Developmental Sentence Scoring for Preschool Language Sample Analysis: A Psychometric Update

PURPOSE: We sought to validate the ability of Developmental Sentence Scoring (DSS) to distinguish age and diagnostic category in a large, open-access corpus of American English-speaking adult-child language interactions at TalkBank.org. We additionally sought to identify whether subscales differed by age and diagnostic category, to explore the potential for adapting DSS to shorter language samples. METHOD: We utilized data from 501 children (ages 2-6 years; 50-utterance sample criterion) and 842 children (25-utterance sample criterion) as well as the open-access DSS utility in Computerized Language Analysis to analyze growth in DSS total and subscale scores and distinguish typical- and late-talker grammatical profiles. RESULTS: DSS total scores were most informative at younger ages; scores of typically developing (TD) children and children with a history of late talking converged over time. The most sensitive subscale was Main Verbs (MV). Shorter samples (25 utterances) were generally as informative as longer ones. Scores derived from our analyses were generally much higher than those reported by Lee (1974). DISCUSSION: Shorter samples (25 utterances) are as informative as longer ones for both diagnostic and therapy planning purposes. Beyond the MV subscale, DSS may have greater utility in identifying therapy intervention targets than in diagnosing language delay. Finally, use of the original percentile scores for the DSS reported by Lee (1974) may underidentify children whose language lags significantly behind that of TD peers. SUPPLEMENTAL MATERIAL: https://doi.org/10.23641/asha.30907109.