In this study we compared the conversations of mothers and fathers with their children at 1;3 and 159, with special attention to breakdown-repair sequences. We found that, overall, children and secondary caregiver fathers experienced more communicative breakdowns than did children and primary caregiver mothers. More specifically, fathers requested clarification of their children more often than did mothers, and they most often used a non-specific query (e.g. What?). Mothers used more specific queries (e.g. Put it where?) and were involved in more ‘looped’ sequences involving multiple requests for clarification. Fathers also failed to acknowledge child utterances more often than did mothers. After a father non-acknowledgement, children tended not to persist and when they did they often received further non-acknowledgements; the dyad did not often return to the child's original topic. After a maternal non-acknowledgement, on the other hand, children persisted and the dyad more often returned to its previous topic. The results are interpreted as support for the Bridge Hypothesis which claims that fathers present children with communicative challenges that help prepare them for communication with less familiar adults.

Young children's competence in using and responding to clarification requests may reflect their desire to achieve goals through the direction of others rather than knowledge that messages must be understood by another mind. This possibility was addressed by examining miscommunication episodes between children aged 2;6 and their parents in videotaped free play sessions. Although the children generally responded appropriately in form to parental clarification requests, they responded more often to and resolved successfully more of those following their own requests than those following their assertions. Thus, the results support the idea that young children are frequently motivated to repair their utterances when they have not achieved some overt behaviour desired from others. These findings are discussed in the light of the conversational experience two-year-olds bring to clarification situations.

This paper explores how children learn the range of aspect inflections to which a verb is amenable. The data considered are from observations of three Japanese boys aged between 1; 10 and 2; 1. Analyses focus on the children's mastery of Aktionsart specific intersentential patterns. For example, in one pattern, a caregiver asks nani shiteiru? ‘What is someone or something doing’. The conventional answer is with the continuative -te iru form of a verb that expresses an activity. Three conclusions are drawn. First, lexical assignment had begun before aspectual inflections were used to signal temporal perspective switch. Secondly, lexical assignment began with anchorpoints tying certain lexical contrasts to the Aktionsart of minimal sentences. Thirdly, overgeneralizations of aspect inflections are due to the difficulty of acquiring devices that signal temporal perspective switch.

This is an exploratory case study of the relation between speech output disturbances (dysfluencies) and the development of language production processes. The data consist of transcribed weekly speech samples of a Dutch boy between 2;4 and 2;11. The period of observation captures the early phase of the transition from ‘pre-grammatical’ to grammatical language. The frequency of occurrence of dysfluencies (i.e. repetitions, revisions and incomplete phrases) shows a significant increase and a subsequent decline. Whereas in the first half of the observation period the dysfluencies are distributed relatively randomly over sentences, in the second half they tend to concentrate in function words and sentence-initial words. The decline of dysfluency rate is shown to be related to an abundant use of a few ‘syntactic frames’. It is argued that these results reflect the emergence of a component in the speech production apparatus which is specifically dedicated to serial-order planning.

In full passive sentences such as The cat was kicked by the dog, the patient (cat) is promoted to subject and the agent is demoted to the by-phrase. Children 2;10 to 4;7 years (mean 3;6) who were taught the form with animate patients and animate agents (The baby is being picked up by the girl) were better able to produce and comprehend passives than children taught with inanimate patients and animate agents (The flower is being picked up by the girl). The finding of comparable post-teaching performance in children taught with perceptually salient (coloured) VS. nonsalient patients argues against a salience explanation for the patient animacy effect. Moreover, equal access to word forms for animate and inanimate nouns did not reduce the effect. The animacy effect is consistent with claims that ‘perspective’ is the cognitive counterpart to the formal category of subject; and, conversely, inconsistent with attempts to understand language acquisition in terms of a language system that operates in isolation from other facets of human cognition.

