Preschool speech and language screening: further validation of the sentence repetition screening test.

A recent review indicated that only one speech and language screening test (the Sentence Repetition Screening Test; SRST) designed for preschoolers has been successfully validated in a representative population. This study sought to replicate the SRST validation study using a somewhat younger age group and to compare predictive indices to typical measures of parent concern and teacher judgment. A sample of (N = 343) prekindergarten children (ages 54 to 66 months) attending school registration in the spring before kindergarten entry was tested with the SRST. A stratified sample of 76 returned for criteria testing within 2 months. Teacher ratings were obtained 5 months later. Outcome measures were standard language (Bankson and Illinois Test of Psycholinguistic Abilities) and speech articulation (Arizona) tests. With a prevalence of 11%, SRST prediction of language outcomes resulted in the following indices: sensitivity, .62; specificity, .91; predictive validity, .44; overreferral, 8.4%; and underreferral, 4%. Prediction of the articulation outcome with a prevalence of 11% resulted in indices as follows: sensitivity, .57; specificity, .95; predictive validity, .75; overreferral, 3.7%; and underreferral, 8.3%. Teacher ratings and a parent questionnaire (Speech and Language Screening Questionnaire) approached the same rate of prediction of articulation but with a higher overreferral rate. The predictive indices from the SRST exceeded the comparison measures for language outcomes.