Structural and contextual constraints on coordination in child speech are discussed. An elicited production study with English- and Japanese-speaking children in which these constraints were manipulated is reported. Results show referential context to be an important determinant of conjunction structure. In addition, structural parameters also influenced the extent to which redundancy is expressed in utterances. A preliminary model of coordination production based on a conjunction rule is proposed.

21 first graders and 20 third graders were told to pick drawings of clowns out of an array on the basis of ambiguous messages. One of the messages could be taken to refer to either of two clowns, depending on which interpretive strategy the child employed. The other message could be given no definite interpretation. Although 23 of 41 children refused to make a definite choice for the second message, 31 specified choices for the first message. First graders applied an Antecedence Maxim strategy. Third graders used a Quantity Maxim strategy.

The present study tested Huttenlocher & Weiner's (1971) hypothesis concerning the role of grammatical function in sentence comprehension. Children arranged objects to correspond to descriptions of transitive relations in two tasks. The patterns of object placements provided evidence of how children understood the various sentence forms. Contrary to Huttenlocher & Weiner's hypothesis, grammatical function was not a critical factor in object placements. This was true based on the group data or individual-based analyses. Two factors were important: logical function and order of mention of the items in the sentence. Whether an item was grammatical subject or grammatical object did not affect subjects' choices. The order of mention effects resulted from inattention to sentence meaning prior to the initial choices on certain trials. If the logical relations in a sentence were understood prior to the response, only logical function influenced subjects' responses.

Four master–dog dyads were studied to determine whether the language that is used in speaking to dogs (doggerel) resembles the language that is used in speaking to children (motherese). The structural properties of doggerel are strikingly similar to those that have been reported for motherese. Certain differences between motherese and doggerel may arise in functional and social areas. The similarities between the two language registers suggest that motherese is not elicited in response to either the linguistic level or the cognitive/intellectual level of the child. Rather, the social responsiveness of the listener may be sufficient to elicit the motherese register.