Subject-verb agreement in German in bilingual children with and without SLI

Abstract

Abstract:We investigated the acquisition of subject-verb agreement (SVA) in German based on spontaneous speech data from both typically developing (TD) and language-impaired (SLI) children learning German as a second language (L2), and from language-impaired monolingual children. Rothweiler et al. (2012) suggested that SVA is vulnerable in SLI. The intent of our study was to determine whether TD child L2 learners (cL2) in contrast to children with SLI do not have difficulties with SVA. Many studies report that cL2 acquisition can differ from monolingual (L1) acquisition and that the differences look similar to those of impaired L1 learners, which makes it difficult to distinguish cL2 learners who have SLI from those who do not. If such similarities were to be found in the acquisition of SVA in the two bilingual groups, SVA would be ruled out as a possible marker of SLI in cL2 learners. Three groups of children were compared: six Turkish-German early cL2 learners without SLI, and twelve children with SLI -six monolingual German children and six Turkish-German early cL2 learners. As in Rothweiler et al. (2012), all children were advanced learners and were therefore expected to have acquired SVA. We found that the unimpaired early cL2 learners had indeed successfully acquired SVA. In contrast, neither the monolingual nor the bilingual children with SLI succeeded in reliably producing correct SVA.

Childhood bilingualism and Specific Language Impairment: A study of the CP-domain in German SLI

“…The SLI-L1 children received individual language therapy and/or attended special language therapy classes and were diagnosed as having SLI by speech therapists. According to the clinicians' reports, their non-verbal cognitive abilities were within the normal range for their chronological age, and there were no reported hearing loss, obvious neurological dysfunction or motor deficits; see Bartke (1998), Clahsen et al (2014), and Rothweiler et al (2017) for more information.…”

“…Furthermore, correctly agreeing -e or -n forms in sentences with overt subjects were coded as ‘finite’ (3d), and non-agreeing -e or -n forms in sentences with overt subjects were treated as ‘non-finite’ (4a); see Clahsen & Penke (1992) and Rothweiler et al (2012) for further justification. Special attention was given to bare unmarked stems ( Ich/du/er kauf ‘I/you/he buy’) which have been argued to be finite by some (Schulz & Schwarze, 2017; Prévost, 2003) and finite or non-finite by others (Rothweiler et al, 2012, 2017). We analysed verb placement for these forms separately.…”

“…Regular affixes are - e , -s(t) , -t , and –n . Here, we reconsidered the data from Rothweiler et al (2012) and Rothweiler et al (2017), which included the same participants and the same recordings as the current study. Since non-parametric methods were employed in these earlier studies, we reanalysed these data to make sure that the results are replicable using the same statistical methods that were applied in the current study for CP-related phenomena.…”

“…Firstly, proportions are inherently bounded between 0 and 1 and their error variance is not independent from the mean (Barr, 2008). Even when such statistical violations are handled by transforming proportions or by employing non-parametric methods (as was done in Rothweiler et al, 2012 and Rothweiler et al, 2017), there are still biases in the associated p -values, which may give rise to spurious significances or null results (Jaeger, 2008). Secondly, the use of logistic regression allows taking into account the actual amount of data that was generated (i.e., the number of utterances produced by each child), whereas an analysis on proportions completely eliminates this information.…”

“…Other researchers have argued against the CP-domain as the core of the grammatical difficulties of German-speaking children with SLI. Rothweiler et al (2012; Rothweiler, Schönenberger & Sterner, 2017) showed that subject-verb agreement is selectively impaired in both monolingual and bilingual German–Turkish children with SLI, even in children who produce well-formed embedded clauses and wh -questions. Consequently these authors proposed subject-verb agreement rather than the CP-domain as a linguistic marker for SLI in bilingual German-speaking children.…”