| The identification of Spanish-English (SE) bilingual children with language impairment (LI) is challenging. Currently there is a dearth of language screening measures designed for Spanish-speaking populations; thus, the construction of a new screening measure is warranted. Non-word repetition (NWR) tasks that measure phonological short-term memory are very successful identifying English-speaking children with LI; however, research on the identification of LI in SE bilingual children using NWR tasks is limited. Further, the impact of non-word characteristics, such as the language that the non-words are based on (language basis), syllable length (length), phonotactic probability (PP) and neighborhood density (ND), on NWR accuracy in bilingual children with LI and with TD have yet to be evaluated. A Spanish-only and an English-only NWR task that controlled for PP, ND, and syllable length were created and administered to 125 SE bilingual children, 79 with typical language development (TD) (Mean age 75.62 months, SD=8.74) and 53 with LI (Mean age 78.33 months, SD=7.88). SE bilingual children with LI accurately repeated significantly less percent phonemes correct on the Spanish and English NWR tasks than same age peers with TD. An evaluation of the non-word characteristics as predictors of NWR accuracy indicated that the linear combination of language basis, length, and ND accounted for more than 60% of the variance in Spanish and English NWR accuracy. Furthermore, the linear combination of language basis, length, and ND significantly accounted for 14% of the variance in the between group (TD vs. LI) difference in Spanish and English NWR accuracy. The clinical utility of the NWR tasks to detect LI in SE bilingual children was evaluated. Results indicated that the Spanish-only and the English-only NWR tasks did not provide adequate classification accuracy for the screening of LI in SE bilingual children. An improvement in classification accuracy was observed when using a combination of non-words from the Spanish-only and English-only NWR tasks; however, the likelihood ratios for the combined task were not adequate to rule in a child with LI. Results of the study do not support using a NWR task as an independent method for identifying LI in SE bilingual children. |
