Menn has suggested that most early phonological rules have the effect of reducing the variety of the child's phonetic output forms. This proposal is compatible with the Interactionist–Discovery (I–D) theory of phonological acquisition. This paper presents one child's unique phonological rule which increases output variety and yet still yields a mismatch with the adult form. Evidence is presented which indicates that the rule was a stable and productive part of the child's phonology and that it was both phonetically and phonologically motivated. It is argued that an addition to the early strategies described by Menn is needed to account, specifically, for the discovery of this unique rule and, generally, for later stages of phonological development.

The tendency of 30 children between the ages of 2; 0 and 2; 1 to overextend and/or underextend newly uttered, but previously understood words in comprehension and production was examined. It was found that both overextensions and underextensions occurred in comprehension and in production. Similarly, in both modes of speech children made more underextensions to peripheral than to central referents, and more overextensions to perceptually similar than to perceptually dissimilar nonreferents. The frequencies of overextensions and underextensions, however, were not the same in comprehension and production. Overextensions occurred more frequently in comprehension than in production, while conversely underextensions occurred more frequently in production than in comprehension. These data are discussed in terms of differences between children's and adult's word meanings, and between comprehension and production.

This study of children's conjunction reduction contrasted the syntactic view of forward and backward deletion of base structure elements with the idea that pragmatic factors of situational redundancy and perceptual grouping account for conjunction reduction. Ninety-four children described an action sequence (putting differently coloured beads into a cup) so that a listener positioned behind a screen could repeat them. Half the children communicated as the action was being carried out (SIMULTANEOUS condition), mitigating against perceptual grouping of beads in the cup. Half communicated after the action was completed (POST condition), permitting perceptual grouping. Backward deletion was more frequent in the post than in the simultaneous condition. Also, the overall high frequency of forward deletion reflected encoding of novelty and omission of repetitive elements. These results suggest syntax is pragmatically motivated.

Modal auxiliaries have an epistemic and deontic sense and range in strength, e.g. must propositions are stronger than may propositions. Children (ages 3; 0–6; 6) heard two contradictory modal propositions of varying strength. In the epistemic condition, the propositions concerned the location of a peanut. In the deontic condition, they were commands by two teachers about what room a puppet should go to. The child was to indicate which command should be followed. The general acquisitional rule was: the greater the difference in the strength of the two modal propositions the earlier the difference was appreciated. Since the acquisitional history was similar across conditions, the two senses probably arose from a single lexical entry.

The present study examines changes with age in interrogative sequences among Dutch-speaking children. Thirty-eight male and female pairs of children were videotaped in a play situation; eight pairs of pre-schoolers and ten pairs of 7-, 9- and 11-year-old schoolchildren, with an equal number of pairs of boys and pairs of girls at each age level. An interrogative sequence consists of: question–listener reaction–confirmation of that reaction. Questions were analysed according to function, content and form, and listener reaction according to how appropriate it was. The main results are: (1) changes with age occur in the use and form but not in the content of questions; (2) a relation exists between the function, content and form of a question and the listener's reaction; (3) listeners frequently do not answer questions (this tendency is stronger within the two younger age groups than within the two older ones); (4) answers are frequently not confirmed.

A case is reported of failure to supply negation in tag questions for a period of nearly two years. The duration of this error shows that it could not have been the result of limited processing capacity, and this provides reason to doubt this explanation for the similar case of children's failure to invert subject and auxiliary verb in wh-questions. It is argued that other cases which have been explained in terms of limited processing capacity are equally compatible with an explanation in terms of context-specific knowledge, which holds that a rule is simply not known for contexts other than those where it is expressed.

