A brief theoretical statement on definition is advanced, followed by a study of the development of definition in children aged 5;0–10;0. The empirical findings suggest that the emergence of superordinate terms in definition cannot be adequately accounted for simply by appeal to changes in underlying knowledge, or to selectional restrictions on the structure of the definiens. Rather, the development of definition can best be characterized as the gradual articulation of a conventional definitional form out of the more general forms of ordinary oral discourse. It is suggested that definitions are expressions of word meaning that reflect the unique requirements of a literate register, and that the increasing dominance of superordinate terms in definitions reflects in part the adoption of this conventional form of expression.

With various descriptions of adults' speech to children in hand, investigators have turned their attention to questions concerning the specific nature and extent of relationships between aspects of input and language development. Answering such questions requires some conceptualization of the potential relationships that may exist. Equally important, though, are the statistical and design issues surrounding the inference of these relationships. An exhaustive discussion of these issues is beyond the scope of this Note. Consequently, we will focus on three specific topics: (1) the use of gain scores; (2) aspects of correlation and alternative procedures that have an impact on the validity and interpretability of statistical findings; and (3) passive observation versus experimental manipulation.

Ninio & Bruner (1978) added an important dimension to the study of early lexical acquisition by drawing attention to the dialogue-like nature of the mother-child interactions where presumably much language-learning takes place. The authors pointed to the well-established findings that much of the child's early speech consists of names for people and objects (Leopold 1949, Werner & Kaplan 1963, Nelson 1973, Greenfield & Smith 1976). They went on to show that in one familiar type of parent–child interaction, joint picture-book reading, labels are used extensively by the adult and are inserted skilfully into a structured interactional sequence that has the texture of a dialogue (Ninio & Bruner 1978: 6). This dialogue, they suggested, ‘seems… to be a format well suited to the teaching of labelling’ (1978: 12). Subsequent research has also been interpreted as pointing to the teaching potential of joint picture-book reading (Wheeler 1983, Ninio 1983) and the opportunities it affords for situation-specific routines (Snow & Goldfield 1983).

Throughout this century observers have commented on the beginning talker's evident preference for words denoting ‘father’ over words representing the meaning of ‘mother’. Bateman (1917) said that ‘the most frequently named object is the paternal parent’. Leopold (1948) stated that ‘papa…appears far more commonly as the first meaningful word than does mama’. In this paper I will review some of the reported data on parent reference and propose a phonetic account of the child's evident preference for paternal terms.