Tailoring Grammatical Test Items in English According to Language Experience for the Classification of Developmental...

This study attempts to discriminate developmental language disorder (DLD) in bilinguals using only the child's second language (L2), English. We grouped bilinguals by their current weekly language experience (high English experience, balanced English/Spanish experience, high Spanish experience) as a way to evaluate which English grammatical test items yielded acceptable levels of diagnostic accuracy at different levels of language experience.

Three hundred seventy-eight Spanish-English bilingual children (65 with DLD) between the ages of 7 years and 9;11 (years;months) responded to English grammatical test items delivered via a sentence repetition and/or a cloze task. For each language experience group, test items with a discriminant value > .30 were retained and used to create a composite of items "tailored" to that experience group. Tailored composites were then entered in discriminant analyses, by group.

Children with DLD consistently scored below those with typical development; nonetheless, the same item set did not identify DLD equally well across language exposure groups. For instance, the item set tailored for children with high English experience yielded sensitivity and specificity of 80% and 94%, respectively, among children with > 60% English exposure. However, those same test items yielded sensitivity of only 20% for children with high Spanish experience.

Testing a bilingual child in both of their languages remains the gold standard for diagnosing DLD. However, our results support the indexing of developmental expectations to measures of L2 experience when testing is constrained to English.