This paper presents a case study of second-language acquisition of Hebrew via immersion in daycare between 1;10 and 3;0. A period of silence was followed by rapid onset of L2 production simultaneously with many references to language itself. Eight types of language awareness were identified, and of these, several types may be prerequisites for starting L2 production. The nature of L2 speech during the first stages of production suggests that to crack the sematic code of L2, the child relies on identifiable contingencies between utterances and subsequent behaviours by speakers and listeners. As a result there are many more imperatives and interrogatives in L2 than are evident in L1 speech, and these appear to be learned by rote in an unanalysed manner. The transition to complex constructions occurs via the juxtaposition of known but syntactically unanalysed chunks, and results in patterns of syntactic errors similar to those of adult second-language learners. Reliance on L1 as a fall-back strategy was also evident. Several implications of these data for cognitive development in general are discussed.

Whenever children hear a novel word, the context supplies information about its meaning. One way children may cope with so much information is to use whatever seems to make sense, given their prior knowledge and beliefs, while ignoring or quickly forgetting the rest. This work examined if and how children's beliefs about word meanings may affect their use of contrastive linguistic information in the input in word learning. In Study 1, some 3- and 4-year-olds were introduced to a novel material or shape name and heard it contrasted with familiar words. Others merely heard the novel word used for referring to an object. These children were then tested to determine what they had learned about their new word meaning. In Study 2, another group of 3-and 4-year-olds were asked to name the materials and shapes used for introducing these novel terms. Children made use of linguistic contrast only in some situations. They benefited more when the novel term did not overlap much in denotation with any terms commonly known by 3-and 4-year-olds. These results suggest that children can use information in the input very efficiently in learning a term for an as-yet-unnamed category, but not in learning a term similar in denotation to a word they already know. Thus, the results are consistent with the claim that children believe every word has a unique denotation.

The transition from slow to rapid word-learning was examined in a longitudinal study of 18 children. Beginning at age 1;2, mothers kept a diary of children's words. Diary entries were discussed during phone calls to the home every 2½ weeks. A chronological record of nouns and other word classes was coded from the diary records.

Thirteen children evidenced a prolonged period of up to three months during which rate of acquisition markedly increased. Almost threequarters of the words learned during this period were nouns. Five children evidenced more gradual word-learning, and acquired a balance of nouns and other word classes. These results suggest that the terms ‘vocabulary spurt’ and ‘naming explosion’ best describe children who focus their early linguistic efforts on a single strategy: learning names for things. Other children may attempt to encode a broad range of experience with a more varied lexicon, a strategy that results in more gradual lexical growth.

Giving good definitions requires controlling both word meaning and definitional form. Definitions from 137 second to fifth graders (i.e. aged seven to eleven) were scored to reflect conformity to the classic Aristotelian form and quality of information provided. Comparisons among children with different backgrounds indicated that school exposure to English was strongly related to per cent formal definitions given and their quality. 63 children were also tested in French, their foreign language. Performance in French was lower than in English; exposure to French at home related to the amount of information the children provided in informal definitions, not to quality of formal definitions. The results suggest that performance on definitions is most strongly affected by opportunities to practise the required form.

In this paper, I review properties and consequences of the PRINCIPLE OF CONTRAST. This principle, which I have argued from the beginning has a pragmatic basis, captures facts about the inferences speakers and addressees make for both conventional and novel words. Along with a PRINCIPLE OF CONVENTIONALITY, it accounts for the pre-emption of novel words by well-established ones. And it holds just as much for morphology as it does for words and larger expressions. In short, Contrast has the major properties Gathercole (1989) proposed as characteristic of her alternative to Contrast.

The input language addressed to 18 language-learning children (MLU 1.00–3.00) was analysed so as to assess the quality of the semanticsyntactic correspondence posited by the semantic bootstrapping hypothesis. The correspondence appears to be quite satisfactory with little variation from the lower to the higher MLUs. All the persons and things referred to in the corpora were labelled by the mothers using nouns. All the actions referred to were labelled using verbs. Most of the attributive information was conveyed by adjectives. Spatial information was expressed through the use of spatial prepositions. As to the functional categories, all agents of actions and causes of events were encoded as subjects of sentences. All patients, themes, sources, goals, locations, and instruments were encoded as objects of sentences (either direct or oblique). This good semantic-syntactic correspondence may make the child's construction of grammatical categories easier.

