Bilingual children acquiring Russian and German in Vienna: nonword repetition correlates with stronger but not with weaker language

Today a majority of children grow up multilingually, because of the many demographic changes. In Austria, German is the official language needed for the daily life and for education. Nevertheless, about 60% of children attending kindergarten in Vienna acquire another first language (L1) [1]. These children often underperform in their second language (L2) Austrian German with regard to the expectations of professionals, like teachers, psychologists, pediatricians and others. This observed lower L2 proficiency compared to monolingual children often leads them to take a developmental language disorder (DLD) into consideration because the first stages of L2 acquisition can resemble the patterns observed in impaired monolingual language development [2, 3]. There is currently a lack of language tests with norms for bilingual children, because most of the available language tests are normed for monolingual children with the aim to distinguish between typical and atypical language development. Nonword repetition tests (NWRT) are an alternative for language tests with bilingual children, because they can also be used with children who have limited exposure to a given language. In NWRT, children are asked to repeat nonwords containing one or more syllables that they have not heard before and therefore they cannot be part of their learned vocabulary. As a consequence, bilingual children with limited exposure to a language are less disadvantaged as in other language tests compared to monolingual children of the same language [4]. NWRTs directly evaluate language proficiency at the phonetic–phonological level, specifically phonological working memory, speech perception and short-term memory and indirectly give insight into phonological awareness, word learning, and overall language acquisition. Therefore, NWRTs have proven to be valuable tools for diagnosing DLD together with other tests like sentence repetition and a parent questionnaire on acquisition of the first language [5, 6]. However, there are still open questions regarding the link between NWRT performance and language development in typically developed children. While NWRT, unlike other language assessments, do not draw directly on knowledge of vocabulary and grammar, it is still influenced by language-specific knowledge, mainly phonology and phonotactics. There is evidence that children are better able to repeat nonwords, which share phonological characteristics of real words in their language. Since such knowledge relies on language exposure, bilingual children should vary in their nonword repetition performance depending on familiarity with lexical phonology of the language on which the nonwords are based upon [4, p. 126]. Up to now in the literature different results have been reported: some studies did not find effects of language exposure on bilingual children’s NWR performance (e.g., [7‐9]), while other studies found a correlation between NWR performance and language exposure [10, 11]. NWR performance was shown to be significantly predicted by language proficiency, specifically by receptive and expressive vocabulary in typically developed children [12]. In a more recent study [13], the significant correlation between (passive) vocabulary and NWR proficiency was again confirmed, whereas a significant correlation between NWR proficiency and language exposure or maternal education could not be found. The authors suggest that NWR performance is an indicator of language proficiency independent of lingual and socioeconomic status.

Thus, it remains an open question whether the amount of language exposure influences not only the development of the stronger and weaker language but also performance in language-specific NWR.

Sociodemographic parameters show that 84.2% of mothers and 57.9% of fathers have a university degree and 42.1% of the households has an income of more than 50,000 € (Table 3).

A descriptive analysis of the children’s language proficiency in Russian and German shows that all children (except child 3 who was excluded from further analyses, see above) could be classified as having a typical language development (Figs. 1 and 2).

Results of the Russian language tests showed that most of the bilingual children scored within the mean range in all subtests. Only in the subtest comprehension of verbs did two children score two standard deviations below average. In the other subtests, none of the children received such low scores. In their L2 German, in contrast, most of the children scored two standard deviations below the mean range in most of the subtests. In German, only one child attained results in the average range or higher in all subtests. In their high exposure language Russian, the variation of test results between children was small, whereas in the low exposure language German the variation was large.

