Identification of Developmental Language Disorder in Bilingual Children: An Accurate and Time-Efficient Combination of Language Measurements.

Purpose: This study addresses the challenge of identifying developmental language disorder (DLD) in bilingual children. Despite the broad range of language measurements documented in the literature, their individual contribution to a DLD diagnosis remains unclear. Administrating a high number of tests will yield a holistic child view, but it needs to be reconciled with a time-efficient protocol that is feasible in clinical practice. Therefore, we aim to evaluate the accuracy and time efficiency of a comprehensive set of measurements, through cross-validated machine learning.

Method: In 50 typically developing bilingual children and 50 bilingual children with DLD aged between 5 and 9 years, background measurements were assessed including hearing, intelligence, language experiences, and socioeconomic status. Alongside standardized language tests, a parental questionnaire on home language, narrative tasks, a nonword repetition task, and a cognitive inhibition task were administered. Both group differences and individual performance were studied.

Results: Significant group differences were observed across most measurements. The most accurate and time-efficient protocol combined four measurements, including sentence repetition, nonword repetition, the parental questionnaire, and the task measuring semantic and morphosyntactic comprehension, achieving 90% classification accuracy. Notably, adding more measurements to the protocol did not enhance accuracy.

Conclusions: This data-driven analysis selected the measurements that are most contributive in identifying DLD in bilingual children. This language assessment protocol successfully combines time efficiency with high accuracy to diagnose DLD, resulting in a useful and feasible protocol for speech-language pathologists in clinical practice.

Supplemental material: https://doi.org/10.23641/asha.29522192.