This investigation studied the performance of fifteen normal and fifteen language-disabled children on experimental pragmatic tasks and on a standardized Piagetian measure of sensorimotor intelligence. The children were matched for mean length of utterance, all subjects performing at the holophrastic level. A series of experimental measures was designed to elicit declarative performatives, imperative performatives, and encoding of the most informative element in a context. All aspects of the child's behavior were observed, and the responses were scored in a dichotomous manner: whether or not they had encoded the most informative, i.e., changed, element. The responses to the experimental presuppositional items were also scored in terms of the level of declarative generated. The children with normal language development at the holophrastic stage could performatively encode the most informative contextual element with either linguistic or non-linguistic means significantly more frequently than a 50-50 chance would permit. The language-disabled children, however, were able to signal the most informative element with a sensorimotor performative but were unable to do so linguistically. The statistical analysis revealed that the language-disabled subjects' pragmatic use of the language was deficient, even though they had been matched for level of linguistic performance with normal subjects. (Author/NCR)