In the present study, language proficiency and its correlation with NWRT was tested in 20 bilingual children who were between 4 and 6 years of age and who simultaneously acquired Russian and Austrian German. All of the bilingual children live in families representing a higher social class of Russian migrants in Vienna. Their stronger language Russian is spoken from one or even both of the parents at home more often than German. The results of this study confirm that the language to which bilingual children are most often exposed evolves into their stronger language in terms of proficiency of language comprehension and production. Results of the Russian language tests showed that most of the bilingual children scored within the mean range. Only in the subtest comprehension of verbs were the scores of two children very conspicuous. In the other subtests, none of the children received such low scores. In their L2 German, in contrast, most of the children scored two standard deviations below the mean range in most of the subtests. The participants live in families who are members of a Russian Facebook group and favor Russian language and culture. Many studies show that input plays an important role for language proficiency since it activates the innate structure of language in children (e.g., [21]). The children of this study have more language exposure in Russian, and develop Russian as their stronger language and German as their weaker language. Furthermore, concerning NWRT, the results show that Russian NWRT and German NWRT correlate, suggesting that the children’s ability to repeat syllables they do not usually hear and that reflect the phonology of the respective language is connected in the two languages of the children. NWRT in German does not reflect the lower input in German as compared to the higher input in Russian. The bilingual children in this study had high scores in both the Russian and the German NWRT. The data, thus, offers evidence that the ability to repeat language-specific nonwords does not differ depending on language exposure (classifying Russian as the high exposure language, and German as the low exposure language). The results for language exposure and NWRT in L2 German to some extent support results previously reported in the literature. Thordadottier and Brandeker [7], for example, found that NWR was less affected by previous language exposure than vocabulary measures in simultaneously bilingual French–English children. Similar results were reported by Thordadottier and Juliusdottier [8], who found that their L2 Icelandic bilingual children achieved high scores in NWRT and low scores in a measure of language. Brandeker and Thordadottier [9] also concluded that NWRT and vocabulary measures are differently associated to language exposure insofar that vocabulary acquisition is associated in a moderate to strong manner to language exposure, whereas the relationship of exposure with NWRT is weak or nonsignificant.

The results of this study also suggest that language exposure influences which of the languages of typically developed simultaneously bilingual children evolves into their stronger or the weaker language, but that language exposure does not influence the performance in NWRT. However, concerning the correlation between language proficiency tests and NWRT, the results of this study suggest a differentiation according to language exposure. Whereas NWRT correlates with language comprehension and production in Russian, i.e., in the language to which the children were most exposed, there is no correlation between NWRT and language comprehension and production (nouns and verbs) in German, namely in the language to which the children were less exposed. Moreover, in the children’s stronger language Russian, language proficiency is linked to NWRT in both L1 Russian and L2 German, whereas language proficiency in their weaker language German is not linked to NWRT in either of the two languages.

An explanation of this discrepancy in the correlation between language proficiency and NWRT in the results could be that Russian phonology and phonotactics are more complex than German phonology and phonotactics (see, for example, [22]), suggesting that acquisition of the complex structures of Russian phonology encloses acquisition of simpler structures of German phonology. Makarova and Therekova [23] who studied the acquisition of Russian phonology in children in a migration context state that these children show similar characteristics concerning the development of their phonetic inventory and phonological processes as compared to their monolingual peers, because they had a high level of Russian language exposure in their families. The children of this study also have a high level of Russian language exposure in migration context and the authors therefore assume that their phonetic inventory and phonological processes are comparable with their monolingual Russian peers. So acquisition of these complex phonological structures in L1 Russian could enclose simpler structures in L2 German. As Ulbrich and Wiese [22] state, Russian shows a more extended patterning of consonant clusters than German. Thus, it differs from German for example in the complexity of possible sound combinations in the syllable coda, in that Russian allows more consonant sequences and more frequent violation of sonority requirements compared to German. We suggest that, at least in the case of the Russian–German bilingual children of this study, language exposure only indirectly influences performance in NWRT, insofar as NWRT in both languages is linked to language development in the stronger language Russian, to which the children are more exposed to and which shows more phonotactic complexity compared to German. Further research desiderata that build on the present study are necessary: studies investigating the correlation between NWRT and language proficiency in L1 and L2 in German–Russian bilingual children whose stronger language is German instead of Russian, studies regarding acquisition of expressive and receptive phonology in bilingual children acquiring Russian and German simultaneously as well as studies investigating the correlation between NWRT and language proficiency in bilingual and typically developing children acquiring other combinations of languages. The aim to distinguish between typical and atypical language development based on NWRT can be reached only when NWRT and all possible influencing factors like, for example, the role of language exposure in typically developed children are analyzed.

The most important limitation of our study is the size of the group which characterizes it as a pilot study. It focuses only on one pair of languages Russian and German. Even though the findings show that NWRT is not affected by language exposure, the authors did not differentiate between cumulative and current exposure and used a cumulative measure of exposure [9]. Further limitations are that language skills in both languages are measured with a screening method instead of standardized tests, and that only a small number of participants are included.

Many bilingual children are suspected to have a developmental language disorder (DLD) because of their lower proficiency their second language (L2). Language evaluation in both languages is complicated by many facts like lack of normed language tests and different amount of exposure in the languages. The nonword repetition test (NWRT) is influenced by language-specific knowledge but does not draw directly on knowledge of vocabulary and grammar. Therefore, we recommend to use language-specific NWRT in bilingual children who are suspected to have DLD, but to interpret test results with caution. There are still many open questions regarding the influence of language exposure and phonological structure of different languages on NWRT performance.