This study investigated mothers' notions of appropriate speech to 40 children aged 1; 6 to 3; 6. Spontaneous maternal action-directives were analysed in terms of the explicitness with which they expressed syntactic and lexical clues to action-directive intentions. Cross-sectional and longitudinal changes, and progressions within maternal repetition sequences, indicated that the children perceived as most limited in inferential skills elicited the most restricted input. Most of the action-directives they heard could be interpreted non-inferentially, because syntactic and lexical clues to intentions were explicitly expressed. External supports were removed gradually in response to the mother's perception of her child's increasing ability to draw extra-linguistic inferences. Several theoretical models of non-literal speech-act comprehension are supported. It is argued that the differences in language use reflect the mother's notions of the speech a listener is likely to understand.

Brown & Hanlon (1970) reported that parents' explicit approval or disapproval of their child's utterances was not contingent on the well-formedness of that utterance. Rather, parents seemed attuned to the semantic value of their child's speech. Using 40 mother–child dyads, 10 from each of four age groups, 2–5 years, we replicated these results. We expanded upon this earlier work, however, to suggest that parents are not totally indifferent to the grammatical form of their child's utterances. Focusing on repetitions, we found that mothers are more inclined to repeat ungrammatical than grammatical sentences generated by 2- year-old subjects. This indicates that the language learning environment does present subtle cues that distinguish between well-formed and ill-formed sentences. Whether the child is capable of capitalizing on these cues remains unanswered.