Language assessment in multilingual children: the Non-word Repetition Task

Abstract

Language assessment in multilingual children is problematic, due to several biases and inaccuracies in traditional norm-referenced knowledge-dependent language measures. There is a need for a task that is able to distinguish between a language difference caused by second language learning, and a specific language impairment (SLI). It was previously shown that the non-word repetition task (NRT) may be able make this distinction (e.g., Archibald, 2008; Campbell et al., 1997; Conti-Ramsden, 2003; Dollaghan & Campbell, 1998; Ellis-Weismer et al., 2000; Girbau & Schwartz, 2008). The present study examines whether the NRT is suitable as a screening tool to detect language disorders in multilingual children from diverse linguistic backgrounds in The Netherlands. A non-word repetition task was adapted specifically for this purpose. In the first study, the screening study, multilingual children were screened for language problems with the NRT together with a traditional knowledge-based language screening tool. Multilingual children with language problems performed significantly worse on the NRT compared to typically developing multilinguals and monolinguals. A follow-up of the language status (two years after testing) of the multilingual children in the language impaired group revealed that the NRT performance on four-syllable non-words is a more accurate measure than the knowledge-based language screening tool. The second study, the assessment study, examined NRT performance in typically developing multilingual children with children diagnosed with SLI. The results of the present study confirmed the hypothesis that the NRT is able to discriminate between multilingual children with and without language impairment. Specifically, the performance on four-syllable non-words showed to be an accurate measure to identify language impairment. The results confirmed that the NRT is a culturally fair method to identify language disorders, and therefore, the NRT is a promising screening tool in the multilingual population. However, the findings indicate that the performance on the NRT alone is never sufficient to rule in or rule out language impairment. Future research is necessary in order to obtain solid reference data on NRT performance by multilinguals. It may be suggested that mainly four-syllable words should be used.