Tracking children from poverty at risk for specific language impairment: a 3-year longitudinal study.

A 3-year longitudinal study of the language performance of children from poverty was designed to address the problem of separating children with a specific language impairment (SLI) from low-scoring normal children in the borderline area on the continuum of language performance where normal ends and abnormal begins. Two approaches to definition were compared: an experimental approach (using story-retelling, rote-memory ability, and invented-morpheme learning) and a traditional approach (using standardized-test discrepancy scores). Results indicated that 6 of 34 children tracked from kindergarten through second grade appeared to be SLI at the end of the study. The best kindergarten predictor for the outcome status of these 6 children was a combination of the score on the Oral Vocabulary subtest of the TOLD-2P and the score on a combination of the experimental tasks. The best single kindergarten predictor of the academic status of the 15 children in the study who received academic remediation was story-retelling. Children's scores on the experimental and standardized tests of language performance and nonverbal intelligence were profiled over the 3 years of the study, and patterns of change in many instances reveal the lifting of the early influences of poverty.