Words artificially isolated from twelve parents' speech to their children (aged 1; 10–3; 0) were significantly less intelligible to adult listeners than words originally spoken to an adult. This effect holds for randomly sampled words and, to a lesser extent, for matched pairs. While parents did not adjust the clarity of word tokens to the linguistic naiveté of the child listeners, they did adjust intelligibility inversely to the observed predictability of the sampled words in their sentence contexts, and words to children proved more redundant in this sense. The relationship of these findings to other work on the clarity of motherese is examined and the implications for the study of children's speech perception outlined.

The study focuses on the language abilities of 30 20-month-old children, using data from two sources: a detailed maternal interview and 90 minutes of videotaped observation. Observed language was coded into the categories used for the interview. Production and comprehension at 28 months (MLU, PPVT and morphology comprehension) were also assessed. Observation and interview data at 20 months were highly intercorrelated. Cluster analyses of both data sets yielded referential, grammatical morpheme and dialogue clusters, providing partial support for the nominal/pronominal and referential/expressive acquisition styles reported in the literature. However, the referential and grammatical morpheme clusters were highly correlated, suggesting that two acquisition strategies are developing in parallel. Only for those children who heavily emphasize one strategy can one speak of a distinctive style. All interview and observation clusters predicted 28-months MLU, but the grammatical morpheme clusters did not predict later performance on a Grammatical Morpheme Test. It is tentatively suggested that holistic processing strategies underlie the pronominal/expressive style.

Children aged 1; 4 to 1; 6 were asked common what/-questions under four different contextual conditions. The presence of a gesture was found to have a significant effect on the nonverbal components of children's responses, while linguistic sophistication and type of question asked affected vocal responses, but did not produce any consistent effects on nonverbal responses. Results are interpreted as suggesting that, at this early age, gestural information is processed relatively independently of speech. These findings cast doubt on the likelihood that maternal gestures facilitate early language responding. Case-study observations of the children's interactional behaviour with mothers suggest instead that prior experience with verbal routines was a factor in the children's experimental vocal performance.

Maternal directives to children aged 10 and 18 months were analysed with respect to both verbal and concomitant nonverbal aspects. The findings emphasize the multimodal nature of mothers' communicative messages and draw attention to the way in which language occurs in an action context and not as an isolated output. No indication was found for this age range that verbal devices come to replace nonverbal ones. At both ages most utterances were accompanied by some form of nonverbal behaviour; the precise nature of the association, however, gives little support to the idea of parallel coding. The child's own behaviour is also shown to form part of the context in which maternal speech takes place.

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.

The existing data concerning the effects of maternal speech on child language development are so conflicting that it is difficult to draw even tentative conclusions about this relationship. One potential contributor to this problem is our lack of information concerning the determinants of maternal speech. The present study was designed to separate those elements of maternal speech which are heavily influenced by child language levels from those that seem to represent the mother's consistent style or strategy for ‘teaching’ language. Syntactic, discourse, and communicative functions of mothers' speech were examined. Mothers of high-and low-level language children were compared when talking to their daughters and to another girl the same age as their daughter. Analysis indicated a striking amount of similarity and consistency in high vs. low mothers' speech to both their daughters and the other child. Those elements of maternal speech that did show differences tended to be features that are fairly easily monitored by mothers or are directly tied to child utterances.

What does the word chair mean? How does the category of objects bearing this name differ between adults and pre-school children? And how does the knowledge of a possible function of an object affect subjects' judgements? Answers to these questions were sought by means of sorting and rank ordering tasks. Subjects were shown drawings of a variety of objects on which one could sit, and were asked to indicate which ones they would call chairs. Those objects so judged to be category members were rank ordered for degree of typicality (or ‘best example’) by a paired-comparisons procedure. Half the subjects saw drawings of the objects alone, while the others saw a person sitting on each object. The results revealed that adults consistently judged some objects to be better examples than others, and that the provision of function information affected the judgements in a characteristic way. A different, less stable typicality structure was found in the children's category. Function cues caused the children's rank order judgements to change greatly. These findings are discussed within the framework of some recent theories of lexical concept formation in young children, and in relation to recent work on the nature of internal representations.

The present study tests the hypothesis that many of the grammatical errors that occur during the course of language development can serve as ‘auto-input’ leading directly to the acquisition of new expressive forms. A free-speech corpus of grammatically incorrect sentences was gathered from each of 4 four-year-old subjects. Three additional sets of sentences were constructed: sentences containing errors similar to those actually produced, sentences containing ‘baby errors’ and correct sentences. The children were asked to judge these sentences as correct or incorrect. Significantly fewer corrections were made in the sentences with subject-generated errors than in the sentences with similar errors or ‘baby errors’. These results can be explained by assuming that children learn their own errors.

Three- and four-year-old children were tested on their comprehension of the locative prepositions in front of, in back of, ahead of, and behind. Results demonstrated that those prepositions characterized as positive by H. Clark (1973) were, contrary to theoretical predictions, comprehended more poorly than their ostensibly negative counterparts. Discussion of the possible reasons for these findings is included.