To be well understood, narratives need to be embedded within appropriate contextual information. The early development of key orientation (participants, location and time) was traced with an 18-month longitudinal study of real-experience narratives produced by 10 children aged approximately 2–3; 6. Listener knowledge or inference was required to decode most named participants and many were not specified at all. There was no developmental improvement. Orientation to when was rare at first and involved formula words indiscriminately applied. There was steady developmental improvement in frequency as well as differentiation of time references. where information was more common at all ages, particularly when the narrated events occurred away from home. It also showed developmental improvement, but only for away-from-home locations. Overall, very young children can produce narratives in an unscaffolded context to adults unfamiliar with their experiences. The potential role of parental scaffolding in teaching orientation skills is discussed.

Comprehension of metaphor in preschoolers was studied through an elicited repetition task. Subjects were 52 children aged 3;0 to 5;2. Repetition performance on metaphors was compared to repetitions of semantically well-formed literal sentences as well as semantically anomalous sentences, all matched for length, vocabulary and sentence structure. Accuracy on literal and metaphoric stimuli was comparable, and both were significantly better than performance on anomalous sentences. There were no effects for age or sex. It was shown that the metaphors were not semantically anomalous to the children and that they were processed on a par with literal language. The argument is advanced from a review of the literature that imitation implicates understanding of the material imitated. If metaphor is thus shown to emerge early in the child's linguistic repertory, figurative language, it may be argued, occupies a more central position in linguistic theory than it has been accorded.

In a previous issue of this Journal, MacWhinney & Snow (1985) laid out the basic sketch for an international system for exchanging and analysing child language transcript data. This system – the Child Language Data Exchange System (CHILDES) – has developed three major tools for child language research: (1) the CHILDES database of transcripts, (2) the CHAT system for transcribing and coding data, and (3) the CLAN programs for analysing CHAT files. Here we sketch out the current shape of these three major tools and the organizational form of the CHILDES system. A forthcoming book (MacWhinney, in press) documents these tools in detail.

Similarity neighbourhoods for words in young children's lexicons were investigated using three computerized databases. These databases were representative of three groups of native English speakers: 5-year-olds, 7-year-olds, and adults. Computations relating to the similarity neighbourhoods of words in the children's and adult's lexicon revealed that words in the 5- and 7-year-olds' lexicons have many fewer similar neighbours than the same words analyzed in the adult lexicon. Thus, young children may employ more global recognition strategies because words are more discriminable in memory. The neighbourhood analyses provide a number of insights into the processes of auditory word recognition in children and the possible structural organization of words in the young child's mental lexicon.

A description is given of the CHILDES/BIB database, its contents, and its relationship to the CHILDES database. CHILDES is a depository of child language corpora, and the publisher of CHAT (a transcription manual), and CLAN (an electronic package for child language research). CHILDES/BIB is an electronic bibliographic database that is an autonomous unit of CHILDES.

Two studies were conducted to examine children's understanding of three terms that denote different degrees of likelihood: possibly, probably and definitely. In Study 1, children in preschool, and first, third and fourth grade completed a comparison task involving pairs of likelihood terms. Twelve stories were created that described the likelihood that each of two children would participate in an activity, and subjects judged which of the two children actually participated. In Study 2, children in preschool, and first, third and fifth grade evaluated separate statements describing three different children, each of whom was said to have a different estimate of the likelihood that a scary event depicted in a movie would actually occur. Subjects rated how scared each of the three children felt. The results of the two studies revealed that preschoolers showed very little comprehension of the meaning of any of the three words, but by fourth grade the majority of children distinguished among all three terms. Children understood the distinction between definitely and both of the other terms (possibly, probably) better than they understood the distinction between possibly and probably. This finding is consistent with research on the development of children's understanding of probability concepts. Unexpectedly, the distinction between probably and definitely seems to emerge at a younger age than the distinction between possibly and definitely. The findings are discussed in terms of their implications for communicating verbal messages to children.