Some subordinating conjunctions reveal whether speakers believe the propositions the conjunctions connect and how speakers believe propositions are connected. Because and although express belief; if + indicative and unless uncertainty, and if + subjunctive, disbelief. Because, if + indicative and if + subjunctive express a positive entailment relation between clauses while although and unless express negative entailment. To validate this analysis a comprehension task was administered to 96 subjects divided among 9-, 11- and 19-year-olds. Nine-year-olds understood although and because while adults also comprehended unless and if + indicative. Generally the more positive beliefs the conjunction embodied, the easier its comprehension. The order of difficulty was because, although, if + indicative, unless, and if + subjunctive. Older subjects more often confused if + subjunctive with if + indicative. Understanding of the uncertainty conveyed by if + indicative and unless was linked to performance on a task of detecting uncertain outcomes.

Seventy-five children, 3, 4, and 5 years old, were interviewed about: (a) toys they had played with just a few minutes earlier, (b) toys they had played with on the preceding day, (c) toys they would play with in a few minutes, and (d) toys reserved for use on the following day. Verb forms indicating past and future time were used as well as the adverbials before and after. The past verb form was understood equally well in reference to the immediate past and the more remote past. However, the future verb form was better understood in reference to the immediate future than in reference to the remote future (the following day). The difference is discussed in terms of the intersection of time and mood in future verb forms. Immediacy of action and certainty of occurrence are suggested as early meaning components of future verb forms.

Kindergarteners, grouped according to number conservation ability, were asked to explain their choices of more and less in a series of parallel tasks requiring comparative judgements of continuous quantities. Errors on less tasks, as well as correct choices of more, were often determined by the non-linguistic strategy of choosing the greater amount. Non-conservers were more likely than conservers or transitional conservers to rely on this strategy. Furthermore, mature semantic knowledge of more was found to precede the development of a similar understanding of less. Non-conservers were less likely than conservers or transitionals to have a mature understanding of either term.

The game of ‘talking backward’, invented independently by two boys aged 8; 10 and 9; 11, is studied with reference to the role of speech play in linguistic, cognitive, and social development. Case reports from 27 adult backward talkers suggest a lower limit of about 7 years for the onset of this ability. The backward speech of the two children illustrates how either phonology or orthography can be used as a basis for backward speech, and demonstrates the boys' knowledge of phonemic units and letter-to-sound correspondences. Talking backward belongs to a larger class of ‘secret’ language games that may be cognitively and pragmatically well suited to children of this age.

Two experiments are reported in each of which eighty children between the ages of two and six years of age were given a series of commands containing relational terms and similar commands in which the relational terms were replaced by nonsense. The results indicated that children are able to use information from a number of sources which help them to interpret such commands. Younger children, particularly, seemed to rely relatively little upon word meaning, per se. Evidence is offered that the children's responses were constrained by the non-linguistic context, by prior repetition of commands, and by information available from the linguistic context.

Fourth-, eighth-, twelfth-graders and expert adults each wrote persuasive messages to three audiences arrayed along a dimension of intimacy to writer. Measures of social cognitive ability were also obtained. Scripts were analysed for (1) fluency, (2) subordination, (3) clause length, (4) logical adverbial clauses, and (5) conjunctions of conclusion. In addition to general developmental trends, results indicated effects of audience adaptation and of mediation of social cognitive ability. Findings support the view that evidence of later language development must be interpreted in light of communicative demands of tasks and subjects' rhetorical skills.

A distinction has been drawn between tag questions which occur within the same temporally defined turn and those in a second turn by the same speaker. A TAG CONSTRUCTION has been defined as a MATRIX CLAUSE (any syntax) and a TAG FORM (reduced interrogative in elliptical constructions; or interrogative expression in stereotypical constructions) produced WITHIN THE SAME TURN. The discoursal functions of tag constructions include requests for information, agreement, permission, compliance, responsiveness and attention. Although stereotypical tag constructions emerge earlier in development than elliptical tag constructions, from the beginning stereotypical tag constructions perform the same discoursal functions as the grammatically more complex elliptical constructions as well as additional ones. Tag constructions are less effective than canonical questions in obtaining turn transfers. This relative ineffectiveness may be due to either the post-positioned clearance signal or the fact that the listener feels less compelled to respond when the speaker seems to have an expectation as to what the response to the request should be.