. (2020). Bilinguals’ and monolinguals’ performance on a non-verbal cognitive control task: How bilingual language experience contributes to cognitive performance by reducing mixing and switching costs. International Journal of Bilingualism. Advanced online publication. https://doi.org/10.1177/1367006920946401.

Lambert, W.E. (1990). Persistent issues in bilingualism. In B. Harley, A. Patrick, J. Cummins, & M. Swain (Eds.), The development of second language proficiency (pp. 201–218). Cambridge: Cambridge University Press.

Laine, M. and M. Lehtonen. (2018). Cognitive consequences of bilingualism: where to go from here? Language, Cognition and Neuroscience, 33(9), pp. 1205–1212.

Moreno, S., E. Bialystok, Z. Wodniecka and C. Alain. (2020). Conflict resolution in sentence processing by bilinguals. Journal of Neurolinguistics, 23(6), pp. 564-579.

Ricciardelli, L.A. (1992). Bilingualism and cognitive development in relation to threshhold theory. Journal of Psycholinguistic Research, 21, pp. 301–316.

Rubino, A. (2010). Multilingualism in Australia: Reflections on current and future research trends. Australian Review of Applied Linguistics, 33(2), pp. 17.11–17.21.

Vaughn, K.A. and A.E. Hernandez. (2018). Becoming a balanced, proficient bilingual: Predictions from age of acquisition & genetic background. Journal of Neurolinguistics, 46, pp. 69–77.

Vygotsky, L.S. (1962). Thought and Language. Cambridge: M.I.T. Press.