Differentiating specific language impairment from normal language development using grammatical analysis

Abstract

The present investigation sought to determine whether preschool children with specific language impairment (SLI) could be differentiated from an age-matched sample of typically developing children on the basis of grammatical features displayed within a 20-min sample of conversational language. Language samples from 47 children, 24-50 months of age, described in a previous study (Klee, Schaffer, May, Membrino and Mougey (1989) were analysed using the LARSP framework (Crystal, Fletcher and Garman, 1989). Composite variables were created from the original LARSP categories and transformed into ratios to adjust for the differing number of utterances in each transcript. A discriminant function which optimized the differences between the two groups was then derived from the grammatical analysis. In addition to age and a constant, the function contained three linguistic variables: stage I-major utterances, three-element NPs and VP errors. The outcomes of the discriminant function and the clinical diagnosis were in agreement in 91·5% of the cases. The predictive validity of the discriminant function was then evaluated in a field test employing a second sample of children (n = 37). The outcome of the discriminant function concurred with the clinical diagnosis in 86·5% of the new cases. © 1993 Informa UK Ltd All rights reserved: reproduction in whole or part not permitted.