Measuring severity of involvement in speech delay: segmental and whole-word measures.

PURPOSE This study examined whether any of a series of segmental and whole-word measures of articulatory competence captured more of the variance in impressionistic ratings of severity of involvement in speech delay. It also examined whether knowing the age of the child affected severity ratings. METHOD Ten very experienced speech-language pathologists rated severity of involvement from conversational speech samples obtained from 17 children with delayed speech. The ratings were then correlated with the candidate measures. The ratings by those who knew the ages of the children were also compared with the ratings by those who did not. RESULTS The severity ratings showed considerable variability. Ratings from 6 clinicians who largely agreed with each other (a "tin standard" group) were significantly associated with several of the candidate measures. Clinicians appeared to pay attention to number, type, and consistency of errors when rating severity. They also attended to both segmental and whole-word levels. Knowledge of the children's ages did not appear to affect the ratings. CONCLUSIONS The observed variability in the severity ratings raises significant questions about their usefulness. Objective measures such as some of those examined herein offer potential as more valid and reliable severity indexes.