Planning Goals for Intervention: Can a Language Test Serve as an Alternative to a Language Sample?

A comparison was made between imitated responses to Carrow's Elicited Language Inventory (CELI) and the spontaneous speech of ten language impaired children (LI). Responses to the CELI and selected utterances from the sample of spontaneous speech were analyzed in a manner similar to the procedure outlined by Carrow (1974a). While there was a high correlation between the proportion of errors on each task, there was also a significant difference between these proportions. All of the LI children had a lower mean proportion of errors on the language sample than on the CELI and discrepancies between the two were greater for the younger children with the lower mean length of utterances. Goals of intervention derived from the two sets of data also indicated differences. Three factors may have accounted for the discrepant scores-differences in lexical items, differences in the structural complexity of the sentences, and differences in the requirements of the two tasks. It is suggested that while standardized elicitation tasks, such as the CELI, serve to identify children with language impairments they may not be as useful as data from free speech samples for determining goals of intervention.