consonant inventory and less complex syllabic structure, while others show adequate complexity of vocalizations (McHale et al., 1980; Stone and Caro-Martinez, 1990; Wetherby and Prutting, 1984; Wetherby et al., 1989).

The vast majority of those who do learn to talk go through a period of using echolalia, the imitation of speech of others, which may be immediate or delayed (Prizant et al., 1997). An echolalic utterance is usually equivalent to a single word or a label for a situation or event. Many children learn to use echolalia purposefully in communicative interactions, and eventually are able to break down the echolalic chunks into smaller meaningful units as part of the process of transitioning to a rule-governed, generative language system (Prizant and Rydell, 1993).

Children with autism who progress beyond echolalia usually acquire more advanced aspects of grammar: that is, they develop grammatical skills in the same general progression as typically developing children, but show persisting problems in following the social rules and shifting between speaker and listener roles of conversation (Baltaxe, 1977; Tager-Flusberg, 1996), which are the pragmatic aspects of language.

In lieu of conventional means of communicating, children with autism may develop idiosyncratic, unconventional, or inappropriate behaviors to communicate, such as self-injurious behavior, aggression, or tantrums. Despite the fact that at least 50 percent of individuals with autism display some functional speech and language skills (Lord and Paul, 1997), challenging behaviors such as aggression, tantrums, and self-injury are often used to procure attention, to escape from a task or situation, to protest against changes of schedule and routine, or to regulate interactions in a predictable manner. Carr and Durand (1985) reported that aggression, tantrums, and self-injury were more likely to occur in situations with a high level of task difficulty and a low level of adult attention. Challenging behaviors need to be considered relative to the child’s repertoire of verbal and nonverbal communicative behaviors and may reflect limitations in symbolic capacity.

Further evidence of a deficit in the symbolic capacity in autism is the limited ability to develop symbolic or pretend play. Although play is a social-cognitive skill, a lack of varied, spontaneous make-believe play is one of the four possible features of the impairment in communication in the most recent Diagnostic and Statistical Manual (DSM-IV) (American Psychiatric Association, 1994). Children with autism show significant deficits in symbolic or make-believe play (i.e., using pretend actions with objects) and limited abilities in functional play (i.e., using objects functionally) (Dawson and Adams, 1984; Sigman and Ungerer, 1984; Wetherby and Prutting, 1984; Wing et al., 1977). Functional and symbolic play skills have been found to be significantly correlated with receptive and expressive language (Mundy et al., 1987; Sigman and Ruskin, 1999). In contrast



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement