Antonymy in childhood: a corpus-based approach to acquisition

This study examines the language to and of three children from age 2 to 5 in order to determine when and how children develop adult-like use of antonyms in discourse. Using the categories of antonym functions developed in Jones 2002, we find that child and child-directed language are similar to adultdirected printed journalism in their use of the two major antonym categories, Ancillary and Coordinated Antonymy—in spite of the fact that Ancillary Antonymy requires more syntactic and semantic complexity than other types. Differences between child and adult language are to be found in the use of minor categories, with Negated Antonymy and (a new category) Interrogative Antonymy playing special roles. Possible reasons for these differences are explored. Background — Children and antonymy Antonymy plays an important role in language use—to the extent that Jones (2002) estimates that as many as one in fifty sentences might include antonym pairs serving one of several possible discourse functions. Antonymy also plays important parts in young children’s lives. Children’s stories, rhymes and songs make heavy use of lexical contrast and parenting handbooks (e.g., Eisenberg et al. 1989) recommend that parents emphasise binary opposition in order to encourage cognitive development. Later, the National Literacy Strategy emphasises antonym awareness throughout primary education, as a means for building the metalinguistic awareness needed for reading (DfES 1998). Nevertheless, very little research has approached the issues of children’s acquisition of antonym pairs and the development of discourse and metalinguistic skills involving antonyms. Research using corpora of adult-directed, mostly written language (Justeson and Katz 1991, Fellbaum 1995, Jones 2002) has shown that antonyms frequently co-occur in sentences, often in recognizably contrastive constructions. The frequency and particular contexts of antonym cooccurrence have led some (Ervin 1963, Charles and Miller 1989, Justeson and Katz 1991) to conclude that we learn that words are antonyms through repeated exposure to pairs in antonymic constructions. However, it’s far from clear that conclusions about child language acquisition can be drawn from observations of formal written language use by adults. Developmental psychologists have found that children appreciate antonymy from very young ages and grasp the notion quite naturally (Kagan 1984, Kreezer and Dallenbach 1929). In a small corpus study of children’s and child-directed language, Murphy (1998, at press) showed that children from ages 2 to 5 not only used antonyms contrastively in their utterances, but that they often did so at higher type and token rates than the adults speaking to them. The children’s antonym use seemed to be independent of adult use, in that the children preferred different antonym pairs than their parents tended to use, and it was more often that the parent parroted the child’s antonym usage than vice versa. Attempts to draw conclusions from this study are hampered by the fact that the corpora involved included only one type childoriented linguistic input: informal interaction with parents. Nevertheless, it does show that children use antonyms contrastively from the early multi-word stages, and raises the question of how the discourse functions of antonymy differ for child and * Steven Jones, University of Central Lancashire, Preston PR1 2HE, [email protected] M Lynne Murphy, University of Sussex, Brighton BN1 9QH, [email protected] adult. The current study uses the same corpora as Murphy 1998 to investigate this problem, and is intended as a pilot for a much larger project on the development of antonym knowledge and use in childhood. Jones (2002), using the Independent newspaper corpus, developed a taxonomy of antonym functions within sentences, leading us to ask the following questions: (a) are Jones’s (2002) major and minor antonym classes found in children’s and child-directed language? (b) are different functional antonym classes to be found in children’s language? (c) what correlations can be found between antonym functions found in childdirected language and those in child language? Methodology To investigate these issues, we searched for instances of antonym co-occurrence in three corpora included in the CHILDES database (MacWhinney 1995). Each of these corpora concerns the longitudinal study of a child’s linguistic development, in the form of transcripts of recorded family interactions between the ages of 2 and 5. The children are Abe (Kuczaj 1976), Naomi (Sachs 1983), and Ross (MacWhinney). Indication of the size of each corpus are given in Table 1, which shows word and turn counts for language spoken by and spoken to the child. Table 1. Corpus information linguistic output linguistic input child age types tokens turns types tokens turns Abe 2;4.24–2;11.3