ABSTRACT
Ten language-normal children, aged 2:1-3:3 (yrs:mos), and eight language-impaired children, 3:7-4:1, in Brown's (1973) Stages II or III received two subtests of the Oral Language Sentence Imitation Screening Test (OLSIST: Zachman, Huisingh, Jorgenson, & Barrett, 1978) in a contextually supported condition and in a standardized test condition. Also, children's generation of eight grammatical morphemes in a spontaneous language sample condition was compared to their imitative control of the same structures on the OLSIST. Results indicated significantly enhanced imitative performance for the language-impaired children but not for the language-normal children in the contextually cued condition. Stage III language-normal children earned significantly better imitative scores than language-impaired subjects and Stage II language-normal children, regardless of test condition, although OLSIST subtests did not differentiate performance of any group. With only one exception, morpheme difficulty orders on the OLSIST did not predict morpheme orders in spontaneous language samples. Presence of cues, moreover, did not significantly increase the OLSIST's predictive ability. With the exception of Stage III language-normal subjects, all children displayed greater productive control of most morphemes in their spontaneous speech. Conclusions were drawn relative to the clinical significance of contextual reliance as well as alternatives to imitation testing for estimating the grammatical skills of the language-impaired child.
