Disentangling SLI and bilingualism using sentence repetition tasks: the impact of L1 and L2 properties

The current study investigated performance on morpho-syntax in Russian-Hebrew sequential bilingual preschool children with and without specific language impairment (SLI) in both languages (L1 Russian and L2 Hebrew) using sentence repetition (SRep) tasks with a fundamental aim to disentangle the language abilities of bilingual children with typical language development (biTLD) from those of bilingual children with SLI (biSLI). Four groups of children participated (N=85) in the study: 45 L1 Russian-L2 Hebrew sequential bilinguals (30 biTLD and 15 biSLI), 20 monolingual Russian-speaking children and 20 monolingual Hebrew-speaking children. The SRep tasks in Russian and in Hebrew were based on Language Impairment Testing in Multilingual Settings (LITMUS) SRep developed within COST Action IS0804 titled Language Impairment in a Multilingual Society: Linguistic Patterns and the Road to Assessment. The tasks in Russian and in Hebrew contained 56 sentences of different length and complexity. The bilingual-monolingual comparisons yielded differences only in L1 Russian, where the monolingual Russian-speaking children outperformed bilingual children with typical language development (TLD) in their first language. In L2 Hebrew, no significant differences emerged between monolingual speakers of Hebrew and bilinguals. Comparisons of bilingual children with and without SLI showed that both the quantity and the quality of errors differentiate the two bilingual groups. In both languages (L1 and L2), bilingual children with TLD outperformed their peers with SLI. Bilingual children with SLI produced specific error patterns like those previously reported for monolingual children with SLI, i.e. omission of coordinators and subordinators, omission of prepositions, and simplification of wh-questions and relative clauses. These error patterns are unique to children with SLI and cannot be attributed to L1-L2 influence, while the errors of children in the biTLD group can be traced to cross-linguistic influence (mostly L2 influence on L1). We conclude that SRep tasks are an effective means to bring us closer to distinguishing bilingual children with SLI from those with TLD. They allow us to make quantitative and qualitative comparisons of performance and errors. But, perhaps more importantly, they point to underlying grammatical representations, in particular those linked to cross-linguistic influence and reduced exposure, as a way of distinguishing bilingual children with and without SLI.