Comprehension and strategy use by autistic children were tested in two experiments. Eighteen autistic subjects were compared with thirty normal 3- and 4-year-olds, matched on the PPVT and Raven's Colored Progressive Matrices. In Experiment I subjects were asked to act out active and passive, biased and reversible sentences. The autistic group's overall comprehension was lower than that of the normal controls; although the autistic subjects did use a word-order strategy, they did not generally use a probable-event strategy. These findings were confirmed in Experiment II, in which the same procedure with anomalous three-word items was used. The results are interpreted as evidence that in autism there is a semantic-cognitive deficit in utilizing conceptual knowledge about relational aspects of the environment and that this deficit underlies the comprehension difficulties of autistic persons.

Experimental studies of autistic children's memory for linguistic materials have yielded what appear to be conflicting results. A review of these studies reveals a pattern of findings consistent with the hypothesis that autistics do, in fact, have a special linguistic coding difficulty. The apparent discrepancies are the result of the use of low-power statistical tests in some experiments. Because the autistic deficit may arise from a failure to use semantic or syntactic knowledge or even from a failure to acquire such knowledge in the first place, future research should be aimed at explicating the precise mechanisms underlying the autistic deficit.

Two hearing children of deaf parents (initially 3;9 and 1;8) had been cared for almost exclusively by their mother, who did not speak or sign to them. Though the older child had heard language from TV and briefly at nursery school, his speech was below age level and structurally idiosyncratic. Intervention led to improvement in his expressive abilities, and by 4;2 the deviant utterance patterns had disappeared. In later years, his spontaneous speech and school performance were normal, though language testing revealed some weak areas. The younger child initially used no speech, but acquired language normally after intervention, with his brother as model. Implications for understanding the role of linguistic input in language development are discussed.

The acquisition of /s/ plus stop clusters by language-delayed children was compared with normal acquisition, first, for phonemic substitutions and second, for differential use of voice onset time (VOT) for the substituted stops. Ten children, diagnosed as language delayed, produced 95 percent of the intended clusters as singleton stops; children who could employ VOT contrastively produced most stops, substituting for clusters with VOT values in the long lag range.

This paper describes a language program designed to allow an autistic 4-year-old child to acquire, in a near simultaneous manner, skill across a variety of sentence types. The training, which was focused on teaching the combinatorial and semantic properties of grammatical morphemes, resulted in the child's developing and using sentences whereas he had previously been limited to single words and rote phrases. The improvement occurred across a variety of settings and encompassed both elicited and spontaneous language. As expected from the autistic syndrome, those areas of language functioning requiring social skill (e.g., sustained dialogue) showed continued deficits.