Utility of a Language Screening Measure for Predicting Risk for Language Impairment in Bilinguals

PURPOSE: This study evaluated the accuracy of an experimental version of the Bilingual English Spanish Oral Screener (BESOS; Peña, Bedore, Iglesias, Gutiérrez-Clellen, & Goldstein, 2008) for predicting the long-term risk for language impairment (LI) for a matched group of preschool-aged Spanish-English bilingual children with and without LI. METHOD: A total of 1,029 Spanish-English bilingual children completed the BESOS before entering kindergarten. A subset of 167 participants completed a follow-up language evaluation in 1st grade. Twenty-one of these children were identified as having LI and were matched to a group of 21 typically developing peers from the larger sample. A series of discriminant analyses were used to determine the combination of scores on the BESOS that most accurately predicted 2 years later which children presented with and without LI. RESULTS: The linear combination of the semantics and morphosyntax scores in the best language resulted in predictive sensitivity of 95.2% and predictive specificity of 71.4%, with an overall accuracy of 81% for predicting risk for LI. CONCLUSION: A bilingual language screener administered before kindergarten can be useful for predicting risk for LI in bilingual children in 1st grade.