Development of the Language-Specific and the Cross-Linguistic Non-word Repetition task for the Serbian Language

Abstract

Research question. Currently, there are no standardized tests for language development assessment for the Serbian language. The nonword repetition task (NWR) is shown to be a promising assessment tool and clinical marker for Specific Language Impairment (SLI) in numerous languages. The aim of this research is to develop two versions of the NWR for the Serbian language: the language-specific (LS-NWR) and the cross-linguistic (CL-NWR), for language development assessment of mono- and bi/multilingual preschool children. Methods. The LS-NWR consists of 48 two to four-syllable nonwords, with systematically varied stress position (1st/2nd syllable) and syllable structure (CV, CCV, CVC, CCVC), constructed according to the characteristics of the Serbian language (Savić et al., 2010; Popović, 2017). The task was administered to 75 3- to 7-year-old typically developing (TD) monolingual Serbian children and 50 4- to 8-year-old monolingual Serbian children with SLI. The CL-NWR was constructed within the COST Action IS0804 framework, and consists of 16 two to five-syllable nonwords with a simple CV syllable structure. Two versions of the task (CLs-NWR – nonwords with even stress on each syllable, and CLp-NWR - nonwords with Serbian prosody) were administered (test-retest) to 46 3- to 6-year-old TD monolingual Serbian children. For all tasks, pre-recorded non-words were presented in two randomized orders. The participants’ repetitions were audio recorded, transcribed and coded for total accuracy and types of errors on the word, syllable and phoneme level. Results. The LS-NWR showed high age and group discrimination, with SLI children performing significantly less successfully on the task compared to their TD peers. While both groups performed less accurately with increasing nonword length and syllable structure complexity, children with SLI had more difficulty with this. The CL-NWR showed poor age discrimination, with only the CLp differing three-year-olds from the older children. The CLs task showed to be more difficult than the CLp, due to the fact that children tended to add Serbian prosody to the CLs nonwords, especially when they were tested with the CLp version first (order: test CLp – retest CLs). The repetition accuracy decreased with the increasing nonword length on both tasks. The easiest task for TD Serbian children is the CLp-NWR, while the LS-NWR is the most difficult. Conclusion. The LS-NWR will be further reviewed and the most discriminative nonwords selected using the IRT analysis. The task will be administered to bi/multilingual TD and children with SLI. The CL-NWR will be reviewed according to the current findings (Chiat & Polisenska, 2016), and administered to monolingual Serbian children with SLI, and bi/multilingual TD and children with